RPCS3/llvm-mirror
1
0
Fork 0
mirror of https://github.com/RPCS3/llvm-mirror.git synced 2024-11-01 16:33:37 +01:00
Code Issues Projects Releases Wiki Activity
b4db8689ee
llvm-mirror/lib/AsmParser/LLParser.h

382 lines
15 KiB
C
Raw Normal View History

Reimplement the old and horrible bison parser for .ll files with a nice and clean recursive descent parser. This change has a couple of ramifications: 1. The parser code is about 400 lines shorter (in what we maintain, not including what is autogenerated). 2. The code should be significantly faster than the old code because we don't have to work around bison's poor handling of datatypes with ctors/dtors. This also makes the code much more resistant to memory leaks. 3. We now get caret diagnostics from the .ll parser, woo. 4. The actual diagnostics emited from the parser are completely different so a bunch of testcases had to be updated. 5. I now disallow "%ty = type opaque %ty = type i32". There was no good reason to support this, it was just an accident of the old implementation. I have no reason to think that anyone is actually using this. 6. The syntax for sticking a global variable has changed to make it unambiguous. I don't think anyone is depending on this since only clang supports this and it is not solid yet, so I'm not worried about anything breaking. 7. This gets rid of the last use of bison, and along with it the .cvs files. I'll prune this from the makefiles as a subsequent commit. There are a few minor cleanups that can be done after this commit (suggestions welcome!) but this passes dejagnu testing and is ready for its time in the limelight. llvm-svn: 61558
2009-01-02 08:01:27 +01:00
//===-- LLParser.h - Parser Class -------------------------------*- C++ -*-===//
//
// The LLVM Compiler Infrastructure
//
// This file is distributed under the University of Illinois Open Source
// License. See LICENSE.TXT for details.
//
//===----------------------------------------------------------------------===//
//
// This file defines the parser class for .ll files.
//
//===----------------------------------------------------------------------===//
#ifndef LLVM_ASMPARSER_LLPARSER_H
#define LLVM_ASMPARSER_LLPARSER_H
#include "LLLexer.h"
Initial implementation of 'fence' instruction, the new C++0x-style replacement for llvm.memory.barrier. This is just a LangRef entry and reading/writing/memory representation; optimizer+codegen support coming soon. llvm-svn: 136009
2011-07-26 01:16:38 +02:00
#include "llvm/Instructions.h"
Try again at converting the LLParser to use LLVMContext, without massive breakage this time. llvm-svn: 74671
2009-07-02 01:57:11 +02:00
#include "llvm/Module.h"
Reimplement the old and horrible bison parser for .ll files with a nice and clean recursive descent parser. This change has a couple of ramifications: 1. The parser code is about 400 lines shorter (in what we maintain, not including what is autogenerated). 2. The code should be significantly faster than the old code because we don't have to work around bison's poor handling of datatypes with ctors/dtors. This also makes the code much more resistant to memory leaks. 3. We now get caret diagnostics from the .ll parser, woo. 4. The actual diagnostics emited from the parser are completely different so a bunch of testcases had to be updated. 5. I now disallow "%ty = type opaque %ty = type i32". There was no good reason to support this, it was just an accident of the old implementation. I have no reason to think that anyone is actually using this. 6. The syntax for sticking a global variable has changed to make it unambiguous. I don't think anyone is depending on this since only clang supports this and it is not solid yet, so I'm not worried about anything breaking. 7. This gets rid of the last use of bison, and along with it the .cvs files. I'll prune this from the makefiles as a subsequent commit. There are a few minor cleanups that can be done after this commit (suggestions welcome!) but this passes dejagnu testing and is ready for its time in the limelight. llvm-svn: 61558
2009-01-02 08:01:27 +01:00
#include "llvm/Type.h"
include header. llvm-svn: 100087
2010-04-01 07:20:21 +02:00
#include "llvm/ADT/DenseMap.h"
Land the long talked about "type system rewrite" patch. This patch brings numerous advantages to LLVM. One way to look at it is through diffstat: 109 files changed, 3005 insertions(+), 5906 deletions(-) Removing almost 3K lines of code is a good thing. Other advantages include: 1. Value::getType() is a simple load that can be CSE'd, not a mutating union-find operation. 2. Types a uniqued and never move once created, defining away PATypeHolder. 3. Structs can be "named" now, and their name is part of the identity that uniques them. This means that the compiler doesn't merge them structurally which makes the IR much less confusing. 4. Now that there is no way to get a cycle in a type graph without a named struct type, "upreferences" go away. 5. Type refinement is completely gone, which should make LTO much MUCH faster in some common cases with C++ code. 6. Types are now generally immutable, so we can use "Type *" instead "const Type *" everywhere. Downsides of this patch are that it removes some functions from the C API, so people using those will have to upgrade to (not yet added) new API. "LLVM 3.0" is the right time to do this. There are still some cleanups pending after this, this patch is large enough as-is. llvm-svn: 134829
2011-07-09 19:41:24 +02:00
#include "llvm/ADT/StringMap.h"
Metadata.h doesn't need to include ValueHandle.h anymore. llvm-svn: 92211
2009-12-28 09:20:46 +01:00
#include "llvm/Support/ValueHandle.h"
switch to TrackingVH instead of WeakVH, since these can never be RAUW'd and go to null. This also gets us some sorely lacking type safety. llvm-svn: 92272
2009-12-29 22:43:58 +01:00
#include <map>
Reimplement the old and horrible bison parser for .ll files with a nice and clean recursive descent parser. This change has a couple of ramifications: 1. The parser code is about 400 lines shorter (in what we maintain, not including what is autogenerated). 2. The code should be significantly faster than the old code because we don't have to work around bison's poor handling of datatypes with ctors/dtors. This also makes the code much more resistant to memory leaks. 3. We now get caret diagnostics from the .ll parser, woo. 4. The actual diagnostics emited from the parser are completely different so a bunch of testcases had to be updated. 5. I now disallow "%ty = type opaque %ty = type i32". There was no good reason to support this, it was just an accident of the old implementation. I have no reason to think that anyone is actually using this. 6. The syntax for sticking a global variable has changed to make it unambiguous. I don't think anyone is depending on this since only clang supports this and it is not solid yet, so I'm not worried about anything breaking. 7. This gets rid of the last use of bison, and along with it the .cvs files. I'll prune this from the makefiles as a subsequent commit. There are a few minor cleanups that can be done after this commit (suggestions welcome!) but this passes dejagnu testing and is ready for its time in the limelight. llvm-svn: 61558
2009-01-02 08:01:27 +01:00
namespace llvm {
class Module;
class OpaqueType;
class Function;
class Value;
class BasicBlock;
class Instruction;
class Constant;
class GlobalValue;
Add support for embedded metadata to LLVM. This introduces two new types of Constant, MDString and MDNode which can only be used by globals with a name that starts with "llvm." or as arguments to a function with the same naming restriction. llvm-svn: 68420
2009-04-04 09:22:01 +02:00
class MDString;
class MDNode;
Land the long talked about "type system rewrite" patch. This patch brings numerous advantages to LLVM. One way to look at it is through diffstat: 109 files changed, 3005 insertions(+), 5906 deletions(-) Removing almost 3K lines of code is a good thing. Other advantages include: 1. Value::getType() is a simple load that can be CSE'd, not a mutating union-find operation. 2. Types a uniqued and never move once created, defining away PATypeHolder. 3. Structs can be "named" now, and their name is part of the identity that uniques them. This means that the compiler doesn't merge them structurally which makes the IR much less confusing. 4. Now that there is no way to get a cycle in a type graph without a named struct type, "upreferences" go away. 5. Type refinement is completely gone, which should make LTO much MUCH faster in some common cases with C++ code. 6. Types are now generally immutable, so we can use "Type *" instead "const Type *" everywhere. Downsides of this patch are that it removes some functions from the C API, so people using those will have to upgrade to (not yet added) new API. "LLVM 3.0" is the right time to do this. There are still some cleanups pending after this, this patch is large enough as-is. llvm-svn: 134829
2011-07-09 19:41:24 +02:00
class StructType;
Down with trailing whitespace! llvm-svn: 61594
2009-01-02 23:46:48 +01:00
full asmparser support for blockaddress. We can now do: $ llvm-as foo.ll -d -disable-output which reads and prints the .ll file. BC encoding is the next project. Testcase will go in once that works. llvm-svn: 85368
2009-10-28 04:39:23 +01:00
/// ValID - Represents a reference of a definition of some sort with no type.
/// There are several cases where we have to parse the value but where the
/// type can depend on later context. This may either be a numeric reference
/// or a symbolic (%var) reference. This is just a discriminated union.
struct ValID {
enum {
t_LocalID, t_GlobalID, // ID in UIntVal.
t_LocalName, t_GlobalName, // Name in StrVal.
t_APSInt, t_APFloat, // Value in APSIntVal/APFloatVal.
t_Null, t_Undef, t_Zero, // No value.
t_EmptyArray, // No value: []
t_Constant, // Value in ConstantVal.
t_InlineAsm, // Value in StrVal/StrVal2/UIntVal.
split t_Metadata into t_MDNode and t_MDString, eliminating some unsafe casting. llvm-svn: 92287
2009-12-30 03:11:14 +01:00
t_MDNode, // Value in MDNodeVal.
Land the long talked about "type system rewrite" patch. This patch brings numerous advantages to LLVM. One way to look at it is through diffstat: 109 files changed, 3005 insertions(+), 5906 deletions(-) Removing almost 3K lines of code is a good thing. Other advantages include: 1. Value::getType() is a simple load that can be CSE'd, not a mutating union-find operation. 2. Types a uniqued and never move once created, defining away PATypeHolder. 3. Structs can be "named" now, and their name is part of the identity that uniques them. This means that the compiler doesn't merge them structurally which makes the IR much less confusing. 4. Now that there is no way to get a cycle in a type graph without a named struct type, "upreferences" go away. 5. Type refinement is completely gone, which should make LTO much MUCH faster in some common cases with C++ code. 6. Types are now generally immutable, so we can use "Type *" instead "const Type *" everywhere. Downsides of this patch are that it removes some functions from the C API, so people using those will have to upgrade to (not yet added) new API. "LLVM 3.0" is the right time to do this. There are still some cleanups pending after this, this patch is large enough as-is. llvm-svn: 134829
2011-07-09 19:41:24 +02:00
t_MDString, // Value in MDStringVal.
t_ConstantStruct, // Value in ConstantStructElts.
t_PackedConstantStruct // Value in ConstantStructElts.
full asmparser support for blockaddress. We can now do: $ llvm-as foo.ll -d -disable-output which reads and prints the .ll file. BC encoding is the next project. Testcase will go in once that works. llvm-svn: 85368
2009-10-28 04:39:23 +01:00
} Kind;
LLLexer::LocTy Loc;
unsigned UIntVal;
std::string StrVal, StrVal2;
APSInt APSIntVal;
APFloat APFloatVal;
Constant *ConstantVal;
split t_Metadata into t_MDNode and t_MDString, eliminating some unsafe casting. llvm-svn: 92287
2009-12-30 03:11:14 +01:00
MDNode *MDNodeVal;
MDString *MDStringVal;
Land the long talked about "type system rewrite" patch. This patch brings numerous advantages to LLVM. One way to look at it is through diffstat: 109 files changed, 3005 insertions(+), 5906 deletions(-) Removing almost 3K lines of code is a good thing. Other advantages include: 1. Value::getType() is a simple load that can be CSE'd, not a mutating union-find operation. 2. Types a uniqued and never move once created, defining away PATypeHolder. 3. Structs can be "named" now, and their name is part of the identity that uniques them. This means that the compiler doesn't merge them structurally which makes the IR much less confusing. 4. Now that there is no way to get a cycle in a type graph without a named struct type, "upreferences" go away. 5. Type refinement is completely gone, which should make LTO much MUCH faster in some common cases with C++ code. 6. Types are now generally immutable, so we can use "Type *" instead "const Type *" everywhere. Downsides of this patch are that it removes some functions from the C API, so people using those will have to upgrade to (not yet added) new API. "LLVM 3.0" is the right time to do this. There are still some cleanups pending after this, this patch is large enough as-is. llvm-svn: 134829
2011-07-09 19:41:24 +02:00
Constant **ConstantStructElts;
ValID() : Kind(t_LocalID), APFloatVal(0.0) {}
~ValID() {
if (Kind == t_ConstantStruct || Kind == t_PackedConstantStruct)
delete [] ConstantStructElts;
}
full asmparser support for blockaddress. We can now do: $ llvm-as foo.ll -d -disable-output which reads and prints the .ll file. BC encoding is the next project. Testcase will go in once that works. llvm-svn: 85368
2009-10-28 04:39:23 +01:00
bool operator<(const ValID &RHS) const {
if (Kind == t_LocalID || Kind == t_GlobalID)
return UIntVal < RHS.UIntVal;
Land the long talked about "type system rewrite" patch. This patch brings numerous advantages to LLVM. One way to look at it is through diffstat: 109 files changed, 3005 insertions(+), 5906 deletions(-) Removing almost 3K lines of code is a good thing. Other advantages include: 1. Value::getType() is a simple load that can be CSE'd, not a mutating union-find operation. 2. Types a uniqued and never move once created, defining away PATypeHolder. 3. Structs can be "named" now, and their name is part of the identity that uniques them. This means that the compiler doesn't merge them structurally which makes the IR much less confusing. 4. Now that there is no way to get a cycle in a type graph without a named struct type, "upreferences" go away. 5. Type refinement is completely gone, which should make LTO much MUCH faster in some common cases with C++ code. 6. Types are now generally immutable, so we can use "Type *" instead "const Type *" everywhere. Downsides of this patch are that it removes some functions from the C API, so people using those will have to upgrade to (not yet added) new API. "LLVM 3.0" is the right time to do this. There are still some cleanups pending after this, this patch is large enough as-is. llvm-svn: 134829
2011-07-09 19:41:24 +02:00
assert((Kind == t_LocalName || Kind == t_GlobalName ||
Kind == t_ConstantStruct || Kind == t_PackedConstantStruct) &&
full asmparser support for blockaddress. We can now do: $ llvm-as foo.ll -d -disable-output which reads and prints the .ll file. BC encoding is the next project. Testcase will go in once that works. llvm-svn: 85368
2009-10-28 04:39:23 +01:00
"Ordering not defined for this ValID kind yet");
return StrVal < RHS.StrVal;
}
};
Reimplement the old and horrible bison parser for .ll files with a nice and clean recursive descent parser. This change has a couple of ramifications: 1. The parser code is about 400 lines shorter (in what we maintain, not including what is autogenerated). 2. The code should be significantly faster than the old code because we don't have to work around bison's poor handling of datatypes with ctors/dtors. This also makes the code much more resistant to memory leaks. 3. We now get caret diagnostics from the .ll parser, woo. 4. The actual diagnostics emited from the parser are completely different so a bunch of testcases had to be updated. 5. I now disallow "%ty = type opaque %ty = type i32". There was no good reason to support this, it was just an accident of the old implementation. I have no reason to think that anyone is actually using this. 6. The syntax for sticking a global variable has changed to make it unambiguous. I don't think anyone is depending on this since only clang supports this and it is not solid yet, so I'm not worried about anything breaking. 7. This gets rid of the last use of bison, and along with it the .cvs files. I'll prune this from the makefiles as a subsequent commit. There are a few minor cleanups that can be done after this commit (suggestions welcome!) but this passes dejagnu testing and is ready for its time in the limelight. llvm-svn: 61558
2009-01-02 08:01:27 +01:00
class LLParser {
public:
typedef LLLexer::LocTy LocTy;
private:
fix a crash on invalid metadata, e.g.: call i32 @foo(), XXXX We would return the error without inserting the new instruction into the program, so it wouldn't get deallocated, and an abort would trigger when the module was deleted. llvm-svn: 100602
2010-04-07 06:08:57 +02:00
LLVMContext &Context;
Reimplement the old and horrible bison parser for .ll files with a nice and clean recursive descent parser. This change has a couple of ramifications: 1. The parser code is about 400 lines shorter (in what we maintain, not including what is autogenerated). 2. The code should be significantly faster than the old code because we don't have to work around bison's poor handling of datatypes with ctors/dtors. This also makes the code much more resistant to memory leaks. 3. We now get caret diagnostics from the .ll parser, woo. 4. The actual diagnostics emited from the parser are completely different so a bunch of testcases had to be updated. 5. I now disallow "%ty = type opaque %ty = type i32". There was no good reason to support this, it was just an accident of the old implementation. I have no reason to think that anyone is actually using this. 6. The syntax for sticking a global variable has changed to make it unambiguous. I don't think anyone is depending on this since only clang supports this and it is not solid yet, so I'm not worried about anything breaking. 7. This gets rid of the last use of bison, and along with it the .cvs files. I'll prune this from the makefiles as a subsequent commit. There are a few minor cleanups that can be done after this commit (suggestions welcome!) but this passes dejagnu testing and is ready for its time in the limelight. llvm-svn: 61558
2009-01-02 08:01:27 +01:00
LLLexer Lex;
Module *M;
rewrite handling of forward ref'd instruction metadata to used deferred resolution instead of creating a temporary node + rauw. There is no reason to create the temporary mdnode, then do rauw, then destroy it. llvm-svn: 100086
2010-04-01 07:14:45 +02:00
// Instruction metadata resolution. Each instruction can have a list of
// MDRef info associated with them.
Add a comment explaining why this code is more complex than it initially seems it should require. llvm-svn: 111913
2010-08-24 16:31:06 +02:00
//
// The simpler approach of just creating temporary MDNodes and then calling
// RAUW on them when the definition is processed doesn't work because some
// instruction metadata kinds, such as dbg, get stored in the IR in an
// "optimized" format which doesn't participate in the normal value use
// lists. This means that RAUW doesn't work, even on temporary MDNodes
// which otherwise support RAUW. Instead, we defer resolving MDNode
// references until the definitions have been processed.
rewrite handling of forward ref'd instruction metadata to used deferred resolution instead of creating a temporary node + rauw. There is no reason to create the temporary mdnode, then do rauw, then destroy it. llvm-svn: 100086
2010-04-01 07:14:45 +02:00
struct MDRef {
SMLoc Loc;
unsigned MDKind, MDSlot;
};
DenseMap<Instruction*, std::vector<MDRef> > ForwardRefInstMetadata;
Down with trailing whitespace! llvm-svn: 61594
2009-01-02 23:46:48 +01:00
Land the long talked about "type system rewrite" patch. This patch brings numerous advantages to LLVM. One way to look at it is through diffstat: 109 files changed, 3005 insertions(+), 5906 deletions(-) Removing almost 3K lines of code is a good thing. Other advantages include: 1. Value::getType() is a simple load that can be CSE'd, not a mutating union-find operation. 2. Types a uniqued and never move once created, defining away PATypeHolder. 3. Structs can be "named" now, and their name is part of the identity that uniques them. This means that the compiler doesn't merge them structurally which makes the IR much less confusing. 4. Now that there is no way to get a cycle in a type graph without a named struct type, "upreferences" go away. 5. Type refinement is completely gone, which should make LTO much MUCH faster in some common cases with C++ code. 6. Types are now generally immutable, so we can use "Type *" instead "const Type *" everywhere. Downsides of this patch are that it removes some functions from the C API, so people using those will have to upgrade to (not yet added) new API. "LLVM 3.0" is the right time to do this. There are still some cleanups pending after this, this patch is large enough as-is. llvm-svn: 134829
2011-07-09 19:41:24 +02:00
// Type resolution handling data structures. The location is set when we
// have processed a use of the type but not a definition yet.
StringMap<std::pair<Type*, LocTy> > NamedTypes;
std::vector<std::pair<Type*, LocTy> > NumberedTypes;
rename MetadataCache -> NumberedMetadata to follow the convention used by other things. Convert it to a vector since it is a dense numbering. llvm-svn: 92293
2009-12-30 05:51:58 +01:00
std::vector<TrackingVH<MDNode> > NumberedMetadata;
switch to TrackingVH instead of WeakVH, since these can never be RAUW'd and go to null. This also gets us some sorely lacking type safety. llvm-svn: 92272
2009-12-29 22:43:58 +01:00
std::map<unsigned, std::pair<TrackingVH<MDNode>, LocTy> > ForwardRefMDNodes;
Reimplement the old and horrible bison parser for .ll files with a nice and clean recursive descent parser. This change has a couple of ramifications: 1. The parser code is about 400 lines shorter (in what we maintain, not including what is autogenerated). 2. The code should be significantly faster than the old code because we don't have to work around bison's poor handling of datatypes with ctors/dtors. This also makes the code much more resistant to memory leaks. 3. We now get caret diagnostics from the .ll parser, woo. 4. The actual diagnostics emited from the parser are completely different so a bunch of testcases had to be updated. 5. I now disallow "%ty = type opaque %ty = type i32". There was no good reason to support this, it was just an accident of the old implementation. I have no reason to think that anyone is actually using this. 6. The syntax for sticking a global variable has changed to make it unambiguous. I don't think anyone is depending on this since only clang supports this and it is not solid yet, so I'm not worried about anything breaking. 7. This gets rid of the last use of bison, and along with it the .cvs files. I'll prune this from the makefiles as a subsequent commit. There are a few minor cleanups that can be done after this commit (suggestions welcome!) but this passes dejagnu testing and is ready for its time in the limelight. llvm-svn: 61558
2009-01-02 08:01:27 +01:00
// Global Value reference information.
std::map<std::string, std::pair<GlobalValue*, LocTy> > ForwardRefVals;
std::map<unsigned, std::pair<GlobalValue*, LocTy> > ForwardRefValIDs;
std::vector<GlobalValue*> NumberedVals;
full asmparser support for blockaddress. We can now do: $ llvm-as foo.ll -d -disable-output which reads and prints the .ll file. BC encoding is the next project. Testcase will go in once that works. llvm-svn: 85368
2009-10-28 04:39:23 +01:00
// References to blockaddress. The key is the function ValID, the value is
// a list of references to blocks in that function.
std::map<ValID, std::vector<std::pair<ValID, GlobalValue*> > >
ForwardRefBlockAddresses;
Reimplement the old and horrible bison parser for .ll files with a nice and clean recursive descent parser. This change has a couple of ramifications: 1. The parser code is about 400 lines shorter (in what we maintain, not including what is autogenerated). 2. The code should be significantly faster than the old code because we don't have to work around bison's poor handling of datatypes with ctors/dtors. This also makes the code much more resistant to memory leaks. 3. We now get caret diagnostics from the .ll parser, woo. 4. The actual diagnostics emited from the parser are completely different so a bunch of testcases had to be updated. 5. I now disallow "%ty = type opaque %ty = type i32". There was no good reason to support this, it was just an accident of the old implementation. I have no reason to think that anyone is actually using this. 6. The syntax for sticking a global variable has changed to make it unambiguous. I don't think anyone is depending on this since only clang supports this and it is not solid yet, so I'm not worried about anything breaking. 7. This gets rid of the last use of bison, and along with it the .cvs files. I'll prune this from the makefiles as a subsequent commit. There are a few minor cleanups that can be done after this commit (suggestions welcome!) but this passes dejagnu testing and is ready for its time in the limelight. llvm-svn: 61558
2009-01-02 08:01:27 +01:00
public:
switch the .ll parser to use SourceMgr. llvm-svn: 74735
2009-07-03 01:08:13 +02:00
LLParser(MemoryBuffer *F, SourceMgr &SM, SMDiagnostic &Err, Module *m) :
Autoupgrade malloc insts to malloc calls. Update testcases that rely on malloc insts being present. Also prematurely remove MallocInst handling from IndMemRemoval and RaiseAllocations to help pass tests in this incremental step. llvm-svn: 84292
2009-10-17 02:00:19 +02:00
Context(m->getContext()), Lex(F, SM, Err, m->getContext()),
Remove old backwards compatibility support from the parser for autoupgrading the old malloc/free instructions, and for 'sext' and 'zext' as function attributes (they are spelled signext/zeroext now), and support for result value attributes being specified after a function. Additionally, diagnose invalid attributes on functions with an error message instead of an abort in the verifier. llvm-svn: 133229
2011-06-17 05:16:47 +02:00
M(m) {}
Refactor some parser interfaces to fix PR3278 and a FIXME: ParseAssemblyString with a specified module would not parse into the module, it would create and return a new one. llvm-svn: 61635
2009-01-04 21:44:11 +01:00
bool Run();
Down with trailing whitespace! llvm-svn: 61594
2009-01-02 23:46:48 +01:00
Land the long talked about "type system rewrite" patch. This patch brings numerous advantages to LLVM. One way to look at it is through diffstat: 109 files changed, 3005 insertions(+), 5906 deletions(-) Removing almost 3K lines of code is a good thing. Other advantages include: 1. Value::getType() is a simple load that can be CSE'd, not a mutating union-find operation. 2. Types a uniqued and never move once created, defining away PATypeHolder. 3. Structs can be "named" now, and their name is part of the identity that uniques them. This means that the compiler doesn't merge them structurally which makes the IR much less confusing. 4. Now that there is no way to get a cycle in a type graph without a named struct type, "upreferences" go away. 5. Type refinement is completely gone, which should make LTO much MUCH faster in some common cases with C++ code. 6. Types are now generally immutable, so we can use "Type *" instead "const Type *" everywhere. Downsides of this patch are that it removes some functions from the C API, so people using those will have to upgrade to (not yet added) new API. "LLVM 3.0" is the right time to do this. There are still some cleanups pending after this, this patch is large enough as-is. llvm-svn: 134829
2011-07-09 19:41:24 +02:00
LLVMContext &getContext() { return Context; }
Use LLVMContext for generating UndefValue constants too! llvm-svn: 74703
2009-07-02 19:04:01 +02:00
Reimplement the old and horrible bison parser for .ll files with a nice and clean recursive descent parser. This change has a couple of ramifications: 1. The parser code is about 400 lines shorter (in what we maintain, not including what is autogenerated). 2. The code should be significantly faster than the old code because we don't have to work around bison's poor handling of datatypes with ctors/dtors. This also makes the code much more resistant to memory leaks. 3. We now get caret diagnostics from the .ll parser, woo. 4. The actual diagnostics emited from the parser are completely different so a bunch of testcases had to be updated. 5. I now disallow "%ty = type opaque %ty = type i32". There was no good reason to support this, it was just an accident of the old implementation. I have no reason to think that anyone is actually using this. 6. The syntax for sticking a global variable has changed to make it unambiguous. I don't think anyone is depending on this since only clang supports this and it is not solid yet, so I'm not worried about anything breaking. 7. This gets rid of the last use of bison, and along with it the .cvs files. I'll prune this from the makefiles as a subsequent commit. There are a few minor cleanups that can be done after this commit (suggestions welcome!) but this passes dejagnu testing and is ready for its time in the limelight. llvm-svn: 61558
2009-01-02 08:01:27 +01:00
private:
Push twines deeper into SourceMgr's error handling methods. llvm-svn: 114847
2010-09-27 19:42:11 +02:00
bool Error(LocTy L, const Twine &Msg) const {
Reimplement the old and horrible bison parser for .ll files with a nice and clean recursive descent parser. This change has a couple of ramifications: 1. The parser code is about 400 lines shorter (in what we maintain, not including what is autogenerated). 2. The code should be significantly faster than the old code because we don't have to work around bison's poor handling of datatypes with ctors/dtors. This also makes the code much more resistant to memory leaks. 3. We now get caret diagnostics from the .ll parser, woo. 4. The actual diagnostics emited from the parser are completely different so a bunch of testcases had to be updated. 5. I now disallow "%ty = type opaque %ty = type i32". There was no good reason to support this, it was just an accident of the old implementation. I have no reason to think that anyone is actually using this. 6. The syntax for sticking a global variable has changed to make it unambiguous. I don't think anyone is depending on this since only clang supports this and it is not solid yet, so I'm not worried about anything breaking. 7. This gets rid of the last use of bison, and along with it the .cvs files. I'll prune this from the makefiles as a subsequent commit. There are a few minor cleanups that can be done after this commit (suggestions welcome!) but this passes dejagnu testing and is ready for its time in the limelight. llvm-svn: 61558
2009-01-02 08:01:27 +01:00
return Lex.Error(L, Msg);
}
Push twines deeper into SourceMgr's error handling methods. llvm-svn: 114847
2010-09-27 19:42:11 +02:00
bool TokError(const Twine &Msg) const {
Reimplement the old and horrible bison parser for .ll files with a nice and clean recursive descent parser. This change has a couple of ramifications: 1. The parser code is about 400 lines shorter (in what we maintain, not including what is autogenerated). 2. The code should be significantly faster than the old code because we don't have to work around bison's poor handling of datatypes with ctors/dtors. This also makes the code much more resistant to memory leaks. 3. We now get caret diagnostics from the .ll parser, woo. 4. The actual diagnostics emited from the parser are completely different so a bunch of testcases had to be updated. 5. I now disallow "%ty = type opaque %ty = type i32". There was no good reason to support this, it was just an accident of the old implementation. I have no reason to think that anyone is actually using this. 6. The syntax for sticking a global variable has changed to make it unambiguous. I don't think anyone is depending on this since only clang supports this and it is not solid yet, so I'm not worried about anything breaking. 7. This gets rid of the last use of bison, and along with it the .cvs files. I'll prune this from the makefiles as a subsequent commit. There are a few minor cleanups that can be done after this commit (suggestions welcome!) but this passes dejagnu testing and is ready for its time in the limelight. llvm-svn: 61558
2009-01-02 08:01:27 +01:00
return Error(Lex.getLoc(), Msg);
}
Down with trailing whitespace! llvm-svn: 61594
2009-01-02 23:46:48 +01:00
Reimplement the old and horrible bison parser for .ll files with a nice and clean recursive descent parser. This change has a couple of ramifications: 1. The parser code is about 400 lines shorter (in what we maintain, not including what is autogenerated). 2. The code should be significantly faster than the old code because we don't have to work around bison's poor handling of datatypes with ctors/dtors. This also makes the code much more resistant to memory leaks. 3. We now get caret diagnostics from the .ll parser, woo. 4. The actual diagnostics emited from the parser are completely different so a bunch of testcases had to be updated. 5. I now disallow "%ty = type opaque %ty = type i32". There was no good reason to support this, it was just an accident of the old implementation. I have no reason to think that anyone is actually using this. 6. The syntax for sticking a global variable has changed to make it unambiguous. I don't think anyone is depending on this since only clang supports this and it is not solid yet, so I'm not worried about anything breaking. 7. This gets rid of the last use of bison, and along with it the .cvs files. I'll prune this from the makefiles as a subsequent commit. There are a few minor cleanups that can be done after this commit (suggestions welcome!) but this passes dejagnu testing and is ready for its time in the limelight. llvm-svn: 61558
2009-01-02 08:01:27 +01:00
/// GetGlobalVal - Get a value with the specified name or ID, creating a
/// forward reference record if needed. This can return null if the value
/// exists but does not have the right type.
land David Blaikie's patch to de-constify Type, with a few tweaks. llvm-svn: 135375
2011-07-18 06:54:35 +02:00
GlobalValue *GetGlobalVal(const std::string &N, Type *Ty, LocTy Loc);
GlobalValue *GetGlobalVal(unsigned ID, Type *Ty, LocTy Loc);
Down with trailing whitespace! llvm-svn: 61594
2009-01-02 23:46:48 +01:00
Reimplement the old and horrible bison parser for .ll files with a nice and clean recursive descent parser. This change has a couple of ramifications: 1. The parser code is about 400 lines shorter (in what we maintain, not including what is autogenerated). 2. The code should be significantly faster than the old code because we don't have to work around bison's poor handling of datatypes with ctors/dtors. This also makes the code much more resistant to memory leaks. 3. We now get caret diagnostics from the .ll parser, woo. 4. The actual diagnostics emited from the parser are completely different so a bunch of testcases had to be updated. 5. I now disallow "%ty = type opaque %ty = type i32". There was no good reason to support this, it was just an accident of the old implementation. I have no reason to think that anyone is actually using this. 6. The syntax for sticking a global variable has changed to make it unambiguous. I don't think anyone is depending on this since only clang supports this and it is not solid yet, so I'm not worried about anything breaking. 7. This gets rid of the last use of bison, and along with it the .cvs files. I'll prune this from the makefiles as a subsequent commit. There are a few minor cleanups that can be done after this commit (suggestions welcome!) but this passes dejagnu testing and is ready for its time in the limelight. llvm-svn: 61558
2009-01-02 08:01:27 +01:00
// Helper Routines.
bool ParseToken(lltok::Kind T, const char *ErrMsg);
minor cleanups and comment improvements. llvm-svn: 61564
2009-01-02 09:05:26 +01:00
bool EatIfPresent(lltok::Kind T) {
if (Lex.getKind() != T) return false;
Lex.Lex();
return true;
}
Reject uses of unnamed_addr in declarations. llvm-svn: 123358
2011-01-13 02:30:30 +01:00
bool ParseOptionalToken(lltok::Kind T, bool &Present, LocTy *Loc = 0) {
Reimplement the old and horrible bison parser for .ll files with a nice and clean recursive descent parser. This change has a couple of ramifications: 1. The parser code is about 400 lines shorter (in what we maintain, not including what is autogenerated). 2. The code should be significantly faster than the old code because we don't have to work around bison's poor handling of datatypes with ctors/dtors. This also makes the code much more resistant to memory leaks. 3. We now get caret diagnostics from the .ll parser, woo. 4. The actual diagnostics emited from the parser are completely different so a bunch of testcases had to be updated. 5. I now disallow "%ty = type opaque %ty = type i32". There was no good reason to support this, it was just an accident of the old implementation. I have no reason to think that anyone is actually using this. 6. The syntax for sticking a global variable has changed to make it unambiguous. I don't think anyone is depending on this since only clang supports this and it is not solid yet, so I'm not worried about anything breaking. 7. This gets rid of the last use of bison, and along with it the .cvs files. I'll prune this from the makefiles as a subsequent commit. There are a few minor cleanups that can be done after this commit (suggestions welcome!) but this passes dejagnu testing and is ready for its time in the limelight. llvm-svn: 61558
2009-01-02 08:01:27 +01:00
if (Lex.getKind() != T) {
Present = false;
} else {
Reject uses of unnamed_addr in declarations. llvm-svn: 123358
2011-01-13 02:30:30 +01:00
if (Loc)
*Loc = Lex.getLoc();
Reimplement the old and horrible bison parser for .ll files with a nice and clean recursive descent parser. This change has a couple of ramifications: 1. The parser code is about 400 lines shorter (in what we maintain, not including what is autogenerated). 2. The code should be significantly faster than the old code because we don't have to work around bison's poor handling of datatypes with ctors/dtors. This also makes the code much more resistant to memory leaks. 3. We now get caret diagnostics from the .ll parser, woo. 4. The actual diagnostics emited from the parser are completely different so a bunch of testcases had to be updated. 5. I now disallow "%ty = type opaque %ty = type i32". There was no good reason to support this, it was just an accident of the old implementation. I have no reason to think that anyone is actually using this. 6. The syntax for sticking a global variable has changed to make it unambiguous. I don't think anyone is depending on this since only clang supports this and it is not solid yet, so I'm not worried about anything breaking. 7. This gets rid of the last use of bison, and along with it the .cvs files. I'll prune this from the makefiles as a subsequent commit. There are a few minor cleanups that can be done after this commit (suggestions welcome!) but this passes dejagnu testing and is ready for its time in the limelight. llvm-svn: 61558
2009-01-02 08:01:27 +01:00
Lex.Lex();
Present = true;
}
return false;
}
minor cleanups and comment improvements. llvm-svn: 61564
2009-01-02 09:05:26 +01:00
bool ParseStringConstant(std::string &Result);
bool ParseUInt32(unsigned &Val);
bool ParseUInt32(unsigned &Val, LocTy &Loc) {
Reimplement the old and horrible bison parser for .ll files with a nice and clean recursive descent parser. This change has a couple of ramifications: 1. The parser code is about 400 lines shorter (in what we maintain, not including what is autogenerated). 2. The code should be significantly faster than the old code because we don't have to work around bison's poor handling of datatypes with ctors/dtors. This also makes the code much more resistant to memory leaks. 3. We now get caret diagnostics from the .ll parser, woo. 4. The actual diagnostics emited from the parser are completely different so a bunch of testcases had to be updated. 5. I now disallow "%ty = type opaque %ty = type i32". There was no good reason to support this, it was just an accident of the old implementation. I have no reason to think that anyone is actually using this. 6. The syntax for sticking a global variable has changed to make it unambiguous. I don't think anyone is depending on this since only clang supports this and it is not solid yet, so I'm not worried about anything breaking. 7. This gets rid of the last use of bison, and along with it the .cvs files. I'll prune this from the makefiles as a subsequent commit. There are a few minor cleanups that can be done after this commit (suggestions welcome!) but this passes dejagnu testing and is ready for its time in the limelight. llvm-svn: 61558
2009-01-02 08:01:27 +01:00
Loc = Lex.getLoc();
minor cleanups and comment improvements. llvm-svn: 61564
2009-01-02 09:05:26 +01:00
return ParseUInt32(Val);
Reimplement the old and horrible bison parser for .ll files with a nice and clean recursive descent parser. This change has a couple of ramifications: 1. The parser code is about 400 lines shorter (in what we maintain, not including what is autogenerated). 2. The code should be significantly faster than the old code because we don't have to work around bison's poor handling of datatypes with ctors/dtors. This also makes the code much more resistant to memory leaks. 3. We now get caret diagnostics from the .ll parser, woo. 4. The actual diagnostics emited from the parser are completely different so a bunch of testcases had to be updated. 5. I now disallow "%ty = type opaque %ty = type i32". There was no good reason to support this, it was just an accident of the old implementation. I have no reason to think that anyone is actually using this. 6. The syntax for sticking a global variable has changed to make it unambiguous. I don't think anyone is depending on this since only clang supports this and it is not solid yet, so I'm not worried about anything breaking. 7. This gets rid of the last use of bison, and along with it the .cvs files. I'll prune this from the makefiles as a subsequent commit. There are a few minor cleanups that can be done after this commit (suggestions welcome!) but this passes dejagnu testing and is ready for its time in the limelight. llvm-svn: 61558
2009-01-02 08:01:27 +01:00
}
bool ParseOptionalAddrSpace(unsigned &AddrSpace);
bool ParseOptionalAttrs(unsigned &Attrs, unsigned AttrKind);
bool ParseOptionalLinkage(unsigned &Linkage, bool &HasLinkage);
bool ParseOptionalLinkage(unsigned &Linkage) {
bool HasLinkage; return ParseOptionalLinkage(Linkage, HasLinkage);
}
bool ParseOptionalVisibility(unsigned &Visibility);
Retype from unsigned to CallingConv::ID accordingly. Approved by Bob Wilson. llvm-svn: 80773
2009-09-02 10:44:58 +02:00
bool ParseOptionalCallingConv(CallingConv::ID &CC);
Reimplement the old and horrible bison parser for .ll files with a nice and clean recursive descent parser. This change has a couple of ramifications: 1. The parser code is about 400 lines shorter (in what we maintain, not including what is autogenerated). 2. The code should be significantly faster than the old code because we don't have to work around bison's poor handling of datatypes with ctors/dtors. This also makes the code much more resistant to memory leaks. 3. We now get caret diagnostics from the .ll parser, woo. 4. The actual diagnostics emited from the parser are completely different so a bunch of testcases had to be updated. 5. I now disallow "%ty = type opaque %ty = type i32". There was no good reason to support this, it was just an accident of the old implementation. I have no reason to think that anyone is actually using this. 6. The syntax for sticking a global variable has changed to make it unambiguous. I don't think anyone is depending on this since only clang supports this and it is not solid yet, so I'm not worried about anything breaking. 7. This gets rid of the last use of bison, and along with it the .cvs files. I'll prune this from the makefiles as a subsequent commit. There are a few minor cleanups that can be done after this commit (suggestions welcome!) but this passes dejagnu testing and is ready for its time in the limelight. llvm-svn: 61558
2009-01-02 08:01:27 +01:00
bool ParseOptionalAlignment(unsigned &Alignment);
Initial implementation of 'fence' instruction, the new C++0x-style replacement for llvm.memory.barrier. This is just a LangRef entry and reading/writing/memory representation; optimizer+codegen support coming soon. llvm-svn: 136009
2011-07-26 01:16:38 +02:00
bool ParseScopeAndOrdering(bool isAtomic, SynchronizationScope &Scope,
AtomicOrdering &Ordering);
Add a new function attribute, 'alignstack'. It will indicate (when the backends implement support for it) that the stack should be forcibly realigned in the prologue (and the process reversed in the epilogue). llvm-svn: 95945
2010-02-12 01:31:15 +01:00
bool ParseOptionalStackAlignment(unsigned &Alignment);
reimplement ParseOptionalInfo as ParseOptionalCommaAlign, correctly handle the comma case for metadata. llvm-svn: 92301
2009-12-30 06:44:30 +01:00
bool ParseOptionalCommaAlign(unsigned &Alignment, bool &AteExtraComma);
reimplement insertvalue/extractvalue metadata handling to not blindly accept invalid input. Actually add a testcase. llvm-svn: 92297
2009-12-30 06:14:00 +01:00
bool ParseIndexList(SmallVectorImpl<unsigned> &Indices,bool &AteExtraComma);
bool ParseIndexList(SmallVectorImpl<unsigned> &Indices) {
bool AteExtraComma;
if (ParseIndexList(Indices, AteExtraComma)) return true;
if (AteExtraComma)
return TokError("expected index");
return false;
}
Down with trailing whitespace! llvm-svn: 61594
2009-01-02 23:46:48 +01:00
Reimplement the old and horrible bison parser for .ll files with a nice and clean recursive descent parser. This change has a couple of ramifications: 1. The parser code is about 400 lines shorter (in what we maintain, not including what is autogenerated). 2. The code should be significantly faster than the old code because we don't have to work around bison's poor handling of datatypes with ctors/dtors. This also makes the code much more resistant to memory leaks. 3. We now get caret diagnostics from the .ll parser, woo. 4. The actual diagnostics emited from the parser are completely different so a bunch of testcases had to be updated. 5. I now disallow "%ty = type opaque %ty = type i32". There was no good reason to support this, it was just an accident of the old implementation. I have no reason to think that anyone is actually using this. 6. The syntax for sticking a global variable has changed to make it unambiguous. I don't think anyone is depending on this since only clang supports this and it is not solid yet, so I'm not worried about anything breaking. 7. This gets rid of the last use of bison, and along with it the .cvs files. I'll prune this from the makefiles as a subsequent commit. There are a few minor cleanups that can be done after this commit (suggestions welcome!) but this passes dejagnu testing and is ready for its time in the limelight. llvm-svn: 61558
2009-01-02 08:01:27 +01:00
// Top-Level Entities
bool ParseTopLevelEntities();
bool ValidateEndOfModule();
bool ParseTargetDefinition();
bool ParseDepLibs();
bool ParseModuleAsm();
bool ParseUnnamedType();
bool ParseNamedType();
bool ParseDeclare();
bool ParseDefine();
Down with trailing whitespace! llvm-svn: 61594
2009-01-02 23:46:48 +01:00
Reimplement the old and horrible bison parser for .ll files with a nice and clean recursive descent parser. This change has a couple of ramifications: 1. The parser code is about 400 lines shorter (in what we maintain, not including what is autogenerated). 2. The code should be significantly faster than the old code because we don't have to work around bison's poor handling of datatypes with ctors/dtors. This also makes the code much more resistant to memory leaks. 3. We now get caret diagnostics from the .ll parser, woo. 4. The actual diagnostics emited from the parser are completely different so a bunch of testcases had to be updated. 5. I now disallow "%ty = type opaque %ty = type i32". There was no good reason to support this, it was just an accident of the old implementation. I have no reason to think that anyone is actually using this. 6. The syntax for sticking a global variable has changed to make it unambiguous. I don't think anyone is depending on this since only clang supports this and it is not solid yet, so I'm not worried about anything breaking. 7. This gets rid of the last use of bison, and along with it the .cvs files. I'll prune this from the makefiles as a subsequent commit. There are a few minor cleanups that can be done after this commit (suggestions welcome!) but this passes dejagnu testing and is ready for its time in the limelight. llvm-svn: 61558
2009-01-02 08:01:27 +01:00
bool ParseGlobalType(bool &IsConstant);
Extend the AsmWriter to print unnamed numbered types as "%0 = type ..." and unnamed numbered global variables as "@0 = global ...". Extend the AsmParser to recognize these forms. llvm-svn: 78859
2009-08-13 01:32:33 +02:00
bool ParseUnnamedGlobal();
Reimplement the old and horrible bison parser for .ll files with a nice and clean recursive descent parser. This change has a couple of ramifications: 1. The parser code is about 400 lines shorter (in what we maintain, not including what is autogenerated). 2. The code should be significantly faster than the old code because we don't have to work around bison's poor handling of datatypes with ctors/dtors. This also makes the code much more resistant to memory leaks. 3. We now get caret diagnostics from the .ll parser, woo. 4. The actual diagnostics emited from the parser are completely different so a bunch of testcases had to be updated. 5. I now disallow "%ty = type opaque %ty = type i32". There was no good reason to support this, it was just an accident of the old implementation. I have no reason to think that anyone is actually using this. 6. The syntax for sticking a global variable has changed to make it unambiguous. I don't think anyone is depending on this since only clang supports this and it is not solid yet, so I'm not worried about anything breaking. 7. This gets rid of the last use of bison, and along with it the .cvs files. I'll prune this from the makefiles as a subsequent commit. There are a few minor cleanups that can be done after this commit (suggestions welcome!) but this passes dejagnu testing and is ready for its time in the limelight. llvm-svn: 61558
2009-01-02 08:01:27 +01:00
bool ParseNamedGlobal();
bool ParseGlobal(const std::string &Name, LocTy Loc, unsigned Linkage,
bool HasLinkage, unsigned Visibility);
bool ParseAlias(const std::string &Name, LocTy Loc, unsigned Visibility);
Support stand alone metadata syntax. !0 = constant metadata !{i32 21, i32 22} @llvm.blah = constant metadata !{i32 1000, i16 200, metadata !0} llvm-svn: 74630
2009-07-01 21:21:12 +02:00
bool ParseStandaloneMetadata();
Parse named metadata. llvm-svn: 77410
2009-07-29 02:34:02 +02:00
bool ParseNamedMetadata();
change ParseMDString and ParseMDNode to take arguments of the right type. This exposed a raft of other problems, which I'll deal with in subsequent patches. llvm-svn: 92273
2009-12-29 22:53:55 +01:00
bool ParseMDString(MDString *&Result);
rename ParseMDNode -> ParseMDNodeID, since it parses !42, not !{... } as you'd expect. llvm-svn: 92291
2009-12-30 05:15:23 +01:00
bool ParseMDNodeID(MDNode *&Result);
rewrite handling of forward ref'd instruction metadata to used deferred resolution instead of creating a temporary node + rauw. There is no reason to create the temporary mdnode, then do rauw, then destroy it. llvm-svn: 100086
2010-04-01 07:14:45 +02:00
bool ParseMDNodeID(MDNode *&Result, unsigned &SlotNo);
Down with trailing whitespace! llvm-svn: 61594
2009-01-02 23:46:48 +01:00
Reimplement the old and horrible bison parser for .ll files with a nice and clean recursive descent parser. This change has a couple of ramifications: 1. The parser code is about 400 lines shorter (in what we maintain, not including what is autogenerated). 2. The code should be significantly faster than the old code because we don't have to work around bison's poor handling of datatypes with ctors/dtors. This also makes the code much more resistant to memory leaks. 3. We now get caret diagnostics from the .ll parser, woo. 4. The actual diagnostics emited from the parser are completely different so a bunch of testcases had to be updated. 5. I now disallow "%ty = type opaque %ty = type i32". There was no good reason to support this, it was just an accident of the old implementation. I have no reason to think that anyone is actually using this. 6. The syntax for sticking a global variable has changed to make it unambiguous. I don't think anyone is depending on this since only clang supports this and it is not solid yet, so I'm not worried about anything breaking. 7. This gets rid of the last use of bison, and along with it the .cvs files. I'll prune this from the makefiles as a subsequent commit. There are a few minor cleanups that can be done after this commit (suggestions welcome!) but this passes dejagnu testing and is ready for its time in the limelight. llvm-svn: 61558
2009-01-02 08:01:27 +01:00
// Type Parsing.
Land the long talked about "type system rewrite" patch. This patch brings numerous advantages to LLVM. One way to look at it is through diffstat: 109 files changed, 3005 insertions(+), 5906 deletions(-) Removing almost 3K lines of code is a good thing. Other advantages include: 1. Value::getType() is a simple load that can be CSE'd, not a mutating union-find operation. 2. Types a uniqued and never move once created, defining away PATypeHolder. 3. Structs can be "named" now, and their name is part of the identity that uniques them. This means that the compiler doesn't merge them structurally which makes the IR much less confusing. 4. Now that there is no way to get a cycle in a type graph without a named struct type, "upreferences" go away. 5. Type refinement is completely gone, which should make LTO much MUCH faster in some common cases with C++ code. 6. Types are now generally immutable, so we can use "Type *" instead "const Type *" everywhere. Downsides of this patch are that it removes some functions from the C API, so people using those will have to upgrade to (not yet added) new API. "LLVM 3.0" is the right time to do this. There are still some cleanups pending after this, this patch is large enough as-is. llvm-svn: 134829
2011-07-09 19:41:24 +02:00
bool ParseType(Type *&Result, bool AllowVoid = false);
bool ParseType(Type *&Result, LocTy &Loc, bool AllowVoid = false) {
Reimplement the old and horrible bison parser for .ll files with a nice and clean recursive descent parser. This change has a couple of ramifications: 1. The parser code is about 400 lines shorter (in what we maintain, not including what is autogenerated). 2. The code should be significantly faster than the old code because we don't have to work around bison's poor handling of datatypes with ctors/dtors. This also makes the code much more resistant to memory leaks. 3. We now get caret diagnostics from the .ll parser, woo. 4. The actual diagnostics emited from the parser are completely different so a bunch of testcases had to be updated. 5. I now disallow "%ty = type opaque %ty = type i32". There was no good reason to support this, it was just an accident of the old implementation. I have no reason to think that anyone is actually using this. 6. The syntax for sticking a global variable has changed to make it unambiguous. I don't think anyone is depending on this since only clang supports this and it is not solid yet, so I'm not worried about anything breaking. 7. This gets rid of the last use of bison, and along with it the .cvs files. I'll prune this from the makefiles as a subsequent commit. There are a few minor cleanups that can be done after this commit (suggestions welcome!) but this passes dejagnu testing and is ready for its time in the limelight. llvm-svn: 61558
2009-01-02 08:01:27 +01:00
Loc = Lex.getLoc();
Fix two classes of bugs. First: validate an invariant so that the asmparser rejects a bad construct instead of the verifier. Before: llvm-as: assembly parsed, but does not verify as correct! Invalid struct return type! i64 (%struct.Type*, %struct.Type*)* @foo after: llvm-as: t.ll:5:8: functions with 'sret' argument must return void define i64 @foo(%struct.Type* noalias nocapture sret %agg.result, %struct.Type* nocapture byval %t) nounwind { ^ Second, check that void is only used where allowed (in function return types) not in arbitrary places, fixing PR3747 - Crash in llvm-as with void field in struct. We now reject that example with: $ llvm-as t.ll llvm-as: t.ll:1:12: struct element can not have void type %x = type {void} ^ llvm-svn: 66394
2009-03-09 05:49:14 +01:00
return ParseType(Result, AllowVoid);
Reimplement the old and horrible bison parser for .ll files with a nice and clean recursive descent parser. This change has a couple of ramifications: 1. The parser code is about 400 lines shorter (in what we maintain, not including what is autogenerated). 2. The code should be significantly faster than the old code because we don't have to work around bison's poor handling of datatypes with ctors/dtors. This also makes the code much more resistant to memory leaks. 3. We now get caret diagnostics from the .ll parser, woo. 4. The actual diagnostics emited from the parser are completely different so a bunch of testcases had to be updated. 5. I now disallow "%ty = type opaque %ty = type i32". There was no good reason to support this, it was just an accident of the old implementation. I have no reason to think that anyone is actually using this. 6. The syntax for sticking a global variable has changed to make it unambiguous. I don't think anyone is depending on this since only clang supports this and it is not solid yet, so I'm not worried about anything breaking. 7. This gets rid of the last use of bison, and along with it the .cvs files. I'll prune this from the makefiles as a subsequent commit. There are a few minor cleanups that can be done after this commit (suggestions welcome!) but this passes dejagnu testing and is ready for its time in the limelight. llvm-svn: 61558
2009-01-02 08:01:27 +01:00
}
Land the long talked about "type system rewrite" patch. This patch brings numerous advantages to LLVM. One way to look at it is through diffstat: 109 files changed, 3005 insertions(+), 5906 deletions(-) Removing almost 3K lines of code is a good thing. Other advantages include: 1. Value::getType() is a simple load that can be CSE'd, not a mutating union-find operation. 2. Types a uniqued and never move once created, defining away PATypeHolder. 3. Structs can be "named" now, and their name is part of the identity that uniques them. This means that the compiler doesn't merge them structurally which makes the IR much less confusing. 4. Now that there is no way to get a cycle in a type graph without a named struct type, "upreferences" go away. 5. Type refinement is completely gone, which should make LTO much MUCH faster in some common cases with C++ code. 6. Types are now generally immutable, so we can use "Type *" instead "const Type *" everywhere. Downsides of this patch are that it removes some functions from the C API, so people using those will have to upgrade to (not yet added) new API. "LLVM 3.0" is the right time to do this. There are still some cleanups pending after this, this patch is large enough as-is. llvm-svn: 134829
2011-07-09 19:41:24 +02:00
bool ParseAnonStructType(Type *&Result, bool Packed);
bool ParseStructBody(SmallVectorImpl<Type*> &Body);
bool ParseStructDefinition(SMLoc TypeLoc, StringRef Name,
std::pair<Type*, LocTy> &Entry,
Type *&ResultTy);
bool ParseArrayVectorType(Type *&Result, bool isVector);
bool ParseFunctionType(Type *&Result);
Reimplement the old and horrible bison parser for .ll files with a nice and clean recursive descent parser. This change has a couple of ramifications: 1. The parser code is about 400 lines shorter (in what we maintain, not including what is autogenerated). 2. The code should be significantly faster than the old code because we don't have to work around bison's poor handling of datatypes with ctors/dtors. This also makes the code much more resistant to memory leaks. 3. We now get caret diagnostics from the .ll parser, woo. 4. The actual diagnostics emited from the parser are completely different so a bunch of testcases had to be updated. 5. I now disallow "%ty = type opaque %ty = type i32". There was no good reason to support this, it was just an accident of the old implementation. I have no reason to think that anyone is actually using this. 6. The syntax for sticking a global variable has changed to make it unambiguous. I don't think anyone is depending on this since only clang supports this and it is not solid yet, so I'm not worried about anything breaking. 7. This gets rid of the last use of bison, and along with it the .cvs files. I'll prune this from the makefiles as a subsequent commit. There are a few minor cleanups that can be done after this commit (suggestions welcome!) but this passes dejagnu testing and is ready for its time in the limelight. llvm-svn: 61558
2009-01-02 08:01:27 +01:00
// Function Semantic Analysis.
class PerFunctionState {
LLParser &P;
Function &F;
std::map<std::string, std::pair<Value*, LocTy> > ForwardRefVals;
std::map<unsigned, std::pair<Value*, LocTy> > ForwardRefValIDs;
std::vector<Value*> NumberedVals;
full asmparser support for blockaddress. We can now do: $ llvm-as foo.ll -d -disable-output which reads and prints the .ll file. BC encoding is the next project. Testcase will go in once that works. llvm-svn: 85368
2009-10-28 04:39:23 +01:00
/// FunctionNumber - If this is an unnamed function, this is the slot
/// number of it, otherwise it is -1.
int FunctionNumber;
Reimplement the old and horrible bison parser for .ll files with a nice and clean recursive descent parser. This change has a couple of ramifications: 1. The parser code is about 400 lines shorter (in what we maintain, not including what is autogenerated). 2. The code should be significantly faster than the old code because we don't have to work around bison's poor handling of datatypes with ctors/dtors. This also makes the code much more resistant to memory leaks. 3. We now get caret diagnostics from the .ll parser, woo. 4. The actual diagnostics emited from the parser are completely different so a bunch of testcases had to be updated. 5. I now disallow "%ty = type opaque %ty = type i32". There was no good reason to support this, it was just an accident of the old implementation. I have no reason to think that anyone is actually using this. 6. The syntax for sticking a global variable has changed to make it unambiguous. I don't think anyone is depending on this since only clang supports this and it is not solid yet, so I'm not worried about anything breaking. 7. This gets rid of the last use of bison, and along with it the .cvs files. I'll prune this from the makefiles as a subsequent commit. There are a few minor cleanups that can be done after this commit (suggestions welcome!) but this passes dejagnu testing and is ready for its time in the limelight. llvm-svn: 61558
2009-01-02 08:01:27 +01:00
public:
full asmparser support for blockaddress. We can now do: $ llvm-as foo.ll -d -disable-output which reads and prints the .ll file. BC encoding is the next project. Testcase will go in once that works. llvm-svn: 85368
2009-10-28 04:39:23 +01:00
PerFunctionState(LLParser &p, Function &f, int FunctionNumber);
Reimplement the old and horrible bison parser for .ll files with a nice and clean recursive descent parser. This change has a couple of ramifications: 1. The parser code is about 400 lines shorter (in what we maintain, not including what is autogenerated). 2. The code should be significantly faster than the old code because we don't have to work around bison's poor handling of datatypes with ctors/dtors. This also makes the code much more resistant to memory leaks. 3. We now get caret diagnostics from the .ll parser, woo. 4. The actual diagnostics emited from the parser are completely different so a bunch of testcases had to be updated. 5. I now disallow "%ty = type opaque %ty = type i32". There was no good reason to support this, it was just an accident of the old implementation. I have no reason to think that anyone is actually using this. 6. The syntax for sticking a global variable has changed to make it unambiguous. I don't think anyone is depending on this since only clang supports this and it is not solid yet, so I'm not worried about anything breaking. 7. This gets rid of the last use of bison, and along with it the .cvs files. I'll prune this from the makefiles as a subsequent commit. There are a few minor cleanups that can be done after this commit (suggestions welcome!) but this passes dejagnu testing and is ready for its time in the limelight. llvm-svn: 61558
2009-01-02 08:01:27 +01:00
~PerFunctionState();
Down with trailing whitespace! llvm-svn: 61594
2009-01-02 23:46:48 +01:00
Reimplement the old and horrible bison parser for .ll files with a nice and clean recursive descent parser. This change has a couple of ramifications: 1. The parser code is about 400 lines shorter (in what we maintain, not including what is autogenerated). 2. The code should be significantly faster than the old code because we don't have to work around bison's poor handling of datatypes with ctors/dtors. This also makes the code much more resistant to memory leaks. 3. We now get caret diagnostics from the .ll parser, woo. 4. The actual diagnostics emited from the parser are completely different so a bunch of testcases had to be updated. 5. I now disallow "%ty = type opaque %ty = type i32". There was no good reason to support this, it was just an accident of the old implementation. I have no reason to think that anyone is actually using this. 6. The syntax for sticking a global variable has changed to make it unambiguous. I don't think anyone is depending on this since only clang supports this and it is not solid yet, so I'm not worried about anything breaking. 7. This gets rid of the last use of bison, and along with it the .cvs files. I'll prune this from the makefiles as a subsequent commit. There are a few minor cleanups that can be done after this commit (suggestions welcome!) but this passes dejagnu testing and is ready for its time in the limelight. llvm-svn: 61558
2009-01-02 08:01:27 +01:00
Function &getFunction() const { return F; }
Down with trailing whitespace! llvm-svn: 61594
2009-01-02 23:46:48 +01:00
full asmparser support for blockaddress. We can now do: $ llvm-as foo.ll -d -disable-output which reads and prints the .ll file. BC encoding is the next project. Testcase will go in once that works. llvm-svn: 85368
2009-10-28 04:39:23 +01:00
bool FinishFunction();
Down with trailing whitespace! llvm-svn: 61594
2009-01-02 23:46:48 +01:00
Reimplement the old and horrible bison parser for .ll files with a nice and clean recursive descent parser. This change has a couple of ramifications: 1. The parser code is about 400 lines shorter (in what we maintain, not including what is autogenerated). 2. The code should be significantly faster than the old code because we don't have to work around bison's poor handling of datatypes with ctors/dtors. This also makes the code much more resistant to memory leaks. 3. We now get caret diagnostics from the .ll parser, woo. 4. The actual diagnostics emited from the parser are completely different so a bunch of testcases had to be updated. 5. I now disallow "%ty = type opaque %ty = type i32". There was no good reason to support this, it was just an accident of the old implementation. I have no reason to think that anyone is actually using this. 6. The syntax for sticking a global variable has changed to make it unambiguous. I don't think anyone is depending on this since only clang supports this and it is not solid yet, so I'm not worried about anything breaking. 7. This gets rid of the last use of bison, and along with it the .cvs files. I'll prune this from the makefiles as a subsequent commit. There are a few minor cleanups that can be done after this commit (suggestions welcome!) but this passes dejagnu testing and is ready for its time in the limelight. llvm-svn: 61558
2009-01-02 08:01:27 +01:00
/// GetVal - Get a value with the specified name or ID, creating a
/// forward reference record if needed. This can return null if the value
/// exists but does not have the right type.
land David Blaikie's patch to de-constify Type, with a few tweaks. llvm-svn: 135375
2011-07-18 06:54:35 +02:00
Value *GetVal(const std::string &Name, Type *Ty, LocTy Loc);
Value *GetVal(unsigned ID, Type *Ty, LocTy Loc);
Down with trailing whitespace! llvm-svn: 61594
2009-01-02 23:46:48 +01:00
Reimplement the old and horrible bison parser for .ll files with a nice and clean recursive descent parser. This change has a couple of ramifications: 1. The parser code is about 400 lines shorter (in what we maintain, not including what is autogenerated). 2. The code should be significantly faster than the old code because we don't have to work around bison's poor handling of datatypes with ctors/dtors. This also makes the code much more resistant to memory leaks. 3. We now get caret diagnostics from the .ll parser, woo. 4. The actual diagnostics emited from the parser are completely different so a bunch of testcases had to be updated. 5. I now disallow "%ty = type opaque %ty = type i32". There was no good reason to support this, it was just an accident of the old implementation. I have no reason to think that anyone is actually using this. 6. The syntax for sticking a global variable has changed to make it unambiguous. I don't think anyone is depending on this since only clang supports this and it is not solid yet, so I'm not worried about anything breaking. 7. This gets rid of the last use of bison, and along with it the .cvs files. I'll prune this from the makefiles as a subsequent commit. There are a few minor cleanups that can be done after this commit (suggestions welcome!) but this passes dejagnu testing and is ready for its time in the limelight. llvm-svn: 61558
2009-01-02 08:01:27 +01:00
/// SetInstName - After an instruction is parsed and inserted into its
/// basic block, this installs its name.
bool SetInstName(int NameID, const std::string &NameStr, LocTy NameLoc,
Instruction *Inst);
Down with trailing whitespace! llvm-svn: 61594
2009-01-02 23:46:48 +01:00
Reimplement the old and horrible bison parser for .ll files with a nice and clean recursive descent parser. This change has a couple of ramifications: 1. The parser code is about 400 lines shorter (in what we maintain, not including what is autogenerated). 2. The code should be significantly faster than the old code because we don't have to work around bison's poor handling of datatypes with ctors/dtors. This also makes the code much more resistant to memory leaks. 3. We now get caret diagnostics from the .ll parser, woo. 4. The actual diagnostics emited from the parser are completely different so a bunch of testcases had to be updated. 5. I now disallow "%ty = type opaque %ty = type i32". There was no good reason to support this, it was just an accident of the old implementation. I have no reason to think that anyone is actually using this. 6. The syntax for sticking a global variable has changed to make it unambiguous. I don't think anyone is depending on this since only clang supports this and it is not solid yet, so I'm not worried about anything breaking. 7. This gets rid of the last use of bison, and along with it the .cvs files. I'll prune this from the makefiles as a subsequent commit. There are a few minor cleanups that can be done after this commit (suggestions welcome!) but this passes dejagnu testing and is ready for its time in the limelight. llvm-svn: 61558
2009-01-02 08:01:27 +01:00
/// GetBB - Get a basic block with the specified name or ID, creating a
/// forward reference record if needed. This can return null if the value
/// is not a BasicBlock.
BasicBlock *GetBB(const std::string &Name, LocTy Loc);
BasicBlock *GetBB(unsigned ID, LocTy Loc);
Down with trailing whitespace! llvm-svn: 61594
2009-01-02 23:46:48 +01:00
Reimplement the old and horrible bison parser for .ll files with a nice and clean recursive descent parser. This change has a couple of ramifications: 1. The parser code is about 400 lines shorter (in what we maintain, not including what is autogenerated). 2. The code should be significantly faster than the old code because we don't have to work around bison's poor handling of datatypes with ctors/dtors. This also makes the code much more resistant to memory leaks. 3. We now get caret diagnostics from the .ll parser, woo. 4. The actual diagnostics emited from the parser are completely different so a bunch of testcases had to be updated. 5. I now disallow "%ty = type opaque %ty = type i32". There was no good reason to support this, it was just an accident of the old implementation. I have no reason to think that anyone is actually using this. 6. The syntax for sticking a global variable has changed to make it unambiguous. I don't think anyone is depending on this since only clang supports this and it is not solid yet, so I'm not worried about anything breaking. 7. This gets rid of the last use of bison, and along with it the .cvs files. I'll prune this from the makefiles as a subsequent commit. There are a few minor cleanups that can be done after this commit (suggestions welcome!) but this passes dejagnu testing and is ready for its time in the limelight. llvm-svn: 61558
2009-01-02 08:01:27 +01:00
/// DefineBB - Define the specified basic block, which is either named or
/// unnamed. If there is an error, this returns null otherwise it returns
/// the block being defined.
BasicBlock *DefineBB(const std::string &Name, LocTy Loc);
};
Down with trailing whitespace! llvm-svn: 61594
2009-01-02 23:46:48 +01:00
land David Blaikie's patch to de-constify Type, with a few tweaks. llvm-svn: 135375
2011-07-18 06:54:35 +02:00
bool ConvertValIDToValue(Type *Ty, ValID &ID, Value *&V,
Consolidate ConvertGlobalValIDToValue, ConvertGlobalOrMetadataValIDToValue, and ConvertValIDToValue into a more powerful ConvertValIDToValue() that does all three's work llvm-svn: 93197
2010-01-11 23:31:58 +01:00
PerFunctionState *PFS);
Down with trailing whitespace! llvm-svn: 61594
2009-01-02 23:46:48 +01:00
land David Blaikie's patch to de-constify Type, with a few tweaks. llvm-svn: 135375
2011-07-18 06:54:35 +02:00
bool ParseValue(Type *Ty, Value *&V, PerFunctionState *PFS);
bool ParseValue(Type *Ty, Value *&V, PerFunctionState &PFS) {
Land the long talked about "type system rewrite" patch. This patch brings numerous advantages to LLVM. One way to look at it is through diffstat: 109 files changed, 3005 insertions(+), 5906 deletions(-) Removing almost 3K lines of code is a good thing. Other advantages include: 1. Value::getType() is a simple load that can be CSE'd, not a mutating union-find operation. 2. Types a uniqued and never move once created, defining away PATypeHolder. 3. Structs can be "named" now, and their name is part of the identity that uniques them. This means that the compiler doesn't merge them structurally which makes the IR much less confusing. 4. Now that there is no way to get a cycle in a type graph without a named struct type, "upreferences" go away. 5. Type refinement is completely gone, which should make LTO much MUCH faster in some common cases with C++ code. 6. Types are now generally immutable, so we can use "Type *" instead "const Type *" everywhere. Downsides of this patch are that it removes some functions from the C API, so people using those will have to upgrade to (not yet added) new API. "LLVM 3.0" is the right time to do this. There are still some cleanups pending after this, this patch is large enough as-is. llvm-svn: 134829
2011-07-09 19:41:24 +02:00
return ParseValue(Ty, V, &PFS);
}
land David Blaikie's patch to de-constify Type, with a few tweaks. llvm-svn: 135375
2011-07-18 06:54:35 +02:00
bool ParseValue(Type *Ty, Value *&V, LocTy &Loc,
Reimplement the old and horrible bison parser for .ll files with a nice and clean recursive descent parser. This change has a couple of ramifications: 1. The parser code is about 400 lines shorter (in what we maintain, not including what is autogenerated). 2. The code should be significantly faster than the old code because we don't have to work around bison's poor handling of datatypes with ctors/dtors. This also makes the code much more resistant to memory leaks. 3. We now get caret diagnostics from the .ll parser, woo. 4. The actual diagnostics emited from the parser are completely different so a bunch of testcases had to be updated. 5. I now disallow "%ty = type opaque %ty = type i32". There was no good reason to support this, it was just an accident of the old implementation. I have no reason to think that anyone is actually using this. 6. The syntax for sticking a global variable has changed to make it unambiguous. I don't think anyone is depending on this since only clang supports this and it is not solid yet, so I'm not worried about anything breaking. 7. This gets rid of the last use of bison, and along with it the .cvs files. I'll prune this from the makefiles as a subsequent commit. There are a few minor cleanups that can be done after this commit (suggestions welcome!) but this passes dejagnu testing and is ready for its time in the limelight. llvm-svn: 61558
2009-01-02 08:01:27 +01:00
PerFunctionState &PFS) {
Loc = Lex.getLoc();
Land the long talked about "type system rewrite" patch. This patch brings numerous advantages to LLVM. One way to look at it is through diffstat: 109 files changed, 3005 insertions(+), 5906 deletions(-) Removing almost 3K lines of code is a good thing. Other advantages include: 1. Value::getType() is a simple load that can be CSE'd, not a mutating union-find operation. 2. Types a uniqued and never move once created, defining away PATypeHolder. 3. Structs can be "named" now, and their name is part of the identity that uniques them. This means that the compiler doesn't merge them structurally which makes the IR much less confusing. 4. Now that there is no way to get a cycle in a type graph without a named struct type, "upreferences" go away. 5. Type refinement is completely gone, which should make LTO much MUCH faster in some common cases with C++ code. 6. Types are now generally immutable, so we can use "Type *" instead "const Type *" everywhere. Downsides of this patch are that it removes some functions from the C API, so people using those will have to upgrade to (not yet added) new API. "LLVM 3.0" is the right time to do this. There are still some cleanups pending after this, this patch is large enough as-is. llvm-svn: 134829
2011-07-09 19:41:24 +02:00
return ParseValue(Ty, V, &PFS);
Reimplement the old and horrible bison parser for .ll files with a nice and clean recursive descent parser. This change has a couple of ramifications: 1. The parser code is about 400 lines shorter (in what we maintain, not including what is autogenerated). 2. The code should be significantly faster than the old code because we don't have to work around bison's poor handling of datatypes with ctors/dtors. This also makes the code much more resistant to memory leaks. 3. We now get caret diagnostics from the .ll parser, woo. 4. The actual diagnostics emited from the parser are completely different so a bunch of testcases had to be updated. 5. I now disallow "%ty = type opaque %ty = type i32". There was no good reason to support this, it was just an accident of the old implementation. I have no reason to think that anyone is actually using this. 6. The syntax for sticking a global variable has changed to make it unambiguous. I don't think anyone is depending on this since only clang supports this and it is not solid yet, so I'm not worried about anything breaking. 7. This gets rid of the last use of bison, and along with it the .cvs files. I'll prune this from the makefiles as a subsequent commit. There are a few minor cleanups that can be done after this commit (suggestions welcome!) but this passes dejagnu testing and is ready for its time in the limelight. llvm-svn: 61558
2009-01-02 08:01:27 +01:00
}
Down with trailing whitespace! llvm-svn: 61594
2009-01-02 23:46:48 +01:00
Land the long talked about "type system rewrite" patch. This patch brings numerous advantages to LLVM. One way to look at it is through diffstat: 109 files changed, 3005 insertions(+), 5906 deletions(-) Removing almost 3K lines of code is a good thing. Other advantages include: 1. Value::getType() is a simple load that can be CSE'd, not a mutating union-find operation. 2. Types a uniqued and never move once created, defining away PATypeHolder. 3. Structs can be "named" now, and their name is part of the identity that uniques them. This means that the compiler doesn't merge them structurally which makes the IR much less confusing. 4. Now that there is no way to get a cycle in a type graph without a named struct type, "upreferences" go away. 5. Type refinement is completely gone, which should make LTO much MUCH faster in some common cases with C++ code. 6. Types are now generally immutable, so we can use "Type *" instead "const Type *" everywhere. Downsides of this patch are that it removes some functions from the C API, so people using those will have to upgrade to (not yet added) new API. "LLVM 3.0" is the right time to do this. There are still some cleanups pending after this, this patch is large enough as-is. llvm-svn: 134829
2011-07-09 19:41:24 +02:00
bool ParseTypeAndValue(Value *&V, PerFunctionState *PFS);
bool ParseTypeAndValue(Value *&V, PerFunctionState &PFS) {
return ParseTypeAndValue(V, &PFS);
}
Reimplement the old and horrible bison parser for .ll files with a nice and clean recursive descent parser. This change has a couple of ramifications: 1. The parser code is about 400 lines shorter (in what we maintain, not including what is autogenerated). 2. The code should be significantly faster than the old code because we don't have to work around bison's poor handling of datatypes with ctors/dtors. This also makes the code much more resistant to memory leaks. 3. We now get caret diagnostics from the .ll parser, woo. 4. The actual diagnostics emited from the parser are completely different so a bunch of testcases had to be updated. 5. I now disallow "%ty = type opaque %ty = type i32". There was no good reason to support this, it was just an accident of the old implementation. I have no reason to think that anyone is actually using this. 6. The syntax for sticking a global variable has changed to make it unambiguous. I don't think anyone is depending on this since only clang supports this and it is not solid yet, so I'm not worried about anything breaking. 7. This gets rid of the last use of bison, and along with it the .cvs files. I'll prune this from the makefiles as a subsequent commit. There are a few minor cleanups that can be done after this commit (suggestions welcome!) but this passes dejagnu testing and is ready for its time in the limelight. llvm-svn: 61558
2009-01-02 08:01:27 +01:00
bool ParseTypeAndValue(Value *&V, LocTy &Loc, PerFunctionState &PFS) {
Loc = Lex.getLoc();
return ParseTypeAndValue(V, PFS);
}
add enough support for indirect branch for the feature test to pass (assembler,asmprinter, bc reader+writer) and document it. Codegen currently aborts on it. llvm-svn: 85274
2009-10-27 20:13:16 +01:00
bool ParseTypeAndBasicBlock(BasicBlock *&BB, LocTy &Loc,
PerFunctionState &PFS);
bool ParseTypeAndBasicBlock(BasicBlock *&BB, PerFunctionState &PFS) {
LocTy Loc;
return ParseTypeAndBasicBlock(BB, Loc, PFS);
}
Add ParseInlineMetadata() which can parses metadata that refers to an instruction. Extend ParseParameterList() to use this new function so that calls to llvm.dbg.declare can pass inline metadata llvm-svn: 90497
2009-12-04 00:40:58 +01:00
Add support for a union type in LLVM IR. Patch by Talin! llvm-svn: 96011
2010-02-12 21:49:41 +01:00
Reimplement the old and horrible bison parser for .ll files with a nice and clean recursive descent parser. This change has a couple of ramifications: 1. The parser code is about 400 lines shorter (in what we maintain, not including what is autogenerated). 2. The code should be significantly faster than the old code because we don't have to work around bison's poor handling of datatypes with ctors/dtors. This also makes the code much more resistant to memory leaks. 3. We now get caret diagnostics from the .ll parser, woo. 4. The actual diagnostics emited from the parser are completely different so a bunch of testcases had to be updated. 5. I now disallow "%ty = type opaque %ty = type i32". There was no good reason to support this, it was just an accident of the old implementation. I have no reason to think that anyone is actually using this. 6. The syntax for sticking a global variable has changed to make it unambiguous. I don't think anyone is depending on this since only clang supports this and it is not solid yet, so I'm not worried about anything breaking. 7. This gets rid of the last use of bison, and along with it the .cvs files. I'll prune this from the makefiles as a subsequent commit. There are a few minor cleanups that can be done after this commit (suggestions welcome!) but this passes dejagnu testing and is ready for its time in the limelight. llvm-svn: 61558
2009-01-02 08:01:27 +01:00
struct ParamInfo {
LocTy Loc;
Value *V;
unsigned Attrs;
Down with trailing whitespace! llvm-svn: 61594
2009-01-02 23:46:48 +01:00
ParamInfo(LocTy loc, Value *v, unsigned attrs)
Reimplement the old and horrible bison parser for .ll files with a nice and clean recursive descent parser. This change has a couple of ramifications: 1. The parser code is about 400 lines shorter (in what we maintain, not including what is autogenerated). 2. The code should be significantly faster than the old code because we don't have to work around bison's poor handling of datatypes with ctors/dtors. This also makes the code much more resistant to memory leaks. 3. We now get caret diagnostics from the .ll parser, woo. 4. The actual diagnostics emited from the parser are completely different so a bunch of testcases had to be updated. 5. I now disallow "%ty = type opaque %ty = type i32". There was no good reason to support this, it was just an accident of the old implementation. I have no reason to think that anyone is actually using this. 6. The syntax for sticking a global variable has changed to make it unambiguous. I don't think anyone is depending on this since only clang supports this and it is not solid yet, so I'm not worried about anything breaking. 7. This gets rid of the last use of bison, and along with it the .cvs files. I'll prune this from the makefiles as a subsequent commit. There are a few minor cleanups that can be done after this commit (suggestions welcome!) but this passes dejagnu testing and is ready for its time in the limelight. llvm-svn: 61558
2009-01-02 08:01:27 +01:00
: Loc(loc), V(v), Attrs(attrs) {}
};
bool ParseParameterList(SmallVectorImpl<ParamInfo> &ArgList,
PerFunctionState &PFS);
Down with trailing whitespace! llvm-svn: 61594
2009-01-02 23:46:48 +01:00
Re-add parsing of function-local metadata; this time with testcase. llvm-svn: 92793
2010-01-05 23:22:14 +01:00
// Constant Parsing.
bool ParseValID(ValID &ID, PerFunctionState *PFS = NULL);
land David Blaikie's patch to de-constify Type, with a few tweaks. llvm-svn: 135375
2011-07-18 06:54:35 +02:00
bool ParseGlobalValue(Type *Ty, Constant *&V);
Re-add parsing of function-local metadata; this time with testcase. llvm-svn: 92793
2010-01-05 23:22:14 +01:00
bool ParseGlobalTypeAndValue(Constant *&V);
bool ParseGlobalValueVector(SmallVectorImpl<Constant*> &Elts);
Extend function-local metadata to be usable as attachments. llvm-svn: 111895
2010-08-24 04:24:03 +02:00
bool ParseMetadataListValue(ValID &ID, PerFunctionState *PFS);
Factor out metadata parsing into a separate function. llvm-svn: 108343
2010-07-14 20:26:50 +02:00
bool ParseMetadataValue(ValID &ID, PerFunctionState *PFS);
Compute isFunctionLocal in MDNode ctor or via argument in new function getWhenValsUnresolved(). Document PFS argument to ParseValID() and ConvertGlobalOrMetadataValIDToValue(). llvm-svn: 93108
2010-01-10 08:14:18 +01:00
bool ParseMDNodeVector(SmallVectorImpl<Value*> &, PerFunctionState *PFS);
Give ParseInstructionMetadata access to the PerFunctionState object. This is in preparation for generalizing its parsing of function-local values. llvm-svn: 111893
2010-08-24 04:05:17 +02:00
bool ParseInstructionMetadata(Instruction *Inst, PerFunctionState *PFS);
Re-add parsing of function-local metadata; this time with testcase. llvm-svn: 92793
2010-01-05 23:22:14 +01:00
Reimplement the old and horrible bison parser for .ll files with a nice and clean recursive descent parser. This change has a couple of ramifications: 1. The parser code is about 400 lines shorter (in what we maintain, not including what is autogenerated). 2. The code should be significantly faster than the old code because we don't have to work around bison's poor handling of datatypes with ctors/dtors. This also makes the code much more resistant to memory leaks. 3. We now get caret diagnostics from the .ll parser, woo. 4. The actual diagnostics emited from the parser are completely different so a bunch of testcases had to be updated. 5. I now disallow "%ty = type opaque %ty = type i32". There was no good reason to support this, it was just an accident of the old implementation. I have no reason to think that anyone is actually using this. 6. The syntax for sticking a global variable has changed to make it unambiguous. I don't think anyone is depending on this since only clang supports this and it is not solid yet, so I'm not worried about anything breaking. 7. This gets rid of the last use of bison, and along with it the .cvs files. I'll prune this from the makefiles as a subsequent commit. There are a few minor cleanups that can be done after this commit (suggestions welcome!) but this passes dejagnu testing and is ready for its time in the limelight. llvm-svn: 61558
2009-01-02 08:01:27 +01:00
// Function Parsing.
struct ArgInfo {
LocTy Loc;
Land the long talked about "type system rewrite" patch. This patch brings numerous advantages to LLVM. One way to look at it is through diffstat: 109 files changed, 3005 insertions(+), 5906 deletions(-) Removing almost 3K lines of code is a good thing. Other advantages include: 1. Value::getType() is a simple load that can be CSE'd, not a mutating union-find operation. 2. Types a uniqued and never move once created, defining away PATypeHolder. 3. Structs can be "named" now, and their name is part of the identity that uniques them. This means that the compiler doesn't merge them structurally which makes the IR much less confusing. 4. Now that there is no way to get a cycle in a type graph without a named struct type, "upreferences" go away. 5. Type refinement is completely gone, which should make LTO much MUCH faster in some common cases with C++ code. 6. Types are now generally immutable, so we can use "Type *" instead "const Type *" everywhere. Downsides of this patch are that it removes some functions from the C API, so people using those will have to upgrade to (not yet added) new API. "LLVM 3.0" is the right time to do this. There are still some cleanups pending after this, this patch is large enough as-is. llvm-svn: 134829
2011-07-09 19:41:24 +02:00
Type *Ty;
Reimplement the old and horrible bison parser for .ll files with a nice and clean recursive descent parser. This change has a couple of ramifications: 1. The parser code is about 400 lines shorter (in what we maintain, not including what is autogenerated). 2. The code should be significantly faster than the old code because we don't have to work around bison's poor handling of datatypes with ctors/dtors. This also makes the code much more resistant to memory leaks. 3. We now get caret diagnostics from the .ll parser, woo. 4. The actual diagnostics emited from the parser are completely different so a bunch of testcases had to be updated. 5. I now disallow "%ty = type opaque %ty = type i32". There was no good reason to support this, it was just an accident of the old implementation. I have no reason to think that anyone is actually using this. 6. The syntax for sticking a global variable has changed to make it unambiguous. I don't think anyone is depending on this since only clang supports this and it is not solid yet, so I'm not worried about anything breaking. 7. This gets rid of the last use of bison, and along with it the .cvs files. I'll prune this from the makefiles as a subsequent commit. There are a few minor cleanups that can be done after this commit (suggestions welcome!) but this passes dejagnu testing and is ready for its time in the limelight. llvm-svn: 61558
2009-01-02 08:01:27 +01:00
unsigned Attrs;
std::string Name;
Land the long talked about "type system rewrite" patch. This patch brings numerous advantages to LLVM. One way to look at it is through diffstat: 109 files changed, 3005 insertions(+), 5906 deletions(-) Removing almost 3K lines of code is a good thing. Other advantages include: 1. Value::getType() is a simple load that can be CSE'd, not a mutating union-find operation. 2. Types a uniqued and never move once created, defining away PATypeHolder. 3. Structs can be "named" now, and their name is part of the identity that uniques them. This means that the compiler doesn't merge them structurally which makes the IR much less confusing. 4. Now that there is no way to get a cycle in a type graph without a named struct type, "upreferences" go away. 5. Type refinement is completely gone, which should make LTO much MUCH faster in some common cases with C++ code. 6. Types are now generally immutable, so we can use "Type *" instead "const Type *" everywhere. Downsides of this patch are that it removes some functions from the C API, so people using those will have to upgrade to (not yet added) new API. "LLVM 3.0" is the right time to do this. There are still some cleanups pending after this, this patch is large enough as-is. llvm-svn: 134829
2011-07-09 19:41:24 +02:00
ArgInfo(LocTy L, Type *ty, unsigned Attr, const std::string &N)
: Loc(L), Ty(ty), Attrs(Attr), Name(N) {}
Reimplement the old and horrible bison parser for .ll files with a nice and clean recursive descent parser. This change has a couple of ramifications: 1. The parser code is about 400 lines shorter (in what we maintain, not including what is autogenerated). 2. The code should be significantly faster than the old code because we don't have to work around bison's poor handling of datatypes with ctors/dtors. This also makes the code much more resistant to memory leaks. 3. We now get caret diagnostics from the .ll parser, woo. 4. The actual diagnostics emited from the parser are completely different so a bunch of testcases had to be updated. 5. I now disallow "%ty = type opaque %ty = type i32". There was no good reason to support this, it was just an accident of the old implementation. I have no reason to think that anyone is actually using this. 6. The syntax for sticking a global variable has changed to make it unambiguous. I don't think anyone is depending on this since only clang supports this and it is not solid yet, so I'm not worried about anything breaking. 7. This gets rid of the last use of bison, and along with it the .cvs files. I'll prune this from the makefiles as a subsequent commit. There are a few minor cleanups that can be done after this commit (suggestions welcome!) but this passes dejagnu testing and is ready for its time in the limelight. llvm-svn: 61558
2009-01-02 08:01:27 +01:00
};
Land the long talked about "type system rewrite" patch. This patch brings numerous advantages to LLVM. One way to look at it is through diffstat: 109 files changed, 3005 insertions(+), 5906 deletions(-) Removing almost 3K lines of code is a good thing. Other advantages include: 1. Value::getType() is a simple load that can be CSE'd, not a mutating union-find operation. 2. Types a uniqued and never move once created, defining away PATypeHolder. 3. Structs can be "named" now, and their name is part of the identity that uniques them. This means that the compiler doesn't merge them structurally which makes the IR much less confusing. 4. Now that there is no way to get a cycle in a type graph without a named struct type, "upreferences" go away. 5. Type refinement is completely gone, which should make LTO much MUCH faster in some common cases with C++ code. 6. Types are now generally immutable, so we can use "Type *" instead "const Type *" everywhere. Downsides of this patch are that it removes some functions from the C API, so people using those will have to upgrade to (not yet added) new API. "LLVM 3.0" is the right time to do this. There are still some cleanups pending after this, this patch is large enough as-is. llvm-svn: 134829
2011-07-09 19:41:24 +02:00
bool ParseArgumentList(SmallVectorImpl<ArgInfo> &ArgList, bool &isVarArg);
Reimplement the old and horrible bison parser for .ll files with a nice and clean recursive descent parser. This change has a couple of ramifications: 1. The parser code is about 400 lines shorter (in what we maintain, not including what is autogenerated). 2. The code should be significantly faster than the old code because we don't have to work around bison's poor handling of datatypes with ctors/dtors. This also makes the code much more resistant to memory leaks. 3. We now get caret diagnostics from the .ll parser, woo. 4. The actual diagnostics emited from the parser are completely different so a bunch of testcases had to be updated. 5. I now disallow "%ty = type opaque %ty = type i32". There was no good reason to support this, it was just an accident of the old implementation. I have no reason to think that anyone is actually using this. 6. The syntax for sticking a global variable has changed to make it unambiguous. I don't think anyone is depending on this since only clang supports this and it is not solid yet, so I'm not worried about anything breaking. 7. This gets rid of the last use of bison, and along with it the .cvs files. I'll prune this from the makefiles as a subsequent commit. There are a few minor cleanups that can be done after this commit (suggestions welcome!) but this passes dejagnu testing and is ready for its time in the limelight. llvm-svn: 61558
2009-01-02 08:01:27 +01:00
bool ParseFunctionHeader(Function *&Fn, bool isDefine);
bool ParseFunctionBody(Function &Fn);
bool ParseBasicBlock(PerFunctionState &PFS);
Down with trailing whitespace! llvm-svn: 61594
2009-01-02 23:46:48 +01:00
add facilities to start factoring instruction metadata parsing out of each opcode's handler. Change ret over so far. llvm-svn: 92298
2009-12-30 06:23:43 +01:00
// Instruction Parsing. Each instruction parsing routine can return with a
// normal result, an error result, or return having eaten an extra comma.
enum InstResult { InstNormal = 0, InstError = 1, InstExtraComma = 2 };
int ParseInstruction(Instruction *&Inst, BasicBlock *BB,
PerFunctionState &PFS);
Reimplement the old and horrible bison parser for .ll files with a nice and clean recursive descent parser. This change has a couple of ramifications: 1. The parser code is about 400 lines shorter (in what we maintain, not including what is autogenerated). 2. The code should be significantly faster than the old code because we don't have to work around bison's poor handling of datatypes with ctors/dtors. This also makes the code much more resistant to memory leaks. 3. We now get caret diagnostics from the .ll parser, woo. 4. The actual diagnostics emited from the parser are completely different so a bunch of testcases had to be updated. 5. I now disallow "%ty = type opaque %ty = type i32". There was no good reason to support this, it was just an accident of the old implementation. I have no reason to think that anyone is actually using this. 6. The syntax for sticking a global variable has changed to make it unambiguous. I don't think anyone is depending on this since only clang supports this and it is not solid yet, so I'm not worried about anything breaking. 7. This gets rid of the last use of bison, and along with it the .cvs files. I'll prune this from the makefiles as a subsequent commit. There are a few minor cleanups that can be done after this commit (suggestions welcome!) but this passes dejagnu testing and is ready for its time in the limelight. llvm-svn: 61558
2009-01-02 08:01:27 +01:00
bool ParseCmpPredicate(unsigned &Pred, unsigned Opc);
Down with trailing whitespace! llvm-svn: 61594
2009-01-02 23:46:48 +01:00
remove parser support for the obsolete "multiple return values" syntax, which was replaced with return of a "first class aggregate". llvm-svn: 133245
2011-06-17 08:49:41 +02:00
bool ParseRet(Instruction *&Inst, BasicBlock *BB, PerFunctionState &PFS);
Reimplement the old and horrible bison parser for .ll files with a nice and clean recursive descent parser. This change has a couple of ramifications: 1. The parser code is about 400 lines shorter (in what we maintain, not including what is autogenerated). 2. The code should be significantly faster than the old code because we don't have to work around bison's poor handling of datatypes with ctors/dtors. This also makes the code much more resistant to memory leaks. 3. We now get caret diagnostics from the .ll parser, woo. 4. The actual diagnostics emited from the parser are completely different so a bunch of testcases had to be updated. 5. I now disallow "%ty = type opaque %ty = type i32". There was no good reason to support this, it was just an accident of the old implementation. I have no reason to think that anyone is actually using this. 6. The syntax for sticking a global variable has changed to make it unambiguous. I don't think anyone is depending on this since only clang supports this and it is not solid yet, so I'm not worried about anything breaking. 7. This gets rid of the last use of bison, and along with it the .cvs files. I'll prune this from the makefiles as a subsequent commit. There are a few minor cleanups that can be done after this commit (suggestions welcome!) but this passes dejagnu testing and is ready for its time in the limelight. llvm-svn: 61558
2009-01-02 08:01:27 +01:00
bool ParseBr(Instruction *&Inst, PerFunctionState &PFS);
bool ParseSwitch(Instruction *&Inst, PerFunctionState &PFS);
rename indbr -> indirectbr to appease the residents of #llvm. llvm-svn: 85351
2009-10-28 01:19:10 +01:00
bool ParseIndirectBr(Instruction *&Inst, PerFunctionState &PFS);
Reimplement the old and horrible bison parser for .ll files with a nice and clean recursive descent parser. This change has a couple of ramifications: 1. The parser code is about 400 lines shorter (in what we maintain, not including what is autogenerated). 2. The code should be significantly faster than the old code because we don't have to work around bison's poor handling of datatypes with ctors/dtors. This also makes the code much more resistant to memory leaks. 3. We now get caret diagnostics from the .ll parser, woo. 4. The actual diagnostics emited from the parser are completely different so a bunch of testcases had to be updated. 5. I now disallow "%ty = type opaque %ty = type i32". There was no good reason to support this, it was just an accident of the old implementation. I have no reason to think that anyone is actually using this. 6. The syntax for sticking a global variable has changed to make it unambiguous. I don't think anyone is depending on this since only clang supports this and it is not solid yet, so I'm not worried about anything breaking. 7. This gets rid of the last use of bison, and along with it the .cvs files. I'll prune this from the makefiles as a subsequent commit. There are a few minor cleanups that can be done after this commit (suggestions welcome!) but this passes dejagnu testing and is ready for its time in the limelight. llvm-svn: 61558
2009-01-02 08:01:27 +01:00
bool ParseInvoke(Instruction *&Inst, PerFunctionState &PFS);
Add the 'resume' instruction for the new EH rewrite. This adds the 'resume' instruction class, IR parsing, and bitcode reading and writing. The 'resume' instruction resumes propagation of an existing (in-flight) exception whose unwinding was interrupted with a 'landingpad' instruction (to be added later). llvm-svn: 136589
2011-07-31 08:30:59 +02:00
bool ParseResume(Instruction *&Inst, PerFunctionState &PFS);
Down with trailing whitespace! llvm-svn: 61594
2009-01-02 23:46:48 +01:00
Fix PR3281:crash08.ll with this diagnostic: llvm-as: crash08.ll:3:15: invalid operand type for instruction "qp" = sdiv fp128 0x1, %30 ^ llvm-svn: 61684
2009-01-05 09:24:46 +01:00
bool ParseArithmetic(Instruction *&I, PerFunctionState &PFS, unsigned Opc,
unsigned OperandType);
Reimplement the old and horrible bison parser for .ll files with a nice and clean recursive descent parser. This change has a couple of ramifications: 1. The parser code is about 400 lines shorter (in what we maintain, not including what is autogenerated). 2. The code should be significantly faster than the old code because we don't have to work around bison's poor handling of datatypes with ctors/dtors. This also makes the code much more resistant to memory leaks. 3. We now get caret diagnostics from the .ll parser, woo. 4. The actual diagnostics emited from the parser are completely different so a bunch of testcases had to be updated. 5. I now disallow "%ty = type opaque %ty = type i32". There was no good reason to support this, it was just an accident of the old implementation. I have no reason to think that anyone is actually using this. 6. The syntax for sticking a global variable has changed to make it unambiguous. I don't think anyone is depending on this since only clang supports this and it is not solid yet, so I'm not worried about anything breaking. 7. This gets rid of the last use of bison, and along with it the .cvs files. I'll prune this from the makefiles as a subsequent commit. There are a few minor cleanups that can be done after this commit (suggestions welcome!) but this passes dejagnu testing and is ready for its time in the limelight. llvm-svn: 61558
2009-01-02 08:01:27 +01:00
bool ParseLogical(Instruction *&I, PerFunctionState &PFS, unsigned Opc);
bool ParseCompare(Instruction *&I, PerFunctionState &PFS, unsigned Opc);
bool ParseCast(Instruction *&I, PerFunctionState &PFS, unsigned Opc);
bool ParseSelect(Instruction *&I, PerFunctionState &PFS);
reject PR3281:crash07.ll with: llvm-as: crash07.ll:2:32: va_arg requires operand with first class type %y = va_arg [52 x <{}>] %43, double (...) sspreq ^ llvm-svn: 61683
2009-01-05 09:18:44 +01:00
bool ParseVA_Arg(Instruction *&I, PerFunctionState &PFS);
Reimplement the old and horrible bison parser for .ll files with a nice and clean recursive descent parser. This change has a couple of ramifications: 1. The parser code is about 400 lines shorter (in what we maintain, not including what is autogenerated). 2. The code should be significantly faster than the old code because we don't have to work around bison's poor handling of datatypes with ctors/dtors. This also makes the code much more resistant to memory leaks. 3. We now get caret diagnostics from the .ll parser, woo. 4. The actual diagnostics emited from the parser are completely different so a bunch of testcases had to be updated. 5. I now disallow "%ty = type opaque %ty = type i32". There was no good reason to support this, it was just an accident of the old implementation. I have no reason to think that anyone is actually using this. 6. The syntax for sticking a global variable has changed to make it unambiguous. I don't think anyone is depending on this since only clang supports this and it is not solid yet, so I'm not worried about anything breaking. 7. This gets rid of the last use of bison, and along with it the .cvs files. I'll prune this from the makefiles as a subsequent commit. There are a few minor cleanups that can be done after this commit (suggestions welcome!) but this passes dejagnu testing and is ready for its time in the limelight. llvm-svn: 61558
2009-01-02 08:01:27 +01:00
bool ParseExtractElement(Instruction *&I, PerFunctionState &PFS);
bool ParseInsertElement(Instruction *&I, PerFunctionState &PFS);
bool ParseShuffleVector(Instruction *&I, PerFunctionState &PFS);
convert 4 more instructions over. llvm-svn: 92299
2009-12-30 06:27:33 +01:00
int ParsePHI(Instruction *&I, PerFunctionState &PFS);
Initial commit of the 'landingpad' instruction. This implements the 'landingpad' instruction. It's used to indicate that a basic block is a landing pad. There are several restrictions on its use (see LangRef.html for more detail). These restrictions allow the exception handling code to gather the information it needs in a much more sane way. This patch has the definition, implementation, C interface, parsing, and bitcode support in it. llvm-svn: 137501
2011-08-12 22:24:12 +02:00
bool ParseLandingPad(Instruction *&I, PerFunctionState &PFS);
Reimplement the old and horrible bison parser for .ll files with a nice and clean recursive descent parser. This change has a couple of ramifications: 1. The parser code is about 400 lines shorter (in what we maintain, not including what is autogenerated). 2. The code should be significantly faster than the old code because we don't have to work around bison's poor handling of datatypes with ctors/dtors. This also makes the code much more resistant to memory leaks. 3. We now get caret diagnostics from the .ll parser, woo. 4. The actual diagnostics emited from the parser are completely different so a bunch of testcases had to be updated. 5. I now disallow "%ty = type opaque %ty = type i32". There was no good reason to support this, it was just an accident of the old implementation. I have no reason to think that anyone is actually using this. 6. The syntax for sticking a global variable has changed to make it unambiguous. I don't think anyone is depending on this since only clang supports this and it is not solid yet, so I'm not worried about anything breaking. 7. This gets rid of the last use of bison, and along with it the .cvs files. I'll prune this from the makefiles as a subsequent commit. There are a few minor cleanups that can be done after this commit (suggestions welcome!) but this passes dejagnu testing and is ready for its time in the limelight. llvm-svn: 61558
2009-01-02 08:01:27 +01:00
bool ParseCall(Instruction *&I, PerFunctionState &PFS, bool isTail);
Remove old backwards compatibility support from the parser for autoupgrading the old malloc/free instructions, and for 'sext' and 'zext' as function attributes (they are spelled signext/zeroext now), and support for result value attributes being specified after a function. Additionally, diagnose invalid attributes on functions with an error message instead of an abort in the verifier. llvm-svn: 133229
2011-06-17 05:16:47 +02:00
int ParseAlloc(Instruction *&I, PerFunctionState &PFS);
remove asmparsing and documentation support for "volatile load", which was only produced by LLVM 2.9 and earlier. LLVM 3.0 and later prefers "load volatile". llvm-svn: 145172
2011-11-27 07:56:53 +01:00
int ParseLoad(Instruction *&I, PerFunctionState &PFS);
int ParseStore(Instruction *&I, PerFunctionState &PFS);
Move "atomic" and "volatile" designations on instructions after the opcode of the instruction. Note that this change affects the existing non-atomic load and store instructions; the parser now accepts both forms, and the change is noted in the release notes. llvm-svn: 137527
2011-08-13 00:50:01 +02:00
int ParseCmpXchg(Instruction *&I, PerFunctionState &PFS);
int ParseAtomicRMW(Instruction *&I, PerFunctionState &PFS);
Initial implementation of 'fence' instruction, the new C++0x-style replacement for llvm.memory.barrier. This is just a LangRef entry and reading/writing/memory representation; optimizer+codegen support coming soon. llvm-svn: 136009
2011-07-26 01:16:38 +02:00
int ParseFence(Instruction *&I, PerFunctionState &PFS);
convert 4 more instructions over. llvm-svn: 92299
2009-12-30 06:27:33 +01:00
int ParseGetElementPtr(Instruction *&I, PerFunctionState &PFS);
int ParseExtractValue(Instruction *&I, PerFunctionState &PFS);
int ParseInsertValue(Instruction *&I, PerFunctionState &PFS);
full asmparser support for blockaddress. We can now do: $ llvm-as foo.ll -d -disable-output which reads and prints the .ll file. BC encoding is the next project. Testcase will go in once that works. llvm-svn: 85368
2009-10-28 04:39:23 +01:00
bool ResolveForwardRefBlockAddresses(Function *TheFn,
std::vector<std::pair<ValID, GlobalValue*> > &Refs,
PerFunctionState *PFS);
Reimplement the old and horrible bison parser for .ll files with a nice and clean recursive descent parser. This change has a couple of ramifications: 1. The parser code is about 400 lines shorter (in what we maintain, not including what is autogenerated). 2. The code should be significantly faster than the old code because we don't have to work around bison's poor handling of datatypes with ctors/dtors. This also makes the code much more resistant to memory leaks. 3. We now get caret diagnostics from the .ll parser, woo. 4. The actual diagnostics emited from the parser are completely different so a bunch of testcases had to be updated. 5. I now disallow "%ty = type opaque %ty = type i32". There was no good reason to support this, it was just an accident of the old implementation. I have no reason to think that anyone is actually using this. 6. The syntax for sticking a global variable has changed to make it unambiguous. I don't think anyone is depending on this since only clang supports this and it is not solid yet, so I'm not worried about anything breaking. 7. This gets rid of the last use of bison, and along with it the .cvs files. I'll prune this from the makefiles as a subsequent commit. There are a few minor cleanups that can be done after this commit (suggestions welcome!) but this passes dejagnu testing and is ready for its time in the limelight. llvm-svn: 61558
2009-01-02 08:01:27 +01:00
};
} // End llvm namespace
#endif
Reference in New Issue Copy Permalink