1
0
mirror of https://github.com/RPCS3/llvm-mirror.git synced 2024-10-19 19:12:56 +02:00
llvm-mirror/lib/IR/Verifier.cpp

5266 lines
198 KiB
C++
Raw Normal View History

//===-- Verifier.cpp - Implement the Module Verifier -----------------------==//
//
// The LLVM Compiler Infrastructure
//
// This file is distributed under the University of Illinois Open Source
// License. See LICENSE.TXT for details.
//
//===----------------------------------------------------------------------===//
2001-06-06 22:29:01 +02:00
//
// This file defines the function verifier interface, that can be used for some
2001-06-06 22:29:01 +02:00
// sanity checking of input to the system.
//
// Note that this does not provide full `Java style' security and verifications,
// instead it just tries to ensure that code is well-formed.
2001-06-06 22:29:01 +02:00
//
// * Both of a binary operator's parameters are of the same type
// * Verify that the indices of mem access instructions match other operands
// * Verify that arithmetic and other things are only performed on first-class
// types. Verify that shifts & logicals only happen on integrals f.e.
// * All of the constants in a switch statement are of the correct type
// * The code is in valid SSA form
// * It should be illegal to put a label into any other type (like a structure)
2001-06-06 22:29:01 +02:00
// or to return one. [except constant arrays!]
// * Only phi nodes can be self referential: 'add i32 %0, %0 ; <int>:0' is bad
// * PHI nodes must have an entry for each predecessor, with no extras.
// * PHI nodes must be the first thing in a basic block, all grouped together
// * PHI nodes must have at least one entry
// * All basic blocks should only end with terminator insts, not contain them
// * The entry node to a function must not have predecessors
2003-10-10 19:54:14 +02:00
// * All Instructions must be embedded into a basic block
// * Functions cannot take a void-typed parameter
// * Verify that a function's argument list agrees with it's declared type.
// * It is illegal to specify a name for a void value.
2003-10-10 19:54:14 +02:00
// * It is illegal to have a internal global value with no initializer
// * It is illegal to have a ret instruction that returns a value that does not
// agree with the function return value type.
// * Function call argument types match the function prototype
// * A landing pad is defined by a landingpad instruction, and can be jumped to
// only by the unwind edge of an invoke instruction.
// * A landingpad instruction must be the first non-PHI instruction in the
// block.
// * Landingpad instructions must be in a function with a personality function.
// * All other things that are tested by asserts spread about the code...
2001-06-06 22:29:01 +02:00
//
//===----------------------------------------------------------------------===//
#include "llvm/IR/Verifier.h"
#include "llvm/ADT/APFloat.h"
#include "llvm/ADT/APInt.h"
#include "llvm/ADT/ArrayRef.h"
#include "llvm/ADT/DenseMap.h"
#include "llvm/ADT/MapVector.h"
#include "llvm/ADT/Optional.h"
#include "llvm/ADT/STLExtras.h"
#include "llvm/ADT/SmallPtrSet.h"
#include "llvm/ADT/SmallSet.h"
#include "llvm/ADT/SmallVector.h"
#include "llvm/ADT/StringExtras.h"
#include "llvm/ADT/StringMap.h"
#include "llvm/ADT/StringRef.h"
#include "llvm/ADT/Twine.h"
#include "llvm/ADT/ilist.h"
#include "llvm/BinaryFormat/Dwarf.h"
#include "llvm/IR/Argument.h"
#include "llvm/IR/Attributes.h"
#include "llvm/IR/BasicBlock.h"
#include "llvm/IR/CFG.h"
#include "llvm/IR/CallSite.h"
#include "llvm/IR/CallingConv.h"
#include "llvm/IR/Comdat.h"
#include "llvm/IR/Constant.h"
#include "llvm/IR/ConstantRange.h"
#include "llvm/IR/Constants.h"
#include "llvm/IR/DataLayout.h"
#include "llvm/IR/DebugInfo.h"
#include "llvm/IR/DebugInfoMetadata.h"
#include "llvm/IR/DebugLoc.h"
#include "llvm/IR/DerivedTypes.h"
#include "llvm/IR/Dominators.h"
#include "llvm/IR/Function.h"
#include "llvm/IR/GlobalAlias.h"
#include "llvm/IR/GlobalValue.h"
#include "llvm/IR/GlobalVariable.h"
#include "llvm/IR/InlineAsm.h"
#include "llvm/IR/InstVisitor.h"
#include "llvm/IR/InstrTypes.h"
#include "llvm/IR/Instruction.h"
#include "llvm/IR/Instructions.h"
#include "llvm/IR/IntrinsicInst.h"
#include "llvm/IR/Intrinsics.h"
#include "llvm/IR/LLVMContext.h"
#include "llvm/IR/Metadata.h"
#include "llvm/IR/Module.h"
#include "llvm/IR/ModuleSlotTracker.h"
#include "llvm/IR/PassManager.h"
#include "llvm/IR/Statepoint.h"
#include "llvm/IR/Type.h"
#include "llvm/IR/Use.h"
#include "llvm/IR/User.h"
#include "llvm/IR/Value.h"
#include "llvm/Pass.h"
#include "llvm/Support/AtomicOrdering.h"
#include "llvm/Support/Casting.h"
#include "llvm/Support/CommandLine.h"
#include "llvm/Support/Debug.h"
#include "llvm/Support/ErrorHandling.h"
#include "llvm/Support/MathExtras.h"
#include "llvm/Support/raw_ostream.h"
#include <algorithm>
#include <cassert>
#include <cstdint>
#include <memory>
#include <string>
#include <utility>
using namespace llvm;
namespace llvm {
struct VerifierSupport {
raw_ostream *OS;
const Module &M;
ModuleSlotTracker MST;
const DataLayout &DL;
LLVMContext &Context;
/// Track the brokenness of the module while recursively visiting.
bool Broken = false;
/// Broken debug info can be "recovered" from by stripping the debug info.
bool BrokenDebugInfo = false;
/// Whether to treat broken debug info as an error.
bool TreatBrokenDebugInfoAsError = true;
explicit VerifierSupport(raw_ostream *OS, const Module &M)
: OS(OS), M(M), MST(&M), DL(M.getDataLayout()), Context(M.getContext()) {}
private:
void Write(const Module *M) {
*OS << "; ModuleID = '" << M->getModuleIdentifier() << "'\n";
}
void Write(const Value *V) {
if (!V)
return;
if (isa<Instruction>(V)) {
V->print(*OS, MST);
*OS << '\n';
} else {
V->printAsOperand(*OS, true, MST);
*OS << '\n';
}
}
void Write(ImmutableCallSite CS) {
Write(CS.getInstruction());
}
void Write(const Metadata *MD) {
IR: Split Metadata from Value Split `Metadata` away from the `Value` class hierarchy, as part of PR21532. Assembly and bitcode changes are in the wings, but this is the bulk of the change for the IR C++ API. I have a follow-up patch prepared for `clang`. If this breaks other sub-projects, I apologize in advance :(. Help me compile it on Darwin I'll try to fix it. FWIW, the errors should be easy to fix, so it may be simpler to just fix it yourself. This breaks the build for all metadata-related code that's out-of-tree. Rest assured the transition is mechanical and the compiler should catch almost all of the problems. Here's a quick guide for updating your code: - `Metadata` is the root of a class hierarchy with three main classes: `MDNode`, `MDString`, and `ValueAsMetadata`. It is distinct from the `Value` class hierarchy. It is typeless -- i.e., instances do *not* have a `Type`. - `MDNode`'s operands are all `Metadata *` (instead of `Value *`). - `TrackingVH<MDNode>` and `WeakVH` referring to metadata can be replaced with `TrackingMDNodeRef` and `TrackingMDRef`, respectively. If you're referring solely to resolved `MDNode`s -- post graph construction -- just use `MDNode*`. - `MDNode` (and the rest of `Metadata`) have only limited support for `replaceAllUsesWith()`. As long as an `MDNode` is pointing at a forward declaration -- the result of `MDNode::getTemporary()` -- it maintains a side map of its uses and can RAUW itself. Once the forward declarations are fully resolved RAUW support is dropped on the ground. This means that uniquing collisions on changing operands cause nodes to become "distinct". (This already happened fairly commonly, whenever an operand went to null.) If you're constructing complex (non self-reference) `MDNode` cycles, you need to call `MDNode::resolveCycles()` on each node (or on a top-level node that somehow references all of the nodes). Also, don't do that. Metadata cycles (and the RAUW machinery needed to construct them) are expensive. - An `MDNode` can only refer to a `Constant` through a bridge called `ConstantAsMetadata` (one of the subclasses of `ValueAsMetadata`). As a side effect, accessing an operand of an `MDNode` that is known to be, e.g., `ConstantInt`, takes three steps: first, cast from `Metadata` to `ConstantAsMetadata`; second, extract the `Constant`; third, cast down to `ConstantInt`. The eventual goal is to introduce `MDInt`/`MDFloat`/etc. and have metadata schema owners transition away from using `Constant`s when the type isn't important (and they don't care about referring to `GlobalValue`s). In the meantime, I've added transitional API to the `mdconst` namespace that matches semantics with the old code, in order to avoid adding the error-prone three-step equivalent to every call site. If your old code was: MDNode *N = foo(); bar(isa <ConstantInt>(N->getOperand(0))); baz(cast <ConstantInt>(N->getOperand(1))); bak(cast_or_null <ConstantInt>(N->getOperand(2))); bat(dyn_cast <ConstantInt>(N->getOperand(3))); bay(dyn_cast_or_null<ConstantInt>(N->getOperand(4))); you can trivially match its semantics with: MDNode *N = foo(); bar(mdconst::hasa <ConstantInt>(N->getOperand(0))); baz(mdconst::extract <ConstantInt>(N->getOperand(1))); bak(mdconst::extract_or_null <ConstantInt>(N->getOperand(2))); bat(mdconst::dyn_extract <ConstantInt>(N->getOperand(3))); bay(mdconst::dyn_extract_or_null<ConstantInt>(N->getOperand(4))); and when you transition your metadata schema to `MDInt`: MDNode *N = foo(); bar(isa <MDInt>(N->getOperand(0))); baz(cast <MDInt>(N->getOperand(1))); bak(cast_or_null <MDInt>(N->getOperand(2))); bat(dyn_cast <MDInt>(N->getOperand(3))); bay(dyn_cast_or_null<MDInt>(N->getOperand(4))); - A `CallInst` -- specifically, intrinsic instructions -- can refer to metadata through a bridge called `MetadataAsValue`. This is a subclass of `Value` where `getType()->isMetadataTy()`. `MetadataAsValue` is the *only* class that can legally refer to a `LocalAsMetadata`, which is a bridged form of non-`Constant` values like `Argument` and `Instruction`. It can also refer to any other `Metadata` subclass. (I'll break all your testcases in a follow-up commit, when I propagate this change to assembly.) llvm-svn: 223802
2014-12-09 19:38:53 +01:00
if (!MD)
return;
MD->print(*OS, MST, &M);
*OS << '\n';
IR: Split Metadata from Value Split `Metadata` away from the `Value` class hierarchy, as part of PR21532. Assembly and bitcode changes are in the wings, but this is the bulk of the change for the IR C++ API. I have a follow-up patch prepared for `clang`. If this breaks other sub-projects, I apologize in advance :(. Help me compile it on Darwin I'll try to fix it. FWIW, the errors should be easy to fix, so it may be simpler to just fix it yourself. This breaks the build for all metadata-related code that's out-of-tree. Rest assured the transition is mechanical and the compiler should catch almost all of the problems. Here's a quick guide for updating your code: - `Metadata` is the root of a class hierarchy with three main classes: `MDNode`, `MDString`, and `ValueAsMetadata`. It is distinct from the `Value` class hierarchy. It is typeless -- i.e., instances do *not* have a `Type`. - `MDNode`'s operands are all `Metadata *` (instead of `Value *`). - `TrackingVH<MDNode>` and `WeakVH` referring to metadata can be replaced with `TrackingMDNodeRef` and `TrackingMDRef`, respectively. If you're referring solely to resolved `MDNode`s -- post graph construction -- just use `MDNode*`. - `MDNode` (and the rest of `Metadata`) have only limited support for `replaceAllUsesWith()`. As long as an `MDNode` is pointing at a forward declaration -- the result of `MDNode::getTemporary()` -- it maintains a side map of its uses and can RAUW itself. Once the forward declarations are fully resolved RAUW support is dropped on the ground. This means that uniquing collisions on changing operands cause nodes to become "distinct". (This already happened fairly commonly, whenever an operand went to null.) If you're constructing complex (non self-reference) `MDNode` cycles, you need to call `MDNode::resolveCycles()` on each node (or on a top-level node that somehow references all of the nodes). Also, don't do that. Metadata cycles (and the RAUW machinery needed to construct them) are expensive. - An `MDNode` can only refer to a `Constant` through a bridge called `ConstantAsMetadata` (one of the subclasses of `ValueAsMetadata`). As a side effect, accessing an operand of an `MDNode` that is known to be, e.g., `ConstantInt`, takes three steps: first, cast from `Metadata` to `ConstantAsMetadata`; second, extract the `Constant`; third, cast down to `ConstantInt`. The eventual goal is to introduce `MDInt`/`MDFloat`/etc. and have metadata schema owners transition away from using `Constant`s when the type isn't important (and they don't care about referring to `GlobalValue`s). In the meantime, I've added transitional API to the `mdconst` namespace that matches semantics with the old code, in order to avoid adding the error-prone three-step equivalent to every call site. If your old code was: MDNode *N = foo(); bar(isa <ConstantInt>(N->getOperand(0))); baz(cast <ConstantInt>(N->getOperand(1))); bak(cast_or_null <ConstantInt>(N->getOperand(2))); bat(dyn_cast <ConstantInt>(N->getOperand(3))); bay(dyn_cast_or_null<ConstantInt>(N->getOperand(4))); you can trivially match its semantics with: MDNode *N = foo(); bar(mdconst::hasa <ConstantInt>(N->getOperand(0))); baz(mdconst::extract <ConstantInt>(N->getOperand(1))); bak(mdconst::extract_or_null <ConstantInt>(N->getOperand(2))); bat(mdconst::dyn_extract <ConstantInt>(N->getOperand(3))); bay(mdconst::dyn_extract_or_null<ConstantInt>(N->getOperand(4))); and when you transition your metadata schema to `MDInt`: MDNode *N = foo(); bar(isa <MDInt>(N->getOperand(0))); baz(cast <MDInt>(N->getOperand(1))); bak(cast_or_null <MDInt>(N->getOperand(2))); bat(dyn_cast <MDInt>(N->getOperand(3))); bay(dyn_cast_or_null<MDInt>(N->getOperand(4))); - A `CallInst` -- specifically, intrinsic instructions -- can refer to metadata through a bridge called `MetadataAsValue`. This is a subclass of `Value` where `getType()->isMetadataTy()`. `MetadataAsValue` is the *only* class that can legally refer to a `LocalAsMetadata`, which is a bridged form of non-`Constant` values like `Argument` and `Instruction`. It can also refer to any other `Metadata` subclass. (I'll break all your testcases in a follow-up commit, when I propagate this change to assembly.) llvm-svn: 223802
2014-12-09 19:38:53 +01:00
}
template <class T> void Write(const MDTupleTypedArrayWrapper<T> &MD) {
Write(MD.get());
}
void Write(const NamedMDNode *NMD) {
if (!NMD)
return;
NMD->print(*OS, MST);
*OS << '\n';
}
void Write(Type *T) {
if (!T)
return;
*OS << ' ' << *T;
}
void Write(const Comdat *C) {
if (!C)
return;
*OS << *C;
}
void Write(const APInt *AI) {
if (!AI)
return;
*OS << *AI << '\n';
}
void Write(const unsigned i) { *OS << i << '\n'; }
template <typename T> void Write(ArrayRef<T> Vs) {
for (const T &V : Vs)
Write(V);
}
template <typename T1, typename... Ts>
void WriteTs(const T1 &V1, const Ts &... Vs) {
Write(V1);
WriteTs(Vs...);
IR: Split Metadata from Value Split `Metadata` away from the `Value` class hierarchy, as part of PR21532. Assembly and bitcode changes are in the wings, but this is the bulk of the change for the IR C++ API. I have a follow-up patch prepared for `clang`. If this breaks other sub-projects, I apologize in advance :(. Help me compile it on Darwin I'll try to fix it. FWIW, the errors should be easy to fix, so it may be simpler to just fix it yourself. This breaks the build for all metadata-related code that's out-of-tree. Rest assured the transition is mechanical and the compiler should catch almost all of the problems. Here's a quick guide for updating your code: - `Metadata` is the root of a class hierarchy with three main classes: `MDNode`, `MDString`, and `ValueAsMetadata`. It is distinct from the `Value` class hierarchy. It is typeless -- i.e., instances do *not* have a `Type`. - `MDNode`'s operands are all `Metadata *` (instead of `Value *`). - `TrackingVH<MDNode>` and `WeakVH` referring to metadata can be replaced with `TrackingMDNodeRef` and `TrackingMDRef`, respectively. If you're referring solely to resolved `MDNode`s -- post graph construction -- just use `MDNode*`. - `MDNode` (and the rest of `Metadata`) have only limited support for `replaceAllUsesWith()`. As long as an `MDNode` is pointing at a forward declaration -- the result of `MDNode::getTemporary()` -- it maintains a side map of its uses and can RAUW itself. Once the forward declarations are fully resolved RAUW support is dropped on the ground. This means that uniquing collisions on changing operands cause nodes to become "distinct". (This already happened fairly commonly, whenever an operand went to null.) If you're constructing complex (non self-reference) `MDNode` cycles, you need to call `MDNode::resolveCycles()` on each node (or on a top-level node that somehow references all of the nodes). Also, don't do that. Metadata cycles (and the RAUW machinery needed to construct them) are expensive. - An `MDNode` can only refer to a `Constant` through a bridge called `ConstantAsMetadata` (one of the subclasses of `ValueAsMetadata`). As a side effect, accessing an operand of an `MDNode` that is known to be, e.g., `ConstantInt`, takes three steps: first, cast from `Metadata` to `ConstantAsMetadata`; second, extract the `Constant`; third, cast down to `ConstantInt`. The eventual goal is to introduce `MDInt`/`MDFloat`/etc. and have metadata schema owners transition away from using `Constant`s when the type isn't important (and they don't care about referring to `GlobalValue`s). In the meantime, I've added transitional API to the `mdconst` namespace that matches semantics with the old code, in order to avoid adding the error-prone three-step equivalent to every call site. If your old code was: MDNode *N = foo(); bar(isa <ConstantInt>(N->getOperand(0))); baz(cast <ConstantInt>(N->getOperand(1))); bak(cast_or_null <ConstantInt>(N->getOperand(2))); bat(dyn_cast <ConstantInt>(N->getOperand(3))); bay(dyn_cast_or_null<ConstantInt>(N->getOperand(4))); you can trivially match its semantics with: MDNode *N = foo(); bar(mdconst::hasa <ConstantInt>(N->getOperand(0))); baz(mdconst::extract <ConstantInt>(N->getOperand(1))); bak(mdconst::extract_or_null <ConstantInt>(N->getOperand(2))); bat(mdconst::dyn_extract <ConstantInt>(N->getOperand(3))); bay(mdconst::dyn_extract_or_null<ConstantInt>(N->getOperand(4))); and when you transition your metadata schema to `MDInt`: MDNode *N = foo(); bar(isa <MDInt>(N->getOperand(0))); baz(cast <MDInt>(N->getOperand(1))); bak(cast_or_null <MDInt>(N->getOperand(2))); bat(dyn_cast <MDInt>(N->getOperand(3))); bay(dyn_cast_or_null<MDInt>(N->getOperand(4))); - A `CallInst` -- specifically, intrinsic instructions -- can refer to metadata through a bridge called `MetadataAsValue`. This is a subclass of `Value` where `getType()->isMetadataTy()`. `MetadataAsValue` is the *only* class that can legally refer to a `LocalAsMetadata`, which is a bridged form of non-`Constant` values like `Argument` and `Instruction`. It can also refer to any other `Metadata` subclass. (I'll break all your testcases in a follow-up commit, when I propagate this change to assembly.) llvm-svn: 223802
2014-12-09 19:38:53 +01:00
}
template <typename... Ts> void WriteTs() {}
IR: Split Metadata from Value Split `Metadata` away from the `Value` class hierarchy, as part of PR21532. Assembly and bitcode changes are in the wings, but this is the bulk of the change for the IR C++ API. I have a follow-up patch prepared for `clang`. If this breaks other sub-projects, I apologize in advance :(. Help me compile it on Darwin I'll try to fix it. FWIW, the errors should be easy to fix, so it may be simpler to just fix it yourself. This breaks the build for all metadata-related code that's out-of-tree. Rest assured the transition is mechanical and the compiler should catch almost all of the problems. Here's a quick guide for updating your code: - `Metadata` is the root of a class hierarchy with three main classes: `MDNode`, `MDString`, and `ValueAsMetadata`. It is distinct from the `Value` class hierarchy. It is typeless -- i.e., instances do *not* have a `Type`. - `MDNode`'s operands are all `Metadata *` (instead of `Value *`). - `TrackingVH<MDNode>` and `WeakVH` referring to metadata can be replaced with `TrackingMDNodeRef` and `TrackingMDRef`, respectively. If you're referring solely to resolved `MDNode`s -- post graph construction -- just use `MDNode*`. - `MDNode` (and the rest of `Metadata`) have only limited support for `replaceAllUsesWith()`. As long as an `MDNode` is pointing at a forward declaration -- the result of `MDNode::getTemporary()` -- it maintains a side map of its uses and can RAUW itself. Once the forward declarations are fully resolved RAUW support is dropped on the ground. This means that uniquing collisions on changing operands cause nodes to become "distinct". (This already happened fairly commonly, whenever an operand went to null.) If you're constructing complex (non self-reference) `MDNode` cycles, you need to call `MDNode::resolveCycles()` on each node (or on a top-level node that somehow references all of the nodes). Also, don't do that. Metadata cycles (and the RAUW machinery needed to construct them) are expensive. - An `MDNode` can only refer to a `Constant` through a bridge called `ConstantAsMetadata` (one of the subclasses of `ValueAsMetadata`). As a side effect, accessing an operand of an `MDNode` that is known to be, e.g., `ConstantInt`, takes three steps: first, cast from `Metadata` to `ConstantAsMetadata`; second, extract the `Constant`; third, cast down to `ConstantInt`. The eventual goal is to introduce `MDInt`/`MDFloat`/etc. and have metadata schema owners transition away from using `Constant`s when the type isn't important (and they don't care about referring to `GlobalValue`s). In the meantime, I've added transitional API to the `mdconst` namespace that matches semantics with the old code, in order to avoid adding the error-prone three-step equivalent to every call site. If your old code was: MDNode *N = foo(); bar(isa <ConstantInt>(N->getOperand(0))); baz(cast <ConstantInt>(N->getOperand(1))); bak(cast_or_null <ConstantInt>(N->getOperand(2))); bat(dyn_cast <ConstantInt>(N->getOperand(3))); bay(dyn_cast_or_null<ConstantInt>(N->getOperand(4))); you can trivially match its semantics with: MDNode *N = foo(); bar(mdconst::hasa <ConstantInt>(N->getOperand(0))); baz(mdconst::extract <ConstantInt>(N->getOperand(1))); bak(mdconst::extract_or_null <ConstantInt>(N->getOperand(2))); bat(mdconst::dyn_extract <ConstantInt>(N->getOperand(3))); bay(mdconst::dyn_extract_or_null<ConstantInt>(N->getOperand(4))); and when you transition your metadata schema to `MDInt`: MDNode *N = foo(); bar(isa <MDInt>(N->getOperand(0))); baz(cast <MDInt>(N->getOperand(1))); bak(cast_or_null <MDInt>(N->getOperand(2))); bat(dyn_cast <MDInt>(N->getOperand(3))); bay(dyn_cast_or_null<MDInt>(N->getOperand(4))); - A `CallInst` -- specifically, intrinsic instructions -- can refer to metadata through a bridge called `MetadataAsValue`. This is a subclass of `Value` where `getType()->isMetadataTy()`. `MetadataAsValue` is the *only* class that can legally refer to a `LocalAsMetadata`, which is a bridged form of non-`Constant` values like `Argument` and `Instruction`. It can also refer to any other `Metadata` subclass. (I'll break all your testcases in a follow-up commit, when I propagate this change to assembly.) llvm-svn: 223802
2014-12-09 19:38:53 +01:00
public:
/// A check failed, so printout out the condition and the message.
///
/// This provides a nice place to put a breakpoint if you want to see why
/// something is not correct.
void CheckFailed(const Twine &Message) {
if (OS)
*OS << Message << '\n';
Broken = true;
}
/// A check failed (with values to print).
///
/// This calls the Message-only version so that the above is easier to set a
/// breakpoint on.
template <typename T1, typename... Ts>
void CheckFailed(const Twine &Message, const T1 &V1, const Ts &... Vs) {
CheckFailed(Message);
if (OS)
WriteTs(V1, Vs...);
}
/// A debug info check failed.
void DebugInfoCheckFailed(const Twine &Message) {
if (OS)
*OS << Message << '\n';
Broken |= TreatBrokenDebugInfoAsError;
BrokenDebugInfo = true;
}
/// A debug info check failed (with values to print).
template <typename T1, typename... Ts>
void DebugInfoCheckFailed(const Twine &Message, const T1 &V1,
const Ts &... Vs) {
DebugInfoCheckFailed(Message);
if (OS)
WriteTs(V1, Vs...);
}
};
} // namespace llvm
namespace {
class Verifier : public InstVisitor<Verifier>, VerifierSupport {
friend class InstVisitor<Verifier>;
DominatorTree DT;
/// When verifying a basic block, keep track of all of the
/// instructions we have seen so far.
///
/// This allows us to do efficient dominance checks for the case when an
/// instruction has an operand that is an instruction in the same block.
SmallPtrSet<Instruction *, 16> InstsInThisBlock;
/// Keep track of the metadata nodes that have been checked already.
SmallPtrSet<const Metadata *, 32> MDNodes;
/// Keep track which DISubprogram is attached to which function.
DenseMap<const DISubprogram *, const Function *> DISubprogramAttachments;
/// Track all DICompileUnits visited.
SmallPtrSet<const Metadata *, 2> CUVisited;
/// The result type for a landingpad.
Type *LandingPadResultTy;
/// Whether we've seen a call to @llvm.localescape in this function
/// already.
bool SawFrameEscape;
/// Whether the current function has a DISubprogram attached to it.
bool HasDebugInfo = false;
/// Whether source was present on the first DIFile encountered in each CU.
DenseMap<const DICompileUnit *, bool> HasSourceDebugInfo;
/// Stores the count of how many objects were passed to llvm.localescape for a
/// given function and the largest index passed to llvm.localrecover.
DenseMap<Function *, std::pair<unsigned, unsigned>> FrameEscapeInfo;
// Maps catchswitches and cleanuppads that unwind to siblings to the
// terminators that indicate the unwind, used to detect cycles therein.
MapVector<Instruction *, Instruction *> SiblingFuncletInfo;
/// Cache of constants visited in search of ConstantExprs.
SmallPtrSet<const Constant *, 32> ConstantExprVisited;
/// Cache of declarations of the llvm.experimental.deoptimize.<ty> intrinsic.
SmallVector<const Function *, 4> DeoptimizeDeclarations;
// Verify that this GlobalValue is only used in this module.
// This map is used to avoid visiting uses twice. We can arrive at a user
// twice, if they have multiple operands. In particular for very large
// constant expressions, we can arrive at a particular user many times.
SmallPtrSet<const Value *, 32> GlobalValueVisited;
// Keeps track of duplicate function argument debug info.
SmallVector<const DILocalVariable *, 16> DebugFnArgs;
TBAAVerifier TBAAVerifyHelper;
void checkAtomicMemAccessSize(Type *Ty, const Instruction *I);
public:
explicit Verifier(raw_ostream *OS, bool ShouldTreatBrokenDebugInfoAsError,
const Module &M)
: VerifierSupport(OS, M), LandingPadResultTy(nullptr),
SawFrameEscape(false), TBAAVerifyHelper(this) {
TreatBrokenDebugInfoAsError = ShouldTreatBrokenDebugInfoAsError;
}
bool hasBrokenDebugInfo() const { return BrokenDebugInfo; }
2001-06-06 22:29:01 +02:00
bool verify(const Function &F) {
assert(F.getParent() == &M &&
"An instance of this class only works with a specific module!");
// First ensure the function is well-enough formed to compute dominance
// information, and directly compute a dominance tree. We don't rely on the
// pass manager to provide this as it isolates us from a potentially
// out-of-date dominator tree and makes it significantly more complex to run
// this code outside of a pass manager.
// FIXME: It's really gross that we have to cast away constness here.
if (!F.empty())
DT.recalculate(const_cast<Function &>(F));
2016-02-25 17:44:27 +01:00
for (const BasicBlock &BB : F) {
if (!BB.empty() && BB.back().isTerminator())
continue;
if (OS) {
*OS << "Basic Block in function '" << F.getName()
<< "' does not have terminator!\n";
BB.printAsOperand(*OS, true, MST);
*OS << "\n";
[PM] Remove the preverifier and directly compute the DominatorTree for the verifier after ensuring the CFG is at least usefully formed. This fixes a number of problems: 1) The PreVerifier was missing the controls the Verifier provides over *how* an invalid module is handled -- it just aborted the program! Now it uses the same logic as the Verifier which is significantly more library-friendly. 2) The DominatorTree used previously could have been cached and not updated due to bugs in prior passes and we would silently use the stale tree. This could cause dominance errors to not be as quickly diagnosed. 3) We can now (in the next patch) pull the functionality of the verifier apart from the pass infrastructure so that you can verify IR without having any form of pass manager. This in turn frees the code to share logic between old and new pass manager variants. Along the way I fixed at least one annoying bug -- the state for 'Broken' wasn't being cleared from run to run causing all functions visited after the first broken function to be marked as broken regardless of whether *they* were a problem. Fortunately, I don't really know much of a way to observe this peculiarity. In case folks are worried about the runtime cost, its negligible. I looked at running the entire regression test suite (which should be a relatively good use of the verifier) before and after but was unable to even measure the time spent on the verifier and there was no regresion from before to after. I checked both with debug builds and optimized builds. llvm-svn: 199487
2014-01-17 11:56:02 +01:00
}
return false;
}
Broken = false;
// FIXME: We strip const here because the inst visitor strips const.
visit(const_cast<Function &>(F));
verifySiblingFuncletUnwinds();
InstsInThisBlock.clear();
DebugFnArgs.clear();
LandingPadResultTy = nullptr;
SawFrameEscape = false;
SiblingFuncletInfo.clear();
return !Broken;
}
/// Verify the module that this instance of \c Verifier was initialized with.
bool verify() {
Broken = false;
// Collect all declarations of the llvm.experimental.deoptimize intrinsic.
for (const Function &F : M)
if (F.getIntrinsicID() == Intrinsic::experimental_deoptimize)
DeoptimizeDeclarations.push_back(&F);
// Now that we've visited every function, verify that we never asked to
// recover a frame index that wasn't escaped.
verifyFrameRecoverIndices();
2016-02-25 17:44:27 +01:00
for (const GlobalVariable &GV : M.globals())
visitGlobalVariable(GV);
2016-02-25 17:44:27 +01:00
for (const GlobalAlias &GA : M.aliases())
visitGlobalAlias(GA);
2016-02-25 17:44:27 +01:00
for (const NamedMDNode &NMD : M.named_metadata())
visitNamedMDNode(NMD);
for (const StringMapEntry<Comdat> &SMEC : M.getComdatSymbolTable())
visitComdat(SMEC.getValue());
visitModuleFlags(M);
visitModuleIdents(M);
visitModuleCommandLines(M);
verifyCompileUnits();
verifyDeoptimizeCallingConvs();
DISubprogramAttachments.clear();
return !Broken;
}
private:
// Verification methods...
void visitGlobalValue(const GlobalValue &GV);
void visitGlobalVariable(const GlobalVariable &GV);
void visitGlobalAlias(const GlobalAlias &GA);
void visitAliaseeSubExpr(const GlobalAlias &A, const Constant &C);
void visitAliaseeSubExpr(SmallPtrSetImpl<const GlobalAlias *> &Visited,
const GlobalAlias &A, const Constant &C);
void visitNamedMDNode(const NamedMDNode &NMD);
void visitMDNode(const MDNode &MD);
void visitMetadataAsValue(const MetadataAsValue &MD, Function *F);
void visitValueAsMetadata(const ValueAsMetadata &MD, Function *F);
void visitComdat(const Comdat &C);
void visitModuleIdents(const Module &M);
void visitModuleCommandLines(const Module &M);
void visitModuleFlags(const Module &M);
void visitModuleFlag(const MDNode *Op,
DenseMap<const MDString *, const MDNode *> &SeenIDs,
SmallVectorImpl<const MDNode *> &Requirements);
void visitModuleFlagCGProfileEntry(const MDOperand &MDO);
void visitFunction(const Function &F);
void visitBasicBlock(BasicBlock &BB);
void visitRangeMetadata(Instruction &I, MDNode *Range, Type *Ty);
void visitDereferenceableMetadata(Instruction &I, MDNode *MD);
template <class Ty> bool isValidMetadataArray(const MDTuple &N);
#define HANDLE_SPECIALIZED_MDNODE_LEAF(CLASS) void visit##CLASS(const CLASS &N);
#include "llvm/IR/Metadata.def"
void visitDIScope(const DIScope &N);
void visitDIVariable(const DIVariable &N);
void visitDILexicalBlockBase(const DILexicalBlockBase &N);
void visitDITemplateParameter(const DITemplateParameter &N);
void visitTemplateParams(const MDNode &N, const Metadata &RawParams);
// InstVisitor overrides...
using InstVisitor<Verifier>::visit;
void visit(Instruction &I);
void visitTruncInst(TruncInst &I);
void visitZExtInst(ZExtInst &I);
void visitSExtInst(SExtInst &I);
void visitFPTruncInst(FPTruncInst &I);
void visitFPExtInst(FPExtInst &I);
void visitFPToUIInst(FPToUIInst &I);
void visitFPToSIInst(FPToSIInst &I);
void visitUIToFPInst(UIToFPInst &I);
void visitSIToFPInst(SIToFPInst &I);
void visitIntToPtrInst(IntToPtrInst &I);
void visitPtrToIntInst(PtrToIntInst &I);
void visitBitCastInst(BitCastInst &I);
void visitAddrSpaceCastInst(AddrSpaceCastInst &I);
void visitPHINode(PHINode &PN);
void visitUnaryOperator(UnaryOperator &U);
void visitBinaryOperator(BinaryOperator &B);
void visitICmpInst(ICmpInst &IC);
void visitFCmpInst(FCmpInst &FC);
void visitExtractElementInst(ExtractElementInst &EI);
void visitInsertElementInst(InsertElementInst &EI);
void visitShuffleVectorInst(ShuffleVectorInst &EI);
void visitVAArgInst(VAArgInst &VAA) { visitInstruction(VAA); }
void visitCallInst(CallInst &CI);
void visitInvokeInst(InvokeInst &II);
void visitGetElementPtrInst(GetElementPtrInst &GEP);
void visitLoadInst(LoadInst &LI);
void visitStoreInst(StoreInst &SI);
void verifyDominatesUse(Instruction &I, unsigned i);
void visitInstruction(Instruction &I);
void visitTerminator(Instruction &I);
void visitBranchInst(BranchInst &BI);
void visitReturnInst(ReturnInst &RI);
void visitSwitchInst(SwitchInst &SI);
void visitIndirectBrInst(IndirectBrInst &BI);
void visitSelectInst(SelectInst &SI);
void visitUserOp1(Instruction &I);
void visitUserOp2(Instruction &I) { visitUserOp1(I); }
void visitIntrinsicCallSite(Intrinsic::ID ID, CallSite CS);
void visitConstrainedFPIntrinsic(ConstrainedFPIntrinsic &FPI);
void visitDbgIntrinsic(StringRef Kind, DbgVariableIntrinsic &DII);
void visitDbgLabelIntrinsic(StringRef Kind, DbgLabelInst &DLI);
void visitAtomicCmpXchgInst(AtomicCmpXchgInst &CXI);
void visitAtomicRMWInst(AtomicRMWInst &RMWI);
void visitFenceInst(FenceInst &FI);
void visitAllocaInst(AllocaInst &AI);
void visitExtractValueInst(ExtractValueInst &EVI);
void visitInsertValueInst(InsertValueInst &IVI);
void visitEHPadPredecessors(Instruction &I);
void visitLandingPadInst(LandingPadInst &LPI);
void visitResumeInst(ResumeInst &RI);
void visitCatchPadInst(CatchPadInst &CPI);
[IR] Reformulate LLVM's EH funclet IR While we have successfully implemented a funclet-oriented EH scheme on top of LLVM IR, our scheme has some notable deficiencies: - catchendpad and cleanupendpad are necessary in the current design but they are difficult to explain to others, even to seasoned LLVM experts. - catchendpad and cleanupendpad are optimization barriers. They cannot be split and force all potentially throwing call-sites to be invokes. This has a noticable effect on the quality of our code generation. - catchpad, while similar in some aspects to invoke, is fairly awkward. It is unsplittable, starts a funclet, and has control flow to other funclets. - The nesting relationship between funclets is currently a property of control flow edges. Because of this, we are forced to carefully analyze the flow graph to see if there might potentially exist illegal nesting among funclets. While we have logic to clone funclets when they are illegally nested, it would be nicer if we had a representation which forbade them upfront. Let's clean this up a bit by doing the following: - Instead, make catchpad more like cleanuppad and landingpad: no control flow, just a bunch of simple operands; catchpad would be splittable. - Introduce catchswitch, a control flow instruction designed to model the constraints of funclet oriented EH. - Make funclet scoping explicit by having funclet instructions consume the token produced by the funclet which contains them. - Remove catchendpad and cleanupendpad. Their presence can be inferred implicitly using coloring information. N.B. The state numbering code for the CLR has been updated but the veracity of it's output cannot be spoken for. An expert should take a look to make sure the results are reasonable. Reviewers: rnk, JosephTremoulet, andrew.w.kaylor Differential Revision: http://reviews.llvm.org/D15139 llvm-svn: 255422
2015-12-12 06:38:55 +01:00
void visitCatchReturnInst(CatchReturnInst &CatchReturn);
void visitCleanupPadInst(CleanupPadInst &CPI);
void visitFuncletPadInst(FuncletPadInst &FPI);
[IR] Reformulate LLVM's EH funclet IR While we have successfully implemented a funclet-oriented EH scheme on top of LLVM IR, our scheme has some notable deficiencies: - catchendpad and cleanupendpad are necessary in the current design but they are difficult to explain to others, even to seasoned LLVM experts. - catchendpad and cleanupendpad are optimization barriers. They cannot be split and force all potentially throwing call-sites to be invokes. This has a noticable effect on the quality of our code generation. - catchpad, while similar in some aspects to invoke, is fairly awkward. It is unsplittable, starts a funclet, and has control flow to other funclets. - The nesting relationship between funclets is currently a property of control flow edges. Because of this, we are forced to carefully analyze the flow graph to see if there might potentially exist illegal nesting among funclets. While we have logic to clone funclets when they are illegally nested, it would be nicer if we had a representation which forbade them upfront. Let's clean this up a bit by doing the following: - Instead, make catchpad more like cleanuppad and landingpad: no control flow, just a bunch of simple operands; catchpad would be splittable. - Introduce catchswitch, a control flow instruction designed to model the constraints of funclet oriented EH. - Make funclet scoping explicit by having funclet instructions consume the token produced by the funclet which contains them. - Remove catchendpad and cleanupendpad. Their presence can be inferred implicitly using coloring information. N.B. The state numbering code for the CLR has been updated but the veracity of it's output cannot be spoken for. An expert should take a look to make sure the results are reasonable. Reviewers: rnk, JosephTremoulet, andrew.w.kaylor Differential Revision: http://reviews.llvm.org/D15139 llvm-svn: 255422
2015-12-12 06:38:55 +01:00
void visitCatchSwitchInst(CatchSwitchInst &CatchSwitch);
void visitCleanupReturnInst(CleanupReturnInst &CRI);
void verifyCallSite(CallSite CS);
void verifySwiftErrorCallSite(CallSite CS, const Value *SwiftErrorVal);
void verifySwiftErrorValue(const Value *SwiftErrorVal);
void verifyMustTailCall(CallInst &CI);
bool performTypeCheck(Intrinsic::ID ID, Function *F, Type *Ty, int VT,
unsigned ArgNo, std::string &Suffix);
bool verifyAttributeCount(AttributeList Attrs, unsigned Params);
void verifyAttributeTypes(AttributeSet Attrs, bool IsFunction,
const Value *V);
void verifyParameterAttrs(AttributeSet Attrs, Type *Ty, const Value *V);
void verifyFunctionAttrs(FunctionType *FT, AttributeList Attrs,
const Value *V);
void verifyFunctionMetadata(ArrayRef<std::pair<unsigned, MDNode *>> MDs);
void visitConstantExprsRecursively(const Constant *EntryC);
void visitConstantExpr(const ConstantExpr *CE);
void verifyStatepoint(ImmutableCallSite CS);
void verifyFrameRecoverIndices();
void verifySiblingFuncletUnwinds();
void verifyFragmentExpression(const DbgVariableIntrinsic &I);
template <typename ValueOrMetadata>
void verifyFragmentExpression(const DIVariable &V,
DIExpression::FragmentInfo Fragment,
ValueOrMetadata *Desc);
void verifyFnArgs(const DbgVariableIntrinsic &I);
/// Module-level debug info verification...
void verifyCompileUnits();
/// Module-level verification that all @llvm.experimental.deoptimize
/// declarations share the same calling convention.
void verifyDeoptimizeCallingConvs();
/// Verify all-or-nothing property of DIFile source attribute within a CU.
void verifySourceDebugInfo(const DICompileUnit &U, const DIFile &F);
};
} // end anonymous namespace
/// We know that cond should be true, if not print an error message.
#define Assert(C, ...) \
do { if (!(C)) { CheckFailed(__VA_ARGS__); return; } } while (false)
/// We know that a debug info condition should be true, if not print
/// an error message.
#define AssertDI(C, ...) \
do { if (!(C)) { DebugInfoCheckFailed(__VA_ARGS__); return; } } while (false)
void Verifier::visit(Instruction &I) {
for (unsigned i = 0, e = I.getNumOperands(); i != e; ++i)
Assert(I.getOperand(i) != nullptr, "Operand is null", &I);
InstVisitor<Verifier>::visit(I);
}
// Helper to recursively iterate over indirect users. By
// returning false, the callback can ask to stop recursing
// further.
static void forEachUser(const Value *User,
SmallPtrSet<const Value *, 32> &Visited,
llvm::function_ref<bool(const Value *)> Callback) {
if (!Visited.insert(User).second)
return;
for (const Value *TheNextUser : User->materialized_users())
if (Callback(TheNextUser))
forEachUser(TheNextUser, Visited, Callback);
}
void Verifier::visitGlobalValue(const GlobalValue &GV) {
Assert(!GV.isDeclaration() || GV.hasValidDeclarationLinkage(),
"Global is external, but doesn't have external or weak linkage!", &GV);
Assert(GV.getAlignment() <= Value::MaximumAlignment,
"huge alignment values are unsupported", &GV);
Assert(!GV.hasAppendingLinkage() || isa<GlobalVariable>(GV),
"Only global variables can have appending linkage!", &GV);
if (GV.hasAppendingLinkage()) {
const GlobalVariable *GVar = dyn_cast<GlobalVariable>(&GV);
Assert(GVar && GVar->getValueType()->isArrayTy(),
"Only global arrays can have appending linkage!", GVar);
}
if (GV.isDeclarationForLinker())
Assert(!GV.hasComdat(), "Declaration may not be in a Comdat!", &GV);
if (GV.hasDLLImportStorageClass()) {
Represent runtime preemption in the IR. Currently we do not represent runtime preemption in the IR, which has several drawbacks: 1) The semantics of GlobalValues differ depending on the object file format you are targeting (as well as the relocation-model and -fPIE value). 2) We have no way of disabling inlining of run time interposable functions, since in the IR we only know if a function is link-time interposable. Because of this llvm cannot support elf-interposition semantics. 3) In LTO builds of executables we will have extra knowledge that a symbol resolved to a local definition and can't be preemptable, but have no way to propagate that knowledge through the compiler. This patch adds preemptability specifiers to the IR with the following meaning: dso_local --> means the compiler may assume the symbol will resolve to a definition within the current linkage unit and the symbol may be accessed directly even if the definition is not within this compilation unit. dso_preemptable --> means that the compiler must assume the GlobalValue may be replaced with a definition from outside the current linkage unit at runtime. To ease transitioning dso_preemptable is treated as a 'default' in that low-level codegen will still do the same checks it did previously to see if a symbol should be accessed indirectly. Eventually when IR producers emit the specifiers on all Globalvalues we can change dso_preemptable to mean 'always access indirectly', and remove the current logic. Differential Revision: https://reviews.llvm.org/D20217 llvm-svn: 316668
2017-10-26 17:00:26 +02:00
Assert(!GV.isDSOLocal(),
"GlobalValue with DLLImport Storage is dso_local!", &GV);
Assert((GV.isDeclaration() && GV.hasExternalLinkage()) ||
GV.hasAvailableExternallyLinkage(),
"Global is marked as dllimport, but not external", &GV);
}
if (GV.hasLocalLinkage())
Assert(GV.isDSOLocal(),
"GlobalValue with private or internal linkage must be dso_local!",
&GV);
if (!GV.hasDefaultVisibility() && !GV.hasExternalWeakLinkage())
Assert(GV.isDSOLocal(),
"GlobalValue with non default visibility must be dso_local!", &GV);
forEachUser(&GV, GlobalValueVisited, [&](const Value *V) -> bool {
if (const Instruction *I = dyn_cast<Instruction>(V)) {
if (!I->getParent() || !I->getParent()->getParent())
CheckFailed("Global is referenced by parentless instruction!", &GV, &M,
I);
else if (I->getParent()->getParent()->getParent() != &M)
CheckFailed("Global is referenced in a different module!", &GV, &M, I,
I->getParent()->getParent(),
I->getParent()->getParent()->getParent());
return false;
} else if (const Function *F = dyn_cast<Function>(V)) {
if (F->getParent() != &M)
CheckFailed("Global is used by function in a different module", &GV, &M,
F, F->getParent());
return false;
}
return true;
});
}
void Verifier::visitGlobalVariable(const GlobalVariable &GV) {
if (GV.hasInitializer()) {
Assert(GV.getInitializer()->getType() == GV.getValueType(),
"Global variable initializer type does not match global "
"variable type!",
&GV);
// If the global has common linkage, it must have a zero initializer and
// cannot be constant.
if (GV.hasCommonLinkage()) {
Assert(GV.getInitializer()->isNullValue(),
"'common' global must have a zero initializer!", &GV);
Assert(!GV.isConstant(), "'common' global may not be marked constant!",
&GV);
Assert(!GV.hasComdat(), "'common' global may not be in a Comdat!", &GV);
}
}
if (GV.hasName() && (GV.getName() == "llvm.global_ctors" ||
GV.getName() == "llvm.global_dtors")) {
Assert(!GV.hasInitializer() || GV.hasAppendingLinkage(),
"invalid linkage for intrinsic global variable", &GV);
// Don't worry about emitting an error for it not being an array,
// visitGlobalValue will complain on appending non-array.
if (ArrayType *ATy = dyn_cast<ArrayType>(GV.getValueType())) {
StructType *STy = dyn_cast<StructType>(ATy->getElementType());
PointerType *FuncPtrTy =
FunctionType::get(Type::getVoidTy(Context), false)->
getPointerTo(DL.getProgramAddressSpace());
// FIXME: Reject the 2-field form in LLVM 4.0.
Assert(STy &&
(STy->getNumElements() == 2 || STy->getNumElements() == 3) &&
STy->getTypeAtIndex(0u)->isIntegerTy(32) &&
STy->getTypeAtIndex(1) == FuncPtrTy,
"wrong type for intrinsic global variable", &GV);
if (STy->getNumElements() == 3) {
Type *ETy = STy->getTypeAtIndex(2);
Assert(ETy->isPointerTy() &&
cast<PointerType>(ETy)->getElementType()->isIntegerTy(8),
"wrong type for intrinsic global variable", &GV);
}
}
}
if (GV.hasName() && (GV.getName() == "llvm.used" ||
GV.getName() == "llvm.compiler.used")) {
Assert(!GV.hasInitializer() || GV.hasAppendingLinkage(),
"invalid linkage for intrinsic global variable", &GV);
Type *GVType = GV.getValueType();
if (ArrayType *ATy = dyn_cast<ArrayType>(GVType)) {
PointerType *PTy = dyn_cast<PointerType>(ATy->getElementType());
Assert(PTy, "wrong type for intrinsic global variable", &GV);
if (GV.hasInitializer()) {
const Constant *Init = GV.getInitializer();
const ConstantArray *InitArray = dyn_cast<ConstantArray>(Init);
Assert(InitArray, "wrong initalizer for intrinsic global variable",
Init);
2016-02-25 17:44:27 +01:00
for (Value *Op : InitArray->operands()) {
Value *V = Op->stripPointerCastsNoFollowAliases();
Assert(isa<GlobalVariable>(V) || isa<Function>(V) ||
isa<GlobalAlias>(V),
"invalid llvm.used member", V);
Assert(V->hasName(), "members of llvm.used must be named", V);
}
}
}
}
// Visit any debug info attachments.
SmallVector<MDNode *, 1> MDs;
GV.getMetadata(LLVMContext::MD_dbg, MDs);
for (auto *MD : MDs) {
if (auto *GVE = dyn_cast<DIGlobalVariableExpression>(MD))
visitDIGlobalVariableExpression(*GVE);
else
AssertDI(false, "!dbg attachment of global variable must be a "
"DIGlobalVariableExpression");
}
if (!GV.hasInitializer()) {
visitGlobalValue(GV);
return;
}
// Walk any aggregate initializers looking for bitcasts between address spaces
visitConstantExprsRecursively(GV.getInitializer());
visitGlobalValue(GV);
}
void Verifier::visitAliaseeSubExpr(const GlobalAlias &GA, const Constant &C) {
SmallPtrSet<const GlobalAlias*, 4> Visited;
Visited.insert(&GA);
visitAliaseeSubExpr(Visited, GA, C);
}
void Verifier::visitAliaseeSubExpr(SmallPtrSetImpl<const GlobalAlias*> &Visited,
const GlobalAlias &GA, const Constant &C) {
if (const auto *GV = dyn_cast<GlobalValue>(&C)) {
Assert(!GV->isDeclarationForLinker(), "Alias must point to a definition",
&GA);
if (const auto *GA2 = dyn_cast<GlobalAlias>(GV)) {
Assert(Visited.insert(GA2).second, "Aliases cannot form a cycle", &GA);
Don't IPO over functions that can be de-refined Summary: Fixes PR26774. If you're aware of the issue, feel free to skip the "Motivation" section and jump directly to "This patch". Motivation: I define "refinement" as discarding behaviors from a program that the optimizer has license to discard. So transforming: ``` void f(unsigned x) { unsigned t = 5 / x; (void)t; } ``` to ``` void f(unsigned x) { } ``` is refinement, since the behavior went from "if x == 0 then undefined else nothing" to "nothing" (the optimizer has license to discard undefined behavior). Refinement is a fundamental aspect of many mid-level optimizations done by LLVM. For instance, transforming `x == (x + 1)` to `false` also involves refinement since the expression's value went from "if x is `undef` then { `true` or `false` } else { `false` }" to "`false`" (by definition, the optimizer has license to fold `undef` to any non-`undef` value). Unfortunately, refinement implies that the optimizer cannot assume that the implementation of a function it can see has all of the behavior an unoptimized or a differently optimized version of the same function can have. This is a problem for functions with comdat linkage, where a function can be replaced by an unoptimized or a differently optimized version of the same source level function. For instance, FunctionAttrs cannot assume a comdat function is actually `readnone` even if it does not have any loads or stores in it; since there may have been loads and stores in the "original function" that were refined out in the currently visible variant, and at the link step the linker may in fact choose an implementation with a load or a store. As an example, consider a function that does two atomic loads from the same memory location, and writes to memory only if the two values are not equal. The optimizer is allowed to refine this function by first CSE'ing the two loads, and the folding the comparision to always report that the two values are equal. Such a refined variant will look like it is `readonly`. However, the unoptimized version of the function can still write to memory (since the two loads //can// result in different values), and selecting the unoptimized version at link time will retroactively invalidate transforms we may have done under the assumption that the function does not write to memory. Note: this is not just a problem with atomics or with linking differently optimized object files. See PR26774 for more realistic examples that involved neither. This patch: This change introduces a new set of linkage types, predicated as `GlobalValue::mayBeDerefined` that returns true if the linkage type allows a function to be replaced by a differently optimized variant at link time. It then changes a set of IPO passes to bail out if they see such a function. Reviewers: chandlerc, hfinkel, dexonsmith, joker.eph, rnk Subscribers: mcrosier, llvm-commits Differential Revision: http://reviews.llvm.org/D18634 llvm-svn: 265762
2016-04-08 02:48:30 +02:00
Assert(!GA2->isInterposable(), "Alias cannot point to an interposable alias",
&GA);
} else {
// Only continue verifying subexpressions of GlobalAliases.
// Do not recurse into global initializers.
return;
}
}
if (const auto *CE = dyn_cast<ConstantExpr>(&C))
visitConstantExprsRecursively(CE);
for (const Use &U : C.operands()) {
Value *V = &*U;
if (const auto *GA2 = dyn_cast<GlobalAlias>(V))
visitAliaseeSubExpr(Visited, GA, *GA2->getAliasee());
else if (const auto *C2 = dyn_cast<Constant>(V))
visitAliaseeSubExpr(Visited, GA, *C2);
}
}
void Verifier::visitGlobalAlias(const GlobalAlias &GA) {
Assert(GlobalAlias::isValidLinkage(GA.getLinkage()),
"Alias should have private, internal, linkonce, weak, linkonce_odr, "
"weak_odr, or external linkage!",
&GA);
const Constant *Aliasee = GA.getAliasee();
Assert(Aliasee, "Aliasee cannot be NULL!", &GA);
Assert(GA.getType() == Aliasee->getType(),
"Alias and aliasee types should match!", &GA);
Assert(isa<GlobalValue>(Aliasee) || isa<ConstantExpr>(Aliasee),
"Aliasee should be either GlobalValue or ConstantExpr", &GA);
visitAliaseeSubExpr(GA, *Aliasee);
visitGlobalValue(GA);
}
void Verifier::visitNamedMDNode(const NamedMDNode &NMD) {
// There used to be various other llvm.dbg.* nodes, but we don't support
// upgrading them and we want to reserve the namespace for future uses.
if (NMD.getName().startswith("llvm.dbg."))
AssertDI(NMD.getName() == "llvm.dbg.cu",
"unrecognized named metadata node in the llvm.dbg namespace",
&NMD);
2016-02-25 17:44:27 +01:00
for (const MDNode *MD : NMD.operands()) {
if (NMD.getName() == "llvm.dbg.cu")
AssertDI(MD && isa<DICompileUnit>(MD), "invalid compile unit", &NMD, MD);
if (!MD)
continue;
IR: Split Metadata from Value Split `Metadata` away from the `Value` class hierarchy, as part of PR21532. Assembly and bitcode changes are in the wings, but this is the bulk of the change for the IR C++ API. I have a follow-up patch prepared for `clang`. If this breaks other sub-projects, I apologize in advance :(. Help me compile it on Darwin I'll try to fix it. FWIW, the errors should be easy to fix, so it may be simpler to just fix it yourself. This breaks the build for all metadata-related code that's out-of-tree. Rest assured the transition is mechanical and the compiler should catch almost all of the problems. Here's a quick guide for updating your code: - `Metadata` is the root of a class hierarchy with three main classes: `MDNode`, `MDString`, and `ValueAsMetadata`. It is distinct from the `Value` class hierarchy. It is typeless -- i.e., instances do *not* have a `Type`. - `MDNode`'s operands are all `Metadata *` (instead of `Value *`). - `TrackingVH<MDNode>` and `WeakVH` referring to metadata can be replaced with `TrackingMDNodeRef` and `TrackingMDRef`, respectively. If you're referring solely to resolved `MDNode`s -- post graph construction -- just use `MDNode*`. - `MDNode` (and the rest of `Metadata`) have only limited support for `replaceAllUsesWith()`. As long as an `MDNode` is pointing at a forward declaration -- the result of `MDNode::getTemporary()` -- it maintains a side map of its uses and can RAUW itself. Once the forward declarations are fully resolved RAUW support is dropped on the ground. This means that uniquing collisions on changing operands cause nodes to become "distinct". (This already happened fairly commonly, whenever an operand went to null.) If you're constructing complex (non self-reference) `MDNode` cycles, you need to call `MDNode::resolveCycles()` on each node (or on a top-level node that somehow references all of the nodes). Also, don't do that. Metadata cycles (and the RAUW machinery needed to construct them) are expensive. - An `MDNode` can only refer to a `Constant` through a bridge called `ConstantAsMetadata` (one of the subclasses of `ValueAsMetadata`). As a side effect, accessing an operand of an `MDNode` that is known to be, e.g., `ConstantInt`, takes three steps: first, cast from `Metadata` to `ConstantAsMetadata`; second, extract the `Constant`; third, cast down to `ConstantInt`. The eventual goal is to introduce `MDInt`/`MDFloat`/etc. and have metadata schema owners transition away from using `Constant`s when the type isn't important (and they don't care about referring to `GlobalValue`s). In the meantime, I've added transitional API to the `mdconst` namespace that matches semantics with the old code, in order to avoid adding the error-prone three-step equivalent to every call site. If your old code was: MDNode *N = foo(); bar(isa <ConstantInt>(N->getOperand(0))); baz(cast <ConstantInt>(N->getOperand(1))); bak(cast_or_null <ConstantInt>(N->getOperand(2))); bat(dyn_cast <ConstantInt>(N->getOperand(3))); bay(dyn_cast_or_null<ConstantInt>(N->getOperand(4))); you can trivially match its semantics with: MDNode *N = foo(); bar(mdconst::hasa <ConstantInt>(N->getOperand(0))); baz(mdconst::extract <ConstantInt>(N->getOperand(1))); bak(mdconst::extract_or_null <ConstantInt>(N->getOperand(2))); bat(mdconst::dyn_extract <ConstantInt>(N->getOperand(3))); bay(mdconst::dyn_extract_or_null<ConstantInt>(N->getOperand(4))); and when you transition your metadata schema to `MDInt`: MDNode *N = foo(); bar(isa <MDInt>(N->getOperand(0))); baz(cast <MDInt>(N->getOperand(1))); bak(cast_or_null <MDInt>(N->getOperand(2))); bat(dyn_cast <MDInt>(N->getOperand(3))); bay(dyn_cast_or_null<MDInt>(N->getOperand(4))); - A `CallInst` -- specifically, intrinsic instructions -- can refer to metadata through a bridge called `MetadataAsValue`. This is a subclass of `Value` where `getType()->isMetadataTy()`. `MetadataAsValue` is the *only* class that can legally refer to a `LocalAsMetadata`, which is a bridged form of non-`Constant` values like `Argument` and `Instruction`. It can also refer to any other `Metadata` subclass. (I'll break all your testcases in a follow-up commit, when I propagate this change to assembly.) llvm-svn: 223802
2014-12-09 19:38:53 +01:00
visitMDNode(*MD);
}
}
void Verifier::visitMDNode(const MDNode &MD) {
// Only visit each node once. Metadata can be mutually recursive, so this
// avoids infinite recursion here, as well as being an optimization.
if (!MDNodes.insert(&MD).second)
return;
switch (MD.getMetadataID()) {
default:
llvm_unreachable("Invalid MDNode subclass");
case Metadata::MDTupleKind:
break;
#define HANDLE_SPECIALIZED_MDNODE_LEAF(CLASS) \
case Metadata::CLASS##Kind: \
visit##CLASS(cast<CLASS>(MD)); \
break;
#include "llvm/IR/Metadata.def"
}
2016-02-25 17:44:27 +01:00
for (const Metadata *Op : MD.operands()) {
if (!Op)
continue;
Assert(!isa<LocalAsMetadata>(Op), "Invalid operand for global metadata!",
&MD, Op);
IR: Split Metadata from Value Split `Metadata` away from the `Value` class hierarchy, as part of PR21532. Assembly and bitcode changes are in the wings, but this is the bulk of the change for the IR C++ API. I have a follow-up patch prepared for `clang`. If this breaks other sub-projects, I apologize in advance :(. Help me compile it on Darwin I'll try to fix it. FWIW, the errors should be easy to fix, so it may be simpler to just fix it yourself. This breaks the build for all metadata-related code that's out-of-tree. Rest assured the transition is mechanical and the compiler should catch almost all of the problems. Here's a quick guide for updating your code: - `Metadata` is the root of a class hierarchy with three main classes: `MDNode`, `MDString`, and `ValueAsMetadata`. It is distinct from the `Value` class hierarchy. It is typeless -- i.e., instances do *not* have a `Type`. - `MDNode`'s operands are all `Metadata *` (instead of `Value *`). - `TrackingVH<MDNode>` and `WeakVH` referring to metadata can be replaced with `TrackingMDNodeRef` and `TrackingMDRef`, respectively. If you're referring solely to resolved `MDNode`s -- post graph construction -- just use `MDNode*`. - `MDNode` (and the rest of `Metadata`) have only limited support for `replaceAllUsesWith()`. As long as an `MDNode` is pointing at a forward declaration -- the result of `MDNode::getTemporary()` -- it maintains a side map of its uses and can RAUW itself. Once the forward declarations are fully resolved RAUW support is dropped on the ground. This means that uniquing collisions on changing operands cause nodes to become "distinct". (This already happened fairly commonly, whenever an operand went to null.) If you're constructing complex (non self-reference) `MDNode` cycles, you need to call `MDNode::resolveCycles()` on each node (or on a top-level node that somehow references all of the nodes). Also, don't do that. Metadata cycles (and the RAUW machinery needed to construct them) are expensive. - An `MDNode` can only refer to a `Constant` through a bridge called `ConstantAsMetadata` (one of the subclasses of `ValueAsMetadata`). As a side effect, accessing an operand of an `MDNode` that is known to be, e.g., `ConstantInt`, takes three steps: first, cast from `Metadata` to `ConstantAsMetadata`; second, extract the `Constant`; third, cast down to `ConstantInt`. The eventual goal is to introduce `MDInt`/`MDFloat`/etc. and have metadata schema owners transition away from using `Constant`s when the type isn't important (and they don't care about referring to `GlobalValue`s). In the meantime, I've added transitional API to the `mdconst` namespace that matches semantics with the old code, in order to avoid adding the error-prone three-step equivalent to every call site. If your old code was: MDNode *N = foo(); bar(isa <ConstantInt>(N->getOperand(0))); baz(cast <ConstantInt>(N->getOperand(1))); bak(cast_or_null <ConstantInt>(N->getOperand(2))); bat(dyn_cast <ConstantInt>(N->getOperand(3))); bay(dyn_cast_or_null<ConstantInt>(N->getOperand(4))); you can trivially match its semantics with: MDNode *N = foo(); bar(mdconst::hasa <ConstantInt>(N->getOperand(0))); baz(mdconst::extract <ConstantInt>(N->getOperand(1))); bak(mdconst::extract_or_null <ConstantInt>(N->getOperand(2))); bat(mdconst::dyn_extract <ConstantInt>(N->getOperand(3))); bay(mdconst::dyn_extract_or_null<ConstantInt>(N->getOperand(4))); and when you transition your metadata schema to `MDInt`: MDNode *N = foo(); bar(isa <MDInt>(N->getOperand(0))); baz(cast <MDInt>(N->getOperand(1))); bak(cast_or_null <MDInt>(N->getOperand(2))); bat(dyn_cast <MDInt>(N->getOperand(3))); bay(dyn_cast_or_null<MDInt>(N->getOperand(4))); - A `CallInst` -- specifically, intrinsic instructions -- can refer to metadata through a bridge called `MetadataAsValue`. This is a subclass of `Value` where `getType()->isMetadataTy()`. `MetadataAsValue` is the *only* class that can legally refer to a `LocalAsMetadata`, which is a bridged form of non-`Constant` values like `Argument` and `Instruction`. It can also refer to any other `Metadata` subclass. (I'll break all your testcases in a follow-up commit, when I propagate this change to assembly.) llvm-svn: 223802
2014-12-09 19:38:53 +01:00
if (auto *N = dyn_cast<MDNode>(Op)) {
visitMDNode(*N);
continue;
IR: Split Metadata from Value Split `Metadata` away from the `Value` class hierarchy, as part of PR21532. Assembly and bitcode changes are in the wings, but this is the bulk of the change for the IR C++ API. I have a follow-up patch prepared for `clang`. If this breaks other sub-projects, I apologize in advance :(. Help me compile it on Darwin I'll try to fix it. FWIW, the errors should be easy to fix, so it may be simpler to just fix it yourself. This breaks the build for all metadata-related code that's out-of-tree. Rest assured the transition is mechanical and the compiler should catch almost all of the problems. Here's a quick guide for updating your code: - `Metadata` is the root of a class hierarchy with three main classes: `MDNode`, `MDString`, and `ValueAsMetadata`. It is distinct from the `Value` class hierarchy. It is typeless -- i.e., instances do *not* have a `Type`. - `MDNode`'s operands are all `Metadata *` (instead of `Value *`). - `TrackingVH<MDNode>` and `WeakVH` referring to metadata can be replaced with `TrackingMDNodeRef` and `TrackingMDRef`, respectively. If you're referring solely to resolved `MDNode`s -- post graph construction -- just use `MDNode*`. - `MDNode` (and the rest of `Metadata`) have only limited support for `replaceAllUsesWith()`. As long as an `MDNode` is pointing at a forward declaration -- the result of `MDNode::getTemporary()` -- it maintains a side map of its uses and can RAUW itself. Once the forward declarations are fully resolved RAUW support is dropped on the ground. This means that uniquing collisions on changing operands cause nodes to become "distinct". (This already happened fairly commonly, whenever an operand went to null.) If you're constructing complex (non self-reference) `MDNode` cycles, you need to call `MDNode::resolveCycles()` on each node (or on a top-level node that somehow references all of the nodes). Also, don't do that. Metadata cycles (and the RAUW machinery needed to construct them) are expensive. - An `MDNode` can only refer to a `Constant` through a bridge called `ConstantAsMetadata` (one of the subclasses of `ValueAsMetadata`). As a side effect, accessing an operand of an `MDNode` that is known to be, e.g., `ConstantInt`, takes three steps: first, cast from `Metadata` to `ConstantAsMetadata`; second, extract the `Constant`; third, cast down to `ConstantInt`. The eventual goal is to introduce `MDInt`/`MDFloat`/etc. and have metadata schema owners transition away from using `Constant`s when the type isn't important (and they don't care about referring to `GlobalValue`s). In the meantime, I've added transitional API to the `mdconst` namespace that matches semantics with the old code, in order to avoid adding the error-prone three-step equivalent to every call site. If your old code was: MDNode *N = foo(); bar(isa <ConstantInt>(N->getOperand(0))); baz(cast <ConstantInt>(N->getOperand(1))); bak(cast_or_null <ConstantInt>(N->getOperand(2))); bat(dyn_cast <ConstantInt>(N->getOperand(3))); bay(dyn_cast_or_null<ConstantInt>(N->getOperand(4))); you can trivially match its semantics with: MDNode *N = foo(); bar(mdconst::hasa <ConstantInt>(N->getOperand(0))); baz(mdconst::extract <ConstantInt>(N->getOperand(1))); bak(mdconst::extract_or_null <ConstantInt>(N->getOperand(2))); bat(mdconst::dyn_extract <ConstantInt>(N->getOperand(3))); bay(mdconst::dyn_extract_or_null<ConstantInt>(N->getOperand(4))); and when you transition your metadata schema to `MDInt`: MDNode *N = foo(); bar(isa <MDInt>(N->getOperand(0))); baz(cast <MDInt>(N->getOperand(1))); bak(cast_or_null <MDInt>(N->getOperand(2))); bat(dyn_cast <MDInt>(N->getOperand(3))); bay(dyn_cast_or_null<MDInt>(N->getOperand(4))); - A `CallInst` -- specifically, intrinsic instructions -- can refer to metadata through a bridge called `MetadataAsValue`. This is a subclass of `Value` where `getType()->isMetadataTy()`. `MetadataAsValue` is the *only* class that can legally refer to a `LocalAsMetadata`, which is a bridged form of non-`Constant` values like `Argument` and `Instruction`. It can also refer to any other `Metadata` subclass. (I'll break all your testcases in a follow-up commit, when I propagate this change to assembly.) llvm-svn: 223802
2014-12-09 19:38:53 +01:00
}
if (auto *V = dyn_cast<ValueAsMetadata>(Op)) {
visitValueAsMetadata(*V, nullptr);
continue;
}
}
IR: Split Metadata from Value Split `Metadata` away from the `Value` class hierarchy, as part of PR21532. Assembly and bitcode changes are in the wings, but this is the bulk of the change for the IR C++ API. I have a follow-up patch prepared for `clang`. If this breaks other sub-projects, I apologize in advance :(. Help me compile it on Darwin I'll try to fix it. FWIW, the errors should be easy to fix, so it may be simpler to just fix it yourself. This breaks the build for all metadata-related code that's out-of-tree. Rest assured the transition is mechanical and the compiler should catch almost all of the problems. Here's a quick guide for updating your code: - `Metadata` is the root of a class hierarchy with three main classes: `MDNode`, `MDString`, and `ValueAsMetadata`. It is distinct from the `Value` class hierarchy. It is typeless -- i.e., instances do *not* have a `Type`. - `MDNode`'s operands are all `Metadata *` (instead of `Value *`). - `TrackingVH<MDNode>` and `WeakVH` referring to metadata can be replaced with `TrackingMDNodeRef` and `TrackingMDRef`, respectively. If you're referring solely to resolved `MDNode`s -- post graph construction -- just use `MDNode*`. - `MDNode` (and the rest of `Metadata`) have only limited support for `replaceAllUsesWith()`. As long as an `MDNode` is pointing at a forward declaration -- the result of `MDNode::getTemporary()` -- it maintains a side map of its uses and can RAUW itself. Once the forward declarations are fully resolved RAUW support is dropped on the ground. This means that uniquing collisions on changing operands cause nodes to become "distinct". (This already happened fairly commonly, whenever an operand went to null.) If you're constructing complex (non self-reference) `MDNode` cycles, you need to call `MDNode::resolveCycles()` on each node (or on a top-level node that somehow references all of the nodes). Also, don't do that. Metadata cycles (and the RAUW machinery needed to construct them) are expensive. - An `MDNode` can only refer to a `Constant` through a bridge called `ConstantAsMetadata` (one of the subclasses of `ValueAsMetadata`). As a side effect, accessing an operand of an `MDNode` that is known to be, e.g., `ConstantInt`, takes three steps: first, cast from `Metadata` to `ConstantAsMetadata`; second, extract the `Constant`; third, cast down to `ConstantInt`. The eventual goal is to introduce `MDInt`/`MDFloat`/etc. and have metadata schema owners transition away from using `Constant`s when the type isn't important (and they don't care about referring to `GlobalValue`s). In the meantime, I've added transitional API to the `mdconst` namespace that matches semantics with the old code, in order to avoid adding the error-prone three-step equivalent to every call site. If your old code was: MDNode *N = foo(); bar(isa <ConstantInt>(N->getOperand(0))); baz(cast <ConstantInt>(N->getOperand(1))); bak(cast_or_null <ConstantInt>(N->getOperand(2))); bat(dyn_cast <ConstantInt>(N->getOperand(3))); bay(dyn_cast_or_null<ConstantInt>(N->getOperand(4))); you can trivially match its semantics with: MDNode *N = foo(); bar(mdconst::hasa <ConstantInt>(N->getOperand(0))); baz(mdconst::extract <ConstantInt>(N->getOperand(1))); bak(mdconst::extract_or_null <ConstantInt>(N->getOperand(2))); bat(mdconst::dyn_extract <ConstantInt>(N->getOperand(3))); bay(mdconst::dyn_extract_or_null<ConstantInt>(N->getOperand(4))); and when you transition your metadata schema to `MDInt`: MDNode *N = foo(); bar(isa <MDInt>(N->getOperand(0))); baz(cast <MDInt>(N->getOperand(1))); bak(cast_or_null <MDInt>(N->getOperand(2))); bat(dyn_cast <MDInt>(N->getOperand(3))); bay(dyn_cast_or_null<MDInt>(N->getOperand(4))); - A `CallInst` -- specifically, intrinsic instructions -- can refer to metadata through a bridge called `MetadataAsValue`. This is a subclass of `Value` where `getType()->isMetadataTy()`. `MetadataAsValue` is the *only* class that can legally refer to a `LocalAsMetadata`, which is a bridged form of non-`Constant` values like `Argument` and `Instruction`. It can also refer to any other `Metadata` subclass. (I'll break all your testcases in a follow-up commit, when I propagate this change to assembly.) llvm-svn: 223802
2014-12-09 19:38:53 +01:00
// Check these last, so we diagnose problems in operands first.
Assert(!MD.isTemporary(), "Expected no forward declarations!", &MD);
Assert(MD.isResolved(), "All nodes should be resolved!", &MD);
IR: Split Metadata from Value Split `Metadata` away from the `Value` class hierarchy, as part of PR21532. Assembly and bitcode changes are in the wings, but this is the bulk of the change for the IR C++ API. I have a follow-up patch prepared for `clang`. If this breaks other sub-projects, I apologize in advance :(. Help me compile it on Darwin I'll try to fix it. FWIW, the errors should be easy to fix, so it may be simpler to just fix it yourself. This breaks the build for all metadata-related code that's out-of-tree. Rest assured the transition is mechanical and the compiler should catch almost all of the problems. Here's a quick guide for updating your code: - `Metadata` is the root of a class hierarchy with three main classes: `MDNode`, `MDString`, and `ValueAsMetadata`. It is distinct from the `Value` class hierarchy. It is typeless -- i.e., instances do *not* have a `Type`. - `MDNode`'s operands are all `Metadata *` (instead of `Value *`). - `TrackingVH<MDNode>` and `WeakVH` referring to metadata can be replaced with `TrackingMDNodeRef` and `TrackingMDRef`, respectively. If you're referring solely to resolved `MDNode`s -- post graph construction -- just use `MDNode*`. - `MDNode` (and the rest of `Metadata`) have only limited support for `replaceAllUsesWith()`. As long as an `MDNode` is pointing at a forward declaration -- the result of `MDNode::getTemporary()` -- it maintains a side map of its uses and can RAUW itself. Once the forward declarations are fully resolved RAUW support is dropped on the ground. This means that uniquing collisions on changing operands cause nodes to become "distinct". (This already happened fairly commonly, whenever an operand went to null.) If you're constructing complex (non self-reference) `MDNode` cycles, you need to call `MDNode::resolveCycles()` on each node (or on a top-level node that somehow references all of the nodes). Also, don't do that. Metadata cycles (and the RAUW machinery needed to construct them) are expensive. - An `MDNode` can only refer to a `Constant` through a bridge called `ConstantAsMetadata` (one of the subclasses of `ValueAsMetadata`). As a side effect, accessing an operand of an `MDNode` that is known to be, e.g., `ConstantInt`, takes three steps: first, cast from `Metadata` to `ConstantAsMetadata`; second, extract the `Constant`; third, cast down to `ConstantInt`. The eventual goal is to introduce `MDInt`/`MDFloat`/etc. and have metadata schema owners transition away from using `Constant`s when the type isn't important (and they don't care about referring to `GlobalValue`s). In the meantime, I've added transitional API to the `mdconst` namespace that matches semantics with the old code, in order to avoid adding the error-prone three-step equivalent to every call site. If your old code was: MDNode *N = foo(); bar(isa <ConstantInt>(N->getOperand(0))); baz(cast <ConstantInt>(N->getOperand(1))); bak(cast_or_null <ConstantInt>(N->getOperand(2))); bat(dyn_cast <ConstantInt>(N->getOperand(3))); bay(dyn_cast_or_null<ConstantInt>(N->getOperand(4))); you can trivially match its semantics with: MDNode *N = foo(); bar(mdconst::hasa <ConstantInt>(N->getOperand(0))); baz(mdconst::extract <ConstantInt>(N->getOperand(1))); bak(mdconst::extract_or_null <ConstantInt>(N->getOperand(2))); bat(mdconst::dyn_extract <ConstantInt>(N->getOperand(3))); bay(mdconst::dyn_extract_or_null<ConstantInt>(N->getOperand(4))); and when you transition your metadata schema to `MDInt`: MDNode *N = foo(); bar(isa <MDInt>(N->getOperand(0))); baz(cast <MDInt>(N->getOperand(1))); bak(cast_or_null <MDInt>(N->getOperand(2))); bat(dyn_cast <MDInt>(N->getOperand(3))); bay(dyn_cast_or_null<MDInt>(N->getOperand(4))); - A `CallInst` -- specifically, intrinsic instructions -- can refer to metadata through a bridge called `MetadataAsValue`. This is a subclass of `Value` where `getType()->isMetadataTy()`. `MetadataAsValue` is the *only* class that can legally refer to a `LocalAsMetadata`, which is a bridged form of non-`Constant` values like `Argument` and `Instruction`. It can also refer to any other `Metadata` subclass. (I'll break all your testcases in a follow-up commit, when I propagate this change to assembly.) llvm-svn: 223802
2014-12-09 19:38:53 +01:00
}
void Verifier::visitValueAsMetadata(const ValueAsMetadata &MD, Function *F) {
Assert(MD.getValue(), "Expected valid value", &MD);
Assert(!MD.getValue()->getType()->isMetadataTy(),
"Unexpected metadata round-trip through values", &MD, MD.getValue());
IR: Split Metadata from Value Split `Metadata` away from the `Value` class hierarchy, as part of PR21532. Assembly and bitcode changes are in the wings, but this is the bulk of the change for the IR C++ API. I have a follow-up patch prepared for `clang`. If this breaks other sub-projects, I apologize in advance :(. Help me compile it on Darwin I'll try to fix it. FWIW, the errors should be easy to fix, so it may be simpler to just fix it yourself. This breaks the build for all metadata-related code that's out-of-tree. Rest assured the transition is mechanical and the compiler should catch almost all of the problems. Here's a quick guide for updating your code: - `Metadata` is the root of a class hierarchy with three main classes: `MDNode`, `MDString`, and `ValueAsMetadata`. It is distinct from the `Value` class hierarchy. It is typeless -- i.e., instances do *not* have a `Type`. - `MDNode`'s operands are all `Metadata *` (instead of `Value *`). - `TrackingVH<MDNode>` and `WeakVH` referring to metadata can be replaced with `TrackingMDNodeRef` and `TrackingMDRef`, respectively. If you're referring solely to resolved `MDNode`s -- post graph construction -- just use `MDNode*`. - `MDNode` (and the rest of `Metadata`) have only limited support for `replaceAllUsesWith()`. As long as an `MDNode` is pointing at a forward declaration -- the result of `MDNode::getTemporary()` -- it maintains a side map of its uses and can RAUW itself. Once the forward declarations are fully resolved RAUW support is dropped on the ground. This means that uniquing collisions on changing operands cause nodes to become "distinct". (This already happened fairly commonly, whenever an operand went to null.) If you're constructing complex (non self-reference) `MDNode` cycles, you need to call `MDNode::resolveCycles()` on each node (or on a top-level node that somehow references all of the nodes). Also, don't do that. Metadata cycles (and the RAUW machinery needed to construct them) are expensive. - An `MDNode` can only refer to a `Constant` through a bridge called `ConstantAsMetadata` (one of the subclasses of `ValueAsMetadata`). As a side effect, accessing an operand of an `MDNode` that is known to be, e.g., `ConstantInt`, takes three steps: first, cast from `Metadata` to `ConstantAsMetadata`; second, extract the `Constant`; third, cast down to `ConstantInt`. The eventual goal is to introduce `MDInt`/`MDFloat`/etc. and have metadata schema owners transition away from using `Constant`s when the type isn't important (and they don't care about referring to `GlobalValue`s). In the meantime, I've added transitional API to the `mdconst` namespace that matches semantics with the old code, in order to avoid adding the error-prone three-step equivalent to every call site. If your old code was: MDNode *N = foo(); bar(isa <ConstantInt>(N->getOperand(0))); baz(cast <ConstantInt>(N->getOperand(1))); bak(cast_or_null <ConstantInt>(N->getOperand(2))); bat(dyn_cast <ConstantInt>(N->getOperand(3))); bay(dyn_cast_or_null<ConstantInt>(N->getOperand(4))); you can trivially match its semantics with: MDNode *N = foo(); bar(mdconst::hasa <ConstantInt>(N->getOperand(0))); baz(mdconst::extract <ConstantInt>(N->getOperand(1))); bak(mdconst::extract_or_null <ConstantInt>(N->getOperand(2))); bat(mdconst::dyn_extract <ConstantInt>(N->getOperand(3))); bay(mdconst::dyn_extract_or_null<ConstantInt>(N->getOperand(4))); and when you transition your metadata schema to `MDInt`: MDNode *N = foo(); bar(isa <MDInt>(N->getOperand(0))); baz(cast <MDInt>(N->getOperand(1))); bak(cast_or_null <MDInt>(N->getOperand(2))); bat(dyn_cast <MDInt>(N->getOperand(3))); bay(dyn_cast_or_null<MDInt>(N->getOperand(4))); - A `CallInst` -- specifically, intrinsic instructions -- can refer to metadata through a bridge called `MetadataAsValue`. This is a subclass of `Value` where `getType()->isMetadataTy()`. `MetadataAsValue` is the *only* class that can legally refer to a `LocalAsMetadata`, which is a bridged form of non-`Constant` values like `Argument` and `Instruction`. It can also refer to any other `Metadata` subclass. (I'll break all your testcases in a follow-up commit, when I propagate this change to assembly.) llvm-svn: 223802
2014-12-09 19:38:53 +01:00
auto *L = dyn_cast<LocalAsMetadata>(&MD);
if (!L)
return;
Assert(F, "function-local metadata used outside a function", L);
IR: Split Metadata from Value Split `Metadata` away from the `Value` class hierarchy, as part of PR21532. Assembly and bitcode changes are in the wings, but this is the bulk of the change for the IR C++ API. I have a follow-up patch prepared for `clang`. If this breaks other sub-projects, I apologize in advance :(. Help me compile it on Darwin I'll try to fix it. FWIW, the errors should be easy to fix, so it may be simpler to just fix it yourself. This breaks the build for all metadata-related code that's out-of-tree. Rest assured the transition is mechanical and the compiler should catch almost all of the problems. Here's a quick guide for updating your code: - `Metadata` is the root of a class hierarchy with three main classes: `MDNode`, `MDString`, and `ValueAsMetadata`. It is distinct from the `Value` class hierarchy. It is typeless -- i.e., instances do *not* have a `Type`. - `MDNode`'s operands are all `Metadata *` (instead of `Value *`). - `TrackingVH<MDNode>` and `WeakVH` referring to metadata can be replaced with `TrackingMDNodeRef` and `TrackingMDRef`, respectively. If you're referring solely to resolved `MDNode`s -- post graph construction -- just use `MDNode*`. - `MDNode` (and the rest of `Metadata`) have only limited support for `replaceAllUsesWith()`. As long as an `MDNode` is pointing at a forward declaration -- the result of `MDNode::getTemporary()` -- it maintains a side map of its uses and can RAUW itself. Once the forward declarations are fully resolved RAUW support is dropped on the ground. This means that uniquing collisions on changing operands cause nodes to become "distinct". (This already happened fairly commonly, whenever an operand went to null.) If you're constructing complex (non self-reference) `MDNode` cycles, you need to call `MDNode::resolveCycles()` on each node (or on a top-level node that somehow references all of the nodes). Also, don't do that. Metadata cycles (and the RAUW machinery needed to construct them) are expensive. - An `MDNode` can only refer to a `Constant` through a bridge called `ConstantAsMetadata` (one of the subclasses of `ValueAsMetadata`). As a side effect, accessing an operand of an `MDNode` that is known to be, e.g., `ConstantInt`, takes three steps: first, cast from `Metadata` to `ConstantAsMetadata`; second, extract the `Constant`; third, cast down to `ConstantInt`. The eventual goal is to introduce `MDInt`/`MDFloat`/etc. and have metadata schema owners transition away from using `Constant`s when the type isn't important (and they don't care about referring to `GlobalValue`s). In the meantime, I've added transitional API to the `mdconst` namespace that matches semantics with the old code, in order to avoid adding the error-prone three-step equivalent to every call site. If your old code was: MDNode *N = foo(); bar(isa <ConstantInt>(N->getOperand(0))); baz(cast <ConstantInt>(N->getOperand(1))); bak(cast_or_null <ConstantInt>(N->getOperand(2))); bat(dyn_cast <ConstantInt>(N->getOperand(3))); bay(dyn_cast_or_null<ConstantInt>(N->getOperand(4))); you can trivially match its semantics with: MDNode *N = foo(); bar(mdconst::hasa <ConstantInt>(N->getOperand(0))); baz(mdconst::extract <ConstantInt>(N->getOperand(1))); bak(mdconst::extract_or_null <ConstantInt>(N->getOperand(2))); bat(mdconst::dyn_extract <ConstantInt>(N->getOperand(3))); bay(mdconst::dyn_extract_or_null<ConstantInt>(N->getOperand(4))); and when you transition your metadata schema to `MDInt`: MDNode *N = foo(); bar(isa <MDInt>(N->getOperand(0))); baz(cast <MDInt>(N->getOperand(1))); bak(cast_or_null <MDInt>(N->getOperand(2))); bat(dyn_cast <MDInt>(N->getOperand(3))); bay(dyn_cast_or_null<MDInt>(N->getOperand(4))); - A `CallInst` -- specifically, intrinsic instructions -- can refer to metadata through a bridge called `MetadataAsValue`. This is a subclass of `Value` where `getType()->isMetadataTy()`. `MetadataAsValue` is the *only* class that can legally refer to a `LocalAsMetadata`, which is a bridged form of non-`Constant` values like `Argument` and `Instruction`. It can also refer to any other `Metadata` subclass. (I'll break all your testcases in a follow-up commit, when I propagate this change to assembly.) llvm-svn: 223802
2014-12-09 19:38:53 +01:00
// If this was an instruction, bb, or argument, verify that it is in the
// function that we expect.
Function *ActualF = nullptr;
if (Instruction *I = dyn_cast<Instruction>(L->getValue())) {
Assert(I->getParent(), "function-local metadata not in basic block", L, I);
IR: Split Metadata from Value Split `Metadata` away from the `Value` class hierarchy, as part of PR21532. Assembly and bitcode changes are in the wings, but this is the bulk of the change for the IR C++ API. I have a follow-up patch prepared for `clang`. If this breaks other sub-projects, I apologize in advance :(. Help me compile it on Darwin I'll try to fix it. FWIW, the errors should be easy to fix, so it may be simpler to just fix it yourself. This breaks the build for all metadata-related code that's out-of-tree. Rest assured the transition is mechanical and the compiler should catch almost all of the problems. Here's a quick guide for updating your code: - `Metadata` is the root of a class hierarchy with three main classes: `MDNode`, `MDString`, and `ValueAsMetadata`. It is distinct from the `Value` class hierarchy. It is typeless -- i.e., instances do *not* have a `Type`. - `MDNode`'s operands are all `Metadata *` (instead of `Value *`). - `TrackingVH<MDNode>` and `WeakVH` referring to metadata can be replaced with `TrackingMDNodeRef` and `TrackingMDRef`, respectively. If you're referring solely to resolved `MDNode`s -- post graph construction -- just use `MDNode*`. - `MDNode` (and the rest of `Metadata`) have only limited support for `replaceAllUsesWith()`. As long as an `MDNode` is pointing at a forward declaration -- the result of `MDNode::getTemporary()` -- it maintains a side map of its uses and can RAUW itself. Once the forward declarations are fully resolved RAUW support is dropped on the ground. This means that uniquing collisions on changing operands cause nodes to become "distinct". (This already happened fairly commonly, whenever an operand went to null.) If you're constructing complex (non self-reference) `MDNode` cycles, you need to call `MDNode::resolveCycles()` on each node (or on a top-level node that somehow references all of the nodes). Also, don't do that. Metadata cycles (and the RAUW machinery needed to construct them) are expensive. - An `MDNode` can only refer to a `Constant` through a bridge called `ConstantAsMetadata` (one of the subclasses of `ValueAsMetadata`). As a side effect, accessing an operand of an `MDNode` that is known to be, e.g., `ConstantInt`, takes three steps: first, cast from `Metadata` to `ConstantAsMetadata`; second, extract the `Constant`; third, cast down to `ConstantInt`. The eventual goal is to introduce `MDInt`/`MDFloat`/etc. and have metadata schema owners transition away from using `Constant`s when the type isn't important (and they don't care about referring to `GlobalValue`s). In the meantime, I've added transitional API to the `mdconst` namespace that matches semantics with the old code, in order to avoid adding the error-prone three-step equivalent to every call site. If your old code was: MDNode *N = foo(); bar(isa <ConstantInt>(N->getOperand(0))); baz(cast <ConstantInt>(N->getOperand(1))); bak(cast_or_null <ConstantInt>(N->getOperand(2))); bat(dyn_cast <ConstantInt>(N->getOperand(3))); bay(dyn_cast_or_null<ConstantInt>(N->getOperand(4))); you can trivially match its semantics with: MDNode *N = foo(); bar(mdconst::hasa <ConstantInt>(N->getOperand(0))); baz(mdconst::extract <ConstantInt>(N->getOperand(1))); bak(mdconst::extract_or_null <ConstantInt>(N->getOperand(2))); bat(mdconst::dyn_extract <ConstantInt>(N->getOperand(3))); bay(mdconst::dyn_extract_or_null<ConstantInt>(N->getOperand(4))); and when you transition your metadata schema to `MDInt`: MDNode *N = foo(); bar(isa <MDInt>(N->getOperand(0))); baz(cast <MDInt>(N->getOperand(1))); bak(cast_or_null <MDInt>(N->getOperand(2))); bat(dyn_cast <MDInt>(N->getOperand(3))); bay(dyn_cast_or_null<MDInt>(N->getOperand(4))); - A `CallInst` -- specifically, intrinsic instructions -- can refer to metadata through a bridge called `MetadataAsValue`. This is a subclass of `Value` where `getType()->isMetadataTy()`. `MetadataAsValue` is the *only* class that can legally refer to a `LocalAsMetadata`, which is a bridged form of non-`Constant` values like `Argument` and `Instruction`. It can also refer to any other `Metadata` subclass. (I'll break all your testcases in a follow-up commit, when I propagate this change to assembly.) llvm-svn: 223802
2014-12-09 19:38:53 +01:00
ActualF = I->getParent()->getParent();
} else if (BasicBlock *BB = dyn_cast<BasicBlock>(L->getValue()))
ActualF = BB->getParent();
else if (Argument *A = dyn_cast<Argument>(L->getValue()))
ActualF = A->getParent();
assert(ActualF && "Unimplemented function local metadata case!");
Assert(ActualF == F, "function-local metadata used in wrong function", L);
IR: Split Metadata from Value Split `Metadata` away from the `Value` class hierarchy, as part of PR21532. Assembly and bitcode changes are in the wings, but this is the bulk of the change for the IR C++ API. I have a follow-up patch prepared for `clang`. If this breaks other sub-projects, I apologize in advance :(. Help me compile it on Darwin I'll try to fix it. FWIW, the errors should be easy to fix, so it may be simpler to just fix it yourself. This breaks the build for all metadata-related code that's out-of-tree. Rest assured the transition is mechanical and the compiler should catch almost all of the problems. Here's a quick guide for updating your code: - `Metadata` is the root of a class hierarchy with three main classes: `MDNode`, `MDString`, and `ValueAsMetadata`. It is distinct from the `Value` class hierarchy. It is typeless -- i.e., instances do *not* have a `Type`. - `MDNode`'s operands are all `Metadata *` (instead of `Value *`). - `TrackingVH<MDNode>` and `WeakVH` referring to metadata can be replaced with `TrackingMDNodeRef` and `TrackingMDRef`, respectively. If you're referring solely to resolved `MDNode`s -- post graph construction -- just use `MDNode*`. - `MDNode` (and the rest of `Metadata`) have only limited support for `replaceAllUsesWith()`. As long as an `MDNode` is pointing at a forward declaration -- the result of `MDNode::getTemporary()` -- it maintains a side map of its uses and can RAUW itself. Once the forward declarations are fully resolved RAUW support is dropped on the ground. This means that uniquing collisions on changing operands cause nodes to become "distinct". (This already happened fairly commonly, whenever an operand went to null.) If you're constructing complex (non self-reference) `MDNode` cycles, you need to call `MDNode::resolveCycles()` on each node (or on a top-level node that somehow references all of the nodes). Also, don't do that. Metadata cycles (and the RAUW machinery needed to construct them) are expensive. - An `MDNode` can only refer to a `Constant` through a bridge called `ConstantAsMetadata` (one of the subclasses of `ValueAsMetadata`). As a side effect, accessing an operand of an `MDNode` that is known to be, e.g., `ConstantInt`, takes three steps: first, cast from `Metadata` to `ConstantAsMetadata`; second, extract the `Constant`; third, cast down to `ConstantInt`. The eventual goal is to introduce `MDInt`/`MDFloat`/etc. and have metadata schema owners transition away from using `Constant`s when the type isn't important (and they don't care about referring to `GlobalValue`s). In the meantime, I've added transitional API to the `mdconst` namespace that matches semantics with the old code, in order to avoid adding the error-prone three-step equivalent to every call site. If your old code was: MDNode *N = foo(); bar(isa <ConstantInt>(N->getOperand(0))); baz(cast <ConstantInt>(N->getOperand(1))); bak(cast_or_null <ConstantInt>(N->getOperand(2))); bat(dyn_cast <ConstantInt>(N->getOperand(3))); bay(dyn_cast_or_null<ConstantInt>(N->getOperand(4))); you can trivially match its semantics with: MDNode *N = foo(); bar(mdconst::hasa <ConstantInt>(N->getOperand(0))); baz(mdconst::extract <ConstantInt>(N->getOperand(1))); bak(mdconst::extract_or_null <ConstantInt>(N->getOperand(2))); bat(mdconst::dyn_extract <ConstantInt>(N->getOperand(3))); bay(mdconst::dyn_extract_or_null<ConstantInt>(N->getOperand(4))); and when you transition your metadata schema to `MDInt`: MDNode *N = foo(); bar(isa <MDInt>(N->getOperand(0))); baz(cast <MDInt>(N->getOperand(1))); bak(cast_or_null <MDInt>(N->getOperand(2))); bat(dyn_cast <MDInt>(N->getOperand(3))); bay(dyn_cast_or_null<MDInt>(N->getOperand(4))); - A `CallInst` -- specifically, intrinsic instructions -- can refer to metadata through a bridge called `MetadataAsValue`. This is a subclass of `Value` where `getType()->isMetadataTy()`. `MetadataAsValue` is the *only* class that can legally refer to a `LocalAsMetadata`, which is a bridged form of non-`Constant` values like `Argument` and `Instruction`. It can also refer to any other `Metadata` subclass. (I'll break all your testcases in a follow-up commit, when I propagate this change to assembly.) llvm-svn: 223802
2014-12-09 19:38:53 +01:00
}
void Verifier::visitMetadataAsValue(const MetadataAsValue &MDV, Function *F) {
IR: Split Metadata from Value Split `Metadata` away from the `Value` class hierarchy, as part of PR21532. Assembly and bitcode changes are in the wings, but this is the bulk of the change for the IR C++ API. I have a follow-up patch prepared for `clang`. If this breaks other sub-projects, I apologize in advance :(. Help me compile it on Darwin I'll try to fix it. FWIW, the errors should be easy to fix, so it may be simpler to just fix it yourself. This breaks the build for all metadata-related code that's out-of-tree. Rest assured the transition is mechanical and the compiler should catch almost all of the problems. Here's a quick guide for updating your code: - `Metadata` is the root of a class hierarchy with three main classes: `MDNode`, `MDString`, and `ValueAsMetadata`. It is distinct from the `Value` class hierarchy. It is typeless -- i.e., instances do *not* have a `Type`. - `MDNode`'s operands are all `Metadata *` (instead of `Value *`). - `TrackingVH<MDNode>` and `WeakVH` referring to metadata can be replaced with `TrackingMDNodeRef` and `TrackingMDRef`, respectively. If you're referring solely to resolved `MDNode`s -- post graph construction -- just use `MDNode*`. - `MDNode` (and the rest of `Metadata`) have only limited support for `replaceAllUsesWith()`. As long as an `MDNode` is pointing at a forward declaration -- the result of `MDNode::getTemporary()` -- it maintains a side map of its uses and can RAUW itself. Once the forward declarations are fully resolved RAUW support is dropped on the ground. This means that uniquing collisions on changing operands cause nodes to become "distinct". (This already happened fairly commonly, whenever an operand went to null.) If you're constructing complex (non self-reference) `MDNode` cycles, you need to call `MDNode::resolveCycles()` on each node (or on a top-level node that somehow references all of the nodes). Also, don't do that. Metadata cycles (and the RAUW machinery needed to construct them) are expensive. - An `MDNode` can only refer to a `Constant` through a bridge called `ConstantAsMetadata` (one of the subclasses of `ValueAsMetadata`). As a side effect, accessing an operand of an `MDNode` that is known to be, e.g., `ConstantInt`, takes three steps: first, cast from `Metadata` to `ConstantAsMetadata`; second, extract the `Constant`; third, cast down to `ConstantInt`. The eventual goal is to introduce `MDInt`/`MDFloat`/etc. and have metadata schema owners transition away from using `Constant`s when the type isn't important (and they don't care about referring to `GlobalValue`s). In the meantime, I've added transitional API to the `mdconst` namespace that matches semantics with the old code, in order to avoid adding the error-prone three-step equivalent to every call site. If your old code was: MDNode *N = foo(); bar(isa <ConstantInt>(N->getOperand(0))); baz(cast <ConstantInt>(N->getOperand(1))); bak(cast_or_null <ConstantInt>(N->getOperand(2))); bat(dyn_cast <ConstantInt>(N->getOperand(3))); bay(dyn_cast_or_null<ConstantInt>(N->getOperand(4))); you can trivially match its semantics with: MDNode *N = foo(); bar(mdconst::hasa <ConstantInt>(N->getOperand(0))); baz(mdconst::extract <ConstantInt>(N->getOperand(1))); bak(mdconst::extract_or_null <ConstantInt>(N->getOperand(2))); bat(mdconst::dyn_extract <ConstantInt>(N->getOperand(3))); bay(mdconst::dyn_extract_or_null<ConstantInt>(N->getOperand(4))); and when you transition your metadata schema to `MDInt`: MDNode *N = foo(); bar(isa <MDInt>(N->getOperand(0))); baz(cast <MDInt>(N->getOperand(1))); bak(cast_or_null <MDInt>(N->getOperand(2))); bat(dyn_cast <MDInt>(N->getOperand(3))); bay(dyn_cast_or_null<MDInt>(N->getOperand(4))); - A `CallInst` -- specifically, intrinsic instructions -- can refer to metadata through a bridge called `MetadataAsValue`. This is a subclass of `Value` where `getType()->isMetadataTy()`. `MetadataAsValue` is the *only* class that can legally refer to a `LocalAsMetadata`, which is a bridged form of non-`Constant` values like `Argument` and `Instruction`. It can also refer to any other `Metadata` subclass. (I'll break all your testcases in a follow-up commit, when I propagate this change to assembly.) llvm-svn: 223802
2014-12-09 19:38:53 +01:00
Metadata *MD = MDV.getMetadata();
if (auto *N = dyn_cast<MDNode>(MD)) {
visitMDNode(*N);
return;
}
// Only visit each node once. Metadata can be mutually recursive, so this
// avoids infinite recursion here, as well as being an optimization.
if (!MDNodes.insert(MD).second)
return;
if (auto *V = dyn_cast<ValueAsMetadata>(MD))
visitValueAsMetadata(*V, F);
}
static bool isType(const Metadata *MD) { return !MD || isa<DIType>(MD); }
static bool isScope(const Metadata *MD) { return !MD || isa<DIScope>(MD); }
static bool isDINode(const Metadata *MD) { return !MD || isa<DINode>(MD); }
void Verifier::visitDILocation(const DILocation &N) {
AssertDI(N.getRawScope() && isa<DILocalScope>(N.getRawScope()),
"location requires a valid scope", &N, N.getRawScope());
if (auto *IA = N.getRawInlinedAt())
AssertDI(isa<DILocation>(IA), "inlined-at should be a location", &N, IA);
if (auto *SP = dyn_cast<DISubprogram>(N.getRawScope()))
AssertDI(SP->isDefinition(), "scope points into the type hierarchy", &N);
}
void Verifier::visitGenericDINode(const GenericDINode &N) {
AssertDI(N.getTag(), "invalid tag", &N);
}
void Verifier::visitDIScope(const DIScope &N) {
if (auto *F = N.getRawFile())
AssertDI(isa<DIFile>(F), "invalid file", &N, F);
}
void Verifier::visitDISubrange(const DISubrange &N) {
AssertDI(N.getTag() == dwarf::DW_TAG_subrange_type, "invalid tag", &N);
auto Count = N.getCount();
AssertDI(Count, "Count must either be a signed constant or a DIVariable",
&N);
AssertDI(!Count.is<ConstantInt*>() ||
Count.get<ConstantInt*>()->getSExtValue() >= -1,
"invalid subrange count", &N);
}
void Verifier::visitDIEnumerator(const DIEnumerator &N) {
AssertDI(N.getTag() == dwarf::DW_TAG_enumerator, "invalid tag", &N);
}
void Verifier::visitDIBasicType(const DIBasicType &N) {
AssertDI(N.getTag() == dwarf::DW_TAG_base_type ||
N.getTag() == dwarf::DW_TAG_unspecified_type,
"invalid tag", &N);
AssertDI(!(N.isBigEndian() && N.isLittleEndian()) ,
"has conflicting flags", &N);
}
void Verifier::visitDIDerivedType(const DIDerivedType &N) {
// Common scope checks.
visitDIScope(N);
AssertDI(N.getTag() == dwarf::DW_TAG_typedef ||
N.getTag() == dwarf::DW_TAG_pointer_type ||
N.getTag() == dwarf::DW_TAG_ptr_to_member_type ||
N.getTag() == dwarf::DW_TAG_reference_type ||
N.getTag() == dwarf::DW_TAG_rvalue_reference_type ||
N.getTag() == dwarf::DW_TAG_const_type ||
N.getTag() == dwarf::DW_TAG_volatile_type ||
N.getTag() == dwarf::DW_TAG_restrict_type ||
N.getTag() == dwarf::DW_TAG_atomic_type ||
N.getTag() == dwarf::DW_TAG_member ||
N.getTag() == dwarf::DW_TAG_inheritance ||
N.getTag() == dwarf::DW_TAG_friend,
"invalid tag", &N);
if (N.getTag() == dwarf::DW_TAG_ptr_to_member_type) {
AssertDI(isType(N.getRawExtraData()), "invalid pointer to member type", &N,
N.getRawExtraData());
}
AssertDI(isScope(N.getRawScope()), "invalid scope", &N, N.getRawScope());
AssertDI(isType(N.getRawBaseType()), "invalid base type", &N,
N.getRawBaseType());
if (N.getDWARFAddressSpace()) {
AssertDI(N.getTag() == dwarf::DW_TAG_pointer_type ||
N.getTag() == dwarf::DW_TAG_reference_type,
"DWARF address space only applies to pointer or reference types",
&N);
}
}
/// Detect mutually exclusive flags.
static bool hasConflictingReferenceFlags(unsigned Flags) {
return ((Flags & DINode::FlagLValueReference) &&
(Flags & DINode::FlagRValueReference)) ||
((Flags & DINode::FlagTypePassByValue) &&
(Flags & DINode::FlagTypePassByReference));
}
void Verifier::visitTemplateParams(const MDNode &N, const Metadata &RawParams) {
auto *Params = dyn_cast<MDTuple>(&RawParams);
AssertDI(Params, "invalid template params", &N, &RawParams);
for (Metadata *Op : Params->operands()) {
AssertDI(Op && isa<DITemplateParameter>(Op), "invalid template parameter",
&N, Params, Op);
}
}
void Verifier::visitDICompositeType(const DICompositeType &N) {
// Common scope checks.
visitDIScope(N);
AssertDI(N.getTag() == dwarf::DW_TAG_array_type ||
N.getTag() == dwarf::DW_TAG_structure_type ||
N.getTag() == dwarf::DW_TAG_union_type ||
N.getTag() == dwarf::DW_TAG_enumeration_type ||
N.getTag() == dwarf::DW_TAG_class_type ||
N.getTag() == dwarf::DW_TAG_variant_part,
"invalid tag", &N);
AssertDI(isScope(N.getRawScope()), "invalid scope", &N, N.getRawScope());
AssertDI(isType(N.getRawBaseType()), "invalid base type", &N,
N.getRawBaseType());
AssertDI(!N.getRawElements() || isa<MDTuple>(N.getRawElements()),
"invalid composite elements", &N, N.getRawElements());
AssertDI(isType(N.getRawVTableHolder()), "invalid vtable holder", &N,
N.getRawVTableHolder());
AssertDI(!hasConflictingReferenceFlags(N.getFlags()),
"invalid reference flags", &N);
if (N.isVector()) {
const DINodeArray Elements = N.getElements();
AssertDI(Elements.size() == 1 &&
Elements[0]->getTag() == dwarf::DW_TAG_subrange_type,
"invalid vector, expected one element of type subrange", &N);
}
if (auto *Params = N.getRawTemplateParams())
visitTemplateParams(N, *Params);
if (N.getTag() == dwarf::DW_TAG_class_type ||
N.getTag() == dwarf::DW_TAG_union_type) {
AssertDI(N.getFile() && !N.getFile()->getFilename().empty(),
"class/union requires a filename", &N, N.getFile());
}
if (auto *D = N.getRawDiscriminator()) {
AssertDI(isa<DIDerivedType>(D) && N.getTag() == dwarf::DW_TAG_variant_part,
"discriminator can only appear on variant part");
}
}
void Verifier::visitDISubroutineType(const DISubroutineType &N) {
AssertDI(N.getTag() == dwarf::DW_TAG_subroutine_type, "invalid tag", &N);
if (auto *Types = N.getRawTypeArray()) {
AssertDI(isa<MDTuple>(Types), "invalid composite elements", &N, Types);
for (Metadata *Ty : N.getTypeArray()->operands()) {
AssertDI(isType(Ty), "invalid subroutine type ref", &N, Types, Ty);
}
}
AssertDI(!hasConflictingReferenceFlags(N.getFlags()),
"invalid reference flags", &N);
}
void Verifier::visitDIFile(const DIFile &N) {
AssertDI(N.getTag() == dwarf::DW_TAG_file_type, "invalid tag", &N);
Optional<DIFile::ChecksumInfo<StringRef>> Checksum = N.getChecksum();
if (Checksum) {
AssertDI(Checksum->Kind <= DIFile::ChecksumKind::CSK_Last,
"invalid checksum kind", &N);
size_t Size;
switch (Checksum->Kind) {
case DIFile::CSK_MD5:
Size = 32;
break;
case DIFile::CSK_SHA1:
Size = 40;
break;
}
AssertDI(Checksum->Value.size() == Size, "invalid checksum length", &N);
AssertDI(Checksum->Value.find_if_not(llvm::isHexDigit) == StringRef::npos,
"invalid checksum", &N);
}
}
void Verifier::visitDICompileUnit(const DICompileUnit &N) {
AssertDI(N.isDistinct(), "compile units must be distinct", &N);
AssertDI(N.getTag() == dwarf::DW_TAG_compile_unit, "invalid tag", &N);
// Don't bother verifying the compilation directory or producer string
// as those could be empty.
AssertDI(N.getRawFile() && isa<DIFile>(N.getRawFile()), "invalid file", &N,
N.getRawFile());
AssertDI(!N.getFile()->getFilename().empty(), "invalid filename", &N,
N.getFile());
verifySourceDebugInfo(N, *N.getFile());
AssertDI((N.getEmissionKind() <= DICompileUnit::LastEmissionKind),
"invalid emission kind", &N);
if (auto *Array = N.getRawEnumTypes()) {
AssertDI(isa<MDTuple>(Array), "invalid enum list", &N, Array);
for (Metadata *Op : N.getEnumTypes()->operands()) {
auto *Enum = dyn_cast_or_null<DICompositeType>(Op);
AssertDI(Enum && Enum->getTag() == dwarf::DW_TAG_enumeration_type,
"invalid enum type", &N, N.getEnumTypes(), Op);
}
}
if (auto *Array = N.getRawRetainedTypes()) {
AssertDI(isa<MDTuple>(Array), "invalid retained type list", &N, Array);
for (Metadata *Op : N.getRetainedTypes()->operands()) {
AssertDI(Op && (isa<DIType>(Op) ||
(isa<DISubprogram>(Op) &&
!cast<DISubprogram>(Op)->isDefinition())),
"invalid retained type", &N, Op);
}
}
if (auto *Array = N.getRawGlobalVariables()) {
AssertDI(isa<MDTuple>(Array), "invalid global variable list", &N, Array);
for (Metadata *Op : N.getGlobalVariables()->operands()) {
AssertDI(Op && (isa<DIGlobalVariableExpression>(Op)),
"invalid global variable ref", &N, Op);
}
}
if (auto *Array = N.getRawImportedEntities()) {
AssertDI(isa<MDTuple>(Array), "invalid imported entity list", &N, Array);
for (Metadata *Op : N.getImportedEntities()->operands()) {
AssertDI(Op && isa<DIImportedEntity>(Op), "invalid imported entity ref",
&N, Op);
}
}
if (auto *Array = N.getRawMacros()) {
AssertDI(isa<MDTuple>(Array), "invalid macro list", &N, Array);
for (Metadata *Op : N.getMacros()->operands()) {
AssertDI(Op && isa<DIMacroNode>(Op), "invalid macro ref", &N, Op);
}
}
CUVisited.insert(&N);
}
void Verifier::visitDISubprogram(const DISubprogram &N) {
AssertDI(N.getTag() == dwarf::DW_TAG_subprogram, "invalid tag", &N);
AssertDI(isScope(N.getRawScope()), "invalid scope", &N, N.getRawScope());
if (auto *F = N.getRawFile())
AssertDI(isa<DIFile>(F), "invalid file", &N, F);
else
AssertDI(N.getLine() == 0, "line specified with no file", &N, N.getLine());
if (auto *T = N.getRawType())
AssertDI(isa<DISubroutineType>(T), "invalid subroutine type", &N, T);
AssertDI(isType(N.getRawContainingType()), "invalid containing type", &N,
N.getRawContainingType());
if (auto *Params = N.getRawTemplateParams())
visitTemplateParams(N, *Params);
if (auto *S = N.getRawDeclaration())
AssertDI(isa<DISubprogram>(S) && !cast<DISubprogram>(S)->isDefinition(),
"invalid subprogram declaration", &N, S);
if (auto *RawNode = N.getRawRetainedNodes()) {
auto *Node = dyn_cast<MDTuple>(RawNode);
AssertDI(Node, "invalid retained nodes list", &N, RawNode);
for (Metadata *Op : Node->operands()) {
AssertDI(Op && (isa<DILocalVariable>(Op) || isa<DILabel>(Op)),
"invalid retained nodes, expected DILocalVariable or DILabel",
&N, Node, Op);
}
}
AssertDI(!hasConflictingReferenceFlags(N.getFlags()),
"invalid reference flags", &N);
auto *Unit = N.getRawUnit();
if (N.isDefinition()) {
// Subprogram definitions (not part of the type hierarchy).
AssertDI(N.isDistinct(), "subprogram definitions must be distinct", &N);
AssertDI(Unit, "subprogram definitions must have a compile unit", &N);
AssertDI(isa<DICompileUnit>(Unit), "invalid unit type", &N, Unit);
if (N.getFile())
verifySourceDebugInfo(*N.getUnit(), *N.getFile());
} else {
// Subprogram declarations (part of the type hierarchy).
AssertDI(!Unit, "subprogram declarations must not have a compile unit", &N);
}
if (auto *RawThrownTypes = N.getRawThrownTypes()) {
auto *ThrownTypes = dyn_cast<MDTuple>(RawThrownTypes);
AssertDI(ThrownTypes, "invalid thrown types list", &N, RawThrownTypes);
for (Metadata *Op : ThrownTypes->operands())
AssertDI(Op && isa<DIType>(Op), "invalid thrown type", &N, ThrownTypes,
Op);
}
if (N.areAllCallsDescribed())
AssertDI(N.isDefinition(),
"DIFlagAllCallsDescribed must be attached to a definition");
}
void Verifier::visitDILexicalBlockBase(const DILexicalBlockBase &N) {
AssertDI(N.getTag() == dwarf::DW_TAG_lexical_block, "invalid tag", &N);
AssertDI(N.getRawScope() && isa<DILocalScope>(N.getRawScope()),
"invalid local scope", &N, N.getRawScope());
if (auto *SP = dyn_cast<DISubprogram>(N.getRawScope()))
AssertDI(SP->isDefinition(), "scope points into the type hierarchy", &N);
}
void Verifier::visitDILexicalBlock(const DILexicalBlock &N) {
visitDILexicalBlockBase(N);
AssertDI(N.getLine() || !N.getColumn(),
"cannot have column info without line info", &N);
}
void Verifier::visitDILexicalBlockFile(const DILexicalBlockFile &N) {
visitDILexicalBlockBase(N);
}
void Verifier::visitDINamespace(const DINamespace &N) {
AssertDI(N.getTag() == dwarf::DW_TAG_namespace, "invalid tag", &N);
if (auto *S = N.getRawScope())
AssertDI(isa<DIScope>(S), "invalid scope ref", &N, S);
}
void Verifier::visitDIMacro(const DIMacro &N) {
AssertDI(N.getMacinfoType() == dwarf::DW_MACINFO_define ||
N.getMacinfoType() == dwarf::DW_MACINFO_undef,
"invalid macinfo type", &N);
AssertDI(!N.getName().empty(), "anonymous macro", &N);
if (!N.getValue().empty()) {
assert(N.getValue().data()[0] != ' ' && "Macro value has a space prefix");
}
}
void Verifier::visitDIMacroFile(const DIMacroFile &N) {
AssertDI(N.getMacinfoType() == dwarf::DW_MACINFO_start_file,
"invalid macinfo type", &N);
if (auto *F = N.getRawFile())
AssertDI(isa<DIFile>(F), "invalid file", &N, F);
if (auto *Array = N.getRawElements()) {
AssertDI(isa<MDTuple>(Array), "invalid macro list", &N, Array);
for (Metadata *Op : N.getElements()->operands()) {
AssertDI(Op && isa<DIMacroNode>(Op), "invalid macro ref", &N, Op);
}
}
}
void Verifier::visitDIModule(const DIModule &N) {
AssertDI(N.getTag() == dwarf::DW_TAG_module, "invalid tag", &N);
AssertDI(!N.getName().empty(), "anonymous module", &N);
}
void Verifier::visitDITemplateParameter(const DITemplateParameter &N) {
AssertDI(isType(N.getRawType()), "invalid type ref", &N, N.getRawType());
}
void Verifier::visitDITemplateTypeParameter(const DITemplateTypeParameter &N) {
visitDITemplateParameter(N);
AssertDI(N.getTag() == dwarf::DW_TAG_template_type_parameter, "invalid tag",
&N);
}
void Verifier::visitDITemplateValueParameter(
const DITemplateValueParameter &N) {
visitDITemplateParameter(N);
AssertDI(N.getTag() == dwarf::DW_TAG_template_value_parameter ||
N.getTag() == dwarf::DW_TAG_GNU_template_template_param ||
N.getTag() == dwarf::DW_TAG_GNU_template_parameter_pack,
"invalid tag", &N);
}
void Verifier::visitDIVariable(const DIVariable &N) {
if (auto *S = N.getRawScope())
AssertDI(isa<DIScope>(S), "invalid scope", &N, S);
if (auto *F = N.getRawFile())
AssertDI(isa<DIFile>(F), "invalid file", &N, F);
}
void Verifier::visitDIGlobalVariable(const DIGlobalVariable &N) {
// Checks common to all variables.
visitDIVariable(N);
AssertDI(N.getTag() == dwarf::DW_TAG_variable, "invalid tag", &N);
AssertDI(!N.getName().empty(), "missing global variable name", &N);
AssertDI(isType(N.getRawType()), "invalid type ref", &N, N.getRawType());
AssertDI(N.getType(), "missing global variable type", &N);
if (auto *Member = N.getRawStaticDataMemberDeclaration()) {
AssertDI(isa<DIDerivedType>(Member),
"invalid static data member declaration", &N, Member);
}
}
void Verifier::visitDILocalVariable(const DILocalVariable &N) {
// Checks common to all variables.
visitDIVariable(N);
AssertDI(isType(N.getRawType()), "invalid type ref", &N, N.getRawType());
AssertDI(N.getTag() == dwarf::DW_TAG_variable, "invalid tag", &N);
AssertDI(N.getRawScope() && isa<DILocalScope>(N.getRawScope()),
"local variable requires a valid scope", &N, N.getRawScope());
}
void Verifier::visitDILabel(const DILabel &N) {
if (auto *S = N.getRawScope())
AssertDI(isa<DIScope>(S), "invalid scope", &N, S);
if (auto *F = N.getRawFile())
AssertDI(isa<DIFile>(F), "invalid file", &N, F);
AssertDI(N.getTag() == dwarf::DW_TAG_label, "invalid tag", &N);
AssertDI(N.getRawScope() && isa<DILocalScope>(N.getRawScope()),
"label requires a valid scope", &N, N.getRawScope());
}
void Verifier::visitDIExpression(const DIExpression &N) {
AssertDI(N.isValid(), "invalid expression", &N);
}
void Verifier::visitDIGlobalVariableExpression(
const DIGlobalVariableExpression &GVE) {
AssertDI(GVE.getVariable(), "missing variable");
if (auto *Var = GVE.getVariable())
visitDIGlobalVariable(*Var);
if (auto *Expr = GVE.getExpression()) {
visitDIExpression(*Expr);
if (auto Fragment = Expr->getFragmentInfo())
verifyFragmentExpression(*GVE.getVariable(), *Fragment, &GVE);
}
}
void Verifier::visitDIObjCProperty(const DIObjCProperty &N) {
AssertDI(N.getTag() == dwarf::DW_TAG_APPLE_property, "invalid tag", &N);
if (auto *T = N.getRawType())
AssertDI(isType(T), "invalid type ref", &N, T);
if (auto *F = N.getRawFile())
AssertDI(isa<DIFile>(F), "invalid file", &N, F);
}
void Verifier::visitDIImportedEntity(const DIImportedEntity &N) {
AssertDI(N.getTag() == dwarf::DW_TAG_imported_module ||
N.getTag() == dwarf::DW_TAG_imported_declaration,
"invalid tag", &N);
if (auto *S = N.getRawScope())
AssertDI(isa<DIScope>(S), "invalid scope for imported entity", &N, S);
AssertDI(isDINode(N.getRawEntity()), "invalid imported entity", &N,
N.getRawEntity());
}
void Verifier::visitComdat(const Comdat &C) {
// The Module is invalid if the GlobalValue has private linkage. Entities
// with private linkage don't have entries in the symbol table.
if (const GlobalValue *GV = M.getNamedValue(C.getName()))
Assert(!GV->hasPrivateLinkage(), "comdat global value has private linkage",
GV);
}
void Verifier::visitModuleIdents(const Module &M) {
const NamedMDNode *Idents = M.getNamedMetadata("llvm.ident");
if (!Idents)
return;
// llvm.ident takes a list of metadata entry. Each entry has only one string.
// Scan each llvm.ident entry and make sure that this requirement is met.
2016-02-25 17:44:27 +01:00
for (const MDNode *N : Idents->operands()) {
Assert(N->getNumOperands() == 1,
"incorrect number of operands in llvm.ident metadata", N);
Assert(dyn_cast_or_null<MDString>(N->getOperand(0)),
("invalid value for llvm.ident metadata entry operand"
"(the operand should be a string)"),
N->getOperand(0));
}
}
void Verifier::visitModuleCommandLines(const Module &M) {
const NamedMDNode *CommandLines = M.getNamedMetadata("llvm.commandline");
if (!CommandLines)
return;
// llvm.commandline takes a list of metadata entry. Each entry has only one
// string. Scan each llvm.commandline entry and make sure that this
// requirement is met.
for (const MDNode *N : CommandLines->operands()) {
Assert(N->getNumOperands() == 1,
"incorrect number of operands in llvm.commandline metadata", N);
Assert(dyn_cast_or_null<MDString>(N->getOperand(0)),
("invalid value for llvm.commandline metadata entry operand"
"(the operand should be a string)"),
N->getOperand(0));
}
}
void Verifier::visitModuleFlags(const Module &M) {
const NamedMDNode *Flags = M.getModuleFlagsMetadata();
if (!Flags) return;
// Scan each flag, and track the flags and requirements.
DenseMap<const MDString*, const MDNode*> SeenIDs;
SmallVector<const MDNode*, 16> Requirements;
2016-02-25 17:44:27 +01:00
for (const MDNode *MDN : Flags->operands())
visitModuleFlag(MDN, SeenIDs, Requirements);
// Validate that the requirements in the module are valid.
2016-02-25 17:44:27 +01:00
for (const MDNode *Requirement : Requirements) {
const MDString *Flag = cast<MDString>(Requirement->getOperand(0));
IR: Split Metadata from Value Split `Metadata` away from the `Value` class hierarchy, as part of PR21532. Assembly and bitcode changes are in the wings, but this is the bulk of the change for the IR C++ API. I have a follow-up patch prepared for `clang`. If this breaks other sub-projects, I apologize in advance :(. Help me compile it on Darwin I'll try to fix it. FWIW, the errors should be easy to fix, so it may be simpler to just fix it yourself. This breaks the build for all metadata-related code that's out-of-tree. Rest assured the transition is mechanical and the compiler should catch almost all of the problems. Here's a quick guide for updating your code: - `Metadata` is the root of a class hierarchy with three main classes: `MDNode`, `MDString`, and `ValueAsMetadata`. It is distinct from the `Value` class hierarchy. It is typeless -- i.e., instances do *not* have a `Type`. - `MDNode`'s operands are all `Metadata *` (instead of `Value *`). - `TrackingVH<MDNode>` and `WeakVH` referring to metadata can be replaced with `TrackingMDNodeRef` and `TrackingMDRef`, respectively. If you're referring solely to resolved `MDNode`s -- post graph construction -- just use `MDNode*`. - `MDNode` (and the rest of `Metadata`) have only limited support for `replaceAllUsesWith()`. As long as an `MDNode` is pointing at a forward declaration -- the result of `MDNode::getTemporary()` -- it maintains a side map of its uses and can RAUW itself. Once the forward declarations are fully resolved RAUW support is dropped on the ground. This means that uniquing collisions on changing operands cause nodes to become "distinct". (This already happened fairly commonly, whenever an operand went to null.) If you're constructing complex (non self-reference) `MDNode` cycles, you need to call `MDNode::resolveCycles()` on each node (or on a top-level node that somehow references all of the nodes). Also, don't do that. Metadata cycles (and the RAUW machinery needed to construct them) are expensive. - An `MDNode` can only refer to a `Constant` through a bridge called `ConstantAsMetadata` (one of the subclasses of `ValueAsMetadata`). As a side effect, accessing an operand of an `MDNode` that is known to be, e.g., `ConstantInt`, takes three steps: first, cast from `Metadata` to `ConstantAsMetadata`; second, extract the `Constant`; third, cast down to `ConstantInt`. The eventual goal is to introduce `MDInt`/`MDFloat`/etc. and have metadata schema owners transition away from using `Constant`s when the type isn't important (and they don't care about referring to `GlobalValue`s). In the meantime, I've added transitional API to the `mdconst` namespace that matches semantics with the old code, in order to avoid adding the error-prone three-step equivalent to every call site. If your old code was: MDNode *N = foo(); bar(isa <ConstantInt>(N->getOperand(0))); baz(cast <ConstantInt>(N->getOperand(1))); bak(cast_or_null <ConstantInt>(N->getOperand(2))); bat(dyn_cast <ConstantInt>(N->getOperand(3))); bay(dyn_cast_or_null<ConstantInt>(N->getOperand(4))); you can trivially match its semantics with: MDNode *N = foo(); bar(mdconst::hasa <ConstantInt>(N->getOperand(0))); baz(mdconst::extract <ConstantInt>(N->getOperand(1))); bak(mdconst::extract_or_null <ConstantInt>(N->getOperand(2))); bat(mdconst::dyn_extract <ConstantInt>(N->getOperand(3))); bay(mdconst::dyn_extract_or_null<ConstantInt>(N->getOperand(4))); and when you transition your metadata schema to `MDInt`: MDNode *N = foo(); bar(isa <MDInt>(N->getOperand(0))); baz(cast <MDInt>(N->getOperand(1))); bak(cast_or_null <MDInt>(N->getOperand(2))); bat(dyn_cast <MDInt>(N->getOperand(3))); bay(dyn_cast_or_null<MDInt>(N->getOperand(4))); - A `CallInst` -- specifically, intrinsic instructions -- can refer to metadata through a bridge called `MetadataAsValue`. This is a subclass of `Value` where `getType()->isMetadataTy()`. `MetadataAsValue` is the *only* class that can legally refer to a `LocalAsMetadata`, which is a bridged form of non-`Constant` values like `Argument` and `Instruction`. It can also refer to any other `Metadata` subclass. (I'll break all your testcases in a follow-up commit, when I propagate this change to assembly.) llvm-svn: 223802
2014-12-09 19:38:53 +01:00
const Metadata *ReqValue = Requirement->getOperand(1);
const MDNode *Op = SeenIDs.lookup(Flag);
if (!Op) {
CheckFailed("invalid requirement on flag, flag is not present in module",
Flag);
continue;
}
if (Op->getOperand(2) != ReqValue) {
CheckFailed(("invalid requirement on flag, "
"flag does not have the required value"),
Flag);
continue;
}
}
}
void
Verifier::visitModuleFlag(const MDNode *Op,
DenseMap<const MDString *, const MDNode *> &SeenIDs,
SmallVectorImpl<const MDNode *> &Requirements) {
// Each module flag should have three arguments, the merge behavior (a
// constant int), the flag ID (an MDString), and the value.
Assert(Op->getNumOperands() == 3,
"incorrect number of operands in module flag", Op);
Module::ModFlagBehavior MFB;
if (!Module::isValidModFlagBehavior(Op->getOperand(0), MFB)) {
Assert(
mdconst::dyn_extract_or_null<ConstantInt>(Op->getOperand(0)),
"invalid behavior operand in module flag (expected constant integer)",
Op->getOperand(0));
Assert(false,
"invalid behavior operand in module flag (unexpected constant)",
Op->getOperand(0));
}
MDString *ID = dyn_cast_or_null<MDString>(Op->getOperand(1));
Assert(ID, "invalid ID operand in module flag (expected metadata string)",
Op->getOperand(1));
// Sanity check the values for behaviors with additional requirements.
switch (MFB) {
case Module::Error:
case Module::Warning:
case Module::Override:
// These behavior types accept any value.
break;
case Module::Max: {
Assert(mdconst::dyn_extract_or_null<ConstantInt>(Op->getOperand(2)),
"invalid value for 'max' module flag (expected constant integer)",
Op->getOperand(2));
break;
}
case Module::Require: {
// The value should itself be an MDNode with two operands, a flag ID (an
// MDString), and a value.
MDNode *Value = dyn_cast<MDNode>(Op->getOperand(2));
Assert(Value && Value->getNumOperands() == 2,
"invalid value for 'require' module flag (expected metadata pair)",
Op->getOperand(2));
Assert(isa<MDString>(Value->getOperand(0)),
("invalid value for 'require' module flag "
"(first value operand should be a string)"),
Value->getOperand(0));
// Append it to the list of requirements, to check once all module flags are
// scanned.
Requirements.push_back(Value);
break;
}
case Module::Append:
case Module::AppendUnique: {
// These behavior types require the operand be an MDNode.
Assert(isa<MDNode>(Op->getOperand(2)),
"invalid value for 'append'-type module flag "
"(expected a metadata node)",
Op->getOperand(2));
break;
}
}
// Unless this is a "requires" flag, check the ID is unique.
if (MFB != Module::Require) {
bool Inserted = SeenIDs.insert(std::make_pair(ID, Op)).second;
Assert(Inserted,
"module flag identifiers must be unique (or of 'require' type)", ID);
}
if (ID->getString() == "wchar_size") {
ConstantInt *Value
= mdconst::dyn_extract_or_null<ConstantInt>(Op->getOperand(2));
Assert(Value, "wchar_size metadata requires constant integer argument");
}
if (ID->getString() == "Linker Options") {
// If the llvm.linker.options named metadata exists, we assume that the
// bitcode reader has upgraded the module flag. Otherwise the flag might
// have been created by a client directly.
Assert(M.getNamedMetadata("llvm.linker.options"),
"'Linker Options' named metadata no longer supported");
}
if (ID->getString() == "CG Profile") {
for (const MDOperand &MDO : cast<MDNode>(Op->getOperand(2))->operands())
visitModuleFlagCGProfileEntry(MDO);
}
}
void Verifier::visitModuleFlagCGProfileEntry(const MDOperand &MDO) {
auto CheckFunction = [&](const MDOperand &FuncMDO) {
if (!FuncMDO)
return;
auto F = dyn_cast<ValueAsMetadata>(FuncMDO);
Assert(F && isa<Function>(F->getValue()), "expected a Function or null",
FuncMDO);
};
auto Node = dyn_cast_or_null<MDNode>(MDO);
Assert(Node && Node->getNumOperands() == 3, "expected a MDNode triple", MDO);
CheckFunction(Node->getOperand(0));
CheckFunction(Node->getOperand(1));
auto Count = dyn_cast_or_null<ConstantAsMetadata>(Node->getOperand(2));
Assert(Count && Count->getType()->isIntegerTy(),
"expected an integer constant", Node->getOperand(2));
}
/// Return true if this attribute kind only applies to functions.
static bool isFuncOnlyAttr(Attribute::AttrKind Kind) {
switch (Kind) {
case Attribute::NoReturn:
case Attribute::NoCfCheck:
case Attribute::NoUnwind:
case Attribute::NoInline:
case Attribute::AlwaysInline:
case Attribute::OptimizeForSize:
case Attribute::StackProtect:
case Attribute::StackProtectReq:
case Attribute::StackProtectStrong:
case Attribute::SafeStack:
case Attribute::ShadowCallStack:
case Attribute::NoRedZone:
case Attribute::NoImplicitFloat:
case Attribute::Naked:
case Attribute::InlineHint:
case Attribute::StackAlignment:
case Attribute::UWTable:
case Attribute::NonLazyBind:
case Attribute::ReturnsTwice:
case Attribute::SanitizeAddress:
case Attribute::SanitizeHWAddress:
case Attribute::SanitizeThread:
case Attribute::SanitizeMemory:
case Attribute::MinSize:
case Attribute::NoDuplicate:
case Attribute::Builtin:
case Attribute::NoBuiltin:
case Attribute::Cold:
case Attribute::OptForFuzzing:
case Attribute::OptimizeNone:
case Attribute::JumpTable:
case Attribute::Convergent:
case Attribute::ArgMemOnly:
case Attribute::NoRecurse:
case Attribute::InaccessibleMemOnly:
case Attribute::InaccessibleMemOrArgMemOnly:
case Attribute::AllocSize:
case Attribute::SpeculativeLoadHardening:
case Attribute::Speculatable:
case Attribute::StrictFP:
return true;
default:
break;
}
return false;
}
/// Return true if this is a function attribute that can also appear on
/// arguments.
static bool isFuncOrArgAttr(Attribute::AttrKind Kind) {
return Kind == Attribute::ReadOnly || Kind == Attribute::WriteOnly ||
Kind == Attribute::ReadNone;
}
void Verifier::verifyAttributeTypes(AttributeSet Attrs, bool IsFunction,
const Value *V) {
for (Attribute A : Attrs) {
if (A.isStringAttribute())
continue;
if (isFuncOnlyAttr(A.getKindAsEnum())) {
if (!IsFunction) {
CheckFailed("Attribute '" + A.getAsString() +
"' only applies to functions!",
V);
return;
}
} else if (IsFunction && !isFuncOrArgAttr(A.getKindAsEnum())) {
CheckFailed("Attribute '" + A.getAsString() +
"' does not apply to functions!",
V);
return;
}
}
}
// VerifyParameterAttrs - Check the given attributes for an argument or return
// value of the specified type. The value V is printed in error messages.
void Verifier::verifyParameterAttrs(AttributeSet Attrs, Type *Ty,
const Value *V) {
if (!Attrs.hasAttributes())
return;
verifyAttributeTypes(Attrs, /*IsFunction=*/false, V);
// Check for mutually incompatible attributes. Only inreg is compatible with
// sret.
unsigned AttrCount = 0;
AttrCount += Attrs.hasAttribute(Attribute::ByVal);
AttrCount += Attrs.hasAttribute(Attribute::InAlloca);
AttrCount += Attrs.hasAttribute(Attribute::StructRet) ||
Attrs.hasAttribute(Attribute::InReg);
AttrCount += Attrs.hasAttribute(Attribute::Nest);
Assert(AttrCount <= 1, "Attributes 'byval', 'inalloca', 'inreg', 'nest', "
"and 'sret' are incompatible!",
V);
Assert(!(Attrs.hasAttribute(Attribute::InAlloca) &&
Attrs.hasAttribute(Attribute::ReadOnly)),
"Attributes "
"'inalloca and readonly' are incompatible!",
V);
Assert(!(Attrs.hasAttribute(Attribute::StructRet) &&
Attrs.hasAttribute(Attribute::Returned)),
"Attributes "
"'sret and returned' are incompatible!",
V);
Assert(!(Attrs.hasAttribute(Attribute::ZExt) &&
Attrs.hasAttribute(Attribute::SExt)),
"Attributes "
"'zeroext and signext' are incompatible!",
V);
Assert(!(Attrs.hasAttribute(Attribute::ReadNone) &&
Attrs.hasAttribute(Attribute::ReadOnly)),
"Attributes "
"'readnone and readonly' are incompatible!",
V);
Assert(!(Attrs.hasAttribute(Attribute::ReadNone) &&
Attrs.hasAttribute(Attribute::WriteOnly)),
"Attributes "
"'readnone and writeonly' are incompatible!",
V);
Assert(!(Attrs.hasAttribute(Attribute::ReadOnly) &&
Attrs.hasAttribute(Attribute::WriteOnly)),
"Attributes "
"'readonly and writeonly' are incompatible!",
V);
Assert(!(Attrs.hasAttribute(Attribute::NoInline) &&
Attrs.hasAttribute(Attribute::AlwaysInline)),
"Attributes "
"'noinline and alwaysinline' are incompatible!",
V);
AttrBuilder IncompatibleAttrs = AttributeFuncs::typeIncompatible(Ty);
Assert(!AttrBuilder(Attrs).overlaps(IncompatibleAttrs),
"Wrong types for attribute: " +
AttributeSet::get(Context, IncompatibleAttrs).getAsString(),
V);
if (PointerType *PTy = dyn_cast<PointerType>(Ty)) {
SmallPtrSet<Type*, 4> Visited;
if (!PTy->getElementType()->isSized(&Visited)) {
Assert(!Attrs.hasAttribute(Attribute::ByVal) &&
!Attrs.hasAttribute(Attribute::InAlloca),
"Attributes 'byval' and 'inalloca' do not support unsized types!",
V);
}
if (!isa<PointerType>(PTy->getElementType()))
Assert(!Attrs.hasAttribute(Attribute::SwiftError),
"Attribute 'swifterror' only applies to parameters "
"with pointer to pointer type!",
V);
} else {
Assert(!Attrs.hasAttribute(Attribute::ByVal),
"Attribute 'byval' only applies to parameters with pointer type!",
V);
Assert(!Attrs.hasAttribute(Attribute::SwiftError),
"Attribute 'swifterror' only applies to parameters "
"with pointer type!",
V);
}
}
// Check parameter attributes against a function type.
// The value V is printed in error messages.
void Verifier::verifyFunctionAttrs(FunctionType *FT, AttributeList Attrs,
const Value *V) {
if (Attrs.isEmpty())
return;
bool SawNest = false;
bool SawReturned = false;
bool SawSRet = false;
bool SawSwiftSelf = false;
bool SawSwiftError = false;
// Verify return value attributes.
AttributeSet RetAttrs = Attrs.getRetAttributes();
Assert((!RetAttrs.hasAttribute(Attribute::ByVal) &&
!RetAttrs.hasAttribute(Attribute::Nest) &&
!RetAttrs.hasAttribute(Attribute::StructRet) &&
!RetAttrs.hasAttribute(Attribute::NoCapture) &&
!RetAttrs.hasAttribute(Attribute::Returned) &&
!RetAttrs.hasAttribute(Attribute::InAlloca) &&
!RetAttrs.hasAttribute(Attribute::SwiftSelf) &&
!RetAttrs.hasAttribute(Attribute::SwiftError)),
"Attributes 'byval', 'inalloca', 'nest', 'sret', 'nocapture', "
"'returned', 'swiftself', and 'swifterror' do not apply to return "
"values!",
V);
Assert((!RetAttrs.hasAttribute(Attribute::ReadOnly) &&
!RetAttrs.hasAttribute(Attribute::WriteOnly) &&
!RetAttrs.hasAttribute(Attribute::ReadNone)),
"Attribute '" + RetAttrs.getAsString() +
"' does not apply to function returns",
V);
verifyParameterAttrs(RetAttrs, FT->getReturnType(), V);
// Verify parameter attributes.
for (unsigned i = 0, e = FT->getNumParams(); i != e; ++i) {
Type *Ty = FT->getParamType(i);
AttributeSet ArgAttrs = Attrs.getParamAttributes(i);
verifyParameterAttrs(ArgAttrs, Ty, V);
if (ArgAttrs.hasAttribute(Attribute::Nest)) {
Assert(!SawNest, "More than one parameter has attribute nest!", V);
SawNest = true;
}
if (ArgAttrs.hasAttribute(Attribute::Returned)) {
Assert(!SawReturned, "More than one parameter has attribute returned!",
V);
Assert(Ty->canLosslesslyBitCastTo(FT->getReturnType()),
"Incompatible argument and return types for 'returned' attribute",
V);
SawReturned = true;
}
if (ArgAttrs.hasAttribute(Attribute::StructRet)) {
Assert(!SawSRet, "Cannot have multiple 'sret' parameters!", V);
Assert(i == 0 || i == 1,
"Attribute 'sret' is not on first or second parameter!", V);
SawSRet = true;
}
if (ArgAttrs.hasAttribute(Attribute::SwiftSelf)) {
Assert(!SawSwiftSelf, "Cannot have multiple 'swiftself' parameters!", V);
SawSwiftSelf = true;
}
if (ArgAttrs.hasAttribute(Attribute::SwiftError)) {
Assert(!SawSwiftError, "Cannot have multiple 'swifterror' parameters!",
V);
SawSwiftError = true;
}
if (ArgAttrs.hasAttribute(Attribute::InAlloca)) {
Assert(i == FT->getNumParams() - 1,
"inalloca isn't on the last parameter!", V);
}
}
if (!Attrs.hasAttributes(AttributeList::FunctionIndex))
return;
verifyAttributeTypes(Attrs.getFnAttributes(), /*IsFunction=*/true, V);
Assert(!(Attrs.hasFnAttribute(Attribute::ReadNone) &&
Attrs.hasFnAttribute(Attribute::ReadOnly)),
"Attributes 'readnone and readonly' are incompatible!", V);
Assert(!(Attrs.hasFnAttribute(Attribute::ReadNone) &&
Attrs.hasFnAttribute(Attribute::WriteOnly)),
"Attributes 'readnone and writeonly' are incompatible!", V);
Assert(!(Attrs.hasFnAttribute(Attribute::ReadOnly) &&
Attrs.hasFnAttribute(Attribute::WriteOnly)),
"Attributes 'readonly and writeonly' are incompatible!", V);
Assert(!(Attrs.hasFnAttribute(Attribute::ReadNone) &&
Attrs.hasFnAttribute(Attribute::InaccessibleMemOrArgMemOnly)),
"Attributes 'readnone and inaccessiblemem_or_argmemonly' are "
"incompatible!",
V);
Assert(!(Attrs.hasFnAttribute(Attribute::ReadNone) &&
Attrs.hasFnAttribute(Attribute::InaccessibleMemOnly)),
"Attributes 'readnone and inaccessiblememonly' are incompatible!", V);
Assert(!(Attrs.hasFnAttribute(Attribute::NoInline) &&
Attrs.hasFnAttribute(Attribute::AlwaysInline)),
"Attributes 'noinline and alwaysinline' are incompatible!", V);
if (Attrs.hasFnAttribute(Attribute::OptimizeNone)) {
Assert(Attrs.hasFnAttribute(Attribute::NoInline),
"Attribute 'optnone' requires 'noinline'!", V);
Assert(!Attrs.hasFnAttribute(Attribute::OptimizeForSize),
"Attributes 'optsize and optnone' are incompatible!", V);
Assert(!Attrs.hasFnAttribute(Attribute::MinSize),
"Attributes 'minsize and optnone' are incompatible!", V);
}
if (Attrs.hasFnAttribute(Attribute::JumpTable)) {
const GlobalValue *GV = cast<GlobalValue>(V);
IR: Introduce local_unnamed_addr attribute. If a local_unnamed_addr attribute is attached to a global, the address is known to be insignificant within the module. It is distinct from the existing unnamed_addr attribute in that it only describes a local property of the module rather than a global property of the symbol. This attribute is intended to be used by the code generator and LTO to allow the linker to decide whether the global needs to be in the symbol table. It is possible to exclude a global from the symbol table if three things are true: - This attribute is present on every instance of the global (which means that the normal rule that the global must have a unique address can be broken without being observable by the program by performing comparisons against the global's address) - The global has linkonce_odr linkage (which means that each linkage unit must have its own copy of the global if it requires one, and the copy in each linkage unit must be the same) - It is a constant or a function (which means that the program cannot observe that the unique-address rule has been broken by writing to the global) Although this attribute could in principle be computed from the module contents, LTO clients (i.e. linkers) will normally need to be able to compute this property as part of symbol resolution, and it would be inefficient to materialize every module just to compute it. See: http://lists.llvm.org/pipermail/llvm-commits/Week-of-Mon-20160509/356401.html http://lists.llvm.org/pipermail/llvm-commits/Week-of-Mon-20160516/356738.html for earlier discussion. Part of the fix for PR27553. Differential Revision: http://reviews.llvm.org/D20348 llvm-svn: 272709
2016-06-14 23:01:22 +02:00
Assert(GV->hasGlobalUnnamedAddr(),
"Attribute 'jumptable' requires 'unnamed_addr'", V);
}
if (Attrs.hasFnAttribute(Attribute::AllocSize)) {
std::pair<unsigned, Optional<unsigned>> Args =
Attrs.getAllocSizeArgs(AttributeList::FunctionIndex);
auto CheckParam = [&](StringRef Name, unsigned ParamNo) {
if (ParamNo >= FT->getNumParams()) {
CheckFailed("'allocsize' " + Name + " argument is out of bounds", V);
return false;
}
if (!FT->getParamType(ParamNo)->isIntegerTy()) {
CheckFailed("'allocsize' " + Name +
" argument must refer to an integer parameter",
V);
return false;
}
return true;
};
if (!CheckParam("element size", Args.first))
return;
if (Args.second && !CheckParam("number of elements", *Args.second))
return;
}
}
void Verifier::verifyFunctionMetadata(
ArrayRef<std::pair<unsigned, MDNode *>> MDs) {
2016-02-25 17:44:27 +01:00
for (const auto &Pair : MDs) {
if (Pair.first == LLVMContext::MD_prof) {
MDNode *MD = Pair.second;
Assert(MD->getNumOperands() >= 2,
"!prof annotations should have no less than 2 operands", MD);
// Check first operand.
Assert(MD->getOperand(0) != nullptr, "first operand should not be null",
MD);
Assert(isa<MDString>(MD->getOperand(0)),
"expected string with name of the !prof annotation", MD);
MDString *MDS = cast<MDString>(MD->getOperand(0));
StringRef ProfName = MDS->getString();
Assert(ProfName.equals("function_entry_count") ||
ProfName.equals("synthetic_function_entry_count"),
"first operand should be 'function_entry_count'"
" or 'synthetic_function_entry_count'",
MD);
// Check second operand.
Assert(MD->getOperand(1) != nullptr, "second operand should not be null",
MD);
Assert(isa<ConstantAsMetadata>(MD->getOperand(1)),
"expected integer argument to function_entry_count", MD);
}
}
}
void Verifier::visitConstantExprsRecursively(const Constant *EntryC) {
if (!ConstantExprVisited.insert(EntryC).second)
return;
SmallVector<const Constant *, 16> Stack;
Stack.push_back(EntryC);
while (!Stack.empty()) {
const Constant *C = Stack.pop_back_val();
// Check this constant expression.
if (const auto *CE = dyn_cast<ConstantExpr>(C))
visitConstantExpr(CE);
if (const auto *GV = dyn_cast<GlobalValue>(C)) {
// Global Values get visited separately, but we do need to make sure
// that the global value is in the correct module
Assert(GV->getParent() == &M, "Referencing global in another module!",
EntryC, &M, GV, GV->getParent());
continue;
}
// Visit all sub-expressions.
for (const Use &U : C->operands()) {
const auto *OpC = dyn_cast<Constant>(U);
if (!OpC)
continue;
if (!ConstantExprVisited.insert(OpC).second)
continue;
Stack.push_back(OpC);
}
}
}
void Verifier::visitConstantExpr(const ConstantExpr *CE) {
if (CE->getOpcode() == Instruction::BitCast)
Assert(CastInst::castIsValid(Instruction::BitCast, CE->getOperand(0),
CE->getType()),
"Invalid bitcast", CE);
if (CE->getOpcode() == Instruction::IntToPtr ||
CE->getOpcode() == Instruction::PtrToInt) {
auto *PtrTy = CE->getOpcode() == Instruction::IntToPtr
? CE->getType()
: CE->getOperand(0)->getType();
StringRef Msg = CE->getOpcode() == Instruction::IntToPtr
? "inttoptr not supported for non-integral pointers"
: "ptrtoint not supported for non-integral pointers";
Assert(
!DL.isNonIntegralPointerType(cast<PointerType>(PtrTy->getScalarType())),
Msg);
}
}
bool Verifier::verifyAttributeCount(AttributeList Attrs, unsigned Params) {
// There shouldn't be more attribute sets than there are parameters plus the
// function and return value.
return Attrs.getNumAttrSets() <= Params + 2;
}
2009-09-07 22:44:51 +02:00
/// Verify that statepoint intrinsic is well formed.
void Verifier::verifyStatepoint(ImmutableCallSite CS) {
assert(CS.getCalledFunction() &&
CS.getCalledFunction()->getIntrinsicID() ==
Intrinsic::experimental_gc_statepoint);
const Instruction &CI = *CS.getInstruction();
Assert(!CS.doesNotAccessMemory() && !CS.onlyReadsMemory() &&
!CS.onlyAccessesArgMemory(),
"gc.statepoint must read and write all memory to preserve "
"reordering restrictions required by safepoint semantics",
&CI);
const Value *IDV = CS.getArgument(0);
Assert(isa<ConstantInt>(IDV), "gc.statepoint ID must be a constant integer",
&CI);
const Value *NumPatchBytesV = CS.getArgument(1);
Assert(isa<ConstantInt>(NumPatchBytesV),
"gc.statepoint number of patchable bytes must be a constant integer",
&CI);
const int64_t NumPatchBytes =
cast<ConstantInt>(NumPatchBytesV)->getSExtValue();
assert(isInt<32>(NumPatchBytes) && "NumPatchBytesV is an i32!");
Assert(NumPatchBytes >= 0, "gc.statepoint number of patchable bytes must be "
"positive",
&CI);
const Value *Target = CS.getArgument(2);
auto *PT = dyn_cast<PointerType>(Target->getType());
Assert(PT && PT->getElementType()->isFunctionTy(),
"gc.statepoint callee must be of function pointer type", &CI, Target);
FunctionType *TargetFuncType = cast<FunctionType>(PT->getElementType());
const Value *NumCallArgsV = CS.getArgument(3);
Assert(isa<ConstantInt>(NumCallArgsV),
"gc.statepoint number of arguments to underlying call "
"must be constant integer",
&CI);
const int NumCallArgs = cast<ConstantInt>(NumCallArgsV)->getZExtValue();
Assert(NumCallArgs >= 0,
"gc.statepoint number of arguments to underlying call "
"must be positive",
&CI);
const int NumParams = (int)TargetFuncType->getNumParams();
if (TargetFuncType->isVarArg()) {
Assert(NumCallArgs >= NumParams,
"gc.statepoint mismatch in number of vararg call args", &CI);
// TODO: Remove this limitation
Assert(TargetFuncType->getReturnType()->isVoidTy(),
"gc.statepoint doesn't support wrapping non-void "
"vararg functions yet",
&CI);
} else
Assert(NumCallArgs == NumParams,
"gc.statepoint mismatch in number of call args", &CI);
const Value *FlagsV = CS.getArgument(4);
Assert(isa<ConstantInt>(FlagsV),
"gc.statepoint flags must be constant integer", &CI);
const uint64_t Flags = cast<ConstantInt>(FlagsV)->getZExtValue();
Assert((Flags & ~(uint64_t)StatepointFlags::MaskAll) == 0,
"unknown flag used in gc.statepoint flags argument", &CI);
// Verify that the types of the call parameter arguments match
// the type of the wrapped callee.
AttributeList Attrs = CS.getAttributes();
for (int i = 0; i < NumParams; i++) {
Type *ParamType = TargetFuncType->getParamType(i);
Type *ArgType = CS.getArgument(5 + i)->getType();
Assert(ArgType == ParamType,
"gc.statepoint call argument does not match wrapped "
"function type",
&CI);
if (TargetFuncType->isVarArg()) {
AttributeSet ArgAttrs = Attrs.getParamAttributes(5 + i);
Assert(!ArgAttrs.hasAttribute(Attribute::StructRet),
"Attribute 'sret' cannot be used for vararg call arguments!", &CI);
}
}
const int EndCallArgsInx = 4 + NumCallArgs;
const Value *NumTransitionArgsV = CS.getArgument(EndCallArgsInx+1);
Assert(isa<ConstantInt>(NumTransitionArgsV),
"gc.statepoint number of transition arguments "
"must be constant integer",
&CI);
const int NumTransitionArgs =
cast<ConstantInt>(NumTransitionArgsV)->getZExtValue();
Assert(NumTransitionArgs >= 0,
"gc.statepoint number of transition arguments must be positive", &CI);
const int EndTransitionArgsInx = EndCallArgsInx + 1 + NumTransitionArgs;
const Value *NumDeoptArgsV = CS.getArgument(EndTransitionArgsInx+1);
Assert(isa<ConstantInt>(NumDeoptArgsV),
"gc.statepoint number of deoptimization arguments "
"must be constant integer",
&CI);
const int NumDeoptArgs = cast<ConstantInt>(NumDeoptArgsV)->getZExtValue();
Assert(NumDeoptArgs >= 0, "gc.statepoint number of deoptimization arguments "
"must be positive",
&CI);
const int ExpectedNumArgs =
7 + NumCallArgs + NumTransitionArgs + NumDeoptArgs;
Assert(ExpectedNumArgs <= (int)CS.arg_size(),
"gc.statepoint too few arguments according to length fields", &CI);
// Check that the only uses of this gc.statepoint are gc.result or
// gc.relocate calls which are tied to this statepoint and thus part
// of the same statepoint sequence
for (const User *U : CI.users()) {
const CallInst *Call = dyn_cast<const CallInst>(U);
Assert(Call, "illegal use of statepoint token", &CI, U);
if (!Call) continue;
Assert(isa<GCRelocateInst>(Call) || isa<GCResultInst>(Call),
"gc.result or gc.relocate are the only value uses "
"of a gc.statepoint",
&CI, U);
if (isa<GCResultInst>(Call)) {
Assert(Call->getArgOperand(0) == &CI,
"gc.result connected to wrong gc.statepoint", &CI, Call);
} else if (isa<GCRelocateInst>(Call)) {
Assert(Call->getArgOperand(0) == &CI,
"gc.relocate connected to wrong gc.statepoint", &CI, Call);
}
}
// Note: It is legal for a single derived pointer to be listed multiple
// times. It's non-optimal, but it is legal. It can also happen after
// insertion if we strip a bitcast away.
// Note: It is really tempting to check that each base is relocated and
// that a derived pointer is never reused as a base pointer. This turns
// out to be problematic since optimizations run after safepoint insertion
// can recognize equality properties that the insertion logic doesn't know
// about. See example statepoint.ll in the verifier subdirectory
}
void Verifier::verifyFrameRecoverIndices() {
for (auto &Counts : FrameEscapeInfo) {
Function *F = Counts.first;
unsigned EscapedObjectCount = Counts.second.first;
unsigned MaxRecoveredIndex = Counts.second.second;
Assert(MaxRecoveredIndex <= EscapedObjectCount,
"all indices passed to llvm.localrecover must be less than the "
"number of arguments passed ot llvm.localescape in the parent "
"function",
F);
}
}
static Instruction *getSuccPad(Instruction *Terminator) {
BasicBlock *UnwindDest;
if (auto *II = dyn_cast<InvokeInst>(Terminator))
UnwindDest = II->getUnwindDest();
else if (auto *CSI = dyn_cast<CatchSwitchInst>(Terminator))
UnwindDest = CSI->getUnwindDest();
else
UnwindDest = cast<CleanupReturnInst>(Terminator)->getUnwindDest();
return UnwindDest->getFirstNonPHI();
}
void Verifier::verifySiblingFuncletUnwinds() {
SmallPtrSet<Instruction *, 8> Visited;
SmallPtrSet<Instruction *, 8> Active;
for (const auto &Pair : SiblingFuncletInfo) {
Instruction *PredPad = Pair.first;
if (Visited.count(PredPad))
continue;
Active.insert(PredPad);
Instruction *Terminator = Pair.second;
do {
Instruction *SuccPad = getSuccPad(Terminator);
if (Active.count(SuccPad)) {
// Found a cycle; report error
Instruction *CyclePad = SuccPad;
SmallVector<Instruction *, 8> CycleNodes;
do {
CycleNodes.push_back(CyclePad);
Instruction *CycleTerminator = SiblingFuncletInfo[CyclePad];
if (CycleTerminator != CyclePad)
CycleNodes.push_back(CycleTerminator);
CyclePad = getSuccPad(CycleTerminator);
} while (CyclePad != SuccPad);
Assert(false, "EH pads can't handle each other's exceptions",
ArrayRef<Instruction *>(CycleNodes));
}
// Don't re-walk a node we've already checked
if (!Visited.insert(SuccPad).second)
break;
// Walk to this successor if it has a map entry.
PredPad = SuccPad;
auto TermI = SiblingFuncletInfo.find(PredPad);
if (TermI == SiblingFuncletInfo.end())
break;
Terminator = TermI->second;
Active.insert(PredPad);
} while (true);
// Each node only has one successor, so we've walked all the active
// nodes' successors.
Active.clear();
}
}
// visitFunction - Verify that a function is ok.
//
void Verifier::visitFunction(const Function &F) {
visitGlobalValue(F);
// Check function arguments.
FunctionType *FT = F.getFunctionType();
2007-08-18 08:13:19 +02:00
unsigned NumArgs = F.arg_size();
Assert(&Context == &F.getContext(),
"Function context does not match Module context!", &F);
Assert(!F.hasCommonLinkage(), "Functions may not have common linkage", &F);
Assert(FT->getNumParams() == NumArgs,
"# formal arguments must match # of arguments for function type!", &F,
FT);
Assert(F.getReturnType()->isFirstClassType() ||
F.getReturnType()->isVoidTy() || F.getReturnType()->isStructTy(),
"Functions cannot return aggregate values!", &F);
Assert(!F.hasStructRetAttr() || F.getReturnType()->isVoidTy(),
"Invalid struct return type!", &F);
AttributeList Attrs = F.getAttributes();
Assert(verifyAttributeCount(Attrs, FT->getNumParams()),
"Attribute after last parameter!", &F);
// Check function attributes.
verifyFunctionAttrs(FT, Attrs, &F);
// On function declarations/definitions, we do not support the builtin
// attribute. We do not check this in VerifyFunctionAttrs since that is
// checking for Attributes that can/can not ever be on functions.
Assert(!Attrs.hasFnAttribute(Attribute::Builtin),
"Attribute 'builtin' can only be applied to a callsite.", &F);
// Check that this function meets the restrictions on this calling convention.
// Sometimes varargs is used for perfectly forwarding thunks, so some of these
// restrictions can be lifted.
switch (F.getCallingConv()) {
default:
case CallingConv::C:
break;
case CallingConv::AMDGPU_KERNEL:
case CallingConv::SPIR_KERNEL:
Assert(F.getReturnType()->isVoidTy(),
"Calling convention requires void return type", &F);
LLVM_FALLTHROUGH;
case CallingConv::AMDGPU_VS:
case CallingConv::AMDGPU_HS:
case CallingConv::AMDGPU_GS:
case CallingConv::AMDGPU_PS:
case CallingConv::AMDGPU_CS:
Assert(!F.hasStructRetAttr(),
"Calling convention does not allow sret", &F);
LLVM_FALLTHROUGH;
case CallingConv::Fast:
case CallingConv::Cold:
case CallingConv::Intel_OCL_BI:
case CallingConv::PTX_Kernel:
case CallingConv::PTX_Device:
Assert(!F.isVarArg(), "Calling convention does not support varargs or "
"perfect forwarding!",
&F);
break;
}
2009-09-07 22:44:51 +02:00
bool isLLVMdotName = F.getName().size() >= 5 &&
F.getName().substr(0, 5) == "llvm.";
// Check that the argument values match the function type for this function...
unsigned i = 0;
2016-02-25 17:44:27 +01:00
for (const Argument &Arg : F.args()) {
Assert(Arg.getType() == FT->getParamType(i),
"Argument value does not match function argument type!", &Arg,
FT->getParamType(i));
2016-02-25 17:44:27 +01:00
Assert(Arg.getType()->isFirstClassType(),
"Function arguments must have first-class types!", &Arg);
[IR] Add token types This introduces the basic functionality to support "token types". The motivation stems from the need to perform operations on a Value whose provenance cannot be obscured. There are several applications for such a type but my immediate motivation stems from WinEH. Our personality routine enforces a single-entry - single-exit regime for cleanups. After several rounds of optimizations, we may be left with a terminator whose "cleanup-entry block" is not entirely clear because control flow has merged two cleanups together. We have experimented with using labels as operands inside of instructions which are not terminators to indicate where we came from but found that LLVM does not expect such exotic uses of BasicBlocks. Instead, we can use this new type to clearly associate the "entry point" and "exit point" of our cleanup. This is done by having the cleanuppad yield a Token and consuming it at the cleanupret. The token type makes it impossible to obscure or otherwise hide the Value, making it trivial to track the relationship between the two points. What is the burden to the optimizer? Well, it turns out we have already paid down this cost by accepting that there are certain calls that we are not permitted to duplicate, optimizations have to watch out for such instructions anyway. There are additional places in the optimizer that we will probably have to update but early examination has given me the impression that this will not be heroic. Differential Revision: http://reviews.llvm.org/D11861 llvm-svn: 245029
2015-08-14 07:09:07 +02:00
if (!isLLVMdotName) {
2016-02-25 17:44:27 +01:00
Assert(!Arg.getType()->isMetadataTy(),
"Function takes metadata but isn't an intrinsic", &Arg, &F);
Assert(!Arg.getType()->isTokenTy(),
"Function takes token but isn't an intrinsic", &Arg, &F);
[IR] Add token types This introduces the basic functionality to support "token types". The motivation stems from the need to perform operations on a Value whose provenance cannot be obscured. There are several applications for such a type but my immediate motivation stems from WinEH. Our personality routine enforces a single-entry - single-exit regime for cleanups. After several rounds of optimizations, we may be left with a terminator whose "cleanup-entry block" is not entirely clear because control flow has merged two cleanups together. We have experimented with using labels as operands inside of instructions which are not terminators to indicate where we came from but found that LLVM does not expect such exotic uses of BasicBlocks. Instead, we can use this new type to clearly associate the "entry point" and "exit point" of our cleanup. This is done by having the cleanuppad yield a Token and consuming it at the cleanupret. The token type makes it impossible to obscure or otherwise hide the Value, making it trivial to track the relationship between the two points. What is the burden to the optimizer? Well, it turns out we have already paid down this cost by accepting that there are certain calls that we are not permitted to duplicate, optimizations have to watch out for such instructions anyway. There are additional places in the optimizer that we will probably have to update but early examination has given me the impression that this will not be heroic. Differential Revision: http://reviews.llvm.org/D11861 llvm-svn: 245029
2015-08-14 07:09:07 +02:00
}
// Check that swifterror argument is only used by loads and stores.
if (Attrs.hasParamAttribute(i, Attribute::SwiftError)) {
verifySwiftErrorValue(&Arg);
}
2016-02-25 17:44:27 +01:00
++i;
}
[IR] Add token types This introduces the basic functionality to support "token types". The motivation stems from the need to perform operations on a Value whose provenance cannot be obscured. There are several applications for such a type but my immediate motivation stems from WinEH. Our personality routine enforces a single-entry - single-exit regime for cleanups. After several rounds of optimizations, we may be left with a terminator whose "cleanup-entry block" is not entirely clear because control flow has merged two cleanups together. We have experimented with using labels as operands inside of instructions which are not terminators to indicate where we came from but found that LLVM does not expect such exotic uses of BasicBlocks. Instead, we can use this new type to clearly associate the "entry point" and "exit point" of our cleanup. This is done by having the cleanuppad yield a Token and consuming it at the cleanupret. The token type makes it impossible to obscure or otherwise hide the Value, making it trivial to track the relationship between the two points. What is the burden to the optimizer? Well, it turns out we have already paid down this cost by accepting that there are certain calls that we are not permitted to duplicate, optimizations have to watch out for such instructions anyway. There are additional places in the optimizer that we will probably have to update but early examination has given me the impression that this will not be heroic. Differential Revision: http://reviews.llvm.org/D11861 llvm-svn: 245029
2015-08-14 07:09:07 +02:00
if (!isLLVMdotName)
Assert(!F.getReturnType()->isTokenTy(),
"Functions returns a token but isn't an intrinsic", &F);
// Get the function metadata attachments.
SmallVector<std::pair<unsigned, MDNode *>, 4> MDs;
F.getAllMetadata(MDs);
assert(F.hasMetadata() != MDs.empty() && "Bit out-of-sync");
verifyFunctionMetadata(MDs);
// Check validity of the personality function
if (F.hasPersonalityFn()) {
auto *Per = dyn_cast<Function>(F.getPersonalityFn()->stripPointerCasts());
if (Per)
Assert(Per->getParent() == F.getParent(),
"Referencing personality function in another module!",
&F, F.getParent(), Per, Per->getParent());
}
if (F.isMaterializable()) {
// Function has a body somewhere we can't see.
Assert(MDs.empty(), "unmaterialized function cannot have metadata", &F,
MDs.empty() ? nullptr : MDs.front().second);
} else if (F.isDeclaration()) {
IR: Allow metadata attachments on declarations, and fix lazy loaded metadata issue with globals. This change is motivated by an upcoming change to the metadata representation used for CFI. The indirect function call checker needs type information for external function declarations in order to correctly generate jump table entries for such declarations. We currently associate such type information with declarations using a global metadata node, but I plan [1] to move all such metadata to global object attachments. In bitcode, metadata attachments for function declarations appear in the global metadata block. This seems reasonable to me because I expect metadata attachments on declarations to be uncommon. In the long term I'd also expect this to be the case for CFI, because we'd want to use some specialized bitcode format for this metadata that could be read as part of the ThinLTO thin-link phase, which would mean that it would not appear in the global metadata block. To solve the lazy loaded metadata issue I was seeing with D20147, I use the same bitcode representation for metadata attachments for global variables as I do for function declarations. Since there's a use case for metadata attachments in the global metadata block, we might as well use that representation for global variables as well, at least until we have a mechanism for lazy loading global variables. In the assembly format, the metadata attachments appear after the "declare" keyword in order to avoid a parsing ambiguity. [1] http://lists.llvm.org/pipermail/llvm-dev/2016-June/100462.html Differential Revision: http://reviews.llvm.org/D21052 llvm-svn: 273336
2016-06-22 01:42:48 +02:00
for (const auto &I : MDs) {
AssertDI(I.first != LLVMContext::MD_dbg,
"function declaration may not have a !dbg attachment", &F);
Assert(I.first != LLVMContext::MD_prof,
"function declaration may not have a !prof attachment", &F);
// Verify the metadata itself.
visitMDNode(*I.second);
}
Assert(!F.hasPersonalityFn(),
"Function declaration shouldn't have a personality routine", &F);
} else {
// Verify that this function (which has a body) is not named "llvm.*". It
// is not legal to define intrinsics.
Assert(!isLLVMdotName, "llvm intrinsics cannot be defined!", &F);
// Check the entry node
const BasicBlock *Entry = &F.getEntryBlock();
Assert(pred_empty(Entry),
"Entry block to function must not have predecessors!", Entry);
// The address of the entry block cannot be taken, unless it is dead.
if (Entry->hasAddressTaken()) {
Assert(!BlockAddress::lookup(Entry)->isConstantUsed(),
"blockaddress may not be used with the entry block!", Entry);
}
unsigned NumDebugAttachments = 0, NumProfAttachments = 0;
// Visit metadata attachments.
for (const auto &I : MDs) {
// Verify that the attachment is legal.
switch (I.first) {
default:
break;
case LLVMContext::MD_dbg: {
++NumDebugAttachments;
AssertDI(NumDebugAttachments == 1,
"function must have a single !dbg attachment", &F, I.second);
AssertDI(isa<DISubprogram>(I.second),
"function !dbg attachment must be a subprogram", &F, I.second);
auto *SP = cast<DISubprogram>(I.second);
const Function *&AttachedTo = DISubprogramAttachments[SP];
AssertDI(!AttachedTo || AttachedTo == &F,
"DISubprogram attached to more than one function", SP, &F);
AttachedTo = &F;
break;
}
case LLVMContext::MD_prof:
++NumProfAttachments;
Assert(NumProfAttachments == 1,
"function must have a single !prof attachment", &F, I.second);
break;
}
// Verify the metadata itself.
visitMDNode(*I.second);
}
}
// If this function is actually an intrinsic, verify that it is only used in
// direct call/invokes, never having its "address taken".
// Only do this if the module is materialized, otherwise we don't have all the
// uses.
if (F.getIntrinsicID() && F.getParent()->isMaterialized()) {
const User *U;
if (F.hasAddressTaken(&U))
Assert(false, "Invalid user of intrinsic instruction!", U);
}
auto *N = F.getSubprogram();
HasDebugInfo = (N != nullptr);
if (!HasDebugInfo)
return;
// Check that all !dbg attachments lead to back to N (or, at least, another
// subprogram that describes the same function).
//
// FIXME: Check this incrementally while visiting !dbg attachments.
// FIXME: Only check when N is the canonical subprogram for F.
SmallPtrSet<const MDNode *, 32> Seen;
for (auto &BB : F)
for (auto &I : BB) {
// Be careful about using DILocation here since we might be dealing with
// broken code (this is the Verifier after all).
DILocation *DL =
dyn_cast_or_null<DILocation>(I.getDebugLoc().getAsMDNode());
if (!DL)
continue;
if (!Seen.insert(DL).second)
continue;
Metadata *Parent = DL->getRawScope();
AssertDI(Parent && isa<DILocalScope>(Parent),
"DILocation's scope must be a DILocalScope", N, &F, &I, DL,
Parent);
DILocalScope *Scope = DL->getInlinedAtScope();
if (Scope && !Seen.insert(Scope).second)
continue;
DISubprogram *SP = Scope ? Scope->getSubprogram() : nullptr;
// Scope and SP could be the same MDNode and we don't want to skip
// validation in that case
if (SP && ((Scope != SP) && !Seen.insert(SP).second))
continue;
// FIXME: Once N is canonical, check "SP == &N".
AssertDI(SP->describes(&F),
"!dbg attachment points at wrong subprogram for function", N, &F,
&I, DL, Scope, SP);
}
}
// verifyBasicBlock - Verify that a basic block is well formed...
//
void Verifier::visitBasicBlock(BasicBlock &BB) {
InstsInThisBlock.clear();
// Ensure that basic blocks have terminators!
Assert(BB.getTerminator(), "Basic Block does not have terminator!", &BB);
// Check constraints that this basic block imposes on all of the PHI nodes in
// it.
if (isa<PHINode>(BB.front())) {
SmallVector<BasicBlock*, 8> Preds(pred_begin(&BB), pred_end(&BB));
SmallVector<std::pair<BasicBlock*, Value*>, 8> Values;
llvm::sort(Preds);
for (const PHINode &PN : BB.phis()) {
// Ensure that PHI nodes have at least one entry!
Assert(PN.getNumIncomingValues() != 0,
"PHI nodes must have at least one entry. If the block is dead, "
"the PHI should be removed!",
&PN);
Assert(PN.getNumIncomingValues() == Preds.size(),
"PHINode should have one entry for each predecessor of its "
"parent basic block!",
&PN);
// Get and sort all incoming values in the PHI node...
Values.clear();
Values.reserve(PN.getNumIncomingValues());
for (unsigned i = 0, e = PN.getNumIncomingValues(); i != e; ++i)
Values.push_back(
std::make_pair(PN.getIncomingBlock(i), PN.getIncomingValue(i)));
llvm::sort(Values);
for (unsigned i = 0, e = Values.size(); i != e; ++i) {
// Check to make sure that if there is more than one entry for a
// particular basic block in this PHI node, that the incoming values are
// all identical.
//
Assert(i == 0 || Values[i].first != Values[i - 1].first ||
Values[i].second == Values[i - 1].second,
"PHI node has multiple entries for the same basic block with "
"different incoming values!",
&PN, Values[i].first, Values[i].second, Values[i - 1].second);
// Check to make sure that the predecessors and PHI node entries are
// matched up.
Assert(Values[i].first == Preds[i],
"PHI node entries do not match predecessors!", &PN,
Values[i].first, Preds[i]);
}
}
}
// Check that all instructions have their parent pointers set up correctly.
for (auto &I : BB)
{
Assert(I.getParent() == &BB, "Instruction has bogus parent pointer!");
}
}
void Verifier::visitTerminator(Instruction &I) {
// Ensure that terminators only exist at the end of the basic block.
Assert(&I == I.getParent()->getTerminator(),
"Terminator found in the middle of a basic block!", I.getParent());
visitInstruction(I);
}
void Verifier::visitBranchInst(BranchInst &BI) {
if (BI.isConditional()) {
Assert(BI.getCondition()->getType()->isIntegerTy(1),
"Branch condition is not 'i1' type!", &BI, BI.getCondition());
}
visitTerminator(BI);
}
void Verifier::visitReturnInst(ReturnInst &RI) {
Function *F = RI.getParent()->getParent();
unsigned N = RI.getNumOperands();
if (F->getReturnType()->isVoidTy())
Assert(N == 0,
"Found return instr that returns non-void in Function of void "
"return type!",
&RI, F->getReturnType());
else
Assert(N == 1 && F->getReturnType() == RI.getOperand(0)->getType(),
"Function return type does not match operand "
"type of return inst!",
&RI, F->getReturnType());
2009-09-07 22:44:51 +02:00
// Check to make sure that the return value has necessary properties for
// terminators...
visitTerminator(RI);
2001-06-06 22:29:01 +02:00
}
void Verifier::visitSwitchInst(SwitchInst &SI) {
// Check to make sure that all of the constants in the switch instruction
// have the same type as the switched-on value.
Type *SwitchTy = SI.getCondition()->getType();
SmallPtrSet<ConstantInt*, 32> Constants;
2016-02-25 17:44:27 +01:00
for (auto &Case : SI.cases()) {
Assert(Case.getCaseValue()->getType() == SwitchTy,
"Switch constants must all be same type as switch value!", &SI);
2016-02-25 17:44:27 +01:00
Assert(Constants.insert(Case.getCaseValue()).second,
"Duplicate integer as switch case", &SI, Case.getCaseValue());
}
visitTerminator(SI);
}
void Verifier::visitIndirectBrInst(IndirectBrInst &BI) {
Assert(BI.getAddress()->getType()->isPointerTy(),
"Indirectbr operand must have pointer type!", &BI);
for (unsigned i = 0, e = BI.getNumDestinations(); i != e; ++i)
Assert(BI.getDestination(i)->getType()->isLabelTy(),
"Indirectbr destinations must all have pointer type!", &BI);
visitTerminator(BI);
}
void Verifier::visitSelectInst(SelectInst &SI) {
Assert(!SelectInst::areInvalidOperands(SI.getOperand(0), SI.getOperand(1),
SI.getOperand(2)),
"Invalid operands for select instruction!", &SI);
Assert(SI.getTrueValue()->getType() == SI.getType(),
"Select values must have same type as select instruction!", &SI);
visitInstruction(SI);
}
2004-03-02 01:22:19 +01:00
/// visitUserOp1 - User defined operators shouldn't live beyond the lifetime of
/// a pass, if any exist, it's an error.
///
void Verifier::visitUserOp1(Instruction &I) {
Assert(false, "User-defined operators should not live outside of a pass!", &I);
}
2001-06-06 22:29:01 +02:00
void Verifier::visitTruncInst(TruncInst &I) {
// Get the source and destination types
Type *SrcTy = I.getOperand(0)->getType();
Type *DestTy = I.getType();
// Get the size of the types in bits, we'll need this later
unsigned SrcBitSize = SrcTy->getScalarSizeInBits();
unsigned DestBitSize = DestTy->getScalarSizeInBits();
Assert(SrcTy->isIntOrIntVectorTy(), "Trunc only operates on integer", &I);
Assert(DestTy->isIntOrIntVectorTy(), "Trunc only produces integer", &I);
Assert(SrcTy->isVectorTy() == DestTy->isVectorTy(),
"trunc source and destination must both be a vector or neither", &I);
Assert(SrcBitSize > DestBitSize, "DestTy too big for Trunc", &I);
visitInstruction(I);
}
void Verifier::visitZExtInst(ZExtInst &I) {
// Get the source and destination types
Type *SrcTy = I.getOperand(0)->getType();
Type *DestTy = I.getType();
// Get the size of the types in bits, we'll need this later
Assert(SrcTy->isIntOrIntVectorTy(), "ZExt only operates on integer", &I);
Assert(DestTy->isIntOrIntVectorTy(), "ZExt only produces an integer", &I);
Assert(SrcTy->isVectorTy() == DestTy->isVectorTy(),
"zext source and destination must both be a vector or neither", &I);
unsigned SrcBitSize = SrcTy->getScalarSizeInBits();
unsigned DestBitSize = DestTy->getScalarSizeInBits();
Assert(SrcBitSize < DestBitSize, "Type too small for ZExt", &I);
visitInstruction(I);
}
void Verifier::visitSExtInst(SExtInst &I) {
// Get the source and destination types
Type *SrcTy = I.getOperand(0)->getType();
Type *DestTy = I.getType();
// Get the size of the types in bits, we'll need this later
unsigned SrcBitSize = SrcTy->getScalarSizeInBits();
unsigned DestBitSize = DestTy->getScalarSizeInBits();
Assert(SrcTy->isIntOrIntVectorTy(), "SExt only operates on integer", &I);
Assert(DestTy->isIntOrIntVectorTy(), "SExt only produces an integer", &I);
Assert(SrcTy->isVectorTy() == DestTy->isVectorTy(),
"sext source and destination must both be a vector or neither", &I);
Assert(SrcBitSize < DestBitSize, "Type too small for SExt", &I);
visitInstruction(I);
}
void Verifier::visitFPTruncInst(FPTruncInst &I) {
// Get the source and destination types
Type *SrcTy = I.getOperand(0)->getType();
Type *DestTy = I.getType();
// Get the size of the types in bits, we'll need this later
unsigned SrcBitSize = SrcTy->getScalarSizeInBits();
unsigned DestBitSize = DestTy->getScalarSizeInBits();
Assert(SrcTy->isFPOrFPVectorTy(), "FPTrunc only operates on FP", &I);
Assert(DestTy->isFPOrFPVectorTy(), "FPTrunc only produces an FP", &I);
Assert(SrcTy->isVectorTy() == DestTy->isVectorTy(),
"fptrunc source and destination must both be a vector or neither", &I);
Assert(SrcBitSize > DestBitSize, "DestTy too big for FPTrunc", &I);
visitInstruction(I);
}
void Verifier::visitFPExtInst(FPExtInst &I) {
// Get the source and destination types
Type *SrcTy = I.getOperand(0)->getType();
Type *DestTy = I.getType();
// Get the size of the types in bits, we'll need this later
unsigned SrcBitSize = SrcTy->getScalarSizeInBits();
unsigned DestBitSize = DestTy->getScalarSizeInBits();
Assert(SrcTy->isFPOrFPVectorTy(), "FPExt only operates on FP", &I);
Assert(DestTy->isFPOrFPVectorTy(), "FPExt only produces an FP", &I);
Assert(SrcTy->isVectorTy() == DestTy->isVectorTy(),
"fpext source and destination must both be a vector or neither", &I);
Assert(SrcBitSize < DestBitSize, "DestTy too small for FPExt", &I);
visitInstruction(I);
}
void Verifier::visitUIToFPInst(UIToFPInst &I) {
// Get the source and destination types
Type *SrcTy = I.getOperand(0)->getType();
Type *DestTy = I.getType();
bool SrcVec = SrcTy->isVectorTy();
bool DstVec = DestTy->isVectorTy();
Assert(SrcVec == DstVec,
"UIToFP source and dest must both be vector or scalar", &I);
Assert(SrcTy->isIntOrIntVectorTy(),
"UIToFP source must be integer or integer vector", &I);
Assert(DestTy->isFPOrFPVectorTy(), "UIToFP result must be FP or FP vector",
&I);
if (SrcVec && DstVec)
Assert(cast<VectorType>(SrcTy)->getNumElements() ==
cast<VectorType>(DestTy)->getNumElements(),
"UIToFP source and dest vector length mismatch", &I);
visitInstruction(I);
}
void Verifier::visitSIToFPInst(SIToFPInst &I) {
// Get the source and destination types
Type *SrcTy = I.getOperand(0)->getType();
Type *DestTy = I.getType();
bool SrcVec = SrcTy->isVectorTy();
bool DstVec = DestTy->isVectorTy();
Assert(SrcVec == DstVec,
"SIToFP source and dest must both be vector or scalar", &I);
Assert(SrcTy->isIntOrIntVectorTy(),
"SIToFP source must be integer or integer vector", &I);
Assert(DestTy->isFPOrFPVectorTy(), "SIToFP result must be FP or FP vector",
&I);
if (SrcVec && DstVec)
Assert(cast<VectorType>(SrcTy)->getNumElements() ==
cast<VectorType>(DestTy)->getNumElements(),
"SIToFP source and dest vector length mismatch", &I);
visitInstruction(I);
}
void Verifier::visitFPToUIInst(FPToUIInst &I) {
// Get the source and destination types
Type *SrcTy = I.getOperand(0)->getType();
Type *DestTy = I.getType();
bool SrcVec = SrcTy->isVectorTy();
bool DstVec = DestTy->isVectorTy();
Assert(SrcVec == DstVec,
"FPToUI source and dest must both be vector or scalar", &I);
Assert(SrcTy->isFPOrFPVectorTy(), "FPToUI source must be FP or FP vector",
&I);
Assert(DestTy->isIntOrIntVectorTy(),
"FPToUI result must be integer or integer vector", &I);
if (SrcVec && DstVec)
Assert(cast<VectorType>(SrcTy)->getNumElements() ==
cast<VectorType>(DestTy)->getNumElements(),
"FPToUI source and dest vector length mismatch", &I);
visitInstruction(I);
}
void Verifier::visitFPToSIInst(FPToSIInst &I) {
// Get the source and destination types
Type *SrcTy = I.getOperand(0)->getType();
Type *DestTy = I.getType();
bool SrcVec = SrcTy->isVectorTy();
bool DstVec = DestTy->isVectorTy();
Assert(SrcVec == DstVec,
"FPToSI source and dest must both be vector or scalar", &I);
Assert(SrcTy->isFPOrFPVectorTy(), "FPToSI source must be FP or FP vector",
&I);
Assert(DestTy->isIntOrIntVectorTy(),
"FPToSI result must be integer or integer vector", &I);
if (SrcVec && DstVec)
Assert(cast<VectorType>(SrcTy)->getNumElements() ==
cast<VectorType>(DestTy)->getNumElements(),
"FPToSI source and dest vector length mismatch", &I);
visitInstruction(I);
}
void Verifier::visitPtrToIntInst(PtrToIntInst &I) {
// Get the source and destination types
Type *SrcTy = I.getOperand(0)->getType();
Type *DestTy = I.getType();
Assert(SrcTy->isPtrOrPtrVectorTy(), "PtrToInt source must be pointer", &I);
if (auto *PTy = dyn_cast<PointerType>(SrcTy->getScalarType()))
Assert(!DL.isNonIntegralPointerType(PTy),
"ptrtoint not supported for non-integral pointers");
Assert(DestTy->isIntOrIntVectorTy(), "PtrToInt result must be integral", &I);
Assert(SrcTy->isVectorTy() == DestTy->isVectorTy(), "PtrToInt type mismatch",
&I);
if (SrcTy->isVectorTy()) {
VectorType *VSrc = dyn_cast<VectorType>(SrcTy);
VectorType *VDest = dyn_cast<VectorType>(DestTy);
Assert(VSrc->getNumElements() == VDest->getNumElements(),
"PtrToInt Vector width mismatch", &I);
}
visitInstruction(I);
}
void Verifier::visitIntToPtrInst(IntToPtrInst &I) {
// Get the source and destination types
Type *SrcTy = I.getOperand(0)->getType();
Type *DestTy = I.getType();
Assert(SrcTy->isIntOrIntVectorTy(),
"IntToPtr source must be an integral", &I);
Assert(DestTy->isPtrOrPtrVectorTy(), "IntToPtr result must be a pointer", &I);
if (auto *PTy = dyn_cast<PointerType>(DestTy->getScalarType()))
Assert(!DL.isNonIntegralPointerType(PTy),
"inttoptr not supported for non-integral pointers");
Assert(SrcTy->isVectorTy() == DestTy->isVectorTy(), "IntToPtr type mismatch",
&I);
if (SrcTy->isVectorTy()) {
VectorType *VSrc = dyn_cast<VectorType>(SrcTy);
VectorType *VDest = dyn_cast<VectorType>(DestTy);
Assert(VSrc->getNumElements() == VDest->getNumElements(),
"IntToPtr Vector width mismatch", &I);
}
visitInstruction(I);
}
void Verifier::visitBitCastInst(BitCastInst &I) {
Assert(
CastInst::castIsValid(Instruction::BitCast, I.getOperand(0), I.getType()),
"Invalid bitcast", &I);
visitInstruction(I);
}
void Verifier::visitAddrSpaceCastInst(AddrSpaceCastInst &I) {
Type *SrcTy = I.getOperand(0)->getType();
Type *DestTy = I.getType();
Assert(SrcTy->isPtrOrPtrVectorTy(), "AddrSpaceCast source must be a pointer",
&I);
Assert(DestTy->isPtrOrPtrVectorTy(), "AddrSpaceCast result must be a pointer",
&I);
Assert(SrcTy->getPointerAddressSpace() != DestTy->getPointerAddressSpace(),
"AddrSpaceCast must be between different address spaces", &I);
if (SrcTy->isVectorTy())
Assert(SrcTy->getVectorNumElements() == DestTy->getVectorNumElements(),
"AddrSpaceCast vector pointer number of elements mismatch", &I);
visitInstruction(I);
}
2004-03-02 01:22:19 +01:00
/// visitPHINode - Ensure that a PHI node is well formed.
///
void Verifier::visitPHINode(PHINode &PN) {
// Ensure that the PHI nodes are all grouped together at the top of the block.
// This can be tested by checking whether the instruction before this is
2003-10-10 19:54:14 +02:00
// either nonexistent (because this is begin()) or is a PHI node. If not,
// then there is some other instruction before a PHI.
Assert(&PN == &PN.getParent()->front() ||
isa<PHINode>(--BasicBlock::iterator(&PN)),
"PHI nodes not grouped at top of basic block!", &PN, PN.getParent());
[IR] Add token types This introduces the basic functionality to support "token types". The motivation stems from the need to perform operations on a Value whose provenance cannot be obscured. There are several applications for such a type but my immediate motivation stems from WinEH. Our personality routine enforces a single-entry - single-exit regime for cleanups. After several rounds of optimizations, we may be left with a terminator whose "cleanup-entry block" is not entirely clear because control flow has merged two cleanups together. We have experimented with using labels as operands inside of instructions which are not terminators to indicate where we came from but found that LLVM does not expect such exotic uses of BasicBlocks. Instead, we can use this new type to clearly associate the "entry point" and "exit point" of our cleanup. This is done by having the cleanuppad yield a Token and consuming it at the cleanupret. The token type makes it impossible to obscure or otherwise hide the Value, making it trivial to track the relationship between the two points. What is the burden to the optimizer? Well, it turns out we have already paid down this cost by accepting that there are certain calls that we are not permitted to duplicate, optimizations have to watch out for such instructions anyway. There are additional places in the optimizer that we will probably have to update but early examination has given me the impression that this will not be heroic. Differential Revision: http://reviews.llvm.org/D11861 llvm-svn: 245029
2015-08-14 07:09:07 +02:00
// Check that a PHI doesn't yield a Token.
Assert(!PN.getType()->isTokenTy(), "PHI nodes cannot have token type!");
// Check that all of the values of the PHI node have the same type as the
// result, and that the incoming blocks are really basic blocks.
for (Value *IncValue : PN.incoming_values()) {
Assert(PN.getType() == IncValue->getType(),
"PHI node operands are not the same type as the result!", &PN);
}
// All other PHI node constraints are checked in the visitBasicBlock method.
visitInstruction(PN);
}
void Verifier::verifyCallSite(CallSite CS) {
Instruction *I = CS.getInstruction();
Assert(CS.getCalledValue()->getType()->isPointerTy(),
"Called function must be a pointer!", I);
PointerType *FPTy = cast<PointerType>(CS.getCalledValue()->getType());
Assert(FPTy->getElementType()->isFunctionTy(),
"Called function is not pointer to function type!", I);
Assert(FPTy->getElementType() == CS.getFunctionType(),
"Called function is not the same type as the call!", I);
FunctionType *FTy = CS.getFunctionType();
// Verify that the correct number of arguments are being passed
if (FTy->isVarArg())
Assert(CS.arg_size() >= FTy->getNumParams(),
"Called function requires more parameters than were provided!", I);
else
Assert(CS.arg_size() == FTy->getNumParams(),
"Incorrect number of arguments passed to called function!", I);
// Verify that all arguments to the call match the function type.
for (unsigned i = 0, e = FTy->getNumParams(); i != e; ++i)
Assert(CS.getArgument(i)->getType() == FTy->getParamType(i),
"Call parameter type does not match function signature!",
CS.getArgument(i), FTy->getParamType(i), I);
AttributeList Attrs = CS.getAttributes();
Assert(verifyAttributeCount(Attrs, CS.arg_size()),
"Attribute after last parameter!", I);
if (Attrs.hasAttribute(AttributeList::FunctionIndex, Attribute::Speculatable)) {
// Don't allow speculatable on call sites, unless the underlying function
// declaration is also speculatable.
Function *Callee
= dyn_cast<Function>(CS.getCalledValue()->stripPointerCasts());
Assert(Callee && Callee->isSpeculatable(),
"speculatable attribute may not apply to call sites", I);
}
// Verify call attributes.
verifyFunctionAttrs(FTy, Attrs, I);
// Conservatively check the inalloca argument.
// We have a bug if we can find that there is an underlying alloca without
// inalloca.
if (CS.hasInAllocaArgument()) {
Value *InAllocaArg = CS.getArgument(FTy->getNumParams() - 1);
if (auto AI = dyn_cast<AllocaInst>(InAllocaArg->stripInBoundsOffsets()))
Assert(AI->isUsedWithInAlloca(),
"inalloca argument for call has mismatched alloca", AI, I);
}
// For each argument of the callsite, if it has the swifterror argument,
// make sure the underlying alloca/parameter it comes from has a swifterror as
// well.
for (unsigned i = 0, e = FTy->getNumParams(); i != e; ++i)
if (CS.paramHasAttr(i, Attribute::SwiftError)) {
Value *SwiftErrorArg = CS.getArgument(i);
if (auto AI = dyn_cast<AllocaInst>(SwiftErrorArg->stripInBoundsOffsets())) {
Assert(AI->isSwiftError(),
"swifterror argument for call has mismatched alloca", AI, I);
continue;
}
auto ArgI = dyn_cast<Argument>(SwiftErrorArg);
Assert(ArgI, "swifterror argument should come from an alloca or parameter", SwiftErrorArg, I);
Assert(ArgI->hasSwiftErrorAttr(),
"swifterror argument for call has mismatched parameter", ArgI, I);
}
if (FTy->isVarArg()) {
// FIXME? is 'nest' even legal here?
bool SawNest = false;
bool SawReturned = false;
for (unsigned Idx = 0; Idx < FTy->getNumParams(); ++Idx) {
if (Attrs.hasParamAttribute(Idx, Attribute::Nest))
SawNest = true;
if (Attrs.hasParamAttribute(Idx, Attribute::Returned))
SawReturned = true;
}
// Check attributes on the varargs part.
for (unsigned Idx = FTy->getNumParams(); Idx < CS.arg_size(); ++Idx) {
Type *Ty = CS.getArgument(Idx)->getType();
AttributeSet ArgAttrs = Attrs.getParamAttributes(Idx);
verifyParameterAttrs(ArgAttrs, Ty, I);
if (ArgAttrs.hasAttribute(Attribute::Nest)) {
Assert(!SawNest, "More than one parameter has attribute nest!", I);
SawNest = true;
}
if (ArgAttrs.hasAttribute(Attribute::Returned)) {
Assert(!SawReturned, "More than one parameter has attribute returned!",
I);
Assert(Ty->canLosslesslyBitCastTo(FTy->getReturnType()),
"Incompatible argument and return types for 'returned' "
"attribute",
I);
SawReturned = true;
}
// Statepoint intrinsic is vararg but the wrapped function may be not.
// Allow sret here and check the wrapped function in verifyStatepoint.
if (CS.getCalledFunction() == nullptr ||
CS.getCalledFunction()->getIntrinsicID() !=
Intrinsic::experimental_gc_statepoint)
Assert(!ArgAttrs.hasAttribute(Attribute::StructRet),
"Attribute 'sret' cannot be used for vararg call arguments!", I);
if (ArgAttrs.hasAttribute(Attribute::InAlloca))
Assert(Idx == CS.arg_size() - 1, "inalloca isn't on the last argument!",
I);
}
}
// Verify that there's no metadata unless it's a direct call to an intrinsic.
if (CS.getCalledFunction() == nullptr ||
!CS.getCalledFunction()->getName().startswith("llvm.")) {
[IR] Add token types This introduces the basic functionality to support "token types". The motivation stems from the need to perform operations on a Value whose provenance cannot be obscured. There are several applications for such a type but my immediate motivation stems from WinEH. Our personality routine enforces a single-entry - single-exit regime for cleanups. After several rounds of optimizations, we may be left with a terminator whose "cleanup-entry block" is not entirely clear because control flow has merged two cleanups together. We have experimented with using labels as operands inside of instructions which are not terminators to indicate where we came from but found that LLVM does not expect such exotic uses of BasicBlocks. Instead, we can use this new type to clearly associate the "entry point" and "exit point" of our cleanup. This is done by having the cleanuppad yield a Token and consuming it at the cleanupret. The token type makes it impossible to obscure or otherwise hide the Value, making it trivial to track the relationship between the two points. What is the burden to the optimizer? Well, it turns out we have already paid down this cost by accepting that there are certain calls that we are not permitted to duplicate, optimizations have to watch out for such instructions anyway. There are additional places in the optimizer that we will probably have to update but early examination has given me the impression that this will not be heroic. Differential Revision: http://reviews.llvm.org/D11861 llvm-svn: 245029
2015-08-14 07:09:07 +02:00
for (Type *ParamTy : FTy->params()) {
Assert(!ParamTy->isMetadataTy(),
"Function has metadata parameter but isn't an intrinsic", I);
[IR] Add token types This introduces the basic functionality to support "token types". The motivation stems from the need to perform operations on a Value whose provenance cannot be obscured. There are several applications for such a type but my immediate motivation stems from WinEH. Our personality routine enforces a single-entry - single-exit regime for cleanups. After several rounds of optimizations, we may be left with a terminator whose "cleanup-entry block" is not entirely clear because control flow has merged two cleanups together. We have experimented with using labels as operands inside of instructions which are not terminators to indicate where we came from but found that LLVM does not expect such exotic uses of BasicBlocks. Instead, we can use this new type to clearly associate the "entry point" and "exit point" of our cleanup. This is done by having the cleanuppad yield a Token and consuming it at the cleanupret. The token type makes it impossible to obscure or otherwise hide the Value, making it trivial to track the relationship between the two points. What is the burden to the optimizer? Well, it turns out we have already paid down this cost by accepting that there are certain calls that we are not permitted to duplicate, optimizations have to watch out for such instructions anyway. There are additional places in the optimizer that we will probably have to update but early examination has given me the impression that this will not be heroic. Differential Revision: http://reviews.llvm.org/D11861 llvm-svn: 245029
2015-08-14 07:09:07 +02:00
Assert(!ParamTy->isTokenTy(),
"Function has token parameter but isn't an intrinsic", I);
}
}
[IR] Add token types This introduces the basic functionality to support "token types". The motivation stems from the need to perform operations on a Value whose provenance cannot be obscured. There are several applications for such a type but my immediate motivation stems from WinEH. Our personality routine enforces a single-entry - single-exit regime for cleanups. After several rounds of optimizations, we may be left with a terminator whose "cleanup-entry block" is not entirely clear because control flow has merged two cleanups together. We have experimented with using labels as operands inside of instructions which are not terminators to indicate where we came from but found that LLVM does not expect such exotic uses of BasicBlocks. Instead, we can use this new type to clearly associate the "entry point" and "exit point" of our cleanup. This is done by having the cleanuppad yield a Token and consuming it at the cleanupret. The token type makes it impossible to obscure or otherwise hide the Value, making it trivial to track the relationship between the two points. What is the burden to the optimizer? Well, it turns out we have already paid down this cost by accepting that there are certain calls that we are not permitted to duplicate, optimizations have to watch out for such instructions anyway. There are additional places in the optimizer that we will probably have to update but early examination has given me the impression that this will not be heroic. Differential Revision: http://reviews.llvm.org/D11861 llvm-svn: 245029
2015-08-14 07:09:07 +02:00
// Verify that indirect calls don't return tokens.
if (CS.getCalledFunction() == nullptr)
Assert(!FTy->getReturnType()->isTokenTy(),
"Return type cannot be token for indirect call!");
if (Function *F = CS.getCalledFunction())
if (Intrinsic::ID ID = (Intrinsic::ID)F->getIntrinsicID())
visitIntrinsicCallSite(ID, CS);
// Verify that a callsite has at most one "deopt", at most one "funclet" and
// at most one "gc-transition" operand bundle.
bool FoundDeoptBundle = false, FoundFuncletBundle = false,
FoundGCTransitionBundle = false;
for (unsigned i = 0, e = CS.getNumOperandBundles(); i < e; ++i) {
OperandBundleUse BU = CS.getOperandBundleAt(i);
uint32_t Tag = BU.getTagID();
if (Tag == LLVMContext::OB_deopt) {
Assert(!FoundDeoptBundle, "Multiple deopt operand bundles", I);
FoundDeoptBundle = true;
} else if (Tag == LLVMContext::OB_gc_transition) {
Assert(!FoundGCTransitionBundle, "Multiple gc-transition operand bundles",
I);
FoundGCTransitionBundle = true;
} else if (Tag == LLVMContext::OB_funclet) {
Assert(!FoundFuncletBundle, "Multiple funclet operand bundles", I);
FoundFuncletBundle = true;
Assert(BU.Inputs.size() == 1,
"Expected exactly one funclet bundle operand", I);
Assert(isa<FuncletPadInst>(BU.Inputs.front()),
"Funclet bundle operands should correspond to a FuncletPadInst",
I);
}
}
// Verify that each inlinable callsite of a debug-info-bearing function in a
// debug-info-bearing function has a debug location attached to it. Failure to
// do so causes assertion failures when the inliner sets up inline scope info.
if (I->getFunction()->getSubprogram() && CS.getCalledFunction() &&
CS.getCalledFunction()->getSubprogram())
AssertDI(I->getDebugLoc(), "inlinable function call in a function with "
"debug info must have a !dbg location",
I);
visitInstruction(*I);
}
/// Two types are "congruent" if they are identical, or if they are both pointer
/// types with different pointee types and the same address space.
static bool isTypeCongruent(Type *L, Type *R) {
if (L == R)
return true;
PointerType *PL = dyn_cast<PointerType>(L);
PointerType *PR = dyn_cast<PointerType>(R);
if (!PL || !PR)
return false;
return PL->getAddressSpace() == PR->getAddressSpace();
}
static AttrBuilder getParameterABIAttributes(int I, AttributeList Attrs) {
static const Attribute::AttrKind ABIAttrs[] = {
Attribute::StructRet, Attribute::ByVal, Attribute::InAlloca,
Attribute::InReg, Attribute::Returned, Attribute::SwiftSelf,
Attribute::SwiftError};
AttrBuilder Copy;
for (auto AK : ABIAttrs) {
if (Attrs.hasParamAttribute(I, AK))
Copy.addAttribute(AK);
}
if (Attrs.hasParamAttribute(I, Attribute::Alignment))
Copy.addAlignmentAttr(Attrs.getParamAlignment(I));
return Copy;
}
void Verifier::verifyMustTailCall(CallInst &CI) {
Assert(!CI.isInlineAsm(), "cannot use musttail call with inline asm", &CI);
// - The caller and callee prototypes must match. Pointer types of
// parameters or return types may differ in pointee type, but not
// address space.
Function *F = CI.getParent()->getParent();
FunctionType *CallerTy = F->getFunctionType();
FunctionType *CalleeTy = CI.getFunctionType();
if (!CI.getCalledFunction() || !CI.getCalledFunction()->isIntrinsic()) {
Assert(CallerTy->getNumParams() == CalleeTy->getNumParams(),
"cannot guarantee tail call due to mismatched parameter counts",
&CI);
for (int I = 0, E = CallerTy->getNumParams(); I != E; ++I) {
Assert(
isTypeCongruent(CallerTy->getParamType(I), CalleeTy->getParamType(I)),
"cannot guarantee tail call due to mismatched parameter types", &CI);
}
}
Assert(CallerTy->isVarArg() == CalleeTy->isVarArg(),
"cannot guarantee tail call due to mismatched varargs", &CI);
Assert(isTypeCongruent(CallerTy->getReturnType(), CalleeTy->getReturnType()),
"cannot guarantee tail call due to mismatched return types", &CI);
// - The calling conventions of the caller and callee must match.
Assert(F->getCallingConv() == CI.getCallingConv(),
"cannot guarantee tail call due to mismatched calling conv", &CI);
// - All ABI-impacting function attributes, such as sret, byval, inreg,
// returned, and inalloca, must match.
AttributeList CallerAttrs = F->getAttributes();
AttributeList CalleeAttrs = CI.getAttributes();
for (int I = 0, E = CallerTy->getNumParams(); I != E; ++I) {
AttrBuilder CallerABIAttrs = getParameterABIAttributes(I, CallerAttrs);
AttrBuilder CalleeABIAttrs = getParameterABIAttributes(I, CalleeAttrs);
Assert(CallerABIAttrs == CalleeABIAttrs,
"cannot guarantee tail call due to mismatched ABI impacting "
"function attributes",
&CI, CI.getOperand(I));
}
// - The call must immediately precede a :ref:`ret <i_ret>` instruction,
// or a pointer bitcast followed by a ret instruction.
// - The ret instruction must return the (possibly bitcasted) value
// produced by the call or void.
Value *RetVal = &CI;
Instruction *Next = CI.getNextNode();
// Handle the optional bitcast.
if (BitCastInst *BI = dyn_cast_or_null<BitCastInst>(Next)) {
Assert(BI->getOperand(0) == RetVal,
"bitcast following musttail call must use the call", BI);
RetVal = BI;
Next = BI->getNextNode();
}
// Check the return.
ReturnInst *Ret = dyn_cast_or_null<ReturnInst>(Next);
Assert(Ret, "musttail call must precede a ret with an optional bitcast",
&CI);
Assert(!Ret->getReturnValue() || Ret->getReturnValue() == RetVal,
"musttail call result must be returned", Ret);
}
void Verifier::visitCallInst(CallInst &CI) {
verifyCallSite(&CI);
if (CI.isMustTailCall())
verifyMustTailCall(CI);
}
void Verifier::visitInvokeInst(InvokeInst &II) {
verifyCallSite(&II);
// Verify that the first non-PHI instruction of the unwind destination is an
// exception handling instruction.
Assert(
II.getUnwindDest()->isEHPad(),
"The unwind destination does not have an exception handling instruction!",
&II);
visitTerminator(II);
}
/// visitUnaryOperator - Check the argument to the unary operator.
///
void Verifier::visitUnaryOperator(UnaryOperator &U) {
Assert(U.getType() == U.getOperand(0)->getType(),
"Unary operators must have same type for"
"operands and result!",
&U);
switch (U.getOpcode()) {
// Check that floating-point arithmetic operators are only used with
// floating-point operands.
case Instruction::FNeg:
Assert(U.getType()->isFPOrFPVectorTy(),
"FNeg operator only works with float types!", &U);
break;
default:
llvm_unreachable("Unknown UnaryOperator opcode!");
}
visitInstruction(U);
}
2004-03-02 01:22:19 +01:00
/// visitBinaryOperator - Check that both arguments to the binary operator are
/// of the same type!
///
void Verifier::visitBinaryOperator(BinaryOperator &B) {
Assert(B.getOperand(0)->getType() == B.getOperand(1)->getType(),
"Both operands to a binary operator are not of the same type!", &B);
switch (B.getOpcode()) {
// Check that integer arithmetic operators are only used with
// integral operands.
case Instruction::Add:
case Instruction::Sub:
case Instruction::Mul:
case Instruction::SDiv:
case Instruction::UDiv:
case Instruction::SRem:
case Instruction::URem:
Assert(B.getType()->isIntOrIntVectorTy(),
"Integer arithmetic operators only work with integral types!", &B);
Assert(B.getType() == B.getOperand(0)->getType(),
"Integer arithmetic operators must have same type "
"for operands and result!",
&B);
break;
// Check that floating-point arithmetic operators are only used with
// floating-point operands.
case Instruction::FAdd:
case Instruction::FSub:
case Instruction::FMul:
case Instruction::FDiv:
case Instruction::FRem:
Assert(B.getType()->isFPOrFPVectorTy(),
"Floating-point arithmetic operators only work with "
"floating-point types!",
&B);
Assert(B.getType() == B.getOperand(0)->getType(),
"Floating-point arithmetic operators must have same type "
"for operands and result!",
&B);
break;
// Check that logical operators are only used with integral operands.
case Instruction::And:
case Instruction::Or:
case Instruction::Xor:
Assert(B.getType()->isIntOrIntVectorTy(),
"Logical operators only work with integral types!", &B);
Assert(B.getType() == B.getOperand(0)->getType(),
"Logical operators must have same type for operands and result!",
&B);
break;
case Instruction::Shl:
case Instruction::LShr:
case Instruction::AShr:
Assert(B.getType()->isIntOrIntVectorTy(),
"Shifts only work with integral types!", &B);
Assert(B.getType() == B.getOperand(0)->getType(),
"Shift return type must be same as operands!", &B);
break;
default:
llvm_unreachable("Unknown BinaryOperator opcode!");
}
visitInstruction(B);
}
void Verifier::visitICmpInst(ICmpInst &IC) {
// Check that the operands are the same type
Type *Op0Ty = IC.getOperand(0)->getType();
Type *Op1Ty = IC.getOperand(1)->getType();
Assert(Op0Ty == Op1Ty,
"Both operands to ICmp instruction are not of the same type!", &IC);
// Check that the operands are the right type
Assert(Op0Ty->isIntOrIntVectorTy() || Op0Ty->isPtrOrPtrVectorTy(),
"Invalid operand types for ICmp instruction", &IC);
// Check that the predicate is valid.
Assert(IC.isIntPredicate(),
"Invalid predicate in ICmp instruction!", &IC);
visitInstruction(IC);
}
void Verifier::visitFCmpInst(FCmpInst &FC) {
// Check that the operands are the same type
Type *Op0Ty = FC.getOperand(0)->getType();
Type *Op1Ty = FC.getOperand(1)->getType();
Assert(Op0Ty == Op1Ty,
"Both operands to FCmp instruction are not of the same type!", &FC);
// Check that the operands are the right type
Assert(Op0Ty->isFPOrFPVectorTy(),
"Invalid operand types for FCmp instruction", &FC);
// Check that the predicate is valid.
Assert(FC.isFPPredicate(),
"Invalid predicate in FCmp instruction!", &FC);
visitInstruction(FC);
}
void Verifier::visitExtractElementInst(ExtractElementInst &EI) {
Assert(
ExtractElementInst::isValidOperands(EI.getOperand(0), EI.getOperand(1)),
"Invalid extractelement operands!", &EI);
visitInstruction(EI);
}
void Verifier::visitInsertElementInst(InsertElementInst &IE) {
Assert(InsertElementInst::isValidOperands(IE.getOperand(0), IE.getOperand(1),
IE.getOperand(2)),
"Invalid insertelement operands!", &IE);
visitInstruction(IE);
}
void Verifier::visitShuffleVectorInst(ShuffleVectorInst &SV) {
Assert(ShuffleVectorInst::isValidOperands(SV.getOperand(0), SV.getOperand(1),
SV.getOperand(2)),
"Invalid shufflevector operands!", &SV);
visitInstruction(SV);
}
void Verifier::visitGetElementPtrInst(GetElementPtrInst &GEP) {
Type *TargetTy = GEP.getPointerOperandType()->getScalarType();
Assert(isa<PointerType>(TargetTy),
"GEP base pointer is not a vector or a vector of pointers", &GEP);
Assert(GEP.getSourceElementType()->isSized(), "GEP into unsized type!", &GEP);
SmallVector<Value*, 16> Idxs(GEP.idx_begin(), GEP.idx_end());
Assert(all_of(
Idxs, [](Value* V) { return V->getType()->isIntOrIntVectorTy(); }),
"GEP indexes must be integers", &GEP);
Type *ElTy =
GetElementPtrInst::getIndexedType(GEP.getSourceElementType(), Idxs);
Assert(ElTy, "Invalid indices for GEP pointer type!", &GEP);
Assert(GEP.getType()->isPtrOrPtrVectorTy() &&
GEP.getResultElementType() == ElTy,
"GEP is not of right type for indices!", &GEP, ElTy);
if (GEP.getType()->isVectorTy()) {
// Additional checks for vector GEPs.
unsigned GEPWidth = GEP.getType()->getVectorNumElements();
if (GEP.getPointerOperandType()->isVectorTy())
Assert(GEPWidth == GEP.getPointerOperandType()->getVectorNumElements(),
"Vector GEP result width doesn't match operand's", &GEP);
2016-02-25 17:44:27 +01:00
for (Value *Idx : Idxs) {
Type *IndexTy = Idx->getType();
if (IndexTy->isVectorTy()) {
unsigned IndexWidth = IndexTy->getVectorNumElements();
Assert(IndexWidth == GEPWidth, "Invalid GEP index vector width", &GEP);
}
Assert(IndexTy->isIntOrIntVectorTy(),
"All GEP indices should be of integer type");
}
}
if (auto *PTy = dyn_cast<PointerType>(GEP.getType())) {
Assert(GEP.getAddressSpace() == PTy->getAddressSpace(),
"GEP address space doesn't match type", &GEP);
}
visitInstruction(GEP);
}
static bool isContiguous(const ConstantRange &A, const ConstantRange &B) {
return A.getUpper() == B.getLower() || A.getLower() == B.getUpper();
}
void Verifier::visitRangeMetadata(Instruction &I, MDNode *Range, Type *Ty) {
assert(Range && Range == I.getMetadata(LLVMContext::MD_range) &&
"precondition violation");
unsigned NumOperands = Range->getNumOperands();
Assert(NumOperands % 2 == 0, "Unfinished range!", Range);
unsigned NumRanges = NumOperands / 2;
Assert(NumRanges >= 1, "It should have at least one range!", Range);
ConstantRange LastRange(1); // Dummy initial value
for (unsigned i = 0; i < NumRanges; ++i) {
IR: Split Metadata from Value Split `Metadata` away from the `Value` class hierarchy, as part of PR21532. Assembly and bitcode changes are in the wings, but this is the bulk of the change for the IR C++ API. I have a follow-up patch prepared for `clang`. If this breaks other sub-projects, I apologize in advance :(. Help me compile it on Darwin I'll try to fix it. FWIW, the errors should be easy to fix, so it may be simpler to just fix it yourself. This breaks the build for all metadata-related code that's out-of-tree. Rest assured the transition is mechanical and the compiler should catch almost all of the problems. Here's a quick guide for updating your code: - `Metadata` is the root of a class hierarchy with three main classes: `MDNode`, `MDString`, and `ValueAsMetadata`. It is distinct from the `Value` class hierarchy. It is typeless -- i.e., instances do *not* have a `Type`. - `MDNode`'s operands are all `Metadata *` (instead of `Value *`). - `TrackingVH<MDNode>` and `WeakVH` referring to metadata can be replaced with `TrackingMDNodeRef` and `TrackingMDRef`, respectively. If you're referring solely to resolved `MDNode`s -- post graph construction -- just use `MDNode*`. - `MDNode` (and the rest of `Metadata`) have only limited support for `replaceAllUsesWith()`. As long as an `MDNode` is pointing at a forward declaration -- the result of `MDNode::getTemporary()` -- it maintains a side map of its uses and can RAUW itself. Once the forward declarations are fully resolved RAUW support is dropped on the ground. This means that uniquing collisions on changing operands cause nodes to become "distinct". (This already happened fairly commonly, whenever an operand went to null.) If you're constructing complex (non self-reference) `MDNode` cycles, you need to call `MDNode::resolveCycles()` on each node (or on a top-level node that somehow references all of the nodes). Also, don't do that. Metadata cycles (and the RAUW machinery needed to construct them) are expensive. - An `MDNode` can only refer to a `Constant` through a bridge called `ConstantAsMetadata` (one of the subclasses of `ValueAsMetadata`). As a side effect, accessing an operand of an `MDNode` that is known to be, e.g., `ConstantInt`, takes three steps: first, cast from `Metadata` to `ConstantAsMetadata`; second, extract the `Constant`; third, cast down to `ConstantInt`. The eventual goal is to introduce `MDInt`/`MDFloat`/etc. and have metadata schema owners transition away from using `Constant`s when the type isn't important (and they don't care about referring to `GlobalValue`s). In the meantime, I've added transitional API to the `mdconst` namespace that matches semantics with the old code, in order to avoid adding the error-prone three-step equivalent to every call site. If your old code was: MDNode *N = foo(); bar(isa <ConstantInt>(N->getOperand(0))); baz(cast <ConstantInt>(N->getOperand(1))); bak(cast_or_null <ConstantInt>(N->getOperand(2))); bat(dyn_cast <ConstantInt>(N->getOperand(3))); bay(dyn_cast_or_null<ConstantInt>(N->getOperand(4))); you can trivially match its semantics with: MDNode *N = foo(); bar(mdconst::hasa <ConstantInt>(N->getOperand(0))); baz(mdconst::extract <ConstantInt>(N->getOperand(1))); bak(mdconst::extract_or_null <ConstantInt>(N->getOperand(2))); bat(mdconst::dyn_extract <ConstantInt>(N->getOperand(3))); bay(mdconst::dyn_extract_or_null<ConstantInt>(N->getOperand(4))); and when you transition your metadata schema to `MDInt`: MDNode *N = foo(); bar(isa <MDInt>(N->getOperand(0))); baz(cast <MDInt>(N->getOperand(1))); bak(cast_or_null <MDInt>(N->getOperand(2))); bat(dyn_cast <MDInt>(N->getOperand(3))); bay(dyn_cast_or_null<MDInt>(N->getOperand(4))); - A `CallInst` -- specifically, intrinsic instructions -- can refer to metadata through a bridge called `MetadataAsValue`. This is a subclass of `Value` where `getType()->isMetadataTy()`. `MetadataAsValue` is the *only* class that can legally refer to a `LocalAsMetadata`, which is a bridged form of non-`Constant` values like `Argument` and `Instruction`. It can also refer to any other `Metadata` subclass. (I'll break all your testcases in a follow-up commit, when I propagate this change to assembly.) llvm-svn: 223802
2014-12-09 19:38:53 +01:00
ConstantInt *Low =
mdconst::dyn_extract<ConstantInt>(Range->getOperand(2 * i));
Assert(Low, "The lower limit must be an integer!", Low);
IR: Split Metadata from Value Split `Metadata` away from the `Value` class hierarchy, as part of PR21532. Assembly and bitcode changes are in the wings, but this is the bulk of the change for the IR C++ API. I have a follow-up patch prepared for `clang`. If this breaks other sub-projects, I apologize in advance :(. Help me compile it on Darwin I'll try to fix it. FWIW, the errors should be easy to fix, so it may be simpler to just fix it yourself. This breaks the build for all metadata-related code that's out-of-tree. Rest assured the transition is mechanical and the compiler should catch almost all of the problems. Here's a quick guide for updating your code: - `Metadata` is the root of a class hierarchy with three main classes: `MDNode`, `MDString`, and `ValueAsMetadata`. It is distinct from the `Value` class hierarchy. It is typeless -- i.e., instances do *not* have a `Type`. - `MDNode`'s operands are all `Metadata *` (instead of `Value *`). - `TrackingVH<MDNode>` and `WeakVH` referring to metadata can be replaced with `TrackingMDNodeRef` and `TrackingMDRef`, respectively. If you're referring solely to resolved `MDNode`s -- post graph construction -- just use `MDNode*`. - `MDNode` (and the rest of `Metadata`) have only limited support for `replaceAllUsesWith()`. As long as an `MDNode` is pointing at a forward declaration -- the result of `MDNode::getTemporary()` -- it maintains a side map of its uses and can RAUW itself. Once the forward declarations are fully resolved RAUW support is dropped on the ground. This means that uniquing collisions on changing operands cause nodes to become "distinct". (This already happened fairly commonly, whenever an operand went to null.) If you're constructing complex (non self-reference) `MDNode` cycles, you need to call `MDNode::resolveCycles()` on each node (or on a top-level node that somehow references all of the nodes). Also, don't do that. Metadata cycles (and the RAUW machinery needed to construct them) are expensive. - An `MDNode` can only refer to a `Constant` through a bridge called `ConstantAsMetadata` (one of the subclasses of `ValueAsMetadata`). As a side effect, accessing an operand of an `MDNode` that is known to be, e.g., `ConstantInt`, takes three steps: first, cast from `Metadata` to `ConstantAsMetadata`; second, extract the `Constant`; third, cast down to `ConstantInt`. The eventual goal is to introduce `MDInt`/`MDFloat`/etc. and have metadata schema owners transition away from using `Constant`s when the type isn't important (and they don't care about referring to `GlobalValue`s). In the meantime, I've added transitional API to the `mdconst` namespace that matches semantics with the old code, in order to avoid adding the error-prone three-step equivalent to every call site. If your old code was: MDNode *N = foo(); bar(isa <ConstantInt>(N->getOperand(0))); baz(cast <ConstantInt>(N->getOperand(1))); bak(cast_or_null <ConstantInt>(N->getOperand(2))); bat(dyn_cast <ConstantInt>(N->getOperand(3))); bay(dyn_cast_or_null<ConstantInt>(N->getOperand(4))); you can trivially match its semantics with: MDNode *N = foo(); bar(mdconst::hasa <ConstantInt>(N->getOperand(0))); baz(mdconst::extract <ConstantInt>(N->getOperand(1))); bak(mdconst::extract_or_null <ConstantInt>(N->getOperand(2))); bat(mdconst::dyn_extract <ConstantInt>(N->getOperand(3))); bay(mdconst::dyn_extract_or_null<ConstantInt>(N->getOperand(4))); and when you transition your metadata schema to `MDInt`: MDNode *N = foo(); bar(isa <MDInt>(N->getOperand(0))); baz(cast <MDInt>(N->getOperand(1))); bak(cast_or_null <MDInt>(N->getOperand(2))); bat(dyn_cast <MDInt>(N->getOperand(3))); bay(dyn_cast_or_null<MDInt>(N->getOperand(4))); - A `CallInst` -- specifically, intrinsic instructions -- can refer to metadata through a bridge called `MetadataAsValue`. This is a subclass of `Value` where `getType()->isMetadataTy()`. `MetadataAsValue` is the *only* class that can legally refer to a `LocalAsMetadata`, which is a bridged form of non-`Constant` values like `Argument` and `Instruction`. It can also refer to any other `Metadata` subclass. (I'll break all your testcases in a follow-up commit, when I propagate this change to assembly.) llvm-svn: 223802
2014-12-09 19:38:53 +01:00
ConstantInt *High =
mdconst::dyn_extract<ConstantInt>(Range->getOperand(2 * i + 1));
Assert(High, "The upper limit must be an integer!", High);
Assert(High->getType() == Low->getType() && High->getType() == Ty,
"Range types must match instruction type!", &I);
APInt HighV = High->getValue();
APInt LowV = Low->getValue();
ConstantRange CurRange(LowV, HighV);
Assert(!CurRange.isEmptySet() && !CurRange.isFullSet(),
"Range must not be empty!", Range);
if (i != 0) {
Assert(CurRange.intersectWith(LastRange).isEmptySet(),
"Intervals are overlapping", Range);
Assert(LowV.sgt(LastRange.getLower()), "Intervals are not in order",
Range);
Assert(!isContiguous(CurRange, LastRange), "Intervals are contiguous",
Range);
}
LastRange = ConstantRange(LowV, HighV);
}
if (NumRanges > 2) {
APInt FirstLow =
IR: Split Metadata from Value Split `Metadata` away from the `Value` class hierarchy, as part of PR21532. Assembly and bitcode changes are in the wings, but this is the bulk of the change for the IR C++ API. I have a follow-up patch prepared for `clang`. If this breaks other sub-projects, I apologize in advance :(. Help me compile it on Darwin I'll try to fix it. FWIW, the errors should be easy to fix, so it may be simpler to just fix it yourself. This breaks the build for all metadata-related code that's out-of-tree. Rest assured the transition is mechanical and the compiler should catch almost all of the problems. Here's a quick guide for updating your code: - `Metadata` is the root of a class hierarchy with three main classes: `MDNode`, `MDString`, and `ValueAsMetadata`. It is distinct from the `Value` class hierarchy. It is typeless -- i.e., instances do *not* have a `Type`. - `MDNode`'s operands are all `Metadata *` (instead of `Value *`). - `TrackingVH<MDNode>` and `WeakVH` referring to metadata can be replaced with `TrackingMDNodeRef` and `TrackingMDRef`, respectively. If you're referring solely to resolved `MDNode`s -- post graph construction -- just use `MDNode*`. - `MDNode` (and the rest of `Metadata`) have only limited support for `replaceAllUsesWith()`. As long as an `MDNode` is pointing at a forward declaration -- the result of `MDNode::getTemporary()` -- it maintains a side map of its uses and can RAUW itself. Once the forward declarations are fully resolved RAUW support is dropped on the ground. This means that uniquing collisions on changing operands cause nodes to become "distinct". (This already happened fairly commonly, whenever an operand went to null.) If you're constructing complex (non self-reference) `MDNode` cycles, you need to call `MDNode::resolveCycles()` on each node (or on a top-level node that somehow references all of the nodes). Also, don't do that. Metadata cycles (and the RAUW machinery needed to construct them) are expensive. - An `MDNode` can only refer to a `Constant` through a bridge called `ConstantAsMetadata` (one of the subclasses of `ValueAsMetadata`). As a side effect, accessing an operand of an `MDNode` that is known to be, e.g., `ConstantInt`, takes three steps: first, cast from `Metadata` to `ConstantAsMetadata`; second, extract the `Constant`; third, cast down to `ConstantInt`. The eventual goal is to introduce `MDInt`/`MDFloat`/etc. and have metadata schema owners transition away from using `Constant`s when the type isn't important (and they don't care about referring to `GlobalValue`s). In the meantime, I've added transitional API to the `mdconst` namespace that matches semantics with the old code, in order to avoid adding the error-prone three-step equivalent to every call site. If your old code was: MDNode *N = foo(); bar(isa <ConstantInt>(N->getOperand(0))); baz(cast <ConstantInt>(N->getOperand(1))); bak(cast_or_null <ConstantInt>(N->getOperand(2))); bat(dyn_cast <ConstantInt>(N->getOperand(3))); bay(dyn_cast_or_null<ConstantInt>(N->getOperand(4))); you can trivially match its semantics with: MDNode *N = foo(); bar(mdconst::hasa <ConstantInt>(N->getOperand(0))); baz(mdconst::extract <ConstantInt>(N->getOperand(1))); bak(mdconst::extract_or_null <ConstantInt>(N->getOperand(2))); bat(mdconst::dyn_extract <ConstantInt>(N->getOperand(3))); bay(mdconst::dyn_extract_or_null<ConstantInt>(N->getOperand(4))); and when you transition your metadata schema to `MDInt`: MDNode *N = foo(); bar(isa <MDInt>(N->getOperand(0))); baz(cast <MDInt>(N->getOperand(1))); bak(cast_or_null <MDInt>(N->getOperand(2))); bat(dyn_cast <MDInt>(N->getOperand(3))); bay(dyn_cast_or_null<MDInt>(N->getOperand(4))); - A `CallInst` -- specifically, intrinsic instructions -- can refer to metadata through a bridge called `MetadataAsValue`. This is a subclass of `Value` where `getType()->isMetadataTy()`. `MetadataAsValue` is the *only* class that can legally refer to a `LocalAsMetadata`, which is a bridged form of non-`Constant` values like `Argument` and `Instruction`. It can also refer to any other `Metadata` subclass. (I'll break all your testcases in a follow-up commit, when I propagate this change to assembly.) llvm-svn: 223802
2014-12-09 19:38:53 +01:00
mdconst::dyn_extract<ConstantInt>(Range->getOperand(0))->getValue();
APInt FirstHigh =
IR: Split Metadata from Value Split `Metadata` away from the `Value` class hierarchy, as part of PR21532. Assembly and bitcode changes are in the wings, but this is the bulk of the change for the IR C++ API. I have a follow-up patch prepared for `clang`. If this breaks other sub-projects, I apologize in advance :(. Help me compile it on Darwin I'll try to fix it. FWIW, the errors should be easy to fix, so it may be simpler to just fix it yourself. This breaks the build for all metadata-related code that's out-of-tree. Rest assured the transition is mechanical and the compiler should catch almost all of the problems. Here's a quick guide for updating your code: - `Metadata` is the root of a class hierarchy with three main classes: `MDNode`, `MDString`, and `ValueAsMetadata`. It is distinct from the `Value` class hierarchy. It is typeless -- i.e., instances do *not* have a `Type`. - `MDNode`'s operands are all `Metadata *` (instead of `Value *`). - `TrackingVH<MDNode>` and `WeakVH` referring to metadata can be replaced with `TrackingMDNodeRef` and `TrackingMDRef`, respectively. If you're referring solely to resolved `MDNode`s -- post graph construction -- just use `MDNode*`. - `MDNode` (and the rest of `Metadata`) have only limited support for `replaceAllUsesWith()`. As long as an `MDNode` is pointing at a forward declaration -- the result of `MDNode::getTemporary()` -- it maintains a side map of its uses and can RAUW itself. Once the forward declarations are fully resolved RAUW support is dropped on the ground. This means that uniquing collisions on changing operands cause nodes to become "distinct". (This already happened fairly commonly, whenever an operand went to null.) If you're constructing complex (non self-reference) `MDNode` cycles, you need to call `MDNode::resolveCycles()` on each node (or on a top-level node that somehow references all of the nodes). Also, don't do that. Metadata cycles (and the RAUW machinery needed to construct them) are expensive. - An `MDNode` can only refer to a `Constant` through a bridge called `ConstantAsMetadata` (one of the subclasses of `ValueAsMetadata`). As a side effect, accessing an operand of an `MDNode` that is known to be, e.g., `ConstantInt`, takes three steps: first, cast from `Metadata` to `ConstantAsMetadata`; second, extract the `Constant`; third, cast down to `ConstantInt`. The eventual goal is to introduce `MDInt`/`MDFloat`/etc. and have metadata schema owners transition away from using `Constant`s when the type isn't important (and they don't care about referring to `GlobalValue`s). In the meantime, I've added transitional API to the `mdconst` namespace that matches semantics with the old code, in order to avoid adding the error-prone three-step equivalent to every call site. If your old code was: MDNode *N = foo(); bar(isa <ConstantInt>(N->getOperand(0))); baz(cast <ConstantInt>(N->getOperand(1))); bak(cast_or_null <ConstantInt>(N->getOperand(2))); bat(dyn_cast <ConstantInt>(N->getOperand(3))); bay(dyn_cast_or_null<ConstantInt>(N->getOperand(4))); you can trivially match its semantics with: MDNode *N = foo(); bar(mdconst::hasa <ConstantInt>(N->getOperand(0))); baz(mdconst::extract <ConstantInt>(N->getOperand(1))); bak(mdconst::extract_or_null <ConstantInt>(N->getOperand(2))); bat(mdconst::dyn_extract <ConstantInt>(N->getOperand(3))); bay(mdconst::dyn_extract_or_null<ConstantInt>(N->getOperand(4))); and when you transition your metadata schema to `MDInt`: MDNode *N = foo(); bar(isa <MDInt>(N->getOperand(0))); baz(cast <MDInt>(N->getOperand(1))); bak(cast_or_null <MDInt>(N->getOperand(2))); bat(dyn_cast <MDInt>(N->getOperand(3))); bay(dyn_cast_or_null<MDInt>(N->getOperand(4))); - A `CallInst` -- specifically, intrinsic instructions -- can refer to metadata through a bridge called `MetadataAsValue`. This is a subclass of `Value` where `getType()->isMetadataTy()`. `MetadataAsValue` is the *only* class that can legally refer to a `LocalAsMetadata`, which is a bridged form of non-`Constant` values like `Argument` and `Instruction`. It can also refer to any other `Metadata` subclass. (I'll break all your testcases in a follow-up commit, when I propagate this change to assembly.) llvm-svn: 223802
2014-12-09 19:38:53 +01:00
mdconst::dyn_extract<ConstantInt>(Range->getOperand(1))->getValue();
ConstantRange FirstRange(FirstLow, FirstHigh);
Assert(FirstRange.intersectWith(LastRange).isEmptySet(),
"Intervals are overlapping", Range);
Assert(!isContiguous(FirstRange, LastRange), "Intervals are contiguous",
Range);
}
}
void Verifier::checkAtomicMemAccessSize(Type *Ty, const Instruction *I) {
unsigned Size = DL.getTypeSizeInBits(Ty);
Assert(Size >= 8, "atomic memory access' size must be byte-sized", Ty, I);
Assert(!(Size & (Size - 1)),
"atomic memory access' operand must have a power-of-two size", Ty, I);
}
void Verifier::visitLoadInst(LoadInst &LI) {
PointerType *PTy = dyn_cast<PointerType>(LI.getOperand(0)->getType());
Assert(PTy, "Load operand must be a pointer.", &LI);
Type *ElTy = LI.getType();
Assert(LI.getAlignment() <= Value::MaximumAlignment,
"huge alignment values are unsupported", &LI);
Assert(ElTy->isSized(), "loading unsized types is not allowed", &LI);
if (LI.isAtomic()) {
Assert(LI.getOrdering() != AtomicOrdering::Release &&
LI.getOrdering() != AtomicOrdering::AcquireRelease,
"Load cannot have Release ordering", &LI);
Assert(LI.getAlignment() != 0,
"Atomic load must specify explicit alignment", &LI);
Assert(ElTy->isIntOrPtrTy() || ElTy->isFloatingPointTy(),
"atomic load operand must have integer, pointer, or floating point "
"type!",
ElTy, &LI);
checkAtomicMemAccessSize(ElTy, &LI);
} else {
Assert(LI.getSyncScopeID() == SyncScope::System,
"Non-atomic load cannot have SynchronizationScope specified", &LI);
}
visitInstruction(LI);
}
void Verifier::visitStoreInst(StoreInst &SI) {
PointerType *PTy = dyn_cast<PointerType>(SI.getOperand(1)->getType());
Assert(PTy, "Store operand must be a pointer.", &SI);
Type *ElTy = PTy->getElementType();
Assert(ElTy == SI.getOperand(0)->getType(),
"Stored value type does not match pointer operand type!", &SI, ElTy);
Assert(SI.getAlignment() <= Value::MaximumAlignment,
"huge alignment values are unsupported", &SI);
Assert(ElTy->isSized(), "storing unsized types is not allowed", &SI);
if (SI.isAtomic()) {
Assert(SI.getOrdering() != AtomicOrdering::Acquire &&
SI.getOrdering() != AtomicOrdering::AcquireRelease,
"Store cannot have Acquire ordering", &SI);
Assert(SI.getAlignment() != 0,
"Atomic store must specify explicit alignment", &SI);
Assert(ElTy->isIntOrPtrTy() || ElTy->isFloatingPointTy(),
"atomic store operand must have integer, pointer, or floating point "
"type!",
ElTy, &SI);
checkAtomicMemAccessSize(ElTy, &SI);
} else {
Assert(SI.getSyncScopeID() == SyncScope::System,
"Non-atomic store cannot have SynchronizationScope specified", &SI);
}
visitInstruction(SI);
}
/// Check that SwiftErrorVal is used as a swifterror argument in CS.
void Verifier::verifySwiftErrorCallSite(CallSite CS,
const Value *SwiftErrorVal) {
unsigned Idx = 0;
for (CallSite::arg_iterator I = CS.arg_begin(), E = CS.arg_end();
I != E; ++I, ++Idx) {
if (*I == SwiftErrorVal) {
Assert(CS.paramHasAttr(Idx, Attribute::SwiftError),
"swifterror value when used in a callsite should be marked "
"with swifterror attribute",
SwiftErrorVal, CS);
}
}
}
void Verifier::verifySwiftErrorValue(const Value *SwiftErrorVal) {
// Check that swifterror value is only used by loads, stores, or as
// a swifterror argument.
for (const User *U : SwiftErrorVal->users()) {
Assert(isa<LoadInst>(U) || isa<StoreInst>(U) || isa<CallInst>(U) ||
isa<InvokeInst>(U),
"swifterror value can only be loaded and stored from, or "
"as a swifterror argument!",
SwiftErrorVal, U);
// If it is used by a store, check it is the second operand.
if (auto StoreI = dyn_cast<StoreInst>(U))
Assert(StoreI->getOperand(1) == SwiftErrorVal,
"swifterror value should be the second operand when used "
"by stores", SwiftErrorVal, U);
if (auto CallI = dyn_cast<CallInst>(U))
verifySwiftErrorCallSite(const_cast<CallInst*>(CallI), SwiftErrorVal);
if (auto II = dyn_cast<InvokeInst>(U))
verifySwiftErrorCallSite(const_cast<InvokeInst*>(II), SwiftErrorVal);
}
}
void Verifier::visitAllocaInst(AllocaInst &AI) {
SmallPtrSet<Type*, 4> Visited;
PointerType *PTy = AI.getType();
// TODO: Relax this restriction?
Assert(PTy->getAddressSpace() == DL.getAllocaAddrSpace(),
"Allocation instruction pointer not in the stack address space!",
&AI);
Assert(AI.getAllocatedType()->isSized(&Visited),
"Cannot allocate unsized type", &AI);
Assert(AI.getArraySize()->getType()->isIntegerTy(),
"Alloca array size must have integer type", &AI);
Assert(AI.getAlignment() <= Value::MaximumAlignment,
"huge alignment values are unsupported", &AI);
if (AI.isSwiftError()) {
verifySwiftErrorValue(&AI);
}
visitInstruction(AI);
}
void Verifier::visitAtomicCmpXchgInst(AtomicCmpXchgInst &CXI) {
// FIXME: more conditions???
Assert(CXI.getSuccessOrdering() != AtomicOrdering::NotAtomic,
"cmpxchg instructions must be atomic.", &CXI);
Assert(CXI.getFailureOrdering() != AtomicOrdering::NotAtomic,
"cmpxchg instructions must be atomic.", &CXI);
Assert(CXI.getSuccessOrdering() != AtomicOrdering::Unordered,
"cmpxchg instructions cannot be unordered.", &CXI);
Assert(CXI.getFailureOrdering() != AtomicOrdering::Unordered,
"cmpxchg instructions cannot be unordered.", &CXI);
Assert(!isStrongerThan(CXI.getFailureOrdering(), CXI.getSuccessOrdering()),
"cmpxchg instructions failure argument shall be no stronger than the "
"success argument",
&CXI);
Assert(CXI.getFailureOrdering() != AtomicOrdering::Release &&
CXI.getFailureOrdering() != AtomicOrdering::AcquireRelease,
"cmpxchg failure ordering cannot include release semantics", &CXI);
PointerType *PTy = dyn_cast<PointerType>(CXI.getOperand(0)->getType());
Assert(PTy, "First cmpxchg operand must be a pointer.", &CXI);
Type *ElTy = PTy->getElementType();
Assert(ElTy->isIntOrPtrTy(),
"cmpxchg operand must have integer or pointer type", ElTy, &CXI);
checkAtomicMemAccessSize(ElTy, &CXI);
Assert(ElTy == CXI.getOperand(1)->getType(),
"Expected value type does not match pointer operand type!", &CXI,
ElTy);
Assert(ElTy == CXI.getOperand(2)->getType(),
"Stored value type does not match pointer operand type!", &CXI, ElTy);
visitInstruction(CXI);
}
void Verifier::visitAtomicRMWInst(AtomicRMWInst &RMWI) {
Assert(RMWI.getOrdering() != AtomicOrdering::NotAtomic,
"atomicrmw instructions must be atomic.", &RMWI);
Assert(RMWI.getOrdering() != AtomicOrdering::Unordered,
"atomicrmw instructions cannot be unordered.", &RMWI);
auto Op = RMWI.getOperation();
PointerType *PTy = dyn_cast<PointerType>(RMWI.getOperand(0)->getType());
Assert(PTy, "First atomicrmw operand must be a pointer.", &RMWI);
Type *ElTy = PTy->getElementType();
Assert(ElTy->isIntegerTy(), "atomicrmw " +
AtomicRMWInst::getOperationName(Op) +
" operand must have integer type!",
&RMWI, ElTy);
checkAtomicMemAccessSize(ElTy, &RMWI);
Assert(ElTy == RMWI.getOperand(1)->getType(),
"Argument value type does not match pointer operand type!", &RMWI,
ElTy);
Assert(AtomicRMWInst::FIRST_BINOP <= Op && Op <= AtomicRMWInst::LAST_BINOP,
"Invalid binary operation!", &RMWI);
visitInstruction(RMWI);
}
void Verifier::visitFenceInst(FenceInst &FI) {
const AtomicOrdering Ordering = FI.getOrdering();
Assert(Ordering == AtomicOrdering::Acquire ||
Ordering == AtomicOrdering::Release ||
Ordering == AtomicOrdering::AcquireRelease ||
Ordering == AtomicOrdering::SequentiallyConsistent,
"fence instructions may only have acquire, release, acq_rel, or "
"seq_cst ordering.",
&FI);
visitInstruction(FI);
}
void Verifier::visitExtractValueInst(ExtractValueInst &EVI) {
Assert(ExtractValueInst::getIndexedType(EVI.getAggregateOperand()->getType(),
EVI.getIndices()) == EVI.getType(),
"Invalid ExtractValueInst operands!", &EVI);
visitInstruction(EVI);
}
void Verifier::visitInsertValueInst(InsertValueInst &IVI) {
Assert(ExtractValueInst::getIndexedType(IVI.getAggregateOperand()->getType(),
IVI.getIndices()) ==
IVI.getOperand(1)->getType(),
"Invalid InsertValueInst operands!", &IVI);
visitInstruction(IVI);
}
static Value *getParentPad(Value *EHPad) {
if (auto *FPI = dyn_cast<FuncletPadInst>(EHPad))
return FPI->getParentPad();
return cast<CatchSwitchInst>(EHPad)->getParentPad();
}
void Verifier::visitEHPadPredecessors(Instruction &I) {
assert(I.isEHPad());
BasicBlock *BB = I.getParent();
Function *F = BB->getParent();
Assert(BB != &F->getEntryBlock(), "EH pad cannot be in entry block.", &I);
if (auto *LPI = dyn_cast<LandingPadInst>(&I)) {
// The landingpad instruction defines its parent as a landing pad block. The
// landing pad block may be branched to only by the unwind edge of an
// invoke.
for (BasicBlock *PredBB : predecessors(BB)) {
const auto *II = dyn_cast<InvokeInst>(PredBB->getTerminator());
Assert(II && II->getUnwindDest() == BB && II->getNormalDest() != BB,
"Block containing LandingPadInst must be jumped to "
"only by the unwind edge of an invoke.",
LPI);
}
return;
}
[IR] Reformulate LLVM's EH funclet IR While we have successfully implemented a funclet-oriented EH scheme on top of LLVM IR, our scheme has some notable deficiencies: - catchendpad and cleanupendpad are necessary in the current design but they are difficult to explain to others, even to seasoned LLVM experts. - catchendpad and cleanupendpad are optimization barriers. They cannot be split and force all potentially throwing call-sites to be invokes. This has a noticable effect on the quality of our code generation. - catchpad, while similar in some aspects to invoke, is fairly awkward. It is unsplittable, starts a funclet, and has control flow to other funclets. - The nesting relationship between funclets is currently a property of control flow edges. Because of this, we are forced to carefully analyze the flow graph to see if there might potentially exist illegal nesting among funclets. While we have logic to clone funclets when they are illegally nested, it would be nicer if we had a representation which forbade them upfront. Let's clean this up a bit by doing the following: - Instead, make catchpad more like cleanuppad and landingpad: no control flow, just a bunch of simple operands; catchpad would be splittable. - Introduce catchswitch, a control flow instruction designed to model the constraints of funclet oriented EH. - Make funclet scoping explicit by having funclet instructions consume the token produced by the funclet which contains them. - Remove catchendpad and cleanupendpad. Their presence can be inferred implicitly using coloring information. N.B. The state numbering code for the CLR has been updated but the veracity of it's output cannot be spoken for. An expert should take a look to make sure the results are reasonable. Reviewers: rnk, JosephTremoulet, andrew.w.kaylor Differential Revision: http://reviews.llvm.org/D15139 llvm-svn: 255422
2015-12-12 06:38:55 +01:00
if (auto *CPI = dyn_cast<CatchPadInst>(&I)) {
if (!pred_empty(BB))
Assert(BB->getUniquePredecessor() == CPI->getCatchSwitch()->getParent(),
"Block containg CatchPadInst must be jumped to "
"only by its catchswitch.",
CPI);
Assert(BB != CPI->getCatchSwitch()->getUnwindDest(),
"Catchswitch cannot unwind to one of its catchpads",
CPI->getCatchSwitch(), CPI);
[IR] Reformulate LLVM's EH funclet IR While we have successfully implemented a funclet-oriented EH scheme on top of LLVM IR, our scheme has some notable deficiencies: - catchendpad and cleanupendpad are necessary in the current design but they are difficult to explain to others, even to seasoned LLVM experts. - catchendpad and cleanupendpad are optimization barriers. They cannot be split and force all potentially throwing call-sites to be invokes. This has a noticable effect on the quality of our code generation. - catchpad, while similar in some aspects to invoke, is fairly awkward. It is unsplittable, starts a funclet, and has control flow to other funclets. - The nesting relationship between funclets is currently a property of control flow edges. Because of this, we are forced to carefully analyze the flow graph to see if there might potentially exist illegal nesting among funclets. While we have logic to clone funclets when they are illegally nested, it would be nicer if we had a representation which forbade them upfront. Let's clean this up a bit by doing the following: - Instead, make catchpad more like cleanuppad and landingpad: no control flow, just a bunch of simple operands; catchpad would be splittable. - Introduce catchswitch, a control flow instruction designed to model the constraints of funclet oriented EH. - Make funclet scoping explicit by having funclet instructions consume the token produced by the funclet which contains them. - Remove catchendpad and cleanupendpad. Their presence can be inferred implicitly using coloring information. N.B. The state numbering code for the CLR has been updated but the veracity of it's output cannot be spoken for. An expert should take a look to make sure the results are reasonable. Reviewers: rnk, JosephTremoulet, andrew.w.kaylor Differential Revision: http://reviews.llvm.org/D15139 llvm-svn: 255422
2015-12-12 06:38:55 +01:00
return;
}
// Verify that each pred has a legal terminator with a legal to/from EH
// pad relationship.
Instruction *ToPad = &I;
Value *ToPadParent = getParentPad(ToPad);
for (BasicBlock *PredBB : predecessors(BB)) {
Instruction *TI = PredBB->getTerminator();
Value *FromPad;
[IR] Reformulate LLVM's EH funclet IR While we have successfully implemented a funclet-oriented EH scheme on top of LLVM IR, our scheme has some notable deficiencies: - catchendpad and cleanupendpad are necessary in the current design but they are difficult to explain to others, even to seasoned LLVM experts. - catchendpad and cleanupendpad are optimization barriers. They cannot be split and force all potentially throwing call-sites to be invokes. This has a noticable effect on the quality of our code generation. - catchpad, while similar in some aspects to invoke, is fairly awkward. It is unsplittable, starts a funclet, and has control flow to other funclets. - The nesting relationship between funclets is currently a property of control flow edges. Because of this, we are forced to carefully analyze the flow graph to see if there might potentially exist illegal nesting among funclets. While we have logic to clone funclets when they are illegally nested, it would be nicer if we had a representation which forbade them upfront. Let's clean this up a bit by doing the following: - Instead, make catchpad more like cleanuppad and landingpad: no control flow, just a bunch of simple operands; catchpad would be splittable. - Introduce catchswitch, a control flow instruction designed to model the constraints of funclet oriented EH. - Make funclet scoping explicit by having funclet instructions consume the token produced by the funclet which contains them. - Remove catchendpad and cleanupendpad. Their presence can be inferred implicitly using coloring information. N.B. The state numbering code for the CLR has been updated but the veracity of it's output cannot be spoken for. An expert should take a look to make sure the results are reasonable. Reviewers: rnk, JosephTremoulet, andrew.w.kaylor Differential Revision: http://reviews.llvm.org/D15139 llvm-svn: 255422
2015-12-12 06:38:55 +01:00
if (auto *II = dyn_cast<InvokeInst>(TI)) {
Assert(II->getUnwindDest() == BB && II->getNormalDest() != BB,
"EH pad must be jumped to via an unwind edge", ToPad, II);
if (auto Bundle = II->getOperandBundle(LLVMContext::OB_funclet))
FromPad = Bundle->Inputs[0];
else
FromPad = ConstantTokenNone::get(II->getContext());
} else if (auto *CRI = dyn_cast<CleanupReturnInst>(TI)) {
FromPad = CRI->getOperand(0);
Assert(FromPad != ToPadParent, "A cleanupret must exit its cleanup", CRI);
} else if (auto *CSI = dyn_cast<CatchSwitchInst>(TI)) {
FromPad = CSI;
} else {
Assert(false, "EH pad must be jumped to via an unwind edge", ToPad, TI);
}
// The edge may exit from zero or more nested pads.
SmallSet<Value *, 8> Seen;
for (;; FromPad = getParentPad(FromPad)) {
Assert(FromPad != ToPad,
"EH pad cannot handle exceptions raised within it", FromPad, TI);
if (FromPad == ToPadParent) {
// This is a legal unwind edge.
break;
}
Assert(!isa<ConstantTokenNone>(FromPad),
"A single unwind edge may only enter one EH pad", TI);
Assert(Seen.insert(FromPad).second,
"EH pad jumps through a cycle of pads", FromPad);
[IR] Reformulate LLVM's EH funclet IR While we have successfully implemented a funclet-oriented EH scheme on top of LLVM IR, our scheme has some notable deficiencies: - catchendpad and cleanupendpad are necessary in the current design but they are difficult to explain to others, even to seasoned LLVM experts. - catchendpad and cleanupendpad are optimization barriers. They cannot be split and force all potentially throwing call-sites to be invokes. This has a noticable effect on the quality of our code generation. - catchpad, while similar in some aspects to invoke, is fairly awkward. It is unsplittable, starts a funclet, and has control flow to other funclets. - The nesting relationship between funclets is currently a property of control flow edges. Because of this, we are forced to carefully analyze the flow graph to see if there might potentially exist illegal nesting among funclets. While we have logic to clone funclets when they are illegally nested, it would be nicer if we had a representation which forbade them upfront. Let's clean this up a bit by doing the following: - Instead, make catchpad more like cleanuppad and landingpad: no control flow, just a bunch of simple operands; catchpad would be splittable. - Introduce catchswitch, a control flow instruction designed to model the constraints of funclet oriented EH. - Make funclet scoping explicit by having funclet instructions consume the token produced by the funclet which contains them. - Remove catchendpad and cleanupendpad. Their presence can be inferred implicitly using coloring information. N.B. The state numbering code for the CLR has been updated but the veracity of it's output cannot be spoken for. An expert should take a look to make sure the results are reasonable. Reviewers: rnk, JosephTremoulet, andrew.w.kaylor Differential Revision: http://reviews.llvm.org/D15139 llvm-svn: 255422
2015-12-12 06:38:55 +01:00
}
}
}
void Verifier::visitLandingPadInst(LandingPadInst &LPI) {
// The landingpad instruction is ill-formed if it doesn't have any clauses and
// isn't a cleanup.
Assert(LPI.getNumClauses() > 0 || LPI.isCleanup(),
"LandingPadInst needs at least one clause or to be a cleanup.", &LPI);
visitEHPadPredecessors(LPI);
if (!LandingPadResultTy)
LandingPadResultTy = LPI.getType();
else
Assert(LandingPadResultTy == LPI.getType(),
"The landingpad instruction should have a consistent result type "
"inside a function.",
&LPI);
Function *F = LPI.getParent()->getParent();
Assert(F->hasPersonalityFn(),
"LandingPadInst needs to be in a function with a personality.", &LPI);
// The landingpad instruction must be the first non-PHI instruction in the
// block.
Assert(LPI.getParent()->getLandingPadInst() == &LPI,
"LandingPadInst not the first non-PHI instruction in the block.",
&LPI);
for (unsigned i = 0, e = LPI.getNumClauses(); i < e; ++i) {
Constant *Clause = LPI.getClause(i);
if (LPI.isCatch(i)) {
Assert(isa<PointerType>(Clause->getType()),
"Catch operand does not have pointer type!", &LPI);
} else {
Assert(LPI.isFilter(i), "Clause is neither catch nor filter!", &LPI);
Assert(isa<ConstantArray>(Clause) || isa<ConstantAggregateZero>(Clause),
"Filter operand is not an array of constants!", &LPI);
}
}
visitInstruction(LPI);
}
void Verifier::visitResumeInst(ResumeInst &RI) {
Assert(RI.getFunction()->hasPersonalityFn(),
"ResumeInst needs to be in a function with a personality.", &RI);
if (!LandingPadResultTy)
LandingPadResultTy = RI.getValue()->getType();
else
Assert(LandingPadResultTy == RI.getValue()->getType(),
"The resume instruction should have a consistent result type "
"inside a function.",
&RI);
visitTerminator(RI);
}
void Verifier::visitCatchPadInst(CatchPadInst &CPI) {
BasicBlock *BB = CPI.getParent();
[IR] Reformulate LLVM's EH funclet IR While we have successfully implemented a funclet-oriented EH scheme on top of LLVM IR, our scheme has some notable deficiencies: - catchendpad and cleanupendpad are necessary in the current design but they are difficult to explain to others, even to seasoned LLVM experts. - catchendpad and cleanupendpad are optimization barriers. They cannot be split and force all potentially throwing call-sites to be invokes. This has a noticable effect on the quality of our code generation. - catchpad, while similar in some aspects to invoke, is fairly awkward. It is unsplittable, starts a funclet, and has control flow to other funclets. - The nesting relationship between funclets is currently a property of control flow edges. Because of this, we are forced to carefully analyze the flow graph to see if there might potentially exist illegal nesting among funclets. While we have logic to clone funclets when they are illegally nested, it would be nicer if we had a representation which forbade them upfront. Let's clean this up a bit by doing the following: - Instead, make catchpad more like cleanuppad and landingpad: no control flow, just a bunch of simple operands; catchpad would be splittable. - Introduce catchswitch, a control flow instruction designed to model the constraints of funclet oriented EH. - Make funclet scoping explicit by having funclet instructions consume the token produced by the funclet which contains them. - Remove catchendpad and cleanupendpad. Their presence can be inferred implicitly using coloring information. N.B. The state numbering code for the CLR has been updated but the veracity of it's output cannot be spoken for. An expert should take a look to make sure the results are reasonable. Reviewers: rnk, JosephTremoulet, andrew.w.kaylor Differential Revision: http://reviews.llvm.org/D15139 llvm-svn: 255422
2015-12-12 06:38:55 +01:00
Function *F = BB->getParent();
Assert(F->hasPersonalityFn(),
"CatchPadInst needs to be in a function with a personality.", &CPI);
[IR] Reformulate LLVM's EH funclet IR While we have successfully implemented a funclet-oriented EH scheme on top of LLVM IR, our scheme has some notable deficiencies: - catchendpad and cleanupendpad are necessary in the current design but they are difficult to explain to others, even to seasoned LLVM experts. - catchendpad and cleanupendpad are optimization barriers. They cannot be split and force all potentially throwing call-sites to be invokes. This has a noticable effect on the quality of our code generation. - catchpad, while similar in some aspects to invoke, is fairly awkward. It is unsplittable, starts a funclet, and has control flow to other funclets. - The nesting relationship between funclets is currently a property of control flow edges. Because of this, we are forced to carefully analyze the flow graph to see if there might potentially exist illegal nesting among funclets. While we have logic to clone funclets when they are illegally nested, it would be nicer if we had a representation which forbade them upfront. Let's clean this up a bit by doing the following: - Instead, make catchpad more like cleanuppad and landingpad: no control flow, just a bunch of simple operands; catchpad would be splittable. - Introduce catchswitch, a control flow instruction designed to model the constraints of funclet oriented EH. - Make funclet scoping explicit by having funclet instructions consume the token produced by the funclet which contains them. - Remove catchendpad and cleanupendpad. Their presence can be inferred implicitly using coloring information. N.B. The state numbering code for the CLR has been updated but the veracity of it's output cannot be spoken for. An expert should take a look to make sure the results are reasonable. Reviewers: rnk, JosephTremoulet, andrew.w.kaylor Differential Revision: http://reviews.llvm.org/D15139 llvm-svn: 255422
2015-12-12 06:38:55 +01:00
Assert(isa<CatchSwitchInst>(CPI.getParentPad()),
"CatchPadInst needs to be directly nested in a CatchSwitchInst.",
CPI.getParentPad());
// The catchpad instruction must be the first non-PHI instruction in the
// block.
Assert(BB->getFirstNonPHI() == &CPI,
[IR] Reformulate LLVM's EH funclet IR While we have successfully implemented a funclet-oriented EH scheme on top of LLVM IR, our scheme has some notable deficiencies: - catchendpad and cleanupendpad are necessary in the current design but they are difficult to explain to others, even to seasoned LLVM experts. - catchendpad and cleanupendpad are optimization barriers. They cannot be split and force all potentially throwing call-sites to be invokes. This has a noticable effect on the quality of our code generation. - catchpad, while similar in some aspects to invoke, is fairly awkward. It is unsplittable, starts a funclet, and has control flow to other funclets. - The nesting relationship between funclets is currently a property of control flow edges. Because of this, we are forced to carefully analyze the flow graph to see if there might potentially exist illegal nesting among funclets. While we have logic to clone funclets when they are illegally nested, it would be nicer if we had a representation which forbade them upfront. Let's clean this up a bit by doing the following: - Instead, make catchpad more like cleanuppad and landingpad: no control flow, just a bunch of simple operands; catchpad would be splittable. - Introduce catchswitch, a control flow instruction designed to model the constraints of funclet oriented EH. - Make funclet scoping explicit by having funclet instructions consume the token produced by the funclet which contains them. - Remove catchendpad and cleanupendpad. Their presence can be inferred implicitly using coloring information. N.B. The state numbering code for the CLR has been updated but the veracity of it's output cannot be spoken for. An expert should take a look to make sure the results are reasonable. Reviewers: rnk, JosephTremoulet, andrew.w.kaylor Differential Revision: http://reviews.llvm.org/D15139 llvm-svn: 255422
2015-12-12 06:38:55 +01:00
"CatchPadInst not the first non-PHI instruction in the block.", &CPI);
visitEHPadPredecessors(CPI);
visitFuncletPadInst(CPI);
}
[IR] Reformulate LLVM's EH funclet IR While we have successfully implemented a funclet-oriented EH scheme on top of LLVM IR, our scheme has some notable deficiencies: - catchendpad and cleanupendpad are necessary in the current design but they are difficult to explain to others, even to seasoned LLVM experts. - catchendpad and cleanupendpad are optimization barriers. They cannot be split and force all potentially throwing call-sites to be invokes. This has a noticable effect on the quality of our code generation. - catchpad, while similar in some aspects to invoke, is fairly awkward. It is unsplittable, starts a funclet, and has control flow to other funclets. - The nesting relationship between funclets is currently a property of control flow edges. Because of this, we are forced to carefully analyze the flow graph to see if there might potentially exist illegal nesting among funclets. While we have logic to clone funclets when they are illegally nested, it would be nicer if we had a representation which forbade them upfront. Let's clean this up a bit by doing the following: - Instead, make catchpad more like cleanuppad and landingpad: no control flow, just a bunch of simple operands; catchpad would be splittable. - Introduce catchswitch, a control flow instruction designed to model the constraints of funclet oriented EH. - Make funclet scoping explicit by having funclet instructions consume the token produced by the funclet which contains them. - Remove catchendpad and cleanupendpad. Their presence can be inferred implicitly using coloring information. N.B. The state numbering code for the CLR has been updated but the veracity of it's output cannot be spoken for. An expert should take a look to make sure the results are reasonable. Reviewers: rnk, JosephTremoulet, andrew.w.kaylor Differential Revision: http://reviews.llvm.org/D15139 llvm-svn: 255422
2015-12-12 06:38:55 +01:00
void Verifier::visitCatchReturnInst(CatchReturnInst &CatchReturn) {
Assert(isa<CatchPadInst>(CatchReturn.getOperand(0)),
"CatchReturnInst needs to be provided a CatchPad", &CatchReturn,
CatchReturn.getOperand(0));
visitTerminator(CatchReturn);
}
void Verifier::visitCleanupPadInst(CleanupPadInst &CPI) {
BasicBlock *BB = CPI.getParent();
Function *F = BB->getParent();
Assert(F->hasPersonalityFn(),
"CleanupPadInst needs to be in a function with a personality.", &CPI);
// The cleanuppad instruction must be the first non-PHI instruction in the
// block.
Assert(BB->getFirstNonPHI() == &CPI,
"CleanupPadInst not the first non-PHI instruction in the block.",
&CPI);
[IR] Reformulate LLVM's EH funclet IR While we have successfully implemented a funclet-oriented EH scheme on top of LLVM IR, our scheme has some notable deficiencies: - catchendpad and cleanupendpad are necessary in the current design but they are difficult to explain to others, even to seasoned LLVM experts. - catchendpad and cleanupendpad are optimization barriers. They cannot be split and force all potentially throwing call-sites to be invokes. This has a noticable effect on the quality of our code generation. - catchpad, while similar in some aspects to invoke, is fairly awkward. It is unsplittable, starts a funclet, and has control flow to other funclets. - The nesting relationship between funclets is currently a property of control flow edges. Because of this, we are forced to carefully analyze the flow graph to see if there might potentially exist illegal nesting among funclets. While we have logic to clone funclets when they are illegally nested, it would be nicer if we had a representation which forbade them upfront. Let's clean this up a bit by doing the following: - Instead, make catchpad more like cleanuppad and landingpad: no control flow, just a bunch of simple operands; catchpad would be splittable. - Introduce catchswitch, a control flow instruction designed to model the constraints of funclet oriented EH. - Make funclet scoping explicit by having funclet instructions consume the token produced by the funclet which contains them. - Remove catchendpad and cleanupendpad. Their presence can be inferred implicitly using coloring information. N.B. The state numbering code for the CLR has been updated but the veracity of it's output cannot be spoken for. An expert should take a look to make sure the results are reasonable. Reviewers: rnk, JosephTremoulet, andrew.w.kaylor Differential Revision: http://reviews.llvm.org/D15139 llvm-svn: 255422
2015-12-12 06:38:55 +01:00
auto *ParentPad = CPI.getParentPad();
Assert(isa<ConstantTokenNone>(ParentPad) || isa<FuncletPadInst>(ParentPad),
[IR] Reformulate LLVM's EH funclet IR While we have successfully implemented a funclet-oriented EH scheme on top of LLVM IR, our scheme has some notable deficiencies: - catchendpad and cleanupendpad are necessary in the current design but they are difficult to explain to others, even to seasoned LLVM experts. - catchendpad and cleanupendpad are optimization barriers. They cannot be split and force all potentially throwing call-sites to be invokes. This has a noticable effect on the quality of our code generation. - catchpad, while similar in some aspects to invoke, is fairly awkward. It is unsplittable, starts a funclet, and has control flow to other funclets. - The nesting relationship between funclets is currently a property of control flow edges. Because of this, we are forced to carefully analyze the flow graph to see if there might potentially exist illegal nesting among funclets. While we have logic to clone funclets when they are illegally nested, it would be nicer if we had a representation which forbade them upfront. Let's clean this up a bit by doing the following: - Instead, make catchpad more like cleanuppad and landingpad: no control flow, just a bunch of simple operands; catchpad would be splittable. - Introduce catchswitch, a control flow instruction designed to model the constraints of funclet oriented EH. - Make funclet scoping explicit by having funclet instructions consume the token produced by the funclet which contains them. - Remove catchendpad and cleanupendpad. Their presence can be inferred implicitly using coloring information. N.B. The state numbering code for the CLR has been updated but the veracity of it's output cannot be spoken for. An expert should take a look to make sure the results are reasonable. Reviewers: rnk, JosephTremoulet, andrew.w.kaylor Differential Revision: http://reviews.llvm.org/D15139 llvm-svn: 255422
2015-12-12 06:38:55 +01:00
"CleanupPadInst has an invalid parent.", &CPI);
visitEHPadPredecessors(CPI);
visitFuncletPadInst(CPI);
}
void Verifier::visitFuncletPadInst(FuncletPadInst &FPI) {
User *FirstUser = nullptr;
Value *FirstUnwindPad = nullptr;
SmallVector<FuncletPadInst *, 8> Worklist({&FPI});
SmallSet<FuncletPadInst *, 8> Seen;
while (!Worklist.empty()) {
FuncletPadInst *CurrentPad = Worklist.pop_back_val();
Assert(Seen.insert(CurrentPad).second,
"FuncletPadInst must not be nested within itself", CurrentPad);
Value *UnresolvedAncestorPad = nullptr;
for (User *U : CurrentPad->users()) {
BasicBlock *UnwindDest;
if (auto *CRI = dyn_cast<CleanupReturnInst>(U)) {
UnwindDest = CRI->getUnwindDest();
} else if (auto *CSI = dyn_cast<CatchSwitchInst>(U)) {
// We allow catchswitch unwind to caller to nest
// within an outer pad that unwinds somewhere else,
// because catchswitch doesn't have a nounwind variant.
// See e.g. SimplifyCFGOpt::SimplifyUnreachable.
if (CSI->unwindsToCaller())
continue;
UnwindDest = CSI->getUnwindDest();
} else if (auto *II = dyn_cast<InvokeInst>(U)) {
UnwindDest = II->getUnwindDest();
} else if (isa<CallInst>(U)) {
// Calls which don't unwind may be found inside funclet
// pads that unwind somewhere else. We don't *require*
// such calls to be annotated nounwind.
continue;
} else if (auto *CPI = dyn_cast<CleanupPadInst>(U)) {
// The unwind dest for a cleanup can only be found by
// recursive search. Add it to the worklist, and we'll
// search for its first use that determines where it unwinds.
Worklist.push_back(CPI);
continue;
} else {
Assert(isa<CatchReturnInst>(U), "Bogus funclet pad use", U);
continue;
}
Value *UnwindPad;
bool ExitsFPI;
if (UnwindDest) {
UnwindPad = UnwindDest->getFirstNonPHI();
if (!cast<Instruction>(UnwindPad)->isEHPad())
continue;
Value *UnwindParent = getParentPad(UnwindPad);
// Ignore unwind edges that don't exit CurrentPad.
if (UnwindParent == CurrentPad)
continue;
// Determine whether the original funclet pad is exited,
// and if we are scanning nested pads determine how many
// of them are exited so we can stop searching their
// children.
Value *ExitedPad = CurrentPad;
ExitsFPI = false;
do {
if (ExitedPad == &FPI) {
ExitsFPI = true;
// Now we can resolve any ancestors of CurrentPad up to
// FPI, but not including FPI since we need to make sure
// to check all direct users of FPI for consistency.
UnresolvedAncestorPad = &FPI;
break;
}
Value *ExitedParent = getParentPad(ExitedPad);
if (ExitedParent == UnwindParent) {
// ExitedPad is the ancestor-most pad which this unwind
// edge exits, so we can resolve up to it, meaning that
// ExitedParent is the first ancestor still unresolved.
UnresolvedAncestorPad = ExitedParent;
break;
}
ExitedPad = ExitedParent;
} while (!isa<ConstantTokenNone>(ExitedPad));
} else {
// Unwinding to caller exits all pads.
UnwindPad = ConstantTokenNone::get(FPI.getContext());
ExitsFPI = true;
UnresolvedAncestorPad = &FPI;
}
if (ExitsFPI) {
// This unwind edge exits FPI. Make sure it agrees with other
// such edges.
if (FirstUser) {
Assert(UnwindPad == FirstUnwindPad, "Unwind edges out of a funclet "
"pad must have the same unwind "
"dest",
&FPI, U, FirstUser);
} else {
FirstUser = U;
FirstUnwindPad = UnwindPad;
// Record cleanup sibling unwinds for verifySiblingFuncletUnwinds
if (isa<CleanupPadInst>(&FPI) && !isa<ConstantTokenNone>(UnwindPad) &&
getParentPad(UnwindPad) == getParentPad(&FPI))
SiblingFuncletInfo[&FPI] = cast<Instruction>(U);
}
}
// Make sure we visit all uses of FPI, but for nested pads stop as
// soon as we know where they unwind to.
if (CurrentPad != &FPI)
break;
}
if (UnresolvedAncestorPad) {
if (CurrentPad == UnresolvedAncestorPad) {
// When CurrentPad is FPI itself, we don't mark it as resolved even if
// we've found an unwind edge that exits it, because we need to verify
// all direct uses of FPI.
assert(CurrentPad == &FPI);
continue;
}
// Pop off the worklist any nested pads that we've found an unwind
// destination for. The pads on the worklist are the uncles,
// great-uncles, etc. of CurrentPad. We've found an unwind destination
// for all ancestors of CurrentPad up to but not including
// UnresolvedAncestorPad.
Value *ResolvedPad = CurrentPad;
while (!Worklist.empty()) {
Value *UnclePad = Worklist.back();
Value *AncestorPad = getParentPad(UnclePad);
// Walk ResolvedPad up the ancestor list until we either find the
// uncle's parent or the last resolved ancestor.
while (ResolvedPad != AncestorPad) {
Value *ResolvedParent = getParentPad(ResolvedPad);
if (ResolvedParent == UnresolvedAncestorPad) {
break;
}
ResolvedPad = ResolvedParent;
}
// If the resolved ancestor search didn't find the uncle's parent,
// then the uncle is not yet resolved.
if (ResolvedPad != AncestorPad)
break;
// This uncle is resolved, so pop it from the worklist.
Worklist.pop_back();
}
}
}
if (FirstUnwindPad) {
if (auto *CatchSwitch = dyn_cast<CatchSwitchInst>(FPI.getParentPad())) {
BasicBlock *SwitchUnwindDest = CatchSwitch->getUnwindDest();
Value *SwitchUnwindPad;
if (SwitchUnwindDest)
SwitchUnwindPad = SwitchUnwindDest->getFirstNonPHI();
else
SwitchUnwindPad = ConstantTokenNone::get(FPI.getContext());
Assert(SwitchUnwindPad == FirstUnwindPad,
"Unwind edges out of a catch must have the same unwind dest as "
"the parent catchswitch",
&FPI, FirstUser, CatchSwitch);
}
}
visitInstruction(FPI);
}
[IR] Reformulate LLVM's EH funclet IR While we have successfully implemented a funclet-oriented EH scheme on top of LLVM IR, our scheme has some notable deficiencies: - catchendpad and cleanupendpad are necessary in the current design but they are difficult to explain to others, even to seasoned LLVM experts. - catchendpad and cleanupendpad are optimization barriers. They cannot be split and force all potentially throwing call-sites to be invokes. This has a noticable effect on the quality of our code generation. - catchpad, while similar in some aspects to invoke, is fairly awkward. It is unsplittable, starts a funclet, and has control flow to other funclets. - The nesting relationship between funclets is currently a property of control flow edges. Because of this, we are forced to carefully analyze the flow graph to see if there might potentially exist illegal nesting among funclets. While we have logic to clone funclets when they are illegally nested, it would be nicer if we had a representation which forbade them upfront. Let's clean this up a bit by doing the following: - Instead, make catchpad more like cleanuppad and landingpad: no control flow, just a bunch of simple operands; catchpad would be splittable. - Introduce catchswitch, a control flow instruction designed to model the constraints of funclet oriented EH. - Make funclet scoping explicit by having funclet instructions consume the token produced by the funclet which contains them. - Remove catchendpad and cleanupendpad. Their presence can be inferred implicitly using coloring information. N.B. The state numbering code for the CLR has been updated but the veracity of it's output cannot be spoken for. An expert should take a look to make sure the results are reasonable. Reviewers: rnk, JosephTremoulet, andrew.w.kaylor Differential Revision: http://reviews.llvm.org/D15139 llvm-svn: 255422
2015-12-12 06:38:55 +01:00
void Verifier::visitCatchSwitchInst(CatchSwitchInst &CatchSwitch) {
BasicBlock *BB = CatchSwitch.getParent();
Function *F = BB->getParent();
Assert(F->hasPersonalityFn(),
[IR] Reformulate LLVM's EH funclet IR While we have successfully implemented a funclet-oriented EH scheme on top of LLVM IR, our scheme has some notable deficiencies: - catchendpad and cleanupendpad are necessary in the current design but they are difficult to explain to others, even to seasoned LLVM experts. - catchendpad and cleanupendpad are optimization barriers. They cannot be split and force all potentially throwing call-sites to be invokes. This has a noticable effect on the quality of our code generation. - catchpad, while similar in some aspects to invoke, is fairly awkward. It is unsplittable, starts a funclet, and has control flow to other funclets. - The nesting relationship between funclets is currently a property of control flow edges. Because of this, we are forced to carefully analyze the flow graph to see if there might potentially exist illegal nesting among funclets. While we have logic to clone funclets when they are illegally nested, it would be nicer if we had a representation which forbade them upfront. Let's clean this up a bit by doing the following: - Instead, make catchpad more like cleanuppad and landingpad: no control flow, just a bunch of simple operands; catchpad would be splittable. - Introduce catchswitch, a control flow instruction designed to model the constraints of funclet oriented EH. - Make funclet scoping explicit by having funclet instructions consume the token produced by the funclet which contains them. - Remove catchendpad and cleanupendpad. Their presence can be inferred implicitly using coloring information. N.B. The state numbering code for the CLR has been updated but the veracity of it's output cannot be spoken for. An expert should take a look to make sure the results are reasonable. Reviewers: rnk, JosephTremoulet, andrew.w.kaylor Differential Revision: http://reviews.llvm.org/D15139 llvm-svn: 255422
2015-12-12 06:38:55 +01:00
"CatchSwitchInst needs to be in a function with a personality.",
&CatchSwitch);
[IR] Reformulate LLVM's EH funclet IR While we have successfully implemented a funclet-oriented EH scheme on top of LLVM IR, our scheme has some notable deficiencies: - catchendpad and cleanupendpad are necessary in the current design but they are difficult to explain to others, even to seasoned LLVM experts. - catchendpad and cleanupendpad are optimization barriers. They cannot be split and force all potentially throwing call-sites to be invokes. This has a noticable effect on the quality of our code generation. - catchpad, while similar in some aspects to invoke, is fairly awkward. It is unsplittable, starts a funclet, and has control flow to other funclets. - The nesting relationship between funclets is currently a property of control flow edges. Because of this, we are forced to carefully analyze the flow graph to see if there might potentially exist illegal nesting among funclets. While we have logic to clone funclets when they are illegally nested, it would be nicer if we had a representation which forbade them upfront. Let's clean this up a bit by doing the following: - Instead, make catchpad more like cleanuppad and landingpad: no control flow, just a bunch of simple operands; catchpad would be splittable. - Introduce catchswitch, a control flow instruction designed to model the constraints of funclet oriented EH. - Make funclet scoping explicit by having funclet instructions consume the token produced by the funclet which contains them. - Remove catchendpad and cleanupendpad. Their presence can be inferred implicitly using coloring information. N.B. The state numbering code for the CLR has been updated but the veracity of it's output cannot be spoken for. An expert should take a look to make sure the results are reasonable. Reviewers: rnk, JosephTremoulet, andrew.w.kaylor Differential Revision: http://reviews.llvm.org/D15139 llvm-svn: 255422
2015-12-12 06:38:55 +01:00
// The catchswitch instruction must be the first non-PHI instruction in the
// block.
[IR] Reformulate LLVM's EH funclet IR While we have successfully implemented a funclet-oriented EH scheme on top of LLVM IR, our scheme has some notable deficiencies: - catchendpad and cleanupendpad are necessary in the current design but they are difficult to explain to others, even to seasoned LLVM experts. - catchendpad and cleanupendpad are optimization barriers. They cannot be split and force all potentially throwing call-sites to be invokes. This has a noticable effect on the quality of our code generation. - catchpad, while similar in some aspects to invoke, is fairly awkward. It is unsplittable, starts a funclet, and has control flow to other funclets. - The nesting relationship between funclets is currently a property of control flow edges. Because of this, we are forced to carefully analyze the flow graph to see if there might potentially exist illegal nesting among funclets. While we have logic to clone funclets when they are illegally nested, it would be nicer if we had a representation which forbade them upfront. Let's clean this up a bit by doing the following: - Instead, make catchpad more like cleanuppad and landingpad: no control flow, just a bunch of simple operands; catchpad would be splittable. - Introduce catchswitch, a control flow instruction designed to model the constraints of funclet oriented EH. - Make funclet scoping explicit by having funclet instructions consume the token produced by the funclet which contains them. - Remove catchendpad and cleanupendpad. Their presence can be inferred implicitly using coloring information. N.B. The state numbering code for the CLR has been updated but the veracity of it's output cannot be spoken for. An expert should take a look to make sure the results are reasonable. Reviewers: rnk, JosephTremoulet, andrew.w.kaylor Differential Revision: http://reviews.llvm.org/D15139 llvm-svn: 255422
2015-12-12 06:38:55 +01:00
Assert(BB->getFirstNonPHI() == &CatchSwitch,
"CatchSwitchInst not the first non-PHI instruction in the block.",
&CatchSwitch);
auto *ParentPad = CatchSwitch.getParentPad();
Assert(isa<ConstantTokenNone>(ParentPad) || isa<FuncletPadInst>(ParentPad),
"CatchSwitchInst has an invalid parent.", ParentPad);
[IR] Reformulate LLVM's EH funclet IR While we have successfully implemented a funclet-oriented EH scheme on top of LLVM IR, our scheme has some notable deficiencies: - catchendpad and cleanupendpad are necessary in the current design but they are difficult to explain to others, even to seasoned LLVM experts. - catchendpad and cleanupendpad are optimization barriers. They cannot be split and force all potentially throwing call-sites to be invokes. This has a noticable effect on the quality of our code generation. - catchpad, while similar in some aspects to invoke, is fairly awkward. It is unsplittable, starts a funclet, and has control flow to other funclets. - The nesting relationship between funclets is currently a property of control flow edges. Because of this, we are forced to carefully analyze the flow graph to see if there might potentially exist illegal nesting among funclets. While we have logic to clone funclets when they are illegally nested, it would be nicer if we had a representation which forbade them upfront. Let's clean this up a bit by doing the following: - Instead, make catchpad more like cleanuppad and landingpad: no control flow, just a bunch of simple operands; catchpad would be splittable. - Introduce catchswitch, a control flow instruction designed to model the constraints of funclet oriented EH. - Make funclet scoping explicit by having funclet instructions consume the token produced by the funclet which contains them. - Remove catchendpad and cleanupendpad. Their presence can be inferred implicitly using coloring information. N.B. The state numbering code for the CLR has been updated but the veracity of it's output cannot be spoken for. An expert should take a look to make sure the results are reasonable. Reviewers: rnk, JosephTremoulet, andrew.w.kaylor Differential Revision: http://reviews.llvm.org/D15139 llvm-svn: 255422
2015-12-12 06:38:55 +01:00
if (BasicBlock *UnwindDest = CatchSwitch.getUnwindDest()) {
Instruction *I = UnwindDest->getFirstNonPHI();
[IR] Reformulate LLVM's EH funclet IR While we have successfully implemented a funclet-oriented EH scheme on top of LLVM IR, our scheme has some notable deficiencies: - catchendpad and cleanupendpad are necessary in the current design but they are difficult to explain to others, even to seasoned LLVM experts. - catchendpad and cleanupendpad are optimization barriers. They cannot be split and force all potentially throwing call-sites to be invokes. This has a noticable effect on the quality of our code generation. - catchpad, while similar in some aspects to invoke, is fairly awkward. It is unsplittable, starts a funclet, and has control flow to other funclets. - The nesting relationship between funclets is currently a property of control flow edges. Because of this, we are forced to carefully analyze the flow graph to see if there might potentially exist illegal nesting among funclets. While we have logic to clone funclets when they are illegally nested, it would be nicer if we had a representation which forbade them upfront. Let's clean this up a bit by doing the following: - Instead, make catchpad more like cleanuppad and landingpad: no control flow, just a bunch of simple operands; catchpad would be splittable. - Introduce catchswitch, a control flow instruction designed to model the constraints of funclet oriented EH. - Make funclet scoping explicit by having funclet instructions consume the token produced by the funclet which contains them. - Remove catchendpad and cleanupendpad. Their presence can be inferred implicitly using coloring information. N.B. The state numbering code for the CLR has been updated but the veracity of it's output cannot be spoken for. An expert should take a look to make sure the results are reasonable. Reviewers: rnk, JosephTremoulet, andrew.w.kaylor Differential Revision: http://reviews.llvm.org/D15139 llvm-svn: 255422
2015-12-12 06:38:55 +01:00
Assert(I->isEHPad() && !isa<LandingPadInst>(I),
"CatchSwitchInst must unwind to an EH block which is not a "
"landingpad.",
&CatchSwitch);
// Record catchswitch sibling unwinds for verifySiblingFuncletUnwinds
if (getParentPad(I) == ParentPad)
SiblingFuncletInfo[&CatchSwitch] = &CatchSwitch;
}
[IR] Reformulate LLVM's EH funclet IR While we have successfully implemented a funclet-oriented EH scheme on top of LLVM IR, our scheme has some notable deficiencies: - catchendpad and cleanupendpad are necessary in the current design but they are difficult to explain to others, even to seasoned LLVM experts. - catchendpad and cleanupendpad are optimization barriers. They cannot be split and force all potentially throwing call-sites to be invokes. This has a noticable effect on the quality of our code generation. - catchpad, while similar in some aspects to invoke, is fairly awkward. It is unsplittable, starts a funclet, and has control flow to other funclets. - The nesting relationship between funclets is currently a property of control flow edges. Because of this, we are forced to carefully analyze the flow graph to see if there might potentially exist illegal nesting among funclets. While we have logic to clone funclets when they are illegally nested, it would be nicer if we had a representation which forbade them upfront. Let's clean this up a bit by doing the following: - Instead, make catchpad more like cleanuppad and landingpad: no control flow, just a bunch of simple operands; catchpad would be splittable. - Introduce catchswitch, a control flow instruction designed to model the constraints of funclet oriented EH. - Make funclet scoping explicit by having funclet instructions consume the token produced by the funclet which contains them. - Remove catchendpad and cleanupendpad. Their presence can be inferred implicitly using coloring information. N.B. The state numbering code for the CLR has been updated but the veracity of it's output cannot be spoken for. An expert should take a look to make sure the results are reasonable. Reviewers: rnk, JosephTremoulet, andrew.w.kaylor Differential Revision: http://reviews.llvm.org/D15139 llvm-svn: 255422
2015-12-12 06:38:55 +01:00
Assert(CatchSwitch.getNumHandlers() != 0,
"CatchSwitchInst cannot have empty handler list", &CatchSwitch);
for (BasicBlock *Handler : CatchSwitch.handlers()) {
Assert(isa<CatchPadInst>(Handler->getFirstNonPHI()),
"CatchSwitchInst handlers must be catchpads", &CatchSwitch, Handler);
}
visitEHPadPredecessors(CatchSwitch);
visitTerminator(CatchSwitch);
}
void Verifier::visitCleanupReturnInst(CleanupReturnInst &CRI) {
[IR] Reformulate LLVM's EH funclet IR While we have successfully implemented a funclet-oriented EH scheme on top of LLVM IR, our scheme has some notable deficiencies: - catchendpad and cleanupendpad are necessary in the current design but they are difficult to explain to others, even to seasoned LLVM experts. - catchendpad and cleanupendpad are optimization barriers. They cannot be split and force all potentially throwing call-sites to be invokes. This has a noticable effect on the quality of our code generation. - catchpad, while similar in some aspects to invoke, is fairly awkward. It is unsplittable, starts a funclet, and has control flow to other funclets. - The nesting relationship between funclets is currently a property of control flow edges. Because of this, we are forced to carefully analyze the flow graph to see if there might potentially exist illegal nesting among funclets. While we have logic to clone funclets when they are illegally nested, it would be nicer if we had a representation which forbade them upfront. Let's clean this up a bit by doing the following: - Instead, make catchpad more like cleanuppad and landingpad: no control flow, just a bunch of simple operands; catchpad would be splittable. - Introduce catchswitch, a control flow instruction designed to model the constraints of funclet oriented EH. - Make funclet scoping explicit by having funclet instructions consume the token produced by the funclet which contains them. - Remove catchendpad and cleanupendpad. Their presence can be inferred implicitly using coloring information. N.B. The state numbering code for the CLR has been updated but the veracity of it's output cannot be spoken for. An expert should take a look to make sure the results are reasonable. Reviewers: rnk, JosephTremoulet, andrew.w.kaylor Differential Revision: http://reviews.llvm.org/D15139 llvm-svn: 255422
2015-12-12 06:38:55 +01:00
Assert(isa<CleanupPadInst>(CRI.getOperand(0)),
"CleanupReturnInst needs to be provided a CleanupPad", &CRI,
CRI.getOperand(0));
if (BasicBlock *UnwindDest = CRI.getUnwindDest()) {
Instruction *I = UnwindDest->getFirstNonPHI();
Assert(I->isEHPad() && !isa<LandingPadInst>(I),
"CleanupReturnInst must unwind to an EH block which is not a "
"landingpad.",
&CRI);
}
visitTerminator(CRI);
}
void Verifier::verifyDominatesUse(Instruction &I, unsigned i) {
Instruction *Op = cast<Instruction>(I.getOperand(i));
// If the we have an invalid invoke, don't try to compute the dominance.
// We already reject it in the invoke specific checks and the dominance
// computation doesn't handle multiple edges.
if (InvokeInst *II = dyn_cast<InvokeInst>(Op)) {
if (II->getNormalDest() == II->getUnwindDest())
return;
}
// Quick check whether the def has already been encountered in the same block.
// PHI nodes are not checked to prevent accepting preceeding PHIs, because PHI
// uses are defined to happen on the incoming edge, not at the instruction.
Reapply "ValueMapper: Treat LocalAsMetadata more like function-local Values" This reverts commit r265765, reapplying r265759 after changing a call from LocalAsMetadata::get to ValueAsMetadata::get (and adding a unit test). When a local value is mapped to a constant (like "i32 %a" => "i32 7"), the new debug intrinsic operand may no longer be pointing at a local. http://lab.llvm.org:8080/green/job/clang-stage1-configure-RA_build/19020/ The previous coommit message follows: -- This is a partial re-commit -- maybe more of a re-implementation -- of r265631 (reverted in r265637). This makes RF_IgnoreMissingLocals behave (almost) consistently between the Value and the Metadata hierarchy. In particular: - MapValue returns nullptr or "metadata !{}" for missing locals in MetadataAsValue/LocalAsMetadata bridging paris, depending on the RF_IgnoreMissingLocals flag. - MapValue doesn't memoize LocalAsMetadata-related results. - MapMetadata no longer deals with LocalAsMetadata or RF_IgnoreMissingLocals at all. (This wasn't in r265631 at all, but I realized during testing it would make the patch simpler with no loss of generality.) r265631 went too far, making both functions universally ignore RF_IgnoreMissingLocals. This broke building (e.g.) compiler-rt. Reassociate (and possibly other passes) don't currently maintain dominates-use invariants for metadata operands, resulting in IR like this: define void @foo(i32 %arg) { call void @llvm.some.intrinsic(metadata i32 %x) %x = add i32 1, i32 %arg } If the inliner chooses to inline @foo into another function, then RemapInstruction will call `MapValue(metadata i32 %x)` and assert that the return is not nullptr. I've filed PR27273 to add a Verifier check and fix the underlying problem in the optimization passes. As a workaround, return `!{}` instead of nullptr for unmapped LocalAsMetadata when RF_IgnoreMissingLocals is unset. Otherwise, match the behaviour of r265631. Original commit message: ValueMapper: Make LocalAsMetadata match function-local Values Start treating LocalAsMetadata similarly to function-local members of the Value hierarchy in MapValue and MapMetadata. - Don't memoize them. - Return nullptr if they are missing. This also cleans up ConstantAsMetadata to stop listening to the RF_IgnoreMissingLocals flag. llvm-svn: 265768
2016-04-08 05:13:22 +02:00
//
// FIXME: If this operand is a MetadataAsValue (wrapping a LocalAsMetadata)
// wrapping an SSA value, assert that we've already encountered it. See
// related FIXME in Mapper::mapLocalAsMetadata in ValueMapper.cpp.
if (!isa<PHINode>(I) && InstsInThisBlock.count(Op))
return;
const Use &U = I.getOperandUse(i);
Assert(DT.dominates(Op, U),
"Instruction does not dominate all uses!", Op, &I);
}
void Verifier::visitDereferenceableMetadata(Instruction& I, MDNode* MD) {
Assert(I.getType()->isPointerTy(), "dereferenceable, dereferenceable_or_null "
"apply only to pointer types", &I);
Assert(isa<LoadInst>(I),
"dereferenceable, dereferenceable_or_null apply only to load"
" instructions, use attributes for calls or invokes", &I);
Assert(MD->getNumOperands() == 1, "dereferenceable, dereferenceable_or_null "
"take one operand!", &I);
ConstantInt *CI = mdconst::dyn_extract<ConstantInt>(MD->getOperand(0));
Assert(CI && CI->getType()->isIntegerTy(64), "dereferenceable, "
"dereferenceable_or_null metadata value must be an i64!", &I);
}
2004-03-02 01:22:19 +01:00
/// verifyInstruction - Verify that an instruction is well formed.
///
void Verifier::visitInstruction(Instruction &I) {
BasicBlock *BB = I.getParent();
Assert(BB, "Instruction not embedded in basic block!", &I);
if (!isa<PHINode>(I)) { // Check that non-phi nodes are not self referential
for (User *U : I.users()) {
Assert(U != (User *)&I || !DT.isReachableFromEntry(BB),
"Only PHI nodes may reference their own value!", &I);
}
}
2009-09-07 22:44:51 +02:00
// Check that void typed values don't have names
Assert(!I.getType()->isVoidTy() || !I.hasName(),
"Instruction has a name, but provides a void value!", &I);
2004-03-29 02:29:36 +02:00
// Check that the return value of the instruction is either void or a legal
// value type.
Assert(I.getType()->isVoidTy() || I.getType()->isFirstClassType(),
"Instruction returns a non-scalar type!", &I);
2004-03-29 02:29:36 +02:00
// Check that the instruction doesn't produce metadata. Calls are already
// checked against the callee type.
Assert(!I.getType()->isMetadataTy() || isa<CallInst>(I) || isa<InvokeInst>(I),
"Invalid use of metadata!", &I);
// Check that all uses of the instruction, if they are instructions
// themselves, actually have parent basic blocks. If the use is not an
// instruction, it is an error!
[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 (Use &U : I.uses()) {
if (Instruction *Used = dyn_cast<Instruction>(U.getUser()))
Assert(Used->getParent() != nullptr,
"Instruction referencing"
" instruction not embedded in a basic block!",
&I, Used);
else {
[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
CheckFailed("Use of instruction is not an instruction!", U);
return;
}
}
[TI removal] Leverage the fact that TerminatorInst is gone to create a normal base class that provides all common "call" functionality. This merges two complex CRTP mixins for the common "call" logic and common operand bundle logic into a single, normal base class of `CallInst` and `InvokeInst`. Going forward, users can typically `dyn_cast<CallBase>` and use the resulting API. No more need for the `CallSite` wrapper. I'm planning to migrate current usage of the wrapper to directly use the base class and then it can be removed, but those are simpler and much more incremental steps. The big change is to introduce this abstraction into the type system. I've tried to do some basic simplifications of the APIs that I couldn't really help but touch as part of this: - I've tried to organize the attribute API and bundle API into groups to make understanding the API of `CallBase` easier. Without this, I wasn't able to navigate the API sanely for all of the ways I needed to modify it. - I've added what seem like more clear and consistent APIs for getting at the called operand. These ended up being especially useful to consolidate the *numerous* duplicated code paths trying to do this. - I've largely reworked the organization and implementation of the APIs for computing the argument operands as they needed to change to work with the new subclass approach. To minimize any cost associated with this abstraction, I've moved the operand layout in memory to store the called operand last. This makes its position relative to the end of the operand array the same, regardless of the subclass. It should make it much cheaper to reference from the `CallBase` abstraction, and this is likely one of the most frequent things to query. We do still pay one abstraction penalty here: we have to branch to determine whether there are 0 or 2 extra operands when computing the end of the argument operand sequence. However, that seems both rare and should optimize well. I've implemented this in a way specifically designed to allow it to optimize fairly well. If this shows up in profiles, we can add overrides of the relevant methods to the subclasses that bypass this penalty. It seems very unlikely that this will be an issue as the code was *already* dealing with an ever present abstraction of whether or not there are operand bundles, so this isn't the first branch to go into the computation. I've tried to remove as much of the obvious vestigial API surface of the old CRTP implementation as I could, but I suspect there is further cleanup that should now be possible, especially around the operand bundle APIs. I'm leaving all of that for future work in this patch as enough things are changing here as-is. One thing that made this harder for me to reason about and debug was the pervasive use of unsigned values in subtraction and other arithmetic computations. I had to debug more than one unintentional wrap. I've switched a few of these to use `int` which seems substantially simpler, but I've held back from doing this more broadly to avoid creating confusing divergence within a single class's API. I also worked to remove all of the magic numbers used to index into operands, putting them behind named constants or putting them into a single method with a comment and strictly using the method elsewhere. This was necessary to be able to re-layout the operands as discussed above. Thanks to Ben for reviewing this (somewhat large and awkward) patch! Differential Revision: https://reviews.llvm.org/D54788 llvm-svn: 347452
2018-11-22 11:31:35 +01:00
// Get a pointer to the call base of the instruction if it is some form of
// call.
const CallBase *CBI = dyn_cast<CallBase>(&I);
for (unsigned i = 0, e = I.getNumOperands(); i != e; ++i) {
Assert(I.getOperand(i) != nullptr, "Instruction has null operand!", &I);
// Check to make sure that only first-class-values are operands to
// instructions.
if (!I.getOperand(i)->getType()->isFirstClassType()) {
Assert(false, "Instruction operands must be first-class values!", &I);
}
if (Function *F = dyn_cast<Function>(I.getOperand(i))) {
// Check to make sure that the "address of" an intrinsic function is never
// taken.
[TI removal] Leverage the fact that TerminatorInst is gone to create a normal base class that provides all common "call" functionality. This merges two complex CRTP mixins for the common "call" logic and common operand bundle logic into a single, normal base class of `CallInst` and `InvokeInst`. Going forward, users can typically `dyn_cast<CallBase>` and use the resulting API. No more need for the `CallSite` wrapper. I'm planning to migrate current usage of the wrapper to directly use the base class and then it can be removed, but those are simpler and much more incremental steps. The big change is to introduce this abstraction into the type system. I've tried to do some basic simplifications of the APIs that I couldn't really help but touch as part of this: - I've tried to organize the attribute API and bundle API into groups to make understanding the API of `CallBase` easier. Without this, I wasn't able to navigate the API sanely for all of the ways I needed to modify it. - I've added what seem like more clear and consistent APIs for getting at the called operand. These ended up being especially useful to consolidate the *numerous* duplicated code paths trying to do this. - I've largely reworked the organization and implementation of the APIs for computing the argument operands as they needed to change to work with the new subclass approach. To minimize any cost associated with this abstraction, I've moved the operand layout in memory to store the called operand last. This makes its position relative to the end of the operand array the same, regardless of the subclass. It should make it much cheaper to reference from the `CallBase` abstraction, and this is likely one of the most frequent things to query. We do still pay one abstraction penalty here: we have to branch to determine whether there are 0 or 2 extra operands when computing the end of the argument operand sequence. However, that seems both rare and should optimize well. I've implemented this in a way specifically designed to allow it to optimize fairly well. If this shows up in profiles, we can add overrides of the relevant methods to the subclasses that bypass this penalty. It seems very unlikely that this will be an issue as the code was *already* dealing with an ever present abstraction of whether or not there are operand bundles, so this isn't the first branch to go into the computation. I've tried to remove as much of the obvious vestigial API surface of the old CRTP implementation as I could, but I suspect there is further cleanup that should now be possible, especially around the operand bundle APIs. I'm leaving all of that for future work in this patch as enough things are changing here as-is. One thing that made this harder for me to reason about and debug was the pervasive use of unsigned values in subtraction and other arithmetic computations. I had to debug more than one unintentional wrap. I've switched a few of these to use `int` which seems substantially simpler, but I've held back from doing this more broadly to avoid creating confusing divergence within a single class's API. I also worked to remove all of the magic numbers used to index into operands, putting them behind named constants or putting them into a single method with a comment and strictly using the method elsewhere. This was necessary to be able to re-layout the operands as discussed above. Thanks to Ben for reviewing this (somewhat large and awkward) patch! Differential Revision: https://reviews.llvm.org/D54788 llvm-svn: 347452
2018-11-22 11:31:35 +01:00
Assert(!F->isIntrinsic() ||
(CBI && &CBI->getCalledOperandUse() == &I.getOperandUse(i)),
"Cannot take the address of an intrinsic!", &I);
Assert(
!F->isIntrinsic() || isa<CallInst>(I) ||
F->getIntrinsicID() == Intrinsic::donothing ||
F->getIntrinsicID() == Intrinsic::coro_resume ||
F->getIntrinsicID() == Intrinsic::coro_destroy ||
F->getIntrinsicID() == Intrinsic::experimental_patchpoint_void ||
F->getIntrinsicID() == Intrinsic::experimental_patchpoint_i64 ||
F->getIntrinsicID() == Intrinsic::experimental_gc_statepoint,
"Cannot invoke an intrinsic other than donothing, patchpoint, "
"statepoint, coro_resume or coro_destroy",
&I);
Assert(F->getParent() == &M, "Referencing function in another module!",
&I, &M, F, F->getParent());
} else if (BasicBlock *OpBB = dyn_cast<BasicBlock>(I.getOperand(i))) {
Assert(OpBB->getParent() == BB->getParent(),
"Referring to a basic block in another function!", &I);
} else if (Argument *OpArg = dyn_cast<Argument>(I.getOperand(i))) {
Assert(OpArg->getParent() == BB->getParent(),
"Referring to an argument in another function!", &I);
} else if (GlobalValue *GV = dyn_cast<GlobalValue>(I.getOperand(i))) {
Assert(GV->getParent() == &M, "Referencing global in another module!", &I,
&M, GV, GV->getParent());
} else if (isa<Instruction>(I.getOperand(i))) {
verifyDominatesUse(I, i);
} else if (isa<InlineAsm>(I.getOperand(i))) {
[TI removal] Leverage the fact that TerminatorInst is gone to create a normal base class that provides all common "call" functionality. This merges two complex CRTP mixins for the common "call" logic and common operand bundle logic into a single, normal base class of `CallInst` and `InvokeInst`. Going forward, users can typically `dyn_cast<CallBase>` and use the resulting API. No more need for the `CallSite` wrapper. I'm planning to migrate current usage of the wrapper to directly use the base class and then it can be removed, but those are simpler and much more incremental steps. The big change is to introduce this abstraction into the type system. I've tried to do some basic simplifications of the APIs that I couldn't really help but touch as part of this: - I've tried to organize the attribute API and bundle API into groups to make understanding the API of `CallBase` easier. Without this, I wasn't able to navigate the API sanely for all of the ways I needed to modify it. - I've added what seem like more clear and consistent APIs for getting at the called operand. These ended up being especially useful to consolidate the *numerous* duplicated code paths trying to do this. - I've largely reworked the organization and implementation of the APIs for computing the argument operands as they needed to change to work with the new subclass approach. To minimize any cost associated with this abstraction, I've moved the operand layout in memory to store the called operand last. This makes its position relative to the end of the operand array the same, regardless of the subclass. It should make it much cheaper to reference from the `CallBase` abstraction, and this is likely one of the most frequent things to query. We do still pay one abstraction penalty here: we have to branch to determine whether there are 0 or 2 extra operands when computing the end of the argument operand sequence. However, that seems both rare and should optimize well. I've implemented this in a way specifically designed to allow it to optimize fairly well. If this shows up in profiles, we can add overrides of the relevant methods to the subclasses that bypass this penalty. It seems very unlikely that this will be an issue as the code was *already* dealing with an ever present abstraction of whether or not there are operand bundles, so this isn't the first branch to go into the computation. I've tried to remove as much of the obvious vestigial API surface of the old CRTP implementation as I could, but I suspect there is further cleanup that should now be possible, especially around the operand bundle APIs. I'm leaving all of that for future work in this patch as enough things are changing here as-is. One thing that made this harder for me to reason about and debug was the pervasive use of unsigned values in subtraction and other arithmetic computations. I had to debug more than one unintentional wrap. I've switched a few of these to use `int` which seems substantially simpler, but I've held back from doing this more broadly to avoid creating confusing divergence within a single class's API. I also worked to remove all of the magic numbers used to index into operands, putting them behind named constants or putting them into a single method with a comment and strictly using the method elsewhere. This was necessary to be able to re-layout the operands as discussed above. Thanks to Ben for reviewing this (somewhat large and awkward) patch! Differential Revision: https://reviews.llvm.org/D54788 llvm-svn: 347452
2018-11-22 11:31:35 +01:00
Assert(CBI && &CBI->getCalledOperandUse() == &I.getOperandUse(i),
"Cannot take the address of an inline asm!", &I);
} else if (ConstantExpr *CE = dyn_cast<ConstantExpr>(I.getOperand(i))) {
if (CE->getType()->isPtrOrPtrVectorTy() ||
!DL.getNonIntegralAddressSpaces().empty()) {
// If we have a ConstantExpr pointer, we need to see if it came from an
// illegal bitcast. If the datalayout string specifies non-integral
// address spaces then we also need to check for illegal ptrtoint and
// inttoptr expressions.
visitConstantExprsRecursively(CE);
}
}
}
if (MDNode *MD = I.getMetadata(LLVMContext::MD_fpmath)) {
Assert(I.getType()->isFPOrFPVectorTy(),
"fpmath requires a floating point result!", &I);
Assert(MD->getNumOperands() == 1, "fpmath takes one operand!", &I);
IR: Split Metadata from Value Split `Metadata` away from the `Value` class hierarchy, as part of PR21532. Assembly and bitcode changes are in the wings, but this is the bulk of the change for the IR C++ API. I have a follow-up patch prepared for `clang`. If this breaks other sub-projects, I apologize in advance :(. Help me compile it on Darwin I'll try to fix it. FWIW, the errors should be easy to fix, so it may be simpler to just fix it yourself. This breaks the build for all metadata-related code that's out-of-tree. Rest assured the transition is mechanical and the compiler should catch almost all of the problems. Here's a quick guide for updating your code: - `Metadata` is the root of a class hierarchy with three main classes: `MDNode`, `MDString`, and `ValueAsMetadata`. It is distinct from the `Value` class hierarchy. It is typeless -- i.e., instances do *not* have a `Type`. - `MDNode`'s operands are all `Metadata *` (instead of `Value *`). - `TrackingVH<MDNode>` and `WeakVH` referring to metadata can be replaced with `TrackingMDNodeRef` and `TrackingMDRef`, respectively. If you're referring solely to resolved `MDNode`s -- post graph construction -- just use `MDNode*`. - `MDNode` (and the rest of `Metadata`) have only limited support for `replaceAllUsesWith()`. As long as an `MDNode` is pointing at a forward declaration -- the result of `MDNode::getTemporary()` -- it maintains a side map of its uses and can RAUW itself. Once the forward declarations are fully resolved RAUW support is dropped on the ground. This means that uniquing collisions on changing operands cause nodes to become "distinct". (This already happened fairly commonly, whenever an operand went to null.) If you're constructing complex (non self-reference) `MDNode` cycles, you need to call `MDNode::resolveCycles()` on each node (or on a top-level node that somehow references all of the nodes). Also, don't do that. Metadata cycles (and the RAUW machinery needed to construct them) are expensive. - An `MDNode` can only refer to a `Constant` through a bridge called `ConstantAsMetadata` (one of the subclasses of `ValueAsMetadata`). As a side effect, accessing an operand of an `MDNode` that is known to be, e.g., `ConstantInt`, takes three steps: first, cast from `Metadata` to `ConstantAsMetadata`; second, extract the `Constant`; third, cast down to `ConstantInt`. The eventual goal is to introduce `MDInt`/`MDFloat`/etc. and have metadata schema owners transition away from using `Constant`s when the type isn't important (and they don't care about referring to `GlobalValue`s). In the meantime, I've added transitional API to the `mdconst` namespace that matches semantics with the old code, in order to avoid adding the error-prone three-step equivalent to every call site. If your old code was: MDNode *N = foo(); bar(isa <ConstantInt>(N->getOperand(0))); baz(cast <ConstantInt>(N->getOperand(1))); bak(cast_or_null <ConstantInt>(N->getOperand(2))); bat(dyn_cast <ConstantInt>(N->getOperand(3))); bay(dyn_cast_or_null<ConstantInt>(N->getOperand(4))); you can trivially match its semantics with: MDNode *N = foo(); bar(mdconst::hasa <ConstantInt>(N->getOperand(0))); baz(mdconst::extract <ConstantInt>(N->getOperand(1))); bak(mdconst::extract_or_null <ConstantInt>(N->getOperand(2))); bat(mdconst::dyn_extract <ConstantInt>(N->getOperand(3))); bay(mdconst::dyn_extract_or_null<ConstantInt>(N->getOperand(4))); and when you transition your metadata schema to `MDInt`: MDNode *N = foo(); bar(isa <MDInt>(N->getOperand(0))); baz(cast <MDInt>(N->getOperand(1))); bak(cast_or_null <MDInt>(N->getOperand(2))); bat(dyn_cast <MDInt>(N->getOperand(3))); bay(dyn_cast_or_null<MDInt>(N->getOperand(4))); - A `CallInst` -- specifically, intrinsic instructions -- can refer to metadata through a bridge called `MetadataAsValue`. This is a subclass of `Value` where `getType()->isMetadataTy()`. `MetadataAsValue` is the *only* class that can legally refer to a `LocalAsMetadata`, which is a bridged form of non-`Constant` values like `Argument` and `Instruction`. It can also refer to any other `Metadata` subclass. (I'll break all your testcases in a follow-up commit, when I propagate this change to assembly.) llvm-svn: 223802
2014-12-09 19:38:53 +01:00
if (ConstantFP *CFP0 =
mdconst::dyn_extract_or_null<ConstantFP>(MD->getOperand(0))) {
const APFloat &Accuracy = CFP0->getValueAPF();
Assert(&Accuracy.getSemantics() == &APFloat::IEEEsingle(),
"fpmath accuracy must have float type", &I);
Assert(Accuracy.isFiniteNonZero() && !Accuracy.isNegative(),
"fpmath accuracy not a positive number!", &I);
} else {
Assert(false, "invalid fpmath accuracy!", &I);
}
}
if (MDNode *Range = I.getMetadata(LLVMContext::MD_range)) {
Assert(isa<LoadInst>(I) || isa<CallInst>(I) || isa<InvokeInst>(I),
"Ranges are only for loads, calls and invokes!", &I);
visitRangeMetadata(I, Range, I.getType());
}
if (I.getMetadata(LLVMContext::MD_nonnull)) {
Assert(I.getType()->isPointerTy(), "nonnull applies only to pointer types",
&I);
Assert(isa<LoadInst>(I),
"nonnull applies only to load instructions, use attributes"
" for calls or invokes",
&I);
}
if (MDNode *MD = I.getMetadata(LLVMContext::MD_dereferenceable))
visitDereferenceableMetadata(I, MD);
if (MDNode *MD = I.getMetadata(LLVMContext::MD_dereferenceable_or_null))
visitDereferenceableMetadata(I, MD);
if (MDNode *TBAA = I.getMetadata(LLVMContext::MD_tbaa))
TBAAVerifyHelper.visitTBAAMetadata(I, TBAA);
if (MDNode *AlignMD = I.getMetadata(LLVMContext::MD_align)) {
Assert(I.getType()->isPointerTy(), "align applies only to pointer types",
&I);
Assert(isa<LoadInst>(I), "align applies only to load instructions, "
"use attributes for calls or invokes", &I);
Assert(AlignMD->getNumOperands() == 1, "align takes one operand!", &I);
ConstantInt *CI = mdconst::dyn_extract<ConstantInt>(AlignMD->getOperand(0));
Assert(CI && CI->getType()->isIntegerTy(64),
"align metadata value must be an i64!", &I);
uint64_t Align = CI->getZExtValue();
Assert(isPowerOf2_64(Align),
"align metadata value must be a power of 2!", &I);
Assert(Align <= Value::MaximumAlignment,
"alignment is larger that implementation defined limit", &I);
}
if (MDNode *N = I.getDebugLoc().getAsMDNode()) {
AssertDI(isa<DILocation>(N), "invalid !dbg metadata attachment", &I, N);
visitMDNode(*N);
}
if (auto *DII = dyn_cast<DbgVariableIntrinsic>(&I))
verifyFragmentExpression(*DII);
InstsInThisBlock.insert(&I);
}
/// Allow intrinsics to be verified in different ways.
void Verifier::visitIntrinsicCallSite(Intrinsic::ID ID, CallSite CS) {
Function *IF = CS.getCalledFunction();
Assert(IF->isDeclaration(), "Intrinsic functions should never be defined!",
IF);
2009-09-07 22:44:51 +02:00
// Verify that the intrinsic prototype lines up with what the .td files
// describe.
FunctionType *IFTy = IF->getFunctionType();
bool IsVarArg = IFTy->isVarArg();
SmallVector<Intrinsic::IITDescriptor, 8> Table;
getIntrinsicInfoTableEntries(ID, Table);
ArrayRef<Intrinsic::IITDescriptor> TableRef = Table;
SmallVector<Type *, 4> ArgTys;
Assert(!Intrinsic::matchIntrinsicType(IFTy->getReturnType(),
TableRef, ArgTys),
"Intrinsic has incorrect return type!", IF);
for (unsigned i = 0, e = IFTy->getNumParams(); i != e; ++i)
Assert(!Intrinsic::matchIntrinsicType(IFTy->getParamType(i),
TableRef, ArgTys),
"Intrinsic has incorrect argument type!", IF);
// Verify if the intrinsic call matches the vararg property.
if (IsVarArg)
Assert(!Intrinsic::matchIntrinsicVarArg(IsVarArg, TableRef),
"Intrinsic was not defined with variable arguments!", IF);
else
Assert(!Intrinsic::matchIntrinsicVarArg(IsVarArg, TableRef),
"Callsite was not defined with variable arguments!", IF);
// All descriptors should be absorbed by now.
Assert(TableRef.empty(), "Intrinsic has too few arguments!", IF);
// Now that we have the intrinsic ID and the actual argument types (and we
// know they are legal for the intrinsic!) get the intrinsic name through the
// usual means. This allows us to verify the mangling of argument types into
// the name.
const std::string ExpectedName = Intrinsic::getName(ID, ArgTys);
Assert(ExpectedName == IF->getName(),
"Intrinsic name not mangled correctly for type arguments! "
"Should be: " +
ExpectedName,
IF);
// If the intrinsic takes MDNode arguments, verify that they are either global
// or are local to *this* function.
for (Value *V : CS.args())
if (auto *MD = dyn_cast<MetadataAsValue>(V))
visitMetadataAsValue(*MD, CS.getCaller());
switch (ID) {
default:
break;
case Intrinsic::coro_id: {
auto *InfoArg = CS.getArgOperand(3)->stripPointerCasts();
if (isa<ConstantPointerNull>(InfoArg))
break;
auto *GV = dyn_cast<GlobalVariable>(InfoArg);
Assert(GV && GV->isConstant() && GV->hasDefinitiveInitializer(),
"info argument of llvm.coro.begin must refer to an initialized "
"constant");
Constant *Init = GV->getInitializer();
Assert(isa<ConstantStruct>(Init) || isa<ConstantArray>(Init),
"info argument of llvm.coro.begin must refer to either a struct or "
"an array");
break;
}
case Intrinsic::ctlz: // llvm.ctlz
case Intrinsic::cttz: // llvm.cttz
Assert(isa<ConstantInt>(CS.getArgOperand(1)),
"is_zero_undef argument of bit counting intrinsics must be a "
"constant int",
CS);
break;
case Intrinsic::experimental_constrained_fadd:
case Intrinsic::experimental_constrained_fsub:
case Intrinsic::experimental_constrained_fmul:
case Intrinsic::experimental_constrained_fdiv:
case Intrinsic::experimental_constrained_frem:
case Intrinsic::experimental_constrained_fma:
case Intrinsic::experimental_constrained_sqrt:
case Intrinsic::experimental_constrained_pow:
case Intrinsic::experimental_constrained_powi:
case Intrinsic::experimental_constrained_sin:
case Intrinsic::experimental_constrained_cos:
case Intrinsic::experimental_constrained_exp:
case Intrinsic::experimental_constrained_exp2:
case Intrinsic::experimental_constrained_log:
case Intrinsic::experimental_constrained_log10:
case Intrinsic::experimental_constrained_log2:
case Intrinsic::experimental_constrained_rint:
case Intrinsic::experimental_constrained_nearbyint:
case Intrinsic::experimental_constrained_maxnum:
case Intrinsic::experimental_constrained_minnum:
case Intrinsic::experimental_constrained_ceil:
case Intrinsic::experimental_constrained_floor:
case Intrinsic::experimental_constrained_round:
case Intrinsic::experimental_constrained_trunc:
visitConstrainedFPIntrinsic(
cast<ConstrainedFPIntrinsic>(*CS.getInstruction()));
break;
case Intrinsic::dbg_declare: // llvm.dbg.declare
Assert(isa<MetadataAsValue>(CS.getArgOperand(0)),
"invalid llvm.dbg.declare intrinsic call 1", CS);
visitDbgIntrinsic("declare", cast<DbgVariableIntrinsic>(*CS.getInstruction()));
break;
case Intrinsic::dbg_addr: // llvm.dbg.addr
visitDbgIntrinsic("addr", cast<DbgVariableIntrinsic>(*CS.getInstruction()));
break;
case Intrinsic::dbg_value: // llvm.dbg.value
visitDbgIntrinsic("value", cast<DbgVariableIntrinsic>(*CS.getInstruction()));
break;
case Intrinsic::dbg_label: // llvm.dbg.label
visitDbgLabelIntrinsic("label", cast<DbgLabelInst>(*CS.getInstruction()));
break;
case Intrinsic::memcpy:
case Intrinsic::memmove:
case Intrinsic::memset: {
const auto *MI = cast<MemIntrinsic>(CS.getInstruction());
auto IsValidAlignment = [&](unsigned Alignment) -> bool {
return Alignment == 0 || isPowerOf2_32(Alignment);
};
Remove alignment argument from memcpy/memmove/memset in favour of alignment attributes (Step 1) Summary: This is a resurrection of work first proposed and discussed in Aug 2015: http://lists.llvm.org/pipermail/llvm-dev/2015-August/089384.html and initially landed (but then backed out) in Nov 2015: http://lists.llvm.org/pipermail/llvm-commits/Week-of-Mon-20151109/312083.html The @llvm.memcpy/memmove/memset intrinsics currently have an explicit argument which is required to be a constant integer. It represents the alignment of the dest (and source), and so must be the minimum of the actual alignment of the two. This change is the first in a series that allows source and dest to each have their own alignments by using the alignment attribute on their arguments. In this change we: 1) Remove the alignment argument. 2) Add alignment attributes to the source & dest arguments. We, temporarily, require that the alignments for source & dest be equal. For example, code which used to read: call void @llvm.memcpy.p0i8.p0i8.i32(i8* %dest, i8* %src, i32 100, i32 4, i1 false) will now read call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 4 %dest, i8* align 4 %src, i32 100, i1 false) Downstream users may have to update their lit tests that check for @llvm.memcpy/memmove/memset call/declaration patterns. The following extended sed script may help with updating the majority of your tests, but it does not catch all possible patterns so some manual checking and updating will be required. s~declare void @llvm\.mem(set|cpy|move)\.p([^(]*)\((.*), i32, i1\)~declare void @llvm.mem\1.p\2(\3, i1)~g s~call void @llvm\.memset\.p([^(]*)i8\(i8([^*]*)\* (.*), i8 (.*), i8 (.*), i32 [01], i1 ([^)]*)\)~call void @llvm.memset.p\1i8(i8\2* \3, i8 \4, i8 \5, i1 \6)~g s~call void @llvm\.memset\.p([^(]*)i16\(i8([^*]*)\* (.*), i8 (.*), i16 (.*), i32 [01], i1 ([^)]*)\)~call void @llvm.memset.p\1i16(i8\2* \3, i8 \4, i16 \5, i1 \6)~g s~call void @llvm\.memset\.p([^(]*)i32\(i8([^*]*)\* (.*), i8 (.*), i32 (.*), i32 [01], i1 ([^)]*)\)~call void @llvm.memset.p\1i32(i8\2* \3, i8 \4, i32 \5, i1 \6)~g s~call void @llvm\.memset\.p([^(]*)i64\(i8([^*]*)\* (.*), i8 (.*), i64 (.*), i32 [01], i1 ([^)]*)\)~call void @llvm.memset.p\1i64(i8\2* \3, i8 \4, i64 \5, i1 \6)~g s~call void @llvm\.memset\.p([^(]*)i128\(i8([^*]*)\* (.*), i8 (.*), i128 (.*), i32 [01], i1 ([^)]*)\)~call void @llvm.memset.p\1i128(i8\2* \3, i8 \4, i128 \5, i1 \6)~g s~call void @llvm\.memset\.p([^(]*)i8\(i8([^*]*)\* (.*), i8 (.*), i8 (.*), i32 ([0-9]*), i1 ([^)]*)\)~call void @llvm.memset.p\1i8(i8\2* align \6 \3, i8 \4, i8 \5, i1 \7)~g s~call void @llvm\.memset\.p([^(]*)i16\(i8([^*]*)\* (.*), i8 (.*), i16 (.*), i32 ([0-9]*), i1 ([^)]*)\)~call void @llvm.memset.p\1i16(i8\2* align \6 \3, i8 \4, i16 \5, i1 \7)~g s~call void @llvm\.memset\.p([^(]*)i32\(i8([^*]*)\* (.*), i8 (.*), i32 (.*), i32 ([0-9]*), i1 ([^)]*)\)~call void @llvm.memset.p\1i32(i8\2* align \6 \3, i8 \4, i32 \5, i1 \7)~g s~call void @llvm\.memset\.p([^(]*)i64\(i8([^*]*)\* (.*), i8 (.*), i64 (.*), i32 ([0-9]*), i1 ([^)]*)\)~call void @llvm.memset.p\1i64(i8\2* align \6 \3, i8 \4, i64 \5, i1 \7)~g s~call void @llvm\.memset\.p([^(]*)i128\(i8([^*]*)\* (.*), i8 (.*), i128 (.*), i32 ([0-9]*), i1 ([^)]*)\)~call void @llvm.memset.p\1i128(i8\2* align \6 \3, i8 \4, i128 \5, i1 \7)~g s~call void @llvm\.mem(cpy|move)\.p([^(]*)i8\(i8([^*]*)\* (.*), i8([^*]*)\* (.*), i8 (.*), i32 [01], i1 ([^)]*)\)~call void @llvm.mem\1.p\2i8(i8\3* \4, i8\5* \6, i8 \7, i1 \8)~g s~call void @llvm\.mem(cpy|move)\.p([^(]*)i16\(i8([^*]*)\* (.*), i8([^*]*)\* (.*), i16 (.*), i32 [01], i1 ([^)]*)\)~call void @llvm.mem\1.p\2i16(i8\3* \4, i8\5* \6, i16 \7, i1 \8)~g s~call void @llvm\.mem(cpy|move)\.p([^(]*)i32\(i8([^*]*)\* (.*), i8([^*]*)\* (.*), i32 (.*), i32 [01], i1 ([^)]*)\)~call void @llvm.mem\1.p\2i32(i8\3* \4, i8\5* \6, i32 \7, i1 \8)~g s~call void @llvm\.mem(cpy|move)\.p([^(]*)i64\(i8([^*]*)\* (.*), i8([^*]*)\* (.*), i64 (.*), i32 [01], i1 ([^)]*)\)~call void @llvm.mem\1.p\2i64(i8\3* \4, i8\5* \6, i64 \7, i1 \8)~g s~call void @llvm\.mem(cpy|move)\.p([^(]*)i128\(i8([^*]*)\* (.*), i8([^*]*)\* (.*), i128 (.*), i32 [01], i1 ([^)]*)\)~call void @llvm.mem\1.p\2i128(i8\3* \4, i8\5* \6, i128 \7, i1 \8)~g s~call void @llvm\.mem(cpy|move)\.p([^(]*)i8\(i8([^*]*)\* (.*), i8([^*]*)\* (.*), i8 (.*), i32 ([0-9]*), i1 ([^)]*)\)~call void @llvm.mem\1.p\2i8(i8\3* align \8 \4, i8\5* align \8 \6, i8 \7, i1 \9)~g s~call void @llvm\.mem(cpy|move)\.p([^(]*)i16\(i8([^*]*)\* (.*), i8([^*]*)\* (.*), i16 (.*), i32 ([0-9]*), i1 ([^)]*)\)~call void @llvm.mem\1.p\2i16(i8\3* align \8 \4, i8\5* align \8 \6, i16 \7, i1 \9)~g s~call void @llvm\.mem(cpy|move)\.p([^(]*)i32\(i8([^*]*)\* (.*), i8([^*]*)\* (.*), i32 (.*), i32 ([0-9]*), i1 ([^)]*)\)~call void @llvm.mem\1.p\2i32(i8\3* align \8 \4, i8\5* align \8 \6, i32 \7, i1 \9)~g s~call void @llvm\.mem(cpy|move)\.p([^(]*)i64\(i8([^*]*)\* (.*), i8([^*]*)\* (.*), i64 (.*), i32 ([0-9]*), i1 ([^)]*)\)~call void @llvm.mem\1.p\2i64(i8\3* align \8 \4, i8\5* align \8 \6, i64 \7, i1 \9)~g s~call void @llvm\.mem(cpy|move)\.p([^(]*)i128\(i8([^*]*)\* (.*), i8([^*]*)\* (.*), i128 (.*), i32 ([0-9]*), i1 ([^)]*)\)~call void @llvm.mem\1.p\2i128(i8\3* align \8 \4, i8\5* align \8 \6, i128 \7, i1 \9)~g The remaining changes in the series will: Step 2) Expand the IRBuilder API to allow creation of memcpy/memmove with differing source and dest alignments. Step 3) Update Clang to use the new IRBuilder API. Step 4) Update Polly to use the new IRBuilder API. Step 5) Update LLVM passes that create memcpy/memmove calls to use the new IRBuilder API, and those that use use MemIntrinsicInst::[get|set]Alignment() to use getDestAlignment() and getSourceAlignment() instead. Step 6) Remove the single-alignment IRBuilder API for memcpy/memmove, and the MemIntrinsicInst::[get|set]Alignment() methods. Reviewers: pete, hfinkel, lhames, reames, bollu Reviewed By: reames Subscribers: niosHD, reames, jholewinski, qcolombet, jfb, sanjoy, arsenm, dschuff, dylanmckay, mehdi_amini, sdardis, nemanjai, david2050, nhaehnle, javed.absar, sbc100, jgravelle-google, eraman, aheejin, kbarton, JDevlieghere, asb, rbar, johnrusso, simoncook, jordy.potman.lists, apazos, sabuasal, llvm-commits Differential Revision: https://reviews.llvm.org/D41675 llvm-svn: 322965
2018-01-19 18:13:12 +01:00
Assert(IsValidAlignment(MI->getDestAlignment()),
"alignment of arg 0 of memory intrinsic must be 0 or a power of 2",
CS);
Remove alignment argument from memcpy/memmove/memset in favour of alignment attributes (Step 1) Summary: This is a resurrection of work first proposed and discussed in Aug 2015: http://lists.llvm.org/pipermail/llvm-dev/2015-August/089384.html and initially landed (but then backed out) in Nov 2015: http://lists.llvm.org/pipermail/llvm-commits/Week-of-Mon-20151109/312083.html The @llvm.memcpy/memmove/memset intrinsics currently have an explicit argument which is required to be a constant integer. It represents the alignment of the dest (and source), and so must be the minimum of the actual alignment of the two. This change is the first in a series that allows source and dest to each have their own alignments by using the alignment attribute on their arguments. In this change we: 1) Remove the alignment argument. 2) Add alignment attributes to the source & dest arguments. We, temporarily, require that the alignments for source & dest be equal. For example, code which used to read: call void @llvm.memcpy.p0i8.p0i8.i32(i8* %dest, i8* %src, i32 100, i32 4, i1 false) will now read call void @llvm.memcpy.p0i8.p0i8.i32(i8* align 4 %dest, i8* align 4 %src, i32 100, i1 false) Downstream users may have to update their lit tests that check for @llvm.memcpy/memmove/memset call/declaration patterns. The following extended sed script may help with updating the majority of your tests, but it does not catch all possible patterns so some manual checking and updating will be required. s~declare void @llvm\.mem(set|cpy|move)\.p([^(]*)\((.*), i32, i1\)~declare void @llvm.mem\1.p\2(\3, i1)~g s~call void @llvm\.memset\.p([^(]*)i8\(i8([^*]*)\* (.*), i8 (.*), i8 (.*), i32 [01], i1 ([^)]*)\)~call void @llvm.memset.p\1i8(i8\2* \3, i8 \4, i8 \5, i1 \6)~g s~call void @llvm\.memset\.p([^(]*)i16\(i8([^*]*)\* (.*), i8 (.*), i16 (.*), i32 [01], i1 ([^)]*)\)~call void @llvm.memset.p\1i16(i8\2* \3, i8 \4, i16 \5, i1 \6)~g s~call void @llvm\.memset\.p([^(]*)i32\(i8([^*]*)\* (.*), i8 (.*), i32 (.*), i32 [01], i1 ([^)]*)\)~call void @llvm.memset.p\1i32(i8\2* \3, i8 \4, i32 \5, i1 \6)~g s~call void @llvm\.memset\.p([^(]*)i64\(i8([^*]*)\* (.*), i8 (.*), i64 (.*), i32 [01], i1 ([^)]*)\)~call void @llvm.memset.p\1i64(i8\2* \3, i8 \4, i64 \5, i1 \6)~g s~call void @llvm\.memset\.p([^(]*)i128\(i8([^*]*)\* (.*), i8 (.*), i128 (.*), i32 [01], i1 ([^)]*)\)~call void @llvm.memset.p\1i128(i8\2* \3, i8 \4, i128 \5, i1 \6)~g s~call void @llvm\.memset\.p([^(]*)i8\(i8([^*]*)\* (.*), i8 (.*), i8 (.*), i32 ([0-9]*), i1 ([^)]*)\)~call void @llvm.memset.p\1i8(i8\2* align \6 \3, i8 \4, i8 \5, i1 \7)~g s~call void @llvm\.memset\.p([^(]*)i16\(i8([^*]*)\* (.*), i8 (.*), i16 (.*), i32 ([0-9]*), i1 ([^)]*)\)~call void @llvm.memset.p\1i16(i8\2* align \6 \3, i8 \4, i16 \5, i1 \7)~g s~call void @llvm\.memset\.p([^(]*)i32\(i8([^*]*)\* (.*), i8 (.*), i32 (.*), i32 ([0-9]*), i1 ([^)]*)\)~call void @llvm.memset.p\1i32(i8\2* align \6 \3, i8 \4, i32 \5, i1 \7)~g s~call void @llvm\.memset\.p([^(]*)i64\(i8([^*]*)\* (.*), i8 (.*), i64 (.*), i32 ([0-9]*), i1 ([^)]*)\)~call void @llvm.memset.p\1i64(i8\2* align \6 \3, i8 \4, i64 \5, i1 \7)~g s~call void @llvm\.memset\.p([^(]*)i128\(i8([^*]*)\* (.*), i8 (.*), i128 (.*), i32 ([0-9]*), i1 ([^)]*)\)~call void @llvm.memset.p\1i128(i8\2* align \6 \3, i8 \4, i128 \5, i1 \7)~g s~call void @llvm\.mem(cpy|move)\.p([^(]*)i8\(i8([^*]*)\* (.*), i8([^*]*)\* (.*), i8 (.*), i32 [01], i1 ([^)]*)\)~call void @llvm.mem\1.p\2i8(i8\3* \4, i8\5* \6, i8 \7, i1 \8)~g s~call void @llvm\.mem(cpy|move)\.p([^(]*)i16\(i8([^*]*)\* (.*), i8([^*]*)\* (.*), i16 (.*), i32 [01], i1 ([^)]*)\)~call void @llvm.mem\1.p\2i16(i8\3* \4, i8\5* \6, i16 \7, i1 \8)~g s~call void @llvm\.mem(cpy|move)\.p([^(]*)i32\(i8([^*]*)\* (.*), i8([^*]*)\* (.*), i32 (.*), i32 [01], i1 ([^)]*)\)~call void @llvm.mem\1.p\2i32(i8\3* \4, i8\5* \6, i32 \7, i1 \8)~g s~call void @llvm\.mem(cpy|move)\.p([^(]*)i64\(i8([^*]*)\* (.*), i8([^*]*)\* (.*), i64 (.*), i32 [01], i1 ([^)]*)\)~call void @llvm.mem\1.p\2i64(i8\3* \4, i8\5* \6, i64 \7, i1 \8)~g s~call void @llvm\.mem(cpy|move)\.p([^(]*)i128\(i8([^*]*)\* (.*), i8([^*]*)\* (.*), i128 (.*), i32 [01], i1 ([^)]*)\)~call void @llvm.mem\1.p\2i128(i8\3* \4, i8\5* \6, i128 \7, i1 \8)~g s~call void @llvm\.mem(cpy|move)\.p([^(]*)i8\(i8([^*]*)\* (.*), i8([^*]*)\* (.*), i8 (.*), i32 ([0-9]*), i1 ([^)]*)\)~call void @llvm.mem\1.p\2i8(i8\3* align \8 \4, i8\5* align \8 \6, i8 \7, i1 \9)~g s~call void @llvm\.mem(cpy|move)\.p([^(]*)i16\(i8([^*]*)\* (.*), i8([^*]*)\* (.*), i16 (.*), i32 ([0-9]*), i1 ([^)]*)\)~call void @llvm.mem\1.p\2i16(i8\3* align \8 \4, i8\5* align \8 \6, i16 \7, i1 \9)~g s~call void @llvm\.mem(cpy|move)\.p([^(]*)i32\(i8([^*]*)\* (.*), i8([^*]*)\* (.*), i32 (.*), i32 ([0-9]*), i1 ([^)]*)\)~call void @llvm.mem\1.p\2i32(i8\3* align \8 \4, i8\5* align \8 \6, i32 \7, i1 \9)~g s~call void @llvm\.mem(cpy|move)\.p([^(]*)i64\(i8([^*]*)\* (.*), i8([^*]*)\* (.*), i64 (.*), i32 ([0-9]*), i1 ([^)]*)\)~call void @llvm.mem\1.p\2i64(i8\3* align \8 \4, i8\5* align \8 \6, i64 \7, i1 \9)~g s~call void @llvm\.mem(cpy|move)\.p([^(]*)i128\(i8([^*]*)\* (.*), i8([^*]*)\* (.*), i128 (.*), i32 ([0-9]*), i1 ([^)]*)\)~call void @llvm.mem\1.p\2i128(i8\3* align \8 \4, i8\5* align \8 \6, i128 \7, i1 \9)~g The remaining changes in the series will: Step 2) Expand the IRBuilder API to allow creation of memcpy/memmove with differing source and dest alignments. Step 3) Update Clang to use the new IRBuilder API. Step 4) Update Polly to use the new IRBuilder API. Step 5) Update LLVM passes that create memcpy/memmove calls to use the new IRBuilder API, and those that use use MemIntrinsicInst::[get|set]Alignment() to use getDestAlignment() and getSourceAlignment() instead. Step 6) Remove the single-alignment IRBuilder API for memcpy/memmove, and the MemIntrinsicInst::[get|set]Alignment() methods. Reviewers: pete, hfinkel, lhames, reames, bollu Reviewed By: reames Subscribers: niosHD, reames, jholewinski, qcolombet, jfb, sanjoy, arsenm, dschuff, dylanmckay, mehdi_amini, sdardis, nemanjai, david2050, nhaehnle, javed.absar, sbc100, jgravelle-google, eraman, aheejin, kbarton, JDevlieghere, asb, rbar, johnrusso, simoncook, jordy.potman.lists, apazos, sabuasal, llvm-commits Differential Revision: https://reviews.llvm.org/D41675 llvm-svn: 322965
2018-01-19 18:13:12 +01:00
if (const auto *MTI = dyn_cast<MemTransferInst>(MI)) {
Assert(IsValidAlignment(MTI->getSourceAlignment()),
"alignment of arg 1 of memory intrinsic must be 0 or a power of 2",
CS);
}
Assert(isa<ConstantInt>(CS.getArgOperand(3)),
"isvolatile argument of memory intrinsics must be a constant int",
CS);
break;
}
case Intrinsic::memcpy_element_unordered_atomic:
case Intrinsic::memmove_element_unordered_atomic:
case Intrinsic::memset_element_unordered_atomic: {
const auto *AMI = cast<AtomicMemIntrinsic>(CS.getInstruction());
ConstantInt *ElementSizeCI =
dyn_cast<ConstantInt>(AMI->getRawElementSizeInBytes());
Assert(ElementSizeCI,
"element size of the element-wise unordered atomic memory "
"intrinsic must be a constant int",
CS);
const APInt &ElementSizeVal = ElementSizeCI->getValue();
Assert(ElementSizeVal.isPowerOf2(),
"element size of the element-wise atomic memory intrinsic "
"must be a power of 2",
CS);
if (auto *LengthCI = dyn_cast<ConstantInt>(AMI->getLength())) {
uint64_t Length = LengthCI->getZExtValue();
uint64_t ElementSize = AMI->getElementSizeInBytes();
Assert((Length % ElementSize) == 0,
"constant length must be a multiple of the element size in the "
"element-wise atomic memory intrinsic",
CS);
}
auto IsValidAlignment = [&](uint64_t Alignment) {
return isPowerOf2_64(Alignment) && ElementSizeVal.ule(Alignment);
};
uint64_t DstAlignment = AMI->getDestAlignment();
Assert(IsValidAlignment(DstAlignment),
"incorrect alignment of the destination argument", CS);
if (const auto *AMT = dyn_cast<AtomicMemTransferInst>(AMI)) {
uint64_t SrcAlignment = AMT->getSourceAlignment();
Assert(IsValidAlignment(SrcAlignment),
"incorrect alignment of the source argument", CS);
}
break;
}
Reverting r55227. This was causing the following failures in the regression tests: Running /Volumes/Sandbox/Buildbot/llvm/full-llvm/build/llvm.HEAD.src/test/Verifier/dg.exp ... FAIL: /Volumes/Sandbox/Buildbot/llvm/full-llvm/build/llvm.HEAD.src/test/Verifier/gcread-ptrptr.ll for PR1633 Failed with exit(1) at line 1 while running: not llvm-as < /Volumes/Sandbox/Buildbot/llvm/full-llvm/build/llvm.HEAD.src/test/Verifier/gcread-ptrptr.ll >& /dev/null child process exited abnormally FAIL: /Volumes/Sandbox/Buildbot/llvm/full-llvm/build/llvm.HEAD.src/test/Verifier/gcroot-alloca.ll for PR1633 Failed with exit(1) at line 1 while running: not llvm-as < /Volumes/Sandbox/Buildbot/llvm/full-llvm/build/llvm.HEAD.src/test/Verifier/gcroot-alloca.ll >& /dev/null child process exited abnormally FAIL: /Volumes/Sandbox/Buildbot/llvm/full-llvm/build/llvm.HEAD.src/test/Verifier/gcroot-meta.ll for PR1633 Failed with exit(1) at line 1 while running: not llvm-as < /Volumes/Sandbox/Buildbot/llvm/full-llvm/build/llvm.HEAD.src/test/Verifier/gcroot-meta.ll >& /dev/null child process exited abnormally FAIL: ndbox/Buildbot/llvm/full-llvm/build/llvm.HEAD.src/test/Verifier/gcroot-ptrptr.ll for PR1633 Failed with exit(1) at line 1 while running: not llvm-as < /Volumes/Sandbox/Buildbot/llvm/full-llvm/build/llvm.HEAD.src/test/Verifier/gcroot-ptrptr.ll >& /dev/null child process exited abnormally FAIL: /Volumes/Sandbox/Buildbot/llvm/full-llvm/build/llvm.HEAD.src/test/Verifier/gcwrite-ptrptr.ll for PR1633 Failed with exit(1) at line 1 while running: not llvm-as < /Volumes/Sandbox/Buildbot/llvm/full-llvm/build/llvm.HEAD.src/test/Verifier/gcwrite-ptrptr.ll >& /dev/null child process exited abnormally === Summary === # of expected passes 3021 # of unexpected failures 6 # of expected failures 16 make[1]: *** [check-local] Error 1 make: *** [check] Error 2 llvm-svn: 55233
2008-08-23 11:46:46 +02:00
case Intrinsic::gcroot:
case Intrinsic::gcwrite:
case Intrinsic::gcread:
if (ID == Intrinsic::gcroot) {
AllocaInst *AI =
dyn_cast<AllocaInst>(CS.getArgOperand(0)->stripPointerCasts());
Assert(AI, "llvm.gcroot parameter #1 must be an alloca.", CS);
Assert(isa<Constant>(CS.getArgOperand(1)),
"llvm.gcroot parameter #2 must be a constant.", CS);
if (!AI->getAllocatedType()->isPointerTy()) {
Assert(!isa<ConstantPointerNull>(CS.getArgOperand(1)),
"llvm.gcroot parameter #1 must either be a pointer alloca, "
"or argument #2 must be a non-null constant.",
CS);
}
}
2009-09-07 22:44:51 +02:00
Assert(CS.getParent()->getParent()->hasGC(),
"Enclosing function does not use GC.", CS);
break;
case Intrinsic::init_trampoline:
Assert(isa<Function>(CS.getArgOperand(1)->stripPointerCasts()),
"llvm.init_trampoline parameter #2 must resolve to a function.",
CS);
break;
case Intrinsic::prefetch:
Assert(isa<ConstantInt>(CS.getArgOperand(1)) &&
isa<ConstantInt>(CS.getArgOperand(2)) &&
cast<ConstantInt>(CS.getArgOperand(1))->getZExtValue() < 2 &&
cast<ConstantInt>(CS.getArgOperand(2))->getZExtValue() < 4,
"invalid arguments to llvm.prefetch", CS);
break;
case Intrinsic::stackprotector:
Assert(isa<AllocaInst>(CS.getArgOperand(1)->stripPointerCasts()),
"llvm.stackprotector parameter #2 must resolve to an alloca.", CS);
break;
case Intrinsic::lifetime_start:
case Intrinsic::lifetime_end:
case Intrinsic::invariant_start:
Assert(isa<ConstantInt>(CS.getArgOperand(0)),
"size argument of memory use markers must be a constant integer",
CS);
break;
case Intrinsic::invariant_end:
Assert(isa<ConstantInt>(CS.getArgOperand(1)),
"llvm.invariant.end parameter #2 must be a constant integer", CS);
break;
case Intrinsic::localescape: {
BasicBlock *BB = CS.getParent();
Assert(BB == &BB->getParent()->front(),
"llvm.localescape used outside of entry block", CS);
Assert(!SawFrameEscape,
"multiple calls to llvm.localescape in one function", CS);
for (Value *Arg : CS.args()) {
if (isa<ConstantPointerNull>(Arg))
continue; // Null values are allowed as placeholders.
auto *AI = dyn_cast<AllocaInst>(Arg->stripPointerCasts());
Assert(AI && AI->isStaticAlloca(),
"llvm.localescape only accepts static allocas", CS);
}
FrameEscapeInfo[BB->getParent()].first = CS.getNumArgOperands();
SawFrameEscape = true;
break;
}
case Intrinsic::localrecover: {
Value *FnArg = CS.getArgOperand(0)->stripPointerCasts();
Function *Fn = dyn_cast<Function>(FnArg);
Assert(Fn && !Fn->isDeclaration(),
"llvm.localrecover first "
"argument must be function defined in this module",
CS);
auto *IdxArg = dyn_cast<ConstantInt>(CS.getArgOperand(2));
Assert(IdxArg, "idx argument of llvm.localrecover must be a constant int",
CS);
auto &Entry = FrameEscapeInfo[Fn];
Entry.second = unsigned(
std::max(uint64_t(Entry.second), IdxArg->getLimitedValue(~0U) + 1));
break;
}
case Intrinsic::experimental_gc_statepoint:
Assert(!CS.isInlineAsm(),
"gc.statepoint support for inline assembly unimplemented", CS);
Assert(CS.getParent()->getParent()->hasGC(),
"Enclosing function does not use GC.", CS);
verifyStatepoint(CS);
break;
case Intrinsic::experimental_gc_result: {
Assert(CS.getParent()->getParent()->hasGC(),
"Enclosing function does not use GC.", CS);
// Are we tied to a statepoint properly?
CallSite StatepointCS(CS.getArgOperand(0));
const Function *StatepointFn =
StatepointCS.getInstruction() ? StatepointCS.getCalledFunction() : nullptr;
Assert(StatepointFn && StatepointFn->isDeclaration() &&
StatepointFn->getIntrinsicID() ==
Intrinsic::experimental_gc_statepoint,
"gc.result operand #1 must be from a statepoint", CS,
CS.getArgOperand(0));
// Assert that result type matches wrapped callee.
const Value *Target = StatepointCS.getArgument(2);
auto *PT = cast<PointerType>(Target->getType());
auto *TargetFuncType = cast<FunctionType>(PT->getElementType());
Assert(CS.getType() == TargetFuncType->getReturnType(),
"gc.result result type does not match wrapped callee", CS);
break;
}
case Intrinsic::experimental_gc_relocate: {
Assert(CS.getNumArgOperands() == 3, "wrong number of arguments", CS);
Assert(isa<PointerType>(CS.getType()->getScalarType()),
"gc.relocate must return a pointer or a vector of pointers", CS);
// Check that this relocate is correctly tied to the statepoint
// This is case for relocate on the unwinding path of an invoke statepoint
if (LandingPadInst *LandingPad =
dyn_cast<LandingPadInst>(CS.getArgOperand(0))) {
const BasicBlock *InvokeBB =
LandingPad->getParent()->getUniquePredecessor();
// Landingpad relocates should have only one predecessor with invoke
// statepoint terminator
Assert(InvokeBB, "safepoints should have unique landingpads",
LandingPad->getParent());
Assert(InvokeBB->getTerminator(), "safepoint block should be well formed",
InvokeBB);
Assert(isStatepoint(InvokeBB->getTerminator()),
"gc relocate should be linked to a statepoint", InvokeBB);
}
else {
// In all other cases relocate should be tied to the statepoint directly.
// This covers relocates on a normal return path of invoke statepoint and
// relocates of a call statepoint.
auto Token = CS.getArgOperand(0);
Assert(isa<Instruction>(Token) && isStatepoint(cast<Instruction>(Token)),
"gc relocate is incorrectly tied to the statepoint", CS, Token);
}
// Verify rest of the relocate arguments.
ImmutableCallSite StatepointCS(
cast<GCRelocateInst>(*CS.getInstruction()).getStatepoint());
// Both the base and derived must be piped through the safepoint.
Value* Base = CS.getArgOperand(1);
Assert(isa<ConstantInt>(Base),
"gc.relocate operand #2 must be integer offset", CS);
Value* Derived = CS.getArgOperand(2);
Assert(isa<ConstantInt>(Derived),
"gc.relocate operand #3 must be integer offset", CS);
const int BaseIndex = cast<ConstantInt>(Base)->getZExtValue();
const int DerivedIndex = cast<ConstantInt>(Derived)->getZExtValue();
// Check the bounds
Assert(0 <= BaseIndex && BaseIndex < (int)StatepointCS.arg_size(),
"gc.relocate: statepoint base index out of bounds", CS);
Assert(0 <= DerivedIndex && DerivedIndex < (int)StatepointCS.arg_size(),
"gc.relocate: statepoint derived index out of bounds", CS);
// Check that BaseIndex and DerivedIndex fall within the 'gc parameters'
// section of the statepoint's argument.
Assert(StatepointCS.arg_size() > 0,
"gc.statepoint: insufficient arguments");
Assert(isa<ConstantInt>(StatepointCS.getArgument(3)),
"gc.statement: number of call arguments must be constant integer");
const unsigned NumCallArgs =
cast<ConstantInt>(StatepointCS.getArgument(3))->getZExtValue();
Assert(StatepointCS.arg_size() > NumCallArgs + 5,
"gc.statepoint: mismatch in number of call arguments");
Assert(isa<ConstantInt>(StatepointCS.getArgument(NumCallArgs + 5)),
"gc.statepoint: number of transition arguments must be "
"a constant integer");
const int NumTransitionArgs =
cast<ConstantInt>(StatepointCS.getArgument(NumCallArgs + 5))
->getZExtValue();
const int DeoptArgsStart = 4 + NumCallArgs + 1 + NumTransitionArgs + 1;
Assert(isa<ConstantInt>(StatepointCS.getArgument(DeoptArgsStart)),
"gc.statepoint: number of deoptimization arguments must be "
"a constant integer");
const int NumDeoptArgs =
cast<ConstantInt>(StatepointCS.getArgument(DeoptArgsStart))
->getZExtValue();
const int GCParamArgsStart = DeoptArgsStart + 1 + NumDeoptArgs;
const int GCParamArgsEnd = StatepointCS.arg_size();
Assert(GCParamArgsStart <= BaseIndex && BaseIndex < GCParamArgsEnd,
"gc.relocate: statepoint base index doesn't fall within the "
"'gc parameters' section of the statepoint call",
CS);
Assert(GCParamArgsStart <= DerivedIndex && DerivedIndex < GCParamArgsEnd,
"gc.relocate: statepoint derived index doesn't fall within the "
"'gc parameters' section of the statepoint call",
CS);
// Relocated value must be either a pointer type or vector-of-pointer type,
// but gc_relocate does not need to return the same pointer type as the
// relocated pointer. It can be casted to the correct type later if it's
// desired. However, they must have the same address space and 'vectorness'
GCRelocateInst &Relocate = cast<GCRelocateInst>(*CS.getInstruction());
Assert(Relocate.getDerivedPtr()->getType()->isPtrOrPtrVectorTy(),
"gc.relocate: relocated value must be a gc pointer", CS);
auto ResultType = CS.getType();
auto DerivedType = Relocate.getDerivedPtr()->getType();
Assert(ResultType->isVectorTy() == DerivedType->isVectorTy(),
"gc.relocate: vector relocates to vector and pointer to pointer",
CS);
Assert(
ResultType->getPointerAddressSpace() ==
DerivedType->getPointerAddressSpace(),
"gc.relocate: relocating a pointer shouldn't change its address space",
CS);
break;
}
case Intrinsic::eh_exceptioncode:
case Intrinsic::eh_exceptionpointer: {
Assert(isa<CatchPadInst>(CS.getArgOperand(0)),
"eh.exceptionpointer argument must be a catchpad", CS);
break;
}
case Intrinsic::masked_load: {
Assert(CS.getType()->isVectorTy(), "masked_load: must return a vector", CS);
Value *Ptr = CS.getArgOperand(0);
//Value *Alignment = CS.getArgOperand(1);
Value *Mask = CS.getArgOperand(2);
Value *PassThru = CS.getArgOperand(3);
Assert(Mask->getType()->isVectorTy(),
"masked_load: mask must be vector", CS);
// DataTy is the overloaded type
Type *DataTy = cast<PointerType>(Ptr->getType())->getElementType();
Assert(DataTy == CS.getType(),
"masked_load: return must match pointer type", CS);
Assert(PassThru->getType() == DataTy,
"masked_load: pass through and data type must match", CS);
Assert(Mask->getType()->getVectorNumElements() ==
DataTy->getVectorNumElements(),
"masked_load: vector mask must be same length as data", CS);
break;
}
case Intrinsic::masked_store: {
Value *Val = CS.getArgOperand(0);
Value *Ptr = CS.getArgOperand(1);
//Value *Alignment = CS.getArgOperand(2);
Value *Mask = CS.getArgOperand(3);
Assert(Mask->getType()->isVectorTy(),
"masked_store: mask must be vector", CS);
// DataTy is the overloaded type
Type *DataTy = cast<PointerType>(Ptr->getType())->getElementType();
Assert(DataTy == Val->getType(),
"masked_store: storee must match pointer type", CS);
Assert(Mask->getType()->getVectorNumElements() ==
DataTy->getVectorNumElements(),
"masked_store: vector mask must be same length as data", CS);
break;
}
case Intrinsic::experimental_guard: {
Assert(CS.isCall(), "experimental_guard cannot be invoked", CS);
Assert(CS.countOperandBundlesOfType(LLVMContext::OB_deopt) == 1,
"experimental_guard must have exactly one "
"\"deopt\" operand bundle");
break;
}
case Intrinsic::experimental_deoptimize: {
Assert(CS.isCall(), "experimental_deoptimize cannot be invoked", CS);
Assert(CS.countOperandBundlesOfType(LLVMContext::OB_deopt) == 1,
"experimental_deoptimize must have exactly one "
"\"deopt\" operand bundle");
Assert(CS.getType() == CS.getInstruction()->getFunction()->getReturnType(),
"experimental_deoptimize return type must match caller return type");
if (CS.isCall()) {
auto *DeoptCI = CS.getInstruction();
auto *RI = dyn_cast<ReturnInst>(DeoptCI->getNextNode());
Assert(RI,
"calls to experimental_deoptimize must be followed by a return");
if (!CS.getType()->isVoidTy() && RI)
Assert(RI->getReturnValue() == DeoptCI,
"calls to experimental_deoptimize must be followed by a return "
"of the value computed by experimental_deoptimize");
}
break;
}
case Intrinsic::sadd_sat:
case Intrinsic::uadd_sat:
case Intrinsic::ssub_sat:
case Intrinsic::usub_sat: {
Value *Op1 = CS.getArgOperand(0);
Value *Op2 = CS.getArgOperand(1);
Assert(Op1->getType()->isIntOrIntVectorTy(),
"first operand of [us][add|sub]_sat must be an int type or vector "
"of ints");
Assert(Op2->getType()->isIntOrIntVectorTy(),
"second operand of [us][add|sub]_sat must be an int type or vector "
"of ints");
break;
}
case Intrinsic::smul_fix: {
Value *Op1 = CS.getArgOperand(0);
Value *Op2 = CS.getArgOperand(1);
Assert(Op1->getType()->isIntOrIntVectorTy(),
"first operand of smul_fix must be an int type or vector "
"of ints");
Assert(Op2->getType()->isIntOrIntVectorTy(),
"second operand of smul_fix must be an int type or vector "
"of ints");
auto *Op3 = dyn_cast<ConstantInt>(CS.getArgOperand(2));
Assert(Op3, "third argument of smul_fix must be a constant integer");
Assert(Op3->getType()->getBitWidth() <= 32,
"third argument of smul_fix must fit within 32 bits");
Assert(Op3->getZExtValue() < Op1->getType()->getScalarSizeInBits(),
"the scale of smul_fix must be less than the width of the operands");
break;
}
};
}
/// Carefully grab the subprogram from a local scope.
///
/// This carefully grabs the subprogram from a local scope, avoiding the
/// built-in assertions that would typically fire.
static DISubprogram *getSubprogram(Metadata *LocalScope) {
if (!LocalScope)
return nullptr;
if (auto *SP = dyn_cast<DISubprogram>(LocalScope))
return SP;
if (auto *LB = dyn_cast<DILexicalBlockBase>(LocalScope))
return getSubprogram(LB->getRawScope());
// Just return null; broken scope chains are checked elsewhere.
assert(!isa<DILocalScope>(LocalScope) && "Unknown type of local scope");
return nullptr;
}
void Verifier::visitConstrainedFPIntrinsic(ConstrainedFPIntrinsic &FPI) {
unsigned NumOperands = FPI.getNumArgOperands();
Assert(((NumOperands == 5 && FPI.isTernaryOp()) ||
(NumOperands == 3 && FPI.isUnaryOp()) || (NumOperands == 4)),
"invalid arguments for constrained FP intrinsic", &FPI);
Assert(isa<MetadataAsValue>(FPI.getArgOperand(NumOperands-1)),
"invalid exception behavior argument", &FPI);
Assert(isa<MetadataAsValue>(FPI.getArgOperand(NumOperands-2)),
"invalid rounding mode argument", &FPI);
Assert(FPI.getRoundingMode() != ConstrainedFPIntrinsic::rmInvalid,
"invalid rounding mode argument", &FPI);
Assert(FPI.getExceptionBehavior() != ConstrainedFPIntrinsic::ebInvalid,
"invalid exception behavior argument", &FPI);
}
void Verifier::visitDbgIntrinsic(StringRef Kind, DbgVariableIntrinsic &DII) {
auto *MD = cast<MetadataAsValue>(DII.getArgOperand(0))->getMetadata();
AssertDI(isa<ValueAsMetadata>(MD) ||
(isa<MDNode>(MD) && !cast<MDNode>(MD)->getNumOperands()),
"invalid llvm.dbg." + Kind + " intrinsic address/value", &DII, MD);
AssertDI(isa<DILocalVariable>(DII.getRawVariable()),
"invalid llvm.dbg." + Kind + " intrinsic variable", &DII,
DII.getRawVariable());
AssertDI(isa<DIExpression>(DII.getRawExpression()),
"invalid llvm.dbg." + Kind + " intrinsic expression", &DII,
DII.getRawExpression());
Verifier: Check that inlined-at locations agree Check that the `MDLocalVariable::getInlinedAt()` in a debug info intrinsic's variable always matches the `MDLocation::getInlinedAt()` of its `!dbg` attachment. The goal here is to get rid of `MDLocalVariable::getInlinedAt()` entirely (PR22778), since it's expensive and unnecessary, but I'll let this verifier check bake for a while (a week maybe?) first. I've updated the testcases that had the wrong value for `inlinedAt:`. This checks that things are sane in the IR, but currently things go out of whack in a few places in the backend. I'll follow shortly with assertions in the backend (with code fixes). If you have out-of-tree testcases that just started failing, here's how I updated these ones: 1. The verifier check gives you the basic block, function, instruction, and relevant metadata arguments (metadata numbering doesn't necessarily match the source file, unfortunately). 2. Look at the `@llvm.dbg.*()` instruction, and compare the `inlinedAt:` fields of the variable argument (second `metadata` argument) and the `!dbg` attachment. 3. Figure out based on the variable `scope:` chain and the functions in the file whether the variable has been inlined (and into what), so you can determine which `inlinedAt:` is actually correct. In all of the in-tree testcases, the `!MDLocation()` was correct and the `!MDLocalVariable()` was wrong, but YMMV. 4. Duplicate the metadata that you're going to change, and add/drop the `inlinedAt:` field from one of them. Be careful that the other references to the same metadata node point at the correct one. llvm-svn: 234021
2015-04-03 18:54:30 +02:00
// Ignore broken !dbg attachments; they're checked elsewhere.
if (MDNode *N = DII.getDebugLoc().getAsMDNode())
if (!isa<DILocation>(N))
Verifier: Check that inlined-at locations agree Check that the `MDLocalVariable::getInlinedAt()` in a debug info intrinsic's variable always matches the `MDLocation::getInlinedAt()` of its `!dbg` attachment. The goal here is to get rid of `MDLocalVariable::getInlinedAt()` entirely (PR22778), since it's expensive and unnecessary, but I'll let this verifier check bake for a while (a week maybe?) first. I've updated the testcases that had the wrong value for `inlinedAt:`. This checks that things are sane in the IR, but currently things go out of whack in a few places in the backend. I'll follow shortly with assertions in the backend (with code fixes). If you have out-of-tree testcases that just started failing, here's how I updated these ones: 1. The verifier check gives you the basic block, function, instruction, and relevant metadata arguments (metadata numbering doesn't necessarily match the source file, unfortunately). 2. Look at the `@llvm.dbg.*()` instruction, and compare the `inlinedAt:` fields of the variable argument (second `metadata` argument) and the `!dbg` attachment. 3. Figure out based on the variable `scope:` chain and the functions in the file whether the variable has been inlined (and into what), so you can determine which `inlinedAt:` is actually correct. In all of the in-tree testcases, the `!MDLocation()` was correct and the `!MDLocalVariable()` was wrong, but YMMV. 4. Duplicate the metadata that you're going to change, and add/drop the `inlinedAt:` field from one of them. Be careful that the other references to the same metadata node point at the correct one. llvm-svn: 234021
2015-04-03 18:54:30 +02:00
return;
BasicBlock *BB = DII.getParent();
Function *F = BB ? BB->getParent() : nullptr;
// The scopes for variables and !dbg attachments must agree.
DILocalVariable *Var = DII.getVariable();
DILocation *Loc = DII.getDebugLoc();
AssertDI(Loc, "llvm.dbg." + Kind + " intrinsic requires a !dbg attachment",
&DII, BB, F);
DISubprogram *VarSP = getSubprogram(Var->getRawScope());
DISubprogram *LocSP = getSubprogram(Loc->getRawScope());
if (!VarSP || !LocSP)
return; // Broken scope chains are checked elsewhere.
AssertDI(VarSP == LocSP, "mismatched subprogram between llvm.dbg." + Kind +
" variable and !dbg attachment",
&DII, BB, F, Var, Var->getScope()->getSubprogram(), Loc,
Loc->getScope()->getSubprogram());
// This check is redundant with one in visitLocalVariable().
AssertDI(isType(Var->getRawType()), "invalid type ref", Var,
Var->getRawType());
if (auto *Type = dyn_cast_or_null<DIType>(Var->getRawType()))
if (Type->isBlockByrefStruct())
AssertDI(DII.getExpression() && DII.getExpression()->getNumElements(),
"BlockByRef variable without complex expression", Var, &DII);
verifyFnArgs(DII);
}
void Verifier::visitDbgLabelIntrinsic(StringRef Kind, DbgLabelInst &DLI) {
AssertDI(isa<DILabel>(DLI.getRawLabel()),
"invalid llvm.dbg." + Kind + " intrinsic variable", &DLI,
DLI.getRawLabel());
// Ignore broken !dbg attachments; they're checked elsewhere.
if (MDNode *N = DLI.getDebugLoc().getAsMDNode())
if (!isa<DILocation>(N))
return;
BasicBlock *BB = DLI.getParent();
Function *F = BB ? BB->getParent() : nullptr;
// The scopes for variables and !dbg attachments must agree.
DILabel *Label = DLI.getLabel();
DILocation *Loc = DLI.getDebugLoc();
Assert(Loc, "llvm.dbg." + Kind + " intrinsic requires a !dbg attachment",
&DLI, BB, F);
DISubprogram *LabelSP = getSubprogram(Label->getRawScope());
DISubprogram *LocSP = getSubprogram(Loc->getRawScope());
if (!LabelSP || !LocSP)
return;
AssertDI(LabelSP == LocSP, "mismatched subprogram between llvm.dbg." + Kind +
" label and !dbg attachment",
&DLI, BB, F, Label, Label->getScope()->getSubprogram(), Loc,
Loc->getScope()->getSubprogram());
}
void Verifier::verifyFragmentExpression(const DbgVariableIntrinsic &I) {
DILocalVariable *V = dyn_cast_or_null<DILocalVariable>(I.getRawVariable());
DIExpression *E = dyn_cast_or_null<DIExpression>(I.getRawExpression());
// We don't know whether this intrinsic verified correctly.
if (!V || !E || !E->isValid())
return;
// Nothing to do if this isn't a DW_OP_LLVM_fragment expression.
auto Fragment = E->getFragmentInfo();
if (!Fragment)
return;
// The frontend helps out GDB by emitting the members of local anonymous
// unions as artificial local variables with shared storage. When SROA splits
// the storage for artificial local variables that are smaller than the entire
// union, the overhang piece will be outside of the allotted space for the
// variable and this check fails.
// FIXME: Remove this check as soon as clang stops doing this; it hides bugs.
if (V->isArtificial())
return;
verifyFragmentExpression(*V, *Fragment, &I);
}
template <typename ValueOrMetadata>
void Verifier::verifyFragmentExpression(const DIVariable &V,
DIExpression::FragmentInfo Fragment,
ValueOrMetadata *Desc) {
// If there's no size, the type is broken, but that should be checked
// elsewhere.
auto VarSize = V.getSizeInBits();
if (!VarSize)
return;
unsigned FragSize = Fragment.SizeInBits;
unsigned FragOffset = Fragment.OffsetInBits;
AssertDI(FragSize + FragOffset <= *VarSize,
"fragment is larger than or outside of variable", Desc, &V);
AssertDI(FragSize != *VarSize, "fragment covers entire variable", Desc, &V);
}
void Verifier::verifyFnArgs(const DbgVariableIntrinsic &I) {
// This function does not take the scope of noninlined function arguments into
// account. Don't run it if current function is nodebug, because it may
// contain inlined debug intrinsics.
if (!HasDebugInfo)
return;
// For performance reasons only check non-inlined ones.
if (I.getDebugLoc()->getInlinedAt())
return;
DILocalVariable *Var = I.getVariable();
AssertDI(Var, "dbg intrinsic without variable");
unsigned ArgNo = Var->getArg();
if (!ArgNo)
return;
// Verify there are no duplicate function argument debug info entries.
// These will cause hard-to-debug assertions in the DWARF backend.
if (DebugFnArgs.size() < ArgNo)
DebugFnArgs.resize(ArgNo, nullptr);
auto *Prev = DebugFnArgs[ArgNo - 1];
DebugFnArgs[ArgNo - 1] = Var;
AssertDI(!Prev || (Prev == Var), "conflicting debug info for argument", &I,
Prev, Var);
}
void Verifier::verifyCompileUnits() {
// When more than one Module is imported into the same context, such as during
// an LTO build before linking the modules, ODR type uniquing may cause types
// to point to a different CU. This check does not make sense in this case.
if (M.getContext().isODRUniquingDebugTypes())
return;
auto *CUs = M.getNamedMetadata("llvm.dbg.cu");
SmallPtrSet<const Metadata *, 2> Listed;
if (CUs)
Listed.insert(CUs->op_begin(), CUs->op_end());
for (auto *CU : CUVisited)
AssertDI(Listed.count(CU), "DICompileUnit not listed in llvm.dbg.cu", CU);
CUVisited.clear();
}
void Verifier::verifyDeoptimizeCallingConvs() {
if (DeoptimizeDeclarations.empty())
return;
const Function *First = DeoptimizeDeclarations[0];
for (auto *F : makeArrayRef(DeoptimizeDeclarations).slice(1)) {
Assert(First->getCallingConv() == F->getCallingConv(),
"All llvm.experimental.deoptimize declarations must have the same "
"calling convention",
First, F);
}
}
void Verifier::verifySourceDebugInfo(const DICompileUnit &U, const DIFile &F) {
bool HasSource = F.getSource().hasValue();
if (!HasSourceDebugInfo.count(&U))
HasSourceDebugInfo[&U] = HasSource;
AssertDI(HasSource == HasSourceDebugInfo[&U],
"inconsistent use of embedded source");
}
//===----------------------------------------------------------------------===//
// Implement the public interfaces to this file...
//===----------------------------------------------------------------------===//
bool llvm::verifyFunction(const Function &f, raw_ostream *OS) {
Function &F = const_cast<Function &>(f);
// Don't use a raw_null_ostream. Printing IR is expensive.
Verifier V(OS, /*ShouldTreatBrokenDebugInfoAsError=*/true, *f.getParent());
// Note that this function's return value is inverted from what you would
// expect of a function called "verify".
return !V.verify(F);
}
bool llvm::verifyModule(const Module &M, raw_ostream *OS,
bool *BrokenDebugInfo) {
// Don't use a raw_null_ostream. Printing IR is expensive.
Verifier V(OS, /*ShouldTreatBrokenDebugInfoAsError=*/!BrokenDebugInfo, M);
bool Broken = false;
2016-02-25 17:44:27 +01:00
for (const Function &F : M)
Broken |= !V.verify(F);
Broken |= !V.verify();
if (BrokenDebugInfo)
*BrokenDebugInfo = V.hasBrokenDebugInfo();
// Note that this function's return value is inverted from what you would
// expect of a function called "verify".
return Broken;
}
namespace {
struct VerifierLegacyPass : public FunctionPass {
static char ID;
std::unique_ptr<Verifier> V;
bool FatalErrors = true;
VerifierLegacyPass() : FunctionPass(ID) {
initializeVerifierLegacyPassPass(*PassRegistry::getPassRegistry());
}
explicit VerifierLegacyPass(bool FatalErrors)
: FunctionPass(ID),
FatalErrors(FatalErrors) {
initializeVerifierLegacyPassPass(*PassRegistry::getPassRegistry());
}
bool doInitialization(Module &M) override {
V = llvm::make_unique<Verifier>(
&dbgs(), /*ShouldTreatBrokenDebugInfoAsError=*/false, M);
return false;
}
bool runOnFunction(Function &F) override {
if (!V->verify(F) && FatalErrors) {
errs() << "in function " << F.getName() << '\n';
report_fatal_error("Broken function found, compilation aborted!");
}
return false;
}
bool doFinalization(Module &M) override {
bool HasErrors = false;
for (Function &F : M)
if (F.isDeclaration())
HasErrors |= !V->verify(F);
HasErrors |= !V->verify();
if (FatalErrors && (HasErrors || V->hasBrokenDebugInfo()))
report_fatal_error("Broken module found, compilation aborted!");
return false;
}
void getAnalysisUsage(AnalysisUsage &AU) const override {
AU.setPreservesAll();
}
};
} // end anonymous namespace
/// Helper to issue failure from the TBAA verification
template <typename... Tys> void TBAAVerifier::CheckFailed(Tys &&... Args) {
if (Diagnostic)
return Diagnostic->CheckFailed(Args...);
}
#define AssertTBAA(C, ...) \
do { \
if (!(C)) { \
CheckFailed(__VA_ARGS__); \
return false; \
} \
} while (false)
/// Verify that \p BaseNode can be used as the "base type" in the struct-path
/// TBAA scheme. This means \p BaseNode is either a scalar node, or a
/// struct-type node describing an aggregate data structure (like a struct).
TBAAVerifier::TBAABaseNodeSummary
TBAAVerifier::verifyTBAABaseNode(Instruction &I, const MDNode *BaseNode,
bool IsNewFormat) {
if (BaseNode->getNumOperands() < 2) {
CheckFailed("Base nodes must have at least two operands", &I, BaseNode);
return {true, ~0u};
}
auto Itr = TBAABaseNodes.find(BaseNode);
if (Itr != TBAABaseNodes.end())
return Itr->second;
auto Result = verifyTBAABaseNodeImpl(I, BaseNode, IsNewFormat);
auto InsertResult = TBAABaseNodes.insert({BaseNode, Result});
(void)InsertResult;
assert(InsertResult.second && "We just checked!");
return Result;
}
TBAAVerifier::TBAABaseNodeSummary
TBAAVerifier::verifyTBAABaseNodeImpl(Instruction &I, const MDNode *BaseNode,
bool IsNewFormat) {
const TBAAVerifier::TBAABaseNodeSummary InvalidNode = {true, ~0u};
if (BaseNode->getNumOperands() == 2) {
// Scalar nodes can only be accessed at offset 0.
return isValidScalarTBAANode(BaseNode)
? TBAAVerifier::TBAABaseNodeSummary({false, 0})
: InvalidNode;
}
if (IsNewFormat) {
if (BaseNode->getNumOperands() % 3 != 0) {
CheckFailed("Access tag nodes must have the number of operands that is a "
"multiple of 3!", BaseNode);
return InvalidNode;
}
} else {
if (BaseNode->getNumOperands() % 2 != 1) {
CheckFailed("Struct tag nodes must have an odd number of operands!",
BaseNode);
return InvalidNode;
}
}
// Check the type size field.
if (IsNewFormat) {
auto *TypeSizeNode = mdconst::dyn_extract_or_null<ConstantInt>(
BaseNode->getOperand(1));
if (!TypeSizeNode) {
CheckFailed("Type size nodes must be constants!", &I, BaseNode);
return InvalidNode;
}
}
// Check the type name field. In the new format it can be anything.
if (!IsNewFormat && !isa<MDString>(BaseNode->getOperand(0))) {
CheckFailed("Struct tag nodes have a string as their first operand",
BaseNode);
return InvalidNode;
}
bool Failed = false;
Optional<APInt> PrevOffset;
unsigned BitWidth = ~0u;
// We've already checked that BaseNode is not a degenerate root node with one
// operand in \c verifyTBAABaseNode, so this loop should run at least once.
unsigned FirstFieldOpNo = IsNewFormat ? 3 : 1;
unsigned NumOpsPerField = IsNewFormat ? 3 : 2;
for (unsigned Idx = FirstFieldOpNo; Idx < BaseNode->getNumOperands();
Idx += NumOpsPerField) {
const MDOperand &FieldTy = BaseNode->getOperand(Idx);
const MDOperand &FieldOffset = BaseNode->getOperand(Idx + 1);
if (!isa<MDNode>(FieldTy)) {
CheckFailed("Incorrect field entry in struct type node!", &I, BaseNode);
Failed = true;
continue;
}
auto *OffsetEntryCI =
mdconst::dyn_extract_or_null<ConstantInt>(FieldOffset);
if (!OffsetEntryCI) {
CheckFailed("Offset entries must be constants!", &I, BaseNode);
Failed = true;
continue;
}
if (BitWidth == ~0u)
BitWidth = OffsetEntryCI->getBitWidth();
if (OffsetEntryCI->getBitWidth() != BitWidth) {
CheckFailed(
"Bitwidth between the offsets and struct type entries must match", &I,
BaseNode);
Failed = true;
continue;
}
// NB! As far as I can tell, we generate a non-strictly increasing offset
// sequence only from structs that have zero size bit fields. When
// recursing into a contained struct in \c getFieldNodeFromTBAABaseNode we
// pick the field lexically the latest in struct type metadata node. This
// mirrors the actual behavior of the alias analysis implementation.
bool IsAscending =
!PrevOffset || PrevOffset->ule(OffsetEntryCI->getValue());
if (!IsAscending) {
CheckFailed("Offsets must be increasing!", &I, BaseNode);
Failed = true;
}
PrevOffset = OffsetEntryCI->getValue();
if (IsNewFormat) {
auto *MemberSizeNode = mdconst::dyn_extract_or_null<ConstantInt>(
BaseNode->getOperand(Idx + 2));
if (!MemberSizeNode) {
CheckFailed("Member size entries must be constants!", &I, BaseNode);
Failed = true;
continue;
}
}
}
return Failed ? InvalidNode
: TBAAVerifier::TBAABaseNodeSummary(false, BitWidth);
}
static bool IsRootTBAANode(const MDNode *MD) {
return MD->getNumOperands() < 2;
}
static bool IsScalarTBAANodeImpl(const MDNode *MD,
SmallPtrSetImpl<const MDNode *> &Visited) {
if (MD->getNumOperands() != 2 && MD->getNumOperands() != 3)
return false;
if (!isa<MDString>(MD->getOperand(0)))
return false;
if (MD->getNumOperands() == 3) {
auto *Offset = mdconst::dyn_extract<ConstantInt>(MD->getOperand(2));
if (!(Offset && Offset->isZero() && isa<MDString>(MD->getOperand(0))))
return false;
}
auto *Parent = dyn_cast_or_null<MDNode>(MD->getOperand(1));
return Parent && Visited.insert(Parent).second &&
(IsRootTBAANode(Parent) || IsScalarTBAANodeImpl(Parent, Visited));
}
bool TBAAVerifier::isValidScalarTBAANode(const MDNode *MD) {
auto ResultIt = TBAAScalarNodes.find(MD);
if (ResultIt != TBAAScalarNodes.end())
return ResultIt->second;
SmallPtrSet<const MDNode *, 4> Visited;
bool Result = IsScalarTBAANodeImpl(MD, Visited);
auto InsertResult = TBAAScalarNodes.insert({MD, Result});
(void)InsertResult;
assert(InsertResult.second && "Just checked!");
return Result;
}
/// Returns the field node at the offset \p Offset in \p BaseNode. Update \p
/// Offset in place to be the offset within the field node returned.
///
/// We assume we've okayed \p BaseNode via \c verifyTBAABaseNode.
MDNode *TBAAVerifier::getFieldNodeFromTBAABaseNode(Instruction &I,
const MDNode *BaseNode,
APInt &Offset,
bool IsNewFormat) {
assert(BaseNode->getNumOperands() >= 2 && "Invalid base node!");
// Scalar nodes have only one possible "field" -- their parent in the access
// hierarchy. Offset must be zero at this point, but our caller is supposed
// to Assert that.
if (BaseNode->getNumOperands() == 2)
return cast<MDNode>(BaseNode->getOperand(1));
unsigned FirstFieldOpNo = IsNewFormat ? 3 : 1;
unsigned NumOpsPerField = IsNewFormat ? 3 : 2;
for (unsigned Idx = FirstFieldOpNo; Idx < BaseNode->getNumOperands();
Idx += NumOpsPerField) {
auto *OffsetEntryCI =
mdconst::extract<ConstantInt>(BaseNode->getOperand(Idx + 1));
if (OffsetEntryCI->getValue().ugt(Offset)) {
if (Idx == FirstFieldOpNo) {
CheckFailed("Could not find TBAA parent in struct type node", &I,
BaseNode, &Offset);
return nullptr;
}
unsigned PrevIdx = Idx - NumOpsPerField;
auto *PrevOffsetEntryCI =
mdconst::extract<ConstantInt>(BaseNode->getOperand(PrevIdx + 1));
Offset -= PrevOffsetEntryCI->getValue();
return cast<MDNode>(BaseNode->getOperand(PrevIdx));
}
}
unsigned LastIdx = BaseNode->getNumOperands() - NumOpsPerField;
auto *LastOffsetEntryCI = mdconst::extract<ConstantInt>(
BaseNode->getOperand(LastIdx + 1));
Offset -= LastOffsetEntryCI->getValue();
return cast<MDNode>(BaseNode->getOperand(LastIdx));
}
static bool isNewFormatTBAATypeNode(llvm::MDNode *Type) {
if (!Type || Type->getNumOperands() < 3)
return false;
// In the new format type nodes shall have a reference to the parent type as
// its first operand.
MDNode *Parent = dyn_cast_or_null<MDNode>(Type->getOperand(0));
if (!Parent)
return false;
return true;
}
bool TBAAVerifier::visitTBAAMetadata(Instruction &I, const MDNode *MD) {
AssertTBAA(isa<LoadInst>(I) || isa<StoreInst>(I) || isa<CallInst>(I) ||
isa<VAArgInst>(I) || isa<AtomicRMWInst>(I) ||
isa<AtomicCmpXchgInst>(I),
"This instruction shall not have a TBAA access tag!", &I);
bool IsStructPathTBAA =
isa<MDNode>(MD->getOperand(0)) && MD->getNumOperands() >= 3;
AssertTBAA(
IsStructPathTBAA,
"Old-style TBAA is no longer allowed, use struct-path TBAA instead", &I);
MDNode *BaseNode = dyn_cast_or_null<MDNode>(MD->getOperand(0));
MDNode *AccessType = dyn_cast_or_null<MDNode>(MD->getOperand(1));
bool IsNewFormat = isNewFormatTBAATypeNode(AccessType);
if (IsNewFormat) {
AssertTBAA(MD->getNumOperands() == 4 || MD->getNumOperands() == 5,
"Access tag metadata must have either 4 or 5 operands", &I, MD);
} else {
AssertTBAA(MD->getNumOperands() < 5,
"Struct tag metadata must have either 3 or 4 operands", &I, MD);
}
// Check the access size field.
if (IsNewFormat) {
auto *AccessSizeNode = mdconst::dyn_extract_or_null<ConstantInt>(
MD->getOperand(3));
AssertTBAA(AccessSizeNode, "Access size field must be a constant", &I, MD);
}
// Check the immutability flag.
unsigned ImmutabilityFlagOpNo = IsNewFormat ? 4 : 3;
if (MD->getNumOperands() == ImmutabilityFlagOpNo + 1) {
auto *IsImmutableCI = mdconst::dyn_extract_or_null<ConstantInt>(
MD->getOperand(ImmutabilityFlagOpNo));
AssertTBAA(IsImmutableCI,
"Immutability tag on struct tag metadata must be a constant",
&I, MD);
AssertTBAA(
IsImmutableCI->isZero() || IsImmutableCI->isOne(),
"Immutability part of the struct tag metadata must be either 0 or 1",
&I, MD);
}
AssertTBAA(BaseNode && AccessType,
"Malformed struct tag metadata: base and access-type "
"should be non-null and point to Metadata nodes",
&I, MD, BaseNode, AccessType);
if (!IsNewFormat) {
AssertTBAA(isValidScalarTBAANode(AccessType),
"Access type node must be a valid scalar type", &I, MD,
AccessType);
}
auto *OffsetCI = mdconst::dyn_extract_or_null<ConstantInt>(MD->getOperand(2));
AssertTBAA(OffsetCI, "Offset must be constant integer", &I, MD);
APInt Offset = OffsetCI->getValue();
bool SeenAccessTypeInPath = false;
SmallPtrSet<MDNode *, 4> StructPath;
for (/* empty */; BaseNode && !IsRootTBAANode(BaseNode);
BaseNode = getFieldNodeFromTBAABaseNode(I, BaseNode, Offset,
IsNewFormat)) {
if (!StructPath.insert(BaseNode).second) {
CheckFailed("Cycle detected in struct path", &I, MD);
return false;
}
bool Invalid;
unsigned BaseNodeBitWidth;
std::tie(Invalid, BaseNodeBitWidth) = verifyTBAABaseNode(I, BaseNode,
IsNewFormat);
// If the base node is invalid in itself, then we've already printed all the
// errors we wanted to print.
if (Invalid)
return false;
SeenAccessTypeInPath |= BaseNode == AccessType;
if (isValidScalarTBAANode(BaseNode) || BaseNode == AccessType)
AssertTBAA(Offset == 0, "Offset not zero at the point of scalar access",
&I, MD, &Offset);
AssertTBAA(BaseNodeBitWidth == Offset.getBitWidth() ||
(BaseNodeBitWidth == 0 && Offset == 0) ||
(IsNewFormat && BaseNodeBitWidth == ~0u),
"Access bit-width not the same as description bit-width", &I, MD,
BaseNodeBitWidth, Offset.getBitWidth());
if (IsNewFormat && SeenAccessTypeInPath)
break;
}
AssertTBAA(SeenAccessTypeInPath, "Did not see access type in access path!",
&I, MD);
return true;
}
char VerifierLegacyPass::ID = 0;
INITIALIZE_PASS(VerifierLegacyPass, "verify", "Module Verifier", false, false)
2009-09-07 22:44:51 +02:00
FunctionPass *llvm::createVerifierPass(bool FatalErrors) {
return new VerifierLegacyPass(FatalErrors);
2001-06-06 22:29:01 +02:00
}
[PM] Change the static object whose address is used to uniquely identify analyses to have a common type which is enforced rather than using a char object and a `void *` type when used as an identifier. This has a number of advantages. First, it at least helps some of the confusion raised in Justin Lebar's code review of why `void *` was being used everywhere by having a stronger type that connects to documentation about this. However, perhaps more importantly, it addresses a serious issue where the alignment of these pointer-like identifiers was unknown. This made it hard to use them in pointer-like data structures. We were already dodging this in dangerous ways to create the "all analyses" entry. In a subsequent patch I attempted to use these with TinyPtrVector and things fell apart in a very bad way. And it isn't just a compile time or type system issue. Worse than that, the actual alignment of these pointer-like opaque identifiers wasn't guaranteed to be a useful alignment as they were just characters. This change introduces a type to use as the "key" object whose address forms the opaque identifier. This both forces the objects to have proper alignment, and provides type checking that we get it right everywhere. It also makes the types somewhat less mysterious than `void *`. We could go one step further and introduce a truly opaque pointer-like type to return from the `ID()` static function rather than returning `AnalysisKey *`, but that didn't seem to be a clear win so this is just the initial change to get to a reliably typed and aligned object serving is a key for all the analyses. Thanks to Richard Smith and Justin Lebar for helping pick plausible names and avoid making this refactoring many times. =] And thanks to Sean for the super fast review! While here, I've tried to move away from the "PassID" nomenclature entirely as it wasn't really helping and is overloaded with old pass manager constructs. Now we have IDs for analyses, and key objects whose address can be used as IDs. Where possible and clear I've shortened this to just "ID". In a few places I kept "AnalysisID" to make it clear what was being identified. Differential Revision: https://reviews.llvm.org/D27031 llvm-svn: 287783
2016-11-23 18:53:26 +01:00
AnalysisKey VerifierAnalysis::Key;
VerifierAnalysis::Result VerifierAnalysis::run(Module &M,
ModuleAnalysisManager &) {
Result Res;
Res.IRBroken = llvm::verifyModule(M, &dbgs(), &Res.DebugInfoBroken);
return Res;
}
VerifierAnalysis::Result VerifierAnalysis::run(Function &F,
FunctionAnalysisManager &) {
return { llvm::verifyFunction(F, &dbgs()), false };
}
PreservedAnalyses VerifierPass::run(Module &M, ModuleAnalysisManager &AM) {
auto Res = AM.getResult<VerifierAnalysis>(M);
if (FatalErrors && (Res.IRBroken || Res.DebugInfoBroken))
report_fatal_error("Broken module found, compilation aborted!");
return PreservedAnalyses::all();
}
PreservedAnalyses VerifierPass::run(Function &F, FunctionAnalysisManager &AM) {
auto res = AM.getResult<VerifierAnalysis>(F);
if (res.IRBroken && FatalErrors)
report_fatal_error("Broken function found, compilation aborted!");
return PreservedAnalyses::all();
}