Files
llvm/lldb/source/Plugins/SymbolFile/DWARF/SymbolFileDWARFDebugMap.cpp

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

1428 lines
54 KiB
C++
Raw Normal View History

//===-- SymbolFileDWARFDebugMap.cpp -----------------------------*- C++ -*-===//
//
// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
// See https://llvm.org/LICENSE.txt for license information.
// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
//
//===----------------------------------------------------------------------===//
#include "SymbolFileDWARFDebugMap.h"
#include "DWARFDebugAranges.h"
#include "lldb/Core/Module.h"
#include "lldb/Core/ModuleList.h"
#include "lldb/Core/PluginManager.h"
<rdar://problem/11757916> Make breakpoint setting by file and line much more efficient by only looking for inlined breakpoint locations if we are setting a breakpoint in anything but a source implementation file. Implementing this complex for a many reasons. Turns out that parsing compile units lazily had some issues with respect to how we need to do things with DWARF in .o files. So the fixes in the checkin for this makes these changes: - Add a new setting called "target.inline-breakpoint-strategy" which can be set to "never", "always", or "headers". "never" will never try and set any inlined breakpoints (fastest). "always" always looks for inlined breakpoint locations (slowest, but most accurate). "headers", which is the default setting, will only look for inlined breakpoint locations if the breakpoint is set in what are consudered to be header files, which is realy defined as "not in an implementation source file". - modify the breakpoint setting by file and line to check the current "target.inline-breakpoint-strategy" setting and act accordingly - Modify compile units to be able to get their language and other info lazily. This allows us to create compile units from the debug map and not have to fill all of the details in, and then lazily discover this information as we go on debuggging. This is needed to avoid parsing all .o files when setting breakpoints in implementation only files (no inlines). Otherwise we would need to parse the .o file, the object file (mach-o in our case) and the symbol file (DWARF in the object file) just to see what the compile unit was. - modify the "SymbolFileDWARFDebugMap" to subclass lldb_private::Module so that the virtual "GetObjectFile()" and "GetSymbolVendor()" functions can be intercepted when the .o file contenst are later lazilly needed. Prior to this fix, when we first instantiated the "SymbolFileDWARFDebugMap" class, we would also make modules, object files and symbol files for every .o file in the debug map because we needed to fix up the sections in the .o files with information that is in the executable debug map. Now we lazily do this in the DebugMapModule::GetObjectFile() Cleaned up header includes a bit as well. llvm-svn: 162860
2012-08-29 21:13:06 +00:00
#include "lldb/Core/Section.h"
#include "lldb/Host/FileSystem.h"
#include "lldb/Utility/RangeMap.h"
#include "lldb/Utility/RegularExpression.h"
#include "lldb/Utility/Timer.h"
//#define DEBUG_OSO_DMAP // DO NOT CHECKIN WITH THIS NOT COMMENTED OUT
<rdar://problem/11757916> Make breakpoint setting by file and line much more efficient by only looking for inlined breakpoint locations if we are setting a breakpoint in anything but a source implementation file. Implementing this complex for a many reasons. Turns out that parsing compile units lazily had some issues with respect to how we need to do things with DWARF in .o files. So the fixes in the checkin for this makes these changes: - Add a new setting called "target.inline-breakpoint-strategy" which can be set to "never", "always", or "headers". "never" will never try and set any inlined breakpoints (fastest). "always" always looks for inlined breakpoint locations (slowest, but most accurate). "headers", which is the default setting, will only look for inlined breakpoint locations if the breakpoint is set in what are consudered to be header files, which is realy defined as "not in an implementation source file". - modify the breakpoint setting by file and line to check the current "target.inline-breakpoint-strategy" setting and act accordingly - Modify compile units to be able to get their language and other info lazily. This allows us to create compile units from the debug map and not have to fill all of the details in, and then lazily discover this information as we go on debuggging. This is needed to avoid parsing all .o files when setting breakpoints in implementation only files (no inlines). Otherwise we would need to parse the .o file, the object file (mach-o in our case) and the symbol file (DWARF in the object file) just to see what the compile unit was. - modify the "SymbolFileDWARFDebugMap" to subclass lldb_private::Module so that the virtual "GetObjectFile()" and "GetSymbolVendor()" functions can be intercepted when the .o file contenst are later lazilly needed. Prior to this fix, when we first instantiated the "SymbolFileDWARFDebugMap" class, we would also make modules, object files and symbol files for every .o file in the debug map because we needed to fix up the sections in the .o files with information that is in the executable debug map. Now we lazily do this in the DebugMapModule::GetObjectFile() Cleaned up header includes a bit as well. llvm-svn: 162860
2012-08-29 21:13:06 +00:00
#if defined(DEBUG_OSO_DMAP)
#include "lldb/Core/StreamFile.h"
<rdar://problem/11757916> Make breakpoint setting by file and line much more efficient by only looking for inlined breakpoint locations if we are setting a breakpoint in anything but a source implementation file. Implementing this complex for a many reasons. Turns out that parsing compile units lazily had some issues with respect to how we need to do things with DWARF in .o files. So the fixes in the checkin for this makes these changes: - Add a new setting called "target.inline-breakpoint-strategy" which can be set to "never", "always", or "headers". "never" will never try and set any inlined breakpoints (fastest). "always" always looks for inlined breakpoint locations (slowest, but most accurate). "headers", which is the default setting, will only look for inlined breakpoint locations if the breakpoint is set in what are consudered to be header files, which is realy defined as "not in an implementation source file". - modify the breakpoint setting by file and line to check the current "target.inline-breakpoint-strategy" setting and act accordingly - Modify compile units to be able to get their language and other info lazily. This allows us to create compile units from the debug map and not have to fill all of the details in, and then lazily discover this information as we go on debuggging. This is needed to avoid parsing all .o files when setting breakpoints in implementation only files (no inlines). Otherwise we would need to parse the .o file, the object file (mach-o in our case) and the symbol file (DWARF in the object file) just to see what the compile unit was. - modify the "SymbolFileDWARFDebugMap" to subclass lldb_private::Module so that the virtual "GetObjectFile()" and "GetSymbolVendor()" functions can be intercepted when the .o file contenst are later lazilly needed. Prior to this fix, when we first instantiated the "SymbolFileDWARFDebugMap" class, we would also make modules, object files and symbol files for every .o file in the debug map because we needed to fix up the sections in the .o files with information that is in the executable debug map. Now we lazily do this in the DebugMapModule::GetObjectFile() Cleaned up header includes a bit as well. llvm-svn: 162860
2012-08-29 21:13:06 +00:00
#endif
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 03:46:26 +00:00
<rdar://problem/11757916> Make breakpoint setting by file and line much more efficient by only looking for inlined breakpoint locations if we are setting a breakpoint in anything but a source implementation file. Implementing this complex for a many reasons. Turns out that parsing compile units lazily had some issues with respect to how we need to do things with DWARF in .o files. So the fixes in the checkin for this makes these changes: - Add a new setting called "target.inline-breakpoint-strategy" which can be set to "never", "always", or "headers". "never" will never try and set any inlined breakpoints (fastest). "always" always looks for inlined breakpoint locations (slowest, but most accurate). "headers", which is the default setting, will only look for inlined breakpoint locations if the breakpoint is set in what are consudered to be header files, which is realy defined as "not in an implementation source file". - modify the breakpoint setting by file and line to check the current "target.inline-breakpoint-strategy" setting and act accordingly - Modify compile units to be able to get their language and other info lazily. This allows us to create compile units from the debug map and not have to fill all of the details in, and then lazily discover this information as we go on debuggging. This is needed to avoid parsing all .o files when setting breakpoints in implementation only files (no inlines). Otherwise we would need to parse the .o file, the object file (mach-o in our case) and the symbol file (DWARF in the object file) just to see what the compile unit was. - modify the "SymbolFileDWARFDebugMap" to subclass lldb_private::Module so that the virtual "GetObjectFile()" and "GetSymbolVendor()" functions can be intercepted when the .o file contenst are later lazilly needed. Prior to this fix, when we first instantiated the "SymbolFileDWARFDebugMap" class, we would also make modules, object files and symbol files for every .o file in the debug map because we needed to fix up the sections in the .o files with information that is in the executable debug map. Now we lazily do this in the DebugMapModule::GetObjectFile() Cleaned up header includes a bit as well. llvm-svn: 162860
2012-08-29 21:13:06 +00:00
#include "lldb/Symbol/CompileUnit.h"
#include "lldb/Symbol/LineTable.h"
#include "lldb/Symbol/ObjectFile.h"
#include "lldb/Symbol/SymbolVendor.h"
#include "lldb/Symbol/TypeMap.h"
#include "lldb/Symbol/VariableList.h"
#include "llvm/Support/ScopedPrinter.h"
Added a mechanism for keeping track of where in the debug information individual Decls came from. We've had a metadata infrastructure for a while, which was intended to solve a problem we've since dealt with in a different way. (It was meant to keep track of which definition of an Objective-C class was the "true" definition, but we now find it by searching the symbols for the class symbol.) The metadata is attached to the ExternalASTSource, which means it has a one-to-one correspondence with AST contexts. I've repurposed the metadata infrastructure to hold the object file and DIE offset for the DWARF information corresponding to a Decl. There are methods in ClangASTContext that get and set this metadata, and the ClangASTImporter is capable of tracking down the metadata for Decls that have been copied out of the debug information into the parser's AST context without using any additional memory. To see the metadata, you just have to enable the expression log: - (lldb) log enable lldb expr - and watch the import messages. The high 32 bits of the metadata indicate the index of the object file in its containing DWARFDebugMap; I have also added a log which you can use to track that mapping: - (lldb) log enable dwarf map - This adds 64 bits per Decl, which in my testing hasn't turned out to be very much (debugging Clang produces around 6500 Decls in my tests). To track how much data is being consumed, I've also added a global variable g_TotalSizeOfMetadata which tracks the total number of Decls that have metadata in all active AST contexts. Right now this metadata is enormously useful for tracking down bugs in the debug info parser. In the future I also want to use this information to provide more intelligent error messages instead of printing empty source lines wherever Clang refers to the location where something is defined. llvm-svn: 154634
2012-04-13 00:10:03 +00:00
#include "LogChannelDWARF.h"
#include "SymbolFileDWARF.h"
#include <memory>
using namespace lldb;
using namespace lldb_private;
<rdar://problem/11757916> Make breakpoint setting by file and line much more efficient by only looking for inlined breakpoint locations if we are setting a breakpoint in anything but a source implementation file. Implementing this complex for a many reasons. Turns out that parsing compile units lazily had some issues with respect to how we need to do things with DWARF in .o files. So the fixes in the checkin for this makes these changes: - Add a new setting called "target.inline-breakpoint-strategy" which can be set to "never", "always", or "headers". "never" will never try and set any inlined breakpoints (fastest). "always" always looks for inlined breakpoint locations (slowest, but most accurate). "headers", which is the default setting, will only look for inlined breakpoint locations if the breakpoint is set in what are consudered to be header files, which is realy defined as "not in an implementation source file". - modify the breakpoint setting by file and line to check the current "target.inline-breakpoint-strategy" setting and act accordingly - Modify compile units to be able to get their language and other info lazily. This allows us to create compile units from the debug map and not have to fill all of the details in, and then lazily discover this information as we go on debuggging. This is needed to avoid parsing all .o files when setting breakpoints in implementation only files (no inlines). Otherwise we would need to parse the .o file, the object file (mach-o in our case) and the symbol file (DWARF in the object file) just to see what the compile unit was. - modify the "SymbolFileDWARFDebugMap" to subclass lldb_private::Module so that the virtual "GetObjectFile()" and "GetSymbolVendor()" functions can be intercepted when the .o file contenst are later lazilly needed. Prior to this fix, when we first instantiated the "SymbolFileDWARFDebugMap" class, we would also make modules, object files and symbol files for every .o file in the debug map because we needed to fix up the sections in the .o files with information that is in the executable debug map. Now we lazily do this in the DebugMapModule::GetObjectFile() Cleaned up header includes a bit as well. llvm-svn: 162860
2012-08-29 21:13:06 +00:00
// Subclass lldb_private::Module so we can intercept the
// "Module::GetObjectFile()" (so we can fixup the object file sections) and
// also for "Module::GetSymbolFile()" (so we can fixup the symbol file id.
<rdar://problem/11757916> Make breakpoint setting by file and line much more efficient by only looking for inlined breakpoint locations if we are setting a breakpoint in anything but a source implementation file. Implementing this complex for a many reasons. Turns out that parsing compile units lazily had some issues with respect to how we need to do things with DWARF in .o files. So the fixes in the checkin for this makes these changes: - Add a new setting called "target.inline-breakpoint-strategy" which can be set to "never", "always", or "headers". "never" will never try and set any inlined breakpoints (fastest). "always" always looks for inlined breakpoint locations (slowest, but most accurate). "headers", which is the default setting, will only look for inlined breakpoint locations if the breakpoint is set in what are consudered to be header files, which is realy defined as "not in an implementation source file". - modify the breakpoint setting by file and line to check the current "target.inline-breakpoint-strategy" setting and act accordingly - Modify compile units to be able to get their language and other info lazily. This allows us to create compile units from the debug map and not have to fill all of the details in, and then lazily discover this information as we go on debuggging. This is needed to avoid parsing all .o files when setting breakpoints in implementation only files (no inlines). Otherwise we would need to parse the .o file, the object file (mach-o in our case) and the symbol file (DWARF in the object file) just to see what the compile unit was. - modify the "SymbolFileDWARFDebugMap" to subclass lldb_private::Module so that the virtual "GetObjectFile()" and "GetSymbolVendor()" functions can be intercepted when the .o file contenst are later lazilly needed. Prior to this fix, when we first instantiated the "SymbolFileDWARFDebugMap" class, we would also make modules, object files and symbol files for every .o file in the debug map because we needed to fix up the sections in the .o files with information that is in the executable debug map. Now we lazily do this in the DebugMapModule::GetObjectFile() Cleaned up header includes a bit as well. llvm-svn: 162860
2012-08-29 21:13:06 +00:00
const SymbolFileDWARFDebugMap::FileRangeMap &
SymbolFileDWARFDebugMap::CompileUnitInfo::GetFileRangeMap(
SymbolFileDWARFDebugMap *exe_symfile) {
if (file_range_map_valid)
return file_range_map;
file_range_map_valid = true;
<rdar://problem/11757916> Make breakpoint setting by file and line much more efficient by only looking for inlined breakpoint locations if we are setting a breakpoint in anything but a source implementation file. Implementing this complex for a many reasons. Turns out that parsing compile units lazily had some issues with respect to how we need to do things with DWARF in .o files. So the fixes in the checkin for this makes these changes: - Add a new setting called "target.inline-breakpoint-strategy" which can be set to "never", "always", or "headers". "never" will never try and set any inlined breakpoints (fastest). "always" always looks for inlined breakpoint locations (slowest, but most accurate). "headers", which is the default setting, will only look for inlined breakpoint locations if the breakpoint is set in what are consudered to be header files, which is realy defined as "not in an implementation source file". - modify the breakpoint setting by file and line to check the current "target.inline-breakpoint-strategy" setting and act accordingly - Modify compile units to be able to get their language and other info lazily. This allows us to create compile units from the debug map and not have to fill all of the details in, and then lazily discover this information as we go on debuggging. This is needed to avoid parsing all .o files when setting breakpoints in implementation only files (no inlines). Otherwise we would need to parse the .o file, the object file (mach-o in our case) and the symbol file (DWARF in the object file) just to see what the compile unit was. - modify the "SymbolFileDWARFDebugMap" to subclass lldb_private::Module so that the virtual "GetObjectFile()" and "GetSymbolVendor()" functions can be intercepted when the .o file contenst are later lazilly needed. Prior to this fix, when we first instantiated the "SymbolFileDWARFDebugMap" class, we would also make modules, object files and symbol files for every .o file in the debug map because we needed to fix up the sections in the .o files with information that is in the executable debug map. Now we lazily do this in the DebugMapModule::GetObjectFile() Cleaned up header includes a bit as well. llvm-svn: 162860
2012-08-29 21:13:06 +00:00
Module *oso_module = exe_symfile->GetModuleByCompUnitInfo(this);
if (!oso_module)
return file_range_map;
<rdar://problem/11757916> Make breakpoint setting by file and line much more efficient by only looking for inlined breakpoint locations if we are setting a breakpoint in anything but a source implementation file. Implementing this complex for a many reasons. Turns out that parsing compile units lazily had some issues with respect to how we need to do things with DWARF in .o files. So the fixes in the checkin for this makes these changes: - Add a new setting called "target.inline-breakpoint-strategy" which can be set to "never", "always", or "headers". "never" will never try and set any inlined breakpoints (fastest). "always" always looks for inlined breakpoint locations (slowest, but most accurate). "headers", which is the default setting, will only look for inlined breakpoint locations if the breakpoint is set in what are consudered to be header files, which is realy defined as "not in an implementation source file". - modify the breakpoint setting by file and line to check the current "target.inline-breakpoint-strategy" setting and act accordingly - Modify compile units to be able to get their language and other info lazily. This allows us to create compile units from the debug map and not have to fill all of the details in, and then lazily discover this information as we go on debuggging. This is needed to avoid parsing all .o files when setting breakpoints in implementation only files (no inlines). Otherwise we would need to parse the .o file, the object file (mach-o in our case) and the symbol file (DWARF in the object file) just to see what the compile unit was. - modify the "SymbolFileDWARFDebugMap" to subclass lldb_private::Module so that the virtual "GetObjectFile()" and "GetSymbolVendor()" functions can be intercepted when the .o file contenst are later lazilly needed. Prior to this fix, when we first instantiated the "SymbolFileDWARFDebugMap" class, we would also make modules, object files and symbol files for every .o file in the debug map because we needed to fix up the sections in the .o files with information that is in the executable debug map. Now we lazily do this in the DebugMapModule::GetObjectFile() Cleaned up header includes a bit as well. llvm-svn: 162860
2012-08-29 21:13:06 +00:00
ObjectFile *oso_objfile = oso_module->GetObjectFile();
if (!oso_objfile)
<rdar://problem/11757916> Make breakpoint setting by file and line much more efficient by only looking for inlined breakpoint locations if we are setting a breakpoint in anything but a source implementation file. Implementing this complex for a many reasons. Turns out that parsing compile units lazily had some issues with respect to how we need to do things with DWARF in .o files. So the fixes in the checkin for this makes these changes: - Add a new setting called "target.inline-breakpoint-strategy" which can be set to "never", "always", or "headers". "never" will never try and set any inlined breakpoints (fastest). "always" always looks for inlined breakpoint locations (slowest, but most accurate). "headers", which is the default setting, will only look for inlined breakpoint locations if the breakpoint is set in what are consudered to be header files, which is realy defined as "not in an implementation source file". - modify the breakpoint setting by file and line to check the current "target.inline-breakpoint-strategy" setting and act accordingly - Modify compile units to be able to get their language and other info lazily. This allows us to create compile units from the debug map and not have to fill all of the details in, and then lazily discover this information as we go on debuggging. This is needed to avoid parsing all .o files when setting breakpoints in implementation only files (no inlines). Otherwise we would need to parse the .o file, the object file (mach-o in our case) and the symbol file (DWARF in the object file) just to see what the compile unit was. - modify the "SymbolFileDWARFDebugMap" to subclass lldb_private::Module so that the virtual "GetObjectFile()" and "GetSymbolVendor()" functions can be intercepted when the .o file contenst are later lazilly needed. Prior to this fix, when we first instantiated the "SymbolFileDWARFDebugMap" class, we would also make modules, object files and symbol files for every .o file in the debug map because we needed to fix up the sections in the .o files with information that is in the executable debug map. Now we lazily do this in the DebugMapModule::GetObjectFile() Cleaned up header includes a bit as well. llvm-svn: 162860
2012-08-29 21:13:06 +00:00
return file_range_map;
Log *log(LogChannelDWARF::GetLogIfAll(DWARF_LOG_DEBUG_MAP));
LLDB_LOGF(
log,
"%p: SymbolFileDWARFDebugMap::CompileUnitInfo::GetFileRangeMap ('%s')",
static_cast<void *>(this),
oso_module->GetSpecificationDescription().c_str());
std::vector<SymbolFileDWARFDebugMap::CompileUnitInfo *> cu_infos;
if (exe_symfile->GetCompUnitInfosForModule(oso_module, cu_infos)) {
for (auto comp_unit_info : cu_infos) {
Symtab *exe_symtab = exe_symfile->GetObjectFile()->GetSymtab();
ModuleSP oso_module_sp(oso_objfile->GetModule());
Symtab *oso_symtab = oso_objfile->GetSymtab();
/// const uint32_t fun_resolve_flags = SymbolContext::Module |
/// eSymbolContextCompUnit | eSymbolContextFunction;
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 03:46:26 +00:00
// SectionList *oso_sections = oso_objfile->Sections();
// Now we need to make sections that map from zero based object file
// addresses to where things ended up in the main executable.
A few of the issue I have been trying to track down and fix have been due to the way LLDB lazily gets complete definitions for types within the debug info. When we run across a class/struct/union definition in the DWARF, we will only parse the full definition if we need to. This works fine for top level types that are assigned directly to variables and arguments, but when we have a variable with a class, lets say "A" for this example, that has a member: "B *m_b". Initially we don't need to hunt down a definition for this class unless we are ever asked to do something with it ("expr m_b->getDecl()" for example). With my previous approach to lazy type completion, we would be able to take a "A *a" and get a complete type for it, but we wouldn't be able to then do an "a->m_b->getDecl()" unless we always expanded all types within a class prior to handing out the type. Expanding everything is very costly and it would be great if there were a better way. A few months ago I worked with the llvm/clang folks to have the ExternalASTSource class be able to complete classes if there weren't completed yet: class ExternalASTSource { .... virtual void CompleteType (clang::TagDecl *Tag); virtual void CompleteType (clang::ObjCInterfaceDecl *Class); }; This was great, because we can now have the class that is producing the AST (SymbolFileDWARF and SymbolFileDWARFDebugMap) sign up as external AST sources and the object that creates the forward declaration types can now also complete them anywhere within the clang type system. This patch makes a few major changes: - lldb_private::Module classes now own the AST context. Previously the TypeList objects did. - The DWARF parsers now sign up as an external AST sources so they can complete types. - All of the pure clang type system wrapper code we have in LLDB (ClangASTContext, ClangASTType, and more) can now be iterating through children of any type, and if a class/union/struct type (clang::RecordType or ObjC interface) is found that is incomplete, we can ask the AST to get the definition. - The SymbolFileDWARFDebugMap class now will create and use a single AST that all child SymbolFileDWARF classes will share (much like what happens when we have a complete linked DWARF for an executable). We will need to modify some of the ClangUserExpression code to take more advantage of this completion ability in the near future. Meanwhile we should be better off now that we can be accessing any children of variables through pointers and always be able to resolve the clang type if needed. llvm-svn: 123613
2011-01-17 03:46:26 +00:00
assert(comp_unit_info->first_symbol_index != UINT32_MAX);
// End index is one past the last valid symbol index
const uint32_t oso_end_idx = comp_unit_info->last_symbol_index + 1;
for (uint32_t idx = comp_unit_info->first_symbol_index +
2; // Skip the N_SO and N_OSO
idx < oso_end_idx; ++idx) {
Symbol *exe_symbol = exe_symtab->SymbolAtIndex(idx);
if (exe_symbol) {
if (!exe_symbol->IsDebug())
continue;
switch (exe_symbol->GetType()) {
default:
break;
case eSymbolTypeCode: {
// For each N_FUN, or function that we run into in the debug map we
// make a new section that we add to the sections found in the .o
// file. This new section has the file address set to what the
// addresses are in the .o file, and the load address is adjusted
// to match where it ended up in the final executable! We do this
// before we parse any dwarf info so that when it goes get parsed
// all section/offset addresses that get registered will resolve
// correctly to the new addresses in the main executable.
// First we find the original symbol in the .o file's symbol table
Symbol *oso_fun_symbol = oso_symtab->FindFirstSymbolWithNameAndType(
exe_symbol->GetMangled().GetName(lldb::eLanguageTypeUnknown,
Mangled::ePreferMangled),
eSymbolTypeCode, Symtab::eDebugNo, Symtab::eVisibilityAny);
if (oso_fun_symbol) {
// Add the inverse OSO file address to debug map entry mapping
exe_symfile->AddOSOFileRange(
this, exe_symbol->GetAddressRef().GetFileAddress(),
exe_symbol->GetByteSize(),
oso_fun_symbol->GetAddressRef().GetFileAddress(),
oso_fun_symbol->GetByteSize());
}
} break;
case eSymbolTypeData: {
// For each N_GSYM we remap the address for the global by making a
// new section that we add to the sections found in the .o file.
// This new section has the file address set to what the addresses
// are in the .o file, and the load address is adjusted to match
// where it ended up in the final executable! We do this before we
// parse any dwarf info so that when it goes get parsed all
// section/offset addresses that get registered will resolve
// correctly to the new addresses in the main executable. We
// initially set the section size to be 1 byte, but will need to
// fix up these addresses further after all globals have been
// parsed to span the gaps, or we can find the global variable
// sizes from the DWARF info as we are parsing.
// Next we find the non-stab entry that corresponds to the N_GSYM
// in the .o file
Symbol *oso_gsym_symbol =
oso_symtab->FindFirstSymbolWithNameAndType(
exe_symbol->GetMangled().GetName(lldb::eLanguageTypeUnknown,
Mangled::ePreferMangled),
eSymbolTypeData, Symtab::eDebugNo, Symtab::eVisibilityAny);
if (exe_symbol && oso_gsym_symbol && exe_symbol->ValueIsAddress() &&
oso_gsym_symbol->ValueIsAddress()) {
// Add the inverse OSO file address to debug map entry mapping
exe_symfile->AddOSOFileRange(
this, exe_symbol->GetAddressRef().GetFileAddress(),
exe_symbol->GetByteSize(),
oso_gsym_symbol->GetAddressRef().GetFileAddress(),
oso_gsym_symbol->GetByteSize());
}
} break;
}
}
}
exe_symfile->FinalizeOSOFileRanges(this);
// We don't need the symbols anymore for the .o files
oso_objfile->ClearSymtab();
}
}
return file_range_map;
}
class DebugMapModule : public Module {
public:
DebugMapModule(const ModuleSP &exe_module_sp, uint32_t cu_idx,
const FileSpec &file_spec, const ArchSpec &arch,
const ConstString *object_name, off_t object_offset,
const llvm::sys::TimePoint<> object_mod_time)
: Module(file_spec, arch, object_name, object_offset, object_mod_time),
m_exe_module_wp(exe_module_sp), m_cu_idx(cu_idx) {}
~DebugMapModule() override = default;
SymbolFile *
GetSymbolFile(bool can_create = true,
lldb_private::Stream *feedback_strm = nullptr) override {
<rdar://problem/11757916> Make breakpoint setting by file and line much more efficient by only looking for inlined breakpoint locations if we are setting a breakpoint in anything but a source implementation file. Implementing this complex for a many reasons. Turns out that parsing compile units lazily had some issues with respect to how we need to do things with DWARF in .o files. So the fixes in the checkin for this makes these changes: - Add a new setting called "target.inline-breakpoint-strategy" which can be set to "never", "always", or "headers". "never" will never try and set any inlined breakpoints (fastest). "always" always looks for inlined breakpoint locations (slowest, but most accurate). "headers", which is the default setting, will only look for inlined breakpoint locations if the breakpoint is set in what are consudered to be header files, which is realy defined as "not in an implementation source file". - modify the breakpoint setting by file and line to check the current "target.inline-breakpoint-strategy" setting and act accordingly - Modify compile units to be able to get their language and other info lazily. This allows us to create compile units from the debug map and not have to fill all of the details in, and then lazily discover this information as we go on debuggging. This is needed to avoid parsing all .o files when setting breakpoints in implementation only files (no inlines). Otherwise we would need to parse the .o file, the object file (mach-o in our case) and the symbol file (DWARF in the object file) just to see what the compile unit was. - modify the "SymbolFileDWARFDebugMap" to subclass lldb_private::Module so that the virtual "GetObjectFile()" and "GetSymbolVendor()" functions can be intercepted when the .o file contenst are later lazilly needed. Prior to this fix, when we first instantiated the "SymbolFileDWARFDebugMap" class, we would also make modules, object files and symbol files for every .o file in the debug map because we needed to fix up the sections in the .o files with information that is in the executable debug map. Now we lazily do this in the DebugMapModule::GetObjectFile() Cleaned up header includes a bit as well. llvm-svn: 162860
2012-08-29 21:13:06 +00:00
// Scope for locker
if (m_symfile_up.get() || !can_create)
return m_symfile_up ? m_symfile_up->GetSymbolFile() : nullptr;
ModuleSP exe_module_sp(m_exe_module_wp.lock());
<rdar://problem/11757916> Make breakpoint setting by file and line much more efficient by only looking for inlined breakpoint locations if we are setting a breakpoint in anything but a source implementation file. Implementing this complex for a many reasons. Turns out that parsing compile units lazily had some issues with respect to how we need to do things with DWARF in .o files. So the fixes in the checkin for this makes these changes: - Add a new setting called "target.inline-breakpoint-strategy" which can be set to "never", "always", or "headers". "never" will never try and set any inlined breakpoints (fastest). "always" always looks for inlined breakpoint locations (slowest, but most accurate). "headers", which is the default setting, will only look for inlined breakpoint locations if the breakpoint is set in what are consudered to be header files, which is realy defined as "not in an implementation source file". - modify the breakpoint setting by file and line to check the current "target.inline-breakpoint-strategy" setting and act accordingly - Modify compile units to be able to get their language and other info lazily. This allows us to create compile units from the debug map and not have to fill all of the details in, and then lazily discover this information as we go on debuggging. This is needed to avoid parsing all .o files when setting breakpoints in implementation only files (no inlines). Otherwise we would need to parse the .o file, the object file (mach-o in our case) and the symbol file (DWARF in the object file) just to see what the compile unit was. - modify the "SymbolFileDWARFDebugMap" to subclass lldb_private::Module so that the virtual "GetObjectFile()" and "GetSymbolVendor()" functions can be intercepted when the .o file contenst are later lazilly needed. Prior to this fix, when we first instantiated the "SymbolFileDWARFDebugMap" class, we would also make modules, object files and symbol files for every .o file in the debug map because we needed to fix up the sections in the .o files with information that is in the executable debug map. Now we lazily do this in the DebugMapModule::GetObjectFile() Cleaned up header includes a bit as well. llvm-svn: 162860
2012-08-29 21:13:06 +00:00
if (exe_module_sp) {
// Now get the object file outside of a locking scope
ObjectFile *oso_objfile = GetObjectFile();
if (oso_objfile) {
std::lock_guard<std::recursive_mutex> guard(m_mutex);
if (SymbolFile *symfile =
Module::GetSymbolFile(can_create, feedback_strm)) {
// Set a pointer to this class to set our OSO DWARF file know that
// the DWARF is being used along with a debug map and that it will
// have the remapped sections that we do below.
SymbolFileDWARF *oso_symfile =
SymbolFileDWARFDebugMap::GetSymbolFileAsSymbolFileDWARF(symfile);
if (!oso_symfile)
return nullptr;
ObjectFile *exe_objfile = exe_module_sp->GetObjectFile();
SymbolFile *exe_symfile = exe_module_sp->GetSymbolFile();
if (exe_objfile && exe_symfile) {
oso_symfile->SetDebugMapModule(exe_module_sp);
// Set the ID of the symbol file DWARF to the index of the OSO
// shifted left by 32 bits to provide a unique prefix for any
// UserID's that get created in the symbol file.
oso_symfile->SetID(((uint64_t)m_cu_idx + 1ull) << 32ull);
}
return symfile;
}
}
}
return nullptr;
}
protected:
ModuleWP m_exe_module_wp;
const uint32_t m_cu_idx;
};
void SymbolFileDWARFDebugMap::Initialize() {
PluginManager::RegisterPlugin(GetPluginNameStatic(),
GetPluginDescriptionStatic(), CreateInstance);
}
void SymbolFileDWARFDebugMap::Terminate() {
PluginManager::UnregisterPlugin(CreateInstance);
}
<rdar://problem/11757916> Make breakpoint setting by file and line much more efficient by only looking for inlined breakpoint locations if we are setting a breakpoint in anything but a source implementation file. Implementing this complex for a many reasons. Turns out that parsing compile units lazily had some issues with respect to how we need to do things with DWARF in .o files. So the fixes in the checkin for this makes these changes: - Add a new setting called "target.inline-breakpoint-strategy" which can be set to "never", "always", or "headers". "never" will never try and set any inlined breakpoints (fastest). "always" always looks for inlined breakpoint locations (slowest, but most accurate). "headers", which is the default setting, will only look for inlined breakpoint locations if the breakpoint is set in what are consudered to be header files, which is realy defined as "not in an implementation source file". - modify the breakpoint setting by file and line to check the current "target.inline-breakpoint-strategy" setting and act accordingly - Modify compile units to be able to get their language and other info lazily. This allows us to create compile units from the debug map and not have to fill all of the details in, and then lazily discover this information as we go on debuggging. This is needed to avoid parsing all .o files when setting breakpoints in implementation only files (no inlines). Otherwise we would need to parse the .o file, the object file (mach-o in our case) and the symbol file (DWARF in the object file) just to see what the compile unit was. - modify the "SymbolFileDWARFDebugMap" to subclass lldb_private::Module so that the virtual "GetObjectFile()" and "GetSymbolVendor()" functions can be intercepted when the .o file contenst are later lazilly needed. Prior to this fix, when we first instantiated the "SymbolFileDWARFDebugMap" class, we would also make modules, object files and symbol files for every .o file in the debug map because we needed to fix up the sections in the .o files with information that is in the executable debug map. Now we lazily do this in the DebugMapModule::GetObjectFile() Cleaned up header includes a bit as well. llvm-svn: 162860
2012-08-29 21:13:06 +00:00
lldb_private::ConstString SymbolFileDWARFDebugMap::GetPluginNameStatic() {
static ConstString g_name("dwarf-debugmap");
return g_name;
}
const char *SymbolFileDWARFDebugMap::GetPluginDescriptionStatic() {
return "DWARF and DWARF3 debug symbol file reader (debug map).";
}
SymbolFile *SymbolFileDWARFDebugMap::CreateInstance(ObjectFileSP objfile_sp) {
return new SymbolFileDWARFDebugMap(std::move(objfile_sp));
}
SymbolFileDWARFDebugMap::SymbolFileDWARFDebugMap(ObjectFileSP objfile_sp)
: SymbolFile(std::move(objfile_sp)), m_flags(), m_compile_unit_infos(),
m_func_indexes(), m_glob_indexes(),
m_supports_DW_AT_APPLE_objc_complete_type(eLazyBoolCalculate) {}
SymbolFileDWARFDebugMap::~SymbolFileDWARFDebugMap() {}
void SymbolFileDWARFDebugMap::InitializeObject() {}
void SymbolFileDWARFDebugMap::InitOSO() {
if (m_flags.test(kHaveInitializedOSOs))
return;
m_flags.set(kHaveInitializedOSOs);
// If the object file has been stripped, there is no sense in looking further
// as all of the debug symbols for the debug map will not be available
if (m_objfile_sp->IsStripped())
return;
// Also make sure the file type is some sort of executable. Core files, debug
// info files (dSYM), object files (.o files), and stub libraries all can
switch (m_objfile_sp->GetType()) {
case ObjectFile::eTypeInvalid:
case ObjectFile::eTypeCoreFile:
case ObjectFile::eTypeDebugInfo:
case ObjectFile::eTypeObjectFile:
case ObjectFile::eTypeStubLibrary:
case ObjectFile::eTypeUnknown:
case ObjectFile::eTypeJIT:
return;
case ObjectFile::eTypeExecutable:
case ObjectFile::eTypeDynamicLinker:
case ObjectFile::eTypeSharedLibrary:
break;
}
// In order to get the abilities of this plug-in, we look at the list of
// N_OSO entries (object files) from the symbol table and make sure that
// these files exist and also contain valid DWARF. If we get any of that then
// we return the abilities of the first N_OSO's DWARF.
Symtab *symtab = m_objfile_sp->GetSymtab();
if (symtab) {
Log *log(LogChannelDWARF::GetLogIfAll(DWARF_LOG_DEBUG_MAP));
std::vector<uint32_t> oso_indexes;
// When a mach-o symbol is encoded, the n_type field is encoded in bits
// 23:16, and the n_desc field is encoded in bits 15:0.
//
// To find all N_OSO entries that are part of the DWARF + debug map we find
// only object file symbols with the flags value as follows: bits 23:16 ==
// 0x66 (N_OSO) bits 15: 0 == 0x0001 (specifies this is a debug map object
// file)
const uint32_t k_oso_symbol_flags_value = 0x660001u;
const uint32_t oso_index_count =
symtab->AppendSymbolIndexesWithTypeAndFlagsValue(
eSymbolTypeObjectFile, k_oso_symbol_flags_value, oso_indexes);
if (oso_index_count > 0) {
symtab->AppendSymbolIndexesWithType(eSymbolTypeCode, Symtab::eDebugYes,
Symtab::eVisibilityAny,
m_func_indexes);
symtab->AppendSymbolIndexesWithType(eSymbolTypeData, Symtab::eDebugYes,
Symtab::eVisibilityAny,
m_glob_indexes);
symtab->SortSymbolIndexesByValue(m_func_indexes, true);
symtab->SortSymbolIndexesByValue(m_glob_indexes, true);
for (uint32_t sym_idx : m_func_indexes) {
const Symbol *symbol = symtab->SymbolAtIndex(sym_idx);
lldb::addr_t file_addr = symbol->GetAddressRef().GetFileAddress();
lldb::addr_t byte_size = symbol->GetByteSize();
DebugMap::Entry debug_map_entry(
file_addr, byte_size, OSOEntry(sym_idx, LLDB_INVALID_ADDRESS));
m_debug_map.Append(debug_map_entry);
}
for (uint32_t sym_idx : m_glob_indexes) {
const Symbol *symbol = symtab->SymbolAtIndex(sym_idx);
lldb::addr_t file_addr = symbol->GetAddressRef().GetFileAddress();
lldb::addr_t byte_size = symbol->GetByteSize();
DebugMap::Entry debug_map_entry(
file_addr, byte_size, OSOEntry(sym_idx, LLDB_INVALID_ADDRESS));
m_debug_map.Append(debug_map_entry);
}
m_debug_map.Sort();
m_compile_unit_infos.resize(oso_index_count);
for (uint32_t i = 0; i < oso_index_count; ++i) {
const uint32_t so_idx = oso_indexes[i] - 1;
const uint32_t oso_idx = oso_indexes[i];
const Symbol *so_symbol = symtab->SymbolAtIndex(so_idx);
const Symbol *oso_symbol = symtab->SymbolAtIndex(oso_idx);
if (so_symbol && oso_symbol &&
so_symbol->GetType() == eSymbolTypeSourceFile &&
oso_symbol->GetType() == eSymbolTypeObjectFile) {
<rdar://problem/11757916> Make breakpoint setting by file and line much more efficient by only looking for inlined breakpoint locations if we are setting a breakpoint in anything but a source implementation file. Implementing this complex for a many reasons. Turns out that parsing compile units lazily had some issues with respect to how we need to do things with DWARF in .o files. So the fixes in the checkin for this makes these changes: - Add a new setting called "target.inline-breakpoint-strategy" which can be set to "never", "always", or "headers". "never" will never try and set any inlined breakpoints (fastest). "always" always looks for inlined breakpoint locations (slowest, but most accurate). "headers", which is the default setting, will only look for inlined breakpoint locations if the breakpoint is set in what are consudered to be header files, which is realy defined as "not in an implementation source file". - modify the breakpoint setting by file and line to check the current "target.inline-breakpoint-strategy" setting and act accordingly - Modify compile units to be able to get their language and other info lazily. This allows us to create compile units from the debug map and not have to fill all of the details in, and then lazily discover this information as we go on debuggging. This is needed to avoid parsing all .o files when setting breakpoints in implementation only files (no inlines). Otherwise we would need to parse the .o file, the object file (mach-o in our case) and the symbol file (DWARF in the object file) just to see what the compile unit was. - modify the "SymbolFileDWARFDebugMap" to subclass lldb_private::Module so that the virtual "GetObjectFile()" and "GetSymbolVendor()" functions can be intercepted when the .o file contenst are later lazilly needed. Prior to this fix, when we first instantiated the "SymbolFileDWARFDebugMap" class, we would also make modules, object files and symbol files for every .o file in the debug map because we needed to fix up the sections in the .o files with information that is in the executable debug map. Now we lazily do this in the DebugMapModule::GetObjectFile() Cleaned up header includes a bit as well. llvm-svn: 162860
2012-08-29 21:13:06 +00:00
m_compile_unit_infos[i].so_file.SetFile(
so_symbol->GetName().AsCString(), FileSpec::Style::native);
m_compile_unit_infos[i].oso_path = oso_symbol->GetName();
m_compile_unit_infos[i].oso_mod_time =
llvm::sys::toTimePoint(oso_symbol->GetIntegerValue(0));
uint32_t sibling_idx = so_symbol->GetSiblingIndex();
// The sibling index can't be less that or equal to the current index
// "i"
if (sibling_idx == UINT32_MAX) {
m_objfile_sp->GetModule()->ReportError(
"N_SO in symbol with UID %u has invalid sibling in debug map, "
"please file a bug and attach the binary listed in this error",
so_symbol->GetID());
} else {
const Symbol *last_symbol = symtab->SymbolAtIndex(sibling_idx - 1);
m_compile_unit_infos[i].first_symbol_index = so_idx;
m_compile_unit_infos[i].last_symbol_index = sibling_idx - 1;
m_compile_unit_infos[i].first_symbol_id = so_symbol->GetID();
m_compile_unit_infos[i].last_symbol_id = last_symbol->GetID();
LLDB_LOGF(log, "Initialized OSO 0x%8.8x: file=%s", i,
oso_symbol->GetName().GetCString());
}
} else {
if (oso_symbol == nullptr)
m_objfile_sp->GetModule()->ReportError(
"N_OSO symbol[%u] can't be found, please file a bug and attach "
"the binary listed in this error",
oso_idx);
else if (so_symbol == nullptr)
m_objfile_sp->GetModule()->ReportError(
"N_SO not found for N_OSO symbol[%u], please file a bug and "
"attach the binary listed in this error",
oso_idx);
else if (so_symbol->GetType() != eSymbolTypeSourceFile)
m_objfile_sp->GetModule()->ReportError(
"N_SO has incorrect symbol type (%u) for N_OSO symbol[%u], "
"please file a bug and attach the binary listed in this error",
so_symbol->GetType(), oso_idx);
else if (oso_symbol->GetType() != eSymbolTypeSourceFile)
m_objfile_sp->GetModule()->ReportError(
"N_OSO has incorrect symbol type (%u) for N_OSO symbol[%u], "
"please file a bug and attach the binary listed in this error",
oso_symbol->GetType(), oso_idx);
}
}
}
}
}
Module *SymbolFileDWARFDebugMap::GetModuleByOSOIndex(uint32_t oso_idx) {
const uint32_t cu_count = GetNumCompileUnits();
if (oso_idx < cu_count)
return GetModuleByCompUnitInfo(&m_compile_unit_infos[oso_idx]);
return nullptr;
}
Module *SymbolFileDWARFDebugMap::GetModuleByCompUnitInfo(
CompileUnitInfo *comp_unit_info) {
if (!comp_unit_info->oso_sp) {
auto pos = m_oso_map.find(
{comp_unit_info->oso_path, comp_unit_info->oso_mod_time});
if (pos != m_oso_map.end()) {
comp_unit_info->oso_sp = pos->second;
} else {
ObjectFile *obj_file = GetObjectFile();
comp_unit_info->oso_sp = std::make_shared<OSOInfo>();
m_oso_map[{comp_unit_info->oso_path, comp_unit_info->oso_mod_time}] =
comp_unit_info->oso_sp;
const char *oso_path = comp_unit_info->oso_path.GetCString();
FileSpec oso_file(oso_path);
ConstString oso_object;
if (FileSystem::Instance().Exists(oso_file)) {
// The modification time returned by the FS can have a higher precision
// than the one from the CU.
auto oso_mod_time = std::chrono::time_point_cast<std::chrono::seconds>(
FileSystem::Instance().GetModificationTime(oso_file));
// A timestamp of 0 means that the linker was in deterministic mode. In
// that case, we should skip the check against the filesystem last
// modification timestamp, since it will never match.
if (comp_unit_info->oso_mod_time != llvm::sys::TimePoint<>() &&
oso_mod_time != comp_unit_info->oso_mod_time) {
obj_file->GetModule()->ReportError(
"debug map object file '%s' has changed (actual time is "
"%s, debug map time is %s"
") since this executable was linked, file will be ignored",
oso_file.GetPath().c_str(), llvm::to_string(oso_mod_time).c_str(),
llvm::to_string(comp_unit_info->oso_mod_time).c_str());
return nullptr;
}
} else {
const bool must_exist = true;
if (!ObjectFile::SplitArchivePathWithObject(oso_path, oso_file,
oso_object, must_exist)) {
return nullptr;
}
}
// Always create a new module for .o files. Why? Because we use the debug
// map, to add new sections to each .o file and even though a .o file
// might not have changed, the sections that get added to the .o file can
// change.
ArchSpec oso_arch;
// Only adopt the architecture from the module (not the vendor or OS)
// since .o files for "i386-apple-ios" will historically show up as "i386
// -apple-macosx" due to the lack of a LC_VERSION_MIN_MACOSX or
// LC_VERSION_MIN_IPHONEOS load command...
oso_arch.SetTriple(m_objfile_sp->GetModule()
->GetArchitecture()
.GetTriple()
.getArchName()
.str()
.c_str());
comp_unit_info->oso_sp->module_sp = std::make_shared<DebugMapModule>(
obj_file->GetModule(), GetCompUnitInfoIndex(comp_unit_info), oso_file,
oso_arch, oso_object ? &oso_object : nullptr, 0,
oso_object ? comp_unit_info->oso_mod_time : llvm::sys::TimePoint<>());
}
}
if (comp_unit_info->oso_sp)
return comp_unit_info->oso_sp->module_sp.get();
return nullptr;
}
bool SymbolFileDWARFDebugMap::GetFileSpecForSO(uint32_t oso_idx,
FileSpec &file_spec) {
if (oso_idx < m_compile_unit_infos.size()) {
<rdar://problem/11757916> Make breakpoint setting by file and line much more efficient by only looking for inlined breakpoint locations if we are setting a breakpoint in anything but a source implementation file. Implementing this complex for a many reasons. Turns out that parsing compile units lazily had some issues with respect to how we need to do things with DWARF in .o files. So the fixes in the checkin for this makes these changes: - Add a new setting called "target.inline-breakpoint-strategy" which can be set to "never", "always", or "headers". "never" will never try and set any inlined breakpoints (fastest). "always" always looks for inlined breakpoint locations (slowest, but most accurate). "headers", which is the default setting, will only look for inlined breakpoint locations if the breakpoint is set in what are consudered to be header files, which is realy defined as "not in an implementation source file". - modify the breakpoint setting by file and line to check the current "target.inline-breakpoint-strategy" setting and act accordingly - Modify compile units to be able to get their language and other info lazily. This allows us to create compile units from the debug map and not have to fill all of the details in, and then lazily discover this information as we go on debuggging. This is needed to avoid parsing all .o files when setting breakpoints in implementation only files (no inlines). Otherwise we would need to parse the .o file, the object file (mach-o in our case) and the symbol file (DWARF in the object file) just to see what the compile unit was. - modify the "SymbolFileDWARFDebugMap" to subclass lldb_private::Module so that the virtual "GetObjectFile()" and "GetSymbolVendor()" functions can be intercepted when the .o file contenst are later lazilly needed. Prior to this fix, when we first instantiated the "SymbolFileDWARFDebugMap" class, we would also make modules, object files and symbol files for every .o file in the debug map because we needed to fix up the sections in the .o files with information that is in the executable debug map. Now we lazily do this in the DebugMapModule::GetObjectFile() Cleaned up header includes a bit as well. llvm-svn: 162860
2012-08-29 21:13:06 +00:00
if (m_compile_unit_infos[oso_idx].so_file) {
file_spec = m_compile_unit_infos[oso_idx].so_file;
return true;
}
}
return false;
}
ObjectFile *SymbolFileDWARFDebugMap::GetObjectFileByOSOIndex(uint32_t oso_idx) {
Module *oso_module = GetModuleByOSOIndex(oso_idx);
if (oso_module)
return oso_module->GetObjectFile();
return nullptr;
}
SymbolFileDWARF *
SymbolFileDWARFDebugMap::GetSymbolFile(const SymbolContext &sc) {
return GetSymbolFile(*sc.comp_unit);
}
SymbolFileDWARF *
SymbolFileDWARFDebugMap::GetSymbolFile(const CompileUnit &comp_unit) {
CompileUnitInfo *comp_unit_info = GetCompUnitInfo(comp_unit);
if (comp_unit_info)
return GetSymbolFileByCompUnitInfo(comp_unit_info);
return nullptr;
}
ObjectFile *SymbolFileDWARFDebugMap::GetObjectFileByCompUnitInfo(
CompileUnitInfo *comp_unit_info) {
Module *oso_module = GetModuleByCompUnitInfo(comp_unit_info);
if (oso_module)
return oso_module->GetObjectFile();
return nullptr;
}
uint32_t SymbolFileDWARFDebugMap::GetCompUnitInfoIndex(
const CompileUnitInfo *comp_unit_info) {
if (!m_compile_unit_infos.empty()) {
const CompileUnitInfo *first_comp_unit_info = &m_compile_unit_infos.front();
const CompileUnitInfo *last_comp_unit_info = &m_compile_unit_infos.back();
if (first_comp_unit_info <= comp_unit_info &&
comp_unit_info <= last_comp_unit_info)
return comp_unit_info - first_comp_unit_info;
}
return UINT32_MAX;
}
SymbolFileDWARF *
SymbolFileDWARFDebugMap::GetSymbolFileByOSOIndex(uint32_t oso_idx) {
unsigned size = m_compile_unit_infos.size();
if (oso_idx < size)
return GetSymbolFileByCompUnitInfo(&m_compile_unit_infos[oso_idx]);
return nullptr;
}
SymbolFileDWARF *
SymbolFileDWARFDebugMap::GetSymbolFileAsSymbolFileDWARF(SymbolFile *sym_file) {
if (sym_file &&
sym_file->GetPluginName() == SymbolFileDWARF::GetPluginNameStatic())
return (SymbolFileDWARF *)sym_file;
return nullptr;
}
SymbolFileDWARF *SymbolFileDWARFDebugMap::GetSymbolFileByCompUnitInfo(
CompileUnitInfo *comp_unit_info) {
if (Module *oso_module = GetModuleByCompUnitInfo(comp_unit_info))
return GetSymbolFileAsSymbolFileDWARF(oso_module->GetSymbolFile());
return nullptr;
}
uint32_t SymbolFileDWARFDebugMap::CalculateAbilities() {
// In order to get the abilities of this plug-in, we look at the list of
// N_OSO entries (object files) from the symbol table and make sure that
// these files exist and also contain valid DWARF. If we get any of that then
// we return the abilities of the first N_OSO's DWARF.
const uint32_t oso_index_count = GetNumCompileUnits();
if (oso_index_count > 0) {
<rdar://problem/11757916> Make breakpoint setting by file and line much more efficient by only looking for inlined breakpoint locations if we are setting a breakpoint in anything but a source implementation file. Implementing this complex for a many reasons. Turns out that parsing compile units lazily had some issues with respect to how we need to do things with DWARF in .o files. So the fixes in the checkin for this makes these changes: - Add a new setting called "target.inline-breakpoint-strategy" which can be set to "never", "always", or "headers". "never" will never try and set any inlined breakpoints (fastest). "always" always looks for inlined breakpoint locations (slowest, but most accurate). "headers", which is the default setting, will only look for inlined breakpoint locations if the breakpoint is set in what are consudered to be header files, which is realy defined as "not in an implementation source file". - modify the breakpoint setting by file and line to check the current "target.inline-breakpoint-strategy" setting and act accordingly - Modify compile units to be able to get their language and other info lazily. This allows us to create compile units from the debug map and not have to fill all of the details in, and then lazily discover this information as we go on debuggging. This is needed to avoid parsing all .o files when setting breakpoints in implementation only files (no inlines). Otherwise we would need to parse the .o file, the object file (mach-o in our case) and the symbol file (DWARF in the object file) just to see what the compile unit was. - modify the "SymbolFileDWARFDebugMap" to subclass lldb_private::Module so that the virtual "GetObjectFile()" and "GetSymbolVendor()" functions can be intercepted when the .o file contenst are later lazilly needed. Prior to this fix, when we first instantiated the "SymbolFileDWARFDebugMap" class, we would also make modules, object files and symbol files for every .o file in the debug map because we needed to fix up the sections in the .o files with information that is in the executable debug map. Now we lazily do this in the DebugMapModule::GetObjectFile() Cleaned up header includes a bit as well. llvm-svn: 162860
2012-08-29 21:13:06 +00:00
InitOSO();
if (!m_compile_unit_infos.empty()) {
return SymbolFile::CompileUnits | SymbolFile::Functions |
SymbolFile::Blocks | SymbolFile::GlobalVariables |
SymbolFile::LocalVariables | SymbolFile::VariableTypes |
SymbolFile::LineTables;
}
}
return 0;
}
2019-07-23 09:24:02 +00:00
uint32_t SymbolFileDWARFDebugMap::CalculateNumCompileUnits() {
InitOSO();
return m_compile_unit_infos.size();
}
CompUnitSP SymbolFileDWARFDebugMap::ParseCompileUnitAtIndex(uint32_t cu_idx) {
<rdar://problem/11757916> Make breakpoint setting by file and line much more efficient by only looking for inlined breakpoint locations if we are setting a breakpoint in anything but a source implementation file. Implementing this complex for a many reasons. Turns out that parsing compile units lazily had some issues with respect to how we need to do things with DWARF in .o files. So the fixes in the checkin for this makes these changes: - Add a new setting called "target.inline-breakpoint-strategy" which can be set to "never", "always", or "headers". "never" will never try and set any inlined breakpoints (fastest). "always" always looks for inlined breakpoint locations (slowest, but most accurate). "headers", which is the default setting, will only look for inlined breakpoint locations if the breakpoint is set in what are consudered to be header files, which is realy defined as "not in an implementation source file". - modify the breakpoint setting by file and line to check the current "target.inline-breakpoint-strategy" setting and act accordingly - Modify compile units to be able to get their language and other info lazily. This allows us to create compile units from the debug map and not have to fill all of the details in, and then lazily discover this information as we go on debuggging. This is needed to avoid parsing all .o files when setting breakpoints in implementation only files (no inlines). Otherwise we would need to parse the .o file, the object file (mach-o in our case) and the symbol file (DWARF in the object file) just to see what the compile unit was. - modify the "SymbolFileDWARFDebugMap" to subclass lldb_private::Module so that the virtual "GetObjectFile()" and "GetSymbolVendor()" functions can be intercepted when the .o file contenst are later lazilly needed. Prior to this fix, when we first instantiated the "SymbolFileDWARFDebugMap" class, we would also make modules, object files and symbol files for every .o file in the debug map because we needed to fix up the sections in the .o files with information that is in the executable debug map. Now we lazily do this in the DebugMapModule::GetObjectFile() Cleaned up header includes a bit as well. llvm-svn: 162860
2012-08-29 21:13:06 +00:00
CompUnitSP comp_unit_sp;
const uint32_t cu_count = GetNumCompileUnits();
if (cu_idx < cu_count) {
Module *oso_module = GetModuleByCompUnitInfo(&m_compile_unit_infos[cu_idx]);
if (oso_module) {
FileSpec so_file_spec;
if (GetFileSpecForSO(cu_idx, so_file_spec)) {
// User zero as the ID to match the compile unit at offset zero in each
// .o file since each .o file can only have one compile unit for now.
lldb::user_id_t cu_id = 0;
m_compile_unit_infos[cu_idx].compile_unit_sp =
std::make_shared<CompileUnit>(
m_objfile_sp->GetModule(), nullptr, so_file_spec, cu_id,
eLanguageTypeUnknown, eLazyBoolCalculate);
if (m_compile_unit_infos[cu_idx].compile_unit_sp) {
2019-07-23 09:24:02 +00:00
SetCompileUnitAtIndex(cu_idx,
m_compile_unit_infos[cu_idx].compile_unit_sp);
}
}
}
comp_unit_sp = m_compile_unit_infos[cu_idx].compile_unit_sp;
}
return comp_unit_sp;
}
SymbolFileDWARFDebugMap::CompileUnitInfo *
SymbolFileDWARFDebugMap::GetCompUnitInfo(const SymbolContext &sc) {
return GetCompUnitInfo(*sc.comp_unit);
}
SymbolFileDWARFDebugMap::CompileUnitInfo *
SymbolFileDWARFDebugMap::GetCompUnitInfo(const CompileUnit &comp_unit) {
const uint32_t cu_count = GetNumCompileUnits();
for (uint32_t i = 0; i < cu_count; ++i) {
if (comp_unit == m_compile_unit_infos[i].compile_unit_sp.get())
return &m_compile_unit_infos[i];
}
return nullptr;
}
size_t SymbolFileDWARFDebugMap::GetCompUnitInfosForModule(
const lldb_private::Module *module,
std::vector<CompileUnitInfo *> &cu_infos) {
<rdar://problem/11757916> Make breakpoint setting by file and line much more efficient by only looking for inlined breakpoint locations if we are setting a breakpoint in anything but a source implementation file. Implementing this complex for a many reasons. Turns out that parsing compile units lazily had some issues with respect to how we need to do things with DWARF in .o files. So the fixes in the checkin for this makes these changes: - Add a new setting called "target.inline-breakpoint-strategy" which can be set to "never", "always", or "headers". "never" will never try and set any inlined breakpoints (fastest). "always" always looks for inlined breakpoint locations (slowest, but most accurate). "headers", which is the default setting, will only look for inlined breakpoint locations if the breakpoint is set in what are consudered to be header files, which is realy defined as "not in an implementation source file". - modify the breakpoint setting by file and line to check the current "target.inline-breakpoint-strategy" setting and act accordingly - Modify compile units to be able to get their language and other info lazily. This allows us to create compile units from the debug map and not have to fill all of the details in, and then lazily discover this information as we go on debuggging. This is needed to avoid parsing all .o files when setting breakpoints in implementation only files (no inlines). Otherwise we would need to parse the .o file, the object file (mach-o in our case) and the symbol file (DWARF in the object file) just to see what the compile unit was. - modify the "SymbolFileDWARFDebugMap" to subclass lldb_private::Module so that the virtual "GetObjectFile()" and "GetSymbolVendor()" functions can be intercepted when the .o file contenst are later lazilly needed. Prior to this fix, when we first instantiated the "SymbolFileDWARFDebugMap" class, we would also make modules, object files and symbol files for every .o file in the debug map because we needed to fix up the sections in the .o files with information that is in the executable debug map. Now we lazily do this in the DebugMapModule::GetObjectFile() Cleaned up header includes a bit as well. llvm-svn: 162860
2012-08-29 21:13:06 +00:00
const uint32_t cu_count = GetNumCompileUnits();
for (uint32_t i = 0; i < cu_count; ++i) {
if (module == GetModuleByCompUnitInfo(&m_compile_unit_infos[i]))
cu_infos.push_back(&m_compile_unit_infos[i]);
<rdar://problem/11757916> Make breakpoint setting by file and line much more efficient by only looking for inlined breakpoint locations if we are setting a breakpoint in anything but a source implementation file. Implementing this complex for a many reasons. Turns out that parsing compile units lazily had some issues with respect to how we need to do things with DWARF in .o files. So the fixes in the checkin for this makes these changes: - Add a new setting called "target.inline-breakpoint-strategy" which can be set to "never", "always", or "headers". "never" will never try and set any inlined breakpoints (fastest). "always" always looks for inlined breakpoint locations (slowest, but most accurate). "headers", which is the default setting, will only look for inlined breakpoint locations if the breakpoint is set in what are consudered to be header files, which is realy defined as "not in an implementation source file". - modify the breakpoint setting by file and line to check the current "target.inline-breakpoint-strategy" setting and act accordingly - Modify compile units to be able to get their language and other info lazily. This allows us to create compile units from the debug map and not have to fill all of the details in, and then lazily discover this information as we go on debuggging. This is needed to avoid parsing all .o files when setting breakpoints in implementation only files (no inlines). Otherwise we would need to parse the .o file, the object file (mach-o in our case) and the symbol file (DWARF in the object file) just to see what the compile unit was. - modify the "SymbolFileDWARFDebugMap" to subclass lldb_private::Module so that the virtual "GetObjectFile()" and "GetSymbolVendor()" functions can be intercepted when the .o file contenst are later lazilly needed. Prior to this fix, when we first instantiated the "SymbolFileDWARFDebugMap" class, we would also make modules, object files and symbol files for every .o file in the debug map because we needed to fix up the sections in the .o files with information that is in the executable debug map. Now we lazily do this in the DebugMapModule::GetObjectFile() Cleaned up header includes a bit as well. llvm-svn: 162860
2012-08-29 21:13:06 +00:00
}
return cu_infos.size();
<rdar://problem/11757916> Make breakpoint setting by file and line much more efficient by only looking for inlined breakpoint locations if we are setting a breakpoint in anything but a source implementation file. Implementing this complex for a many reasons. Turns out that parsing compile units lazily had some issues with respect to how we need to do things with DWARF in .o files. So the fixes in the checkin for this makes these changes: - Add a new setting called "target.inline-breakpoint-strategy" which can be set to "never", "always", or "headers". "never" will never try and set any inlined breakpoints (fastest). "always" always looks for inlined breakpoint locations (slowest, but most accurate). "headers", which is the default setting, will only look for inlined breakpoint locations if the breakpoint is set in what are consudered to be header files, which is realy defined as "not in an implementation source file". - modify the breakpoint setting by file and line to check the current "target.inline-breakpoint-strategy" setting and act accordingly - Modify compile units to be able to get their language and other info lazily. This allows us to create compile units from the debug map and not have to fill all of the details in, and then lazily discover this information as we go on debuggging. This is needed to avoid parsing all .o files when setting breakpoints in implementation only files (no inlines). Otherwise we would need to parse the .o file, the object file (mach-o in our case) and the symbol file (DWARF in the object file) just to see what the compile unit was. - modify the "SymbolFileDWARFDebugMap" to subclass lldb_private::Module so that the virtual "GetObjectFile()" and "GetSymbolVendor()" functions can be intercepted when the .o file contenst are later lazilly needed. Prior to this fix, when we first instantiated the "SymbolFileDWARFDebugMap" class, we would also make modules, object files and symbol files for every .o file in the debug map because we needed to fix up the sections in the .o files with information that is in the executable debug map. Now we lazily do this in the DebugMapModule::GetObjectFile() Cleaned up header includes a bit as well. llvm-svn: 162860
2012-08-29 21:13:06 +00:00
}
lldb::LanguageType
SymbolFileDWARFDebugMap::ParseLanguage(CompileUnit &comp_unit) {
std::lock_guard<std::recursive_mutex> guard(GetModuleMutex());
SymbolFileDWARF *oso_dwarf = GetSymbolFile(comp_unit);
<rdar://problem/11757916> Make breakpoint setting by file and line much more efficient by only looking for inlined breakpoint locations if we are setting a breakpoint in anything but a source implementation file. Implementing this complex for a many reasons. Turns out that parsing compile units lazily had some issues with respect to how we need to do things with DWARF in .o files. So the fixes in the checkin for this makes these changes: - Add a new setting called "target.inline-breakpoint-strategy" which can be set to "never", "always", or "headers". "never" will never try and set any inlined breakpoints (fastest). "always" always looks for inlined breakpoint locations (slowest, but most accurate). "headers", which is the default setting, will only look for inlined breakpoint locations if the breakpoint is set in what are consudered to be header files, which is realy defined as "not in an implementation source file". - modify the breakpoint setting by file and line to check the current "target.inline-breakpoint-strategy" setting and act accordingly - Modify compile units to be able to get their language and other info lazily. This allows us to create compile units from the debug map and not have to fill all of the details in, and then lazily discover this information as we go on debuggging. This is needed to avoid parsing all .o files when setting breakpoints in implementation only files (no inlines). Otherwise we would need to parse the .o file, the object file (mach-o in our case) and the symbol file (DWARF in the object file) just to see what the compile unit was. - modify the "SymbolFileDWARFDebugMap" to subclass lldb_private::Module so that the virtual "GetObjectFile()" and "GetSymbolVendor()" functions can be intercepted when the .o file contenst are later lazilly needed. Prior to this fix, when we first instantiated the "SymbolFileDWARFDebugMap" class, we would also make modules, object files and symbol files for every .o file in the debug map because we needed to fix up the sections in the .o files with information that is in the executable debug map. Now we lazily do this in the DebugMapModule::GetObjectFile() Cleaned up header includes a bit as well. llvm-svn: 162860
2012-08-29 21:13:06 +00:00
if (oso_dwarf)
return oso_dwarf->ParseLanguage(comp_unit);
<rdar://problem/11757916> Make breakpoint setting by file and line much more efficient by only looking for inlined breakpoint locations if we are setting a breakpoint in anything but a source implementation file. Implementing this complex for a many reasons. Turns out that parsing compile units lazily had some issues with respect to how we need to do things with DWARF in .o files. So the fixes in the checkin for this makes these changes: - Add a new setting called "target.inline-breakpoint-strategy" which can be set to "never", "always", or "headers". "never" will never try and set any inlined breakpoints (fastest). "always" always looks for inlined breakpoint locations (slowest, but most accurate). "headers", which is the default setting, will only look for inlined breakpoint locations if the breakpoint is set in what are consudered to be header files, which is realy defined as "not in an implementation source file". - modify the breakpoint setting by file and line to check the current "target.inline-breakpoint-strategy" setting and act accordingly - Modify compile units to be able to get their language and other info lazily. This allows us to create compile units from the debug map and not have to fill all of the details in, and then lazily discover this information as we go on debuggging. This is needed to avoid parsing all .o files when setting breakpoints in implementation only files (no inlines). Otherwise we would need to parse the .o file, the object file (mach-o in our case) and the symbol file (DWARF in the object file) just to see what the compile unit was. - modify the "SymbolFileDWARFDebugMap" to subclass lldb_private::Module so that the virtual "GetObjectFile()" and "GetSymbolVendor()" functions can be intercepted when the .o file contenst are later lazilly needed. Prior to this fix, when we first instantiated the "SymbolFileDWARFDebugMap" class, we would also make modules, object files and symbol files for every .o file in the debug map because we needed to fix up the sections in the .o files with information that is in the executable debug map. Now we lazily do this in the DebugMapModule::GetObjectFile() Cleaned up header includes a bit as well. llvm-svn: 162860
2012-08-29 21:13:06 +00:00
return eLanguageTypeUnknown;
}
size_t SymbolFileDWARFDebugMap::ParseFunctions(CompileUnit &comp_unit) {
std::lock_guard<std::recursive_mutex> guard(GetModuleMutex());
SymbolFileDWARF *oso_dwarf = GetSymbolFile(comp_unit);
if (oso_dwarf)
return oso_dwarf->ParseFunctions(comp_unit);
return 0;
}
bool SymbolFileDWARFDebugMap::ParseLineTable(CompileUnit &comp_unit) {
std::lock_guard<std::recursive_mutex> guard(GetModuleMutex());
SymbolFileDWARF *oso_dwarf = GetSymbolFile(comp_unit);
if (oso_dwarf)
return oso_dwarf->ParseLineTable(comp_unit);
return false;
}
bool SymbolFileDWARFDebugMap::ParseDebugMacros(CompileUnit &comp_unit) {
std::lock_guard<std::recursive_mutex> guard(GetModuleMutex());
SymbolFileDWARF *oso_dwarf = GetSymbolFile(comp_unit);
if (oso_dwarf)
return oso_dwarf->ParseDebugMacros(comp_unit);
return false;
}
[lldb] Decouple importing the std C++ module from the way the program is compiled Summary: At the moment, when trying to import the `std` module in LLDB, we look at the imported modules used in the compiled program and try to infer the Clang configuration we need from the DWARF module-import. That was the initial idea but turned out to cause a few problems or inconveniences: * It requires that users compile their programs with C++ modules. Given how experimental C++ modules are makes this feature inaccessible for many users. Also it means that people can't just get the benefits of this feature for free when we activate it by default (and we can't just close all the associated bug reports). * Relying on DWARF's imported module tags (that are only emitted by default on macOS) means this can only be used when using DWARF (and with -glldb on Linux). * We essentially hardcoded the C standard library paths on some platforms (Linux) or just couldn't support this feature on other platforms (macOS). This patch drops the whole idea of looking at the imported module DWARF tags and instead just uses the support files of the compilation unit. If we look at the support files and see file paths that indicate where the C standard library and libc++ are, we can just create the module configuration this information. This fixes all the problems above which means we can enable all the tests now on Linux, macOS and with other debug information than what we currently had. The only debug information specific code is now the iteration over external type module when -gmodules is used (as `std` and also the `Darwin` module are their own external type module with their own files). The meat of this patch is the CppModuleConfiguration which looks at the file paths from the compilation unit and then figures out the include paths based on those paths. It's quite conservative in that it only enables modules if we find a single C library and single libc++ library. It's still missing some test mode where we try to compile an expression before we actually activate the config for the user (which probably also needs some caching mechanism), but for now it works and makes the feature usable. Reviewers: aprantl, shafik, jdoerfert Reviewed By: aprantl Subscribers: mgorny, abidh, JDevlieghere, lldb-commits Tags: #c_modules_in_lldb, #lldb Differential Revision: https://reviews.llvm.org/D67760 llvm-svn: 372716
2019-09-24 10:08:18 +00:00
void SymbolFileDWARFDebugMap::ForEachExternalModule(
CompileUnit &comp_unit, llvm::function_ref<void(ModuleSP)> f) {
std::lock_guard<std::recursive_mutex> guard(GetModuleMutex());
SymbolFileDWARF *oso_dwarf = GetSymbolFile(comp_unit);
if (oso_dwarf)
oso_dwarf->ForEachExternalModule(comp_unit, f);
}
bool SymbolFileDWARFDebugMap::ParseSupportFiles(CompileUnit &comp_unit,
FileSpecList &support_files) {
std::lock_guard<std::recursive_mutex> guard(GetModuleMutex());
SymbolFileDWARF *oso_dwarf = GetSymbolFile(comp_unit);
if (oso_dwarf)
return oso_dwarf->ParseSupportFiles(comp_unit, support_files);
return false;
}
bool SymbolFileDWARFDebugMap::ParseIsOptimized(CompileUnit &comp_unit) {
std::lock_guard<std::recursive_mutex> guard(GetModuleMutex());
SymbolFileDWARF *oso_dwarf = GetSymbolFile(comp_unit);
if (oso_dwarf)
return oso_dwarf->ParseIsOptimized(comp_unit);
return false;
}
bool SymbolFileDWARFDebugMap::ParseImportedModules(
const SymbolContext &sc, std::vector<SourceModule> &imported_modules) {
std::lock_guard<std::recursive_mutex> guard(GetModuleMutex());
SymbolFileDWARF *oso_dwarf = GetSymbolFile(sc);
if (oso_dwarf)
return oso_dwarf->ParseImportedModules(sc, imported_modules);
return false;
}
size_t SymbolFileDWARFDebugMap::ParseBlocksRecursive(Function &func) {
std::lock_guard<std::recursive_mutex> guard(GetModuleMutex());
CompileUnit *comp_unit = func.GetCompileUnit();
if (!comp_unit)
return 0;
SymbolFileDWARF *oso_dwarf = GetSymbolFile(*comp_unit);
if (oso_dwarf)
return oso_dwarf->ParseBlocksRecursive(func);
return 0;
}
size_t SymbolFileDWARFDebugMap::ParseTypes(CompileUnit &comp_unit) {
std::lock_guard<std::recursive_mutex> guard(GetModuleMutex());
SymbolFileDWARF *oso_dwarf = GetSymbolFile(comp_unit);
if (oso_dwarf)
return oso_dwarf->ParseTypes(comp_unit);
return 0;
}
size_t
SymbolFileDWARFDebugMap::ParseVariablesForContext(const SymbolContext &sc) {
std::lock_guard<std::recursive_mutex> guard(GetModuleMutex());
SymbolFileDWARF *oso_dwarf = GetSymbolFile(sc);
if (oso_dwarf)
<rdar://problem/11757916> Make breakpoint setting by file and line much more efficient by only looking for inlined breakpoint locations if we are setting a breakpoint in anything but a source implementation file. Implementing this complex for a many reasons. Turns out that parsing compile units lazily had some issues with respect to how we need to do things with DWARF in .o files. So the fixes in the checkin for this makes these changes: - Add a new setting called "target.inline-breakpoint-strategy" which can be set to "never", "always", or "headers". "never" will never try and set any inlined breakpoints (fastest). "always" always looks for inlined breakpoint locations (slowest, but most accurate). "headers", which is the default setting, will only look for inlined breakpoint locations if the breakpoint is set in what are consudered to be header files, which is realy defined as "not in an implementation source file". - modify the breakpoint setting by file and line to check the current "target.inline-breakpoint-strategy" setting and act accordingly - Modify compile units to be able to get their language and other info lazily. This allows us to create compile units from the debug map and not have to fill all of the details in, and then lazily discover this information as we go on debuggging. This is needed to avoid parsing all .o files when setting breakpoints in implementation only files (no inlines). Otherwise we would need to parse the .o file, the object file (mach-o in our case) and the symbol file (DWARF in the object file) just to see what the compile unit was. - modify the "SymbolFileDWARFDebugMap" to subclass lldb_private::Module so that the virtual "GetObjectFile()" and "GetSymbolVendor()" functions can be intercepted when the .o file contenst are later lazilly needed. Prior to this fix, when we first instantiated the "SymbolFileDWARFDebugMap" class, we would also make modules, object files and symbol files for every .o file in the debug map because we needed to fix up the sections in the .o files with information that is in the executable debug map. Now we lazily do this in the DebugMapModule::GetObjectFile() Cleaned up header includes a bit as well. llvm-svn: 162860
2012-08-29 21:13:06 +00:00
return oso_dwarf->ParseVariablesForContext(sc);
return 0;
}
Type *SymbolFileDWARFDebugMap::ResolveTypeUID(lldb::user_id_t type_uid) {
std::lock_guard<std::recursive_mutex> guard(GetModuleMutex());
const uint64_t oso_idx = GetOSOIndexFromUserID(type_uid);
SymbolFileDWARF *oso_dwarf = GetSymbolFileByOSOIndex(oso_idx);
if (oso_dwarf)
return oso_dwarf->ResolveTypeUID(type_uid);
return nullptr;
}
llvm::Optional<SymbolFile::ArrayInfo>
SymbolFileDWARFDebugMap::GetDynamicArrayInfoForUID(
lldb::user_id_t type_uid, const lldb_private::ExecutionContext *exe_ctx) {
const uint64_t oso_idx = GetOSOIndexFromUserID(type_uid);
SymbolFileDWARF *oso_dwarf = GetSymbolFileByOSOIndex(oso_idx);
if (oso_dwarf)
return oso_dwarf->GetDynamicArrayInfoForUID(type_uid, exe_ctx);
return llvm::None;
}
bool SymbolFileDWARFDebugMap::CompleteType(CompilerType &compiler_type) {
bool success = false;
if (compiler_type) {
ForEachSymbolFile([&](SymbolFileDWARF *oso_dwarf) -> bool {
if (oso_dwarf->HasForwardDeclForClangType(compiler_type)) {
oso_dwarf->CompleteType(compiler_type);
success = true;
return true;
}
return false;
});
}
return success;
}
uint32_t
SymbolFileDWARFDebugMap::ResolveSymbolContext(const Address &exe_so_addr,
SymbolContextItem resolve_scope,
SymbolContext &sc) {
std::lock_guard<std::recursive_mutex> guard(GetModuleMutex());
uint32_t resolved_flags = 0;
Symtab *symtab = m_objfile_sp->GetSymtab();
if (symtab) {
const addr_t exe_file_addr = exe_so_addr.GetFileAddress();
const DebugMap::Entry *debug_map_entry =
m_debug_map.FindEntryThatContains(exe_file_addr);
if (debug_map_entry) {
sc.symbol =
symtab->SymbolAtIndex(debug_map_entry->data.GetExeSymbolIndex());
if (sc.symbol != nullptr) {
resolved_flags |= eSymbolContextSymbol;
uint32_t oso_idx = 0;
CompileUnitInfo *comp_unit_info =
GetCompileUnitInfoForSymbolWithID(sc.symbol->GetID(), &oso_idx);
if (comp_unit_info) {
comp_unit_info->GetFileRangeMap(this);
Module *oso_module = GetModuleByCompUnitInfo(comp_unit_info);
if (oso_module) {
lldb::addr_t oso_file_addr =
exe_file_addr - debug_map_entry->GetRangeBase() +
debug_map_entry->data.GetOSOFileAddress();
Address oso_so_addr;
if (oso_module->ResolveFileAddress(oso_file_addr, oso_so_addr)) {
resolved_flags |=
oso_module->GetSymbolFile()->ResolveSymbolContext(
oso_so_addr, resolve_scope, sc);
}
}
}
}
}
}
return resolved_flags;
}
uint32_t SymbolFileDWARFDebugMap::ResolveSymbolContext(
const FileSpec &file_spec, uint32_t line, bool check_inlines,
SymbolContextItem resolve_scope, SymbolContextList &sc_list) {
std::lock_guard<std::recursive_mutex> guard(GetModuleMutex());
const uint32_t initial = sc_list.GetSize();
const uint32_t cu_count = GetNumCompileUnits();
for (uint32_t i = 0; i < cu_count; ++i) {
// If we are checking for inlines, then we need to look through all compile
// units no matter if "file_spec" matches.
<rdar://problem/11757916> Make breakpoint setting by file and line much more efficient by only looking for inlined breakpoint locations if we are setting a breakpoint in anything but a source implementation file. Implementing this complex for a many reasons. Turns out that parsing compile units lazily had some issues with respect to how we need to do things with DWARF in .o files. So the fixes in the checkin for this makes these changes: - Add a new setting called "target.inline-breakpoint-strategy" which can be set to "never", "always", or "headers". "never" will never try and set any inlined breakpoints (fastest). "always" always looks for inlined breakpoint locations (slowest, but most accurate). "headers", which is the default setting, will only look for inlined breakpoint locations if the breakpoint is set in what are consudered to be header files, which is realy defined as "not in an implementation source file". - modify the breakpoint setting by file and line to check the current "target.inline-breakpoint-strategy" setting and act accordingly - Modify compile units to be able to get their language and other info lazily. This allows us to create compile units from the debug map and not have to fill all of the details in, and then lazily discover this information as we go on debuggging. This is needed to avoid parsing all .o files when setting breakpoints in implementation only files (no inlines). Otherwise we would need to parse the .o file, the object file (mach-o in our case) and the symbol file (DWARF in the object file) just to see what the compile unit was. - modify the "SymbolFileDWARFDebugMap" to subclass lldb_private::Module so that the virtual "GetObjectFile()" and "GetSymbolVendor()" functions can be intercepted when the .o file contenst are later lazilly needed. Prior to this fix, when we first instantiated the "SymbolFileDWARFDebugMap" class, we would also make modules, object files and symbol files for every .o file in the debug map because we needed to fix up the sections in the .o files with information that is in the executable debug map. Now we lazily do this in the DebugMapModule::GetObjectFile() Cleaned up header includes a bit as well. llvm-svn: 162860
2012-08-29 21:13:06 +00:00
bool resolve = check_inlines;
<rdar://problem/11757916> Make breakpoint setting by file and line much more efficient by only looking for inlined breakpoint locations if we are setting a breakpoint in anything but a source implementation file. Implementing this complex for a many reasons. Turns out that parsing compile units lazily had some issues with respect to how we need to do things with DWARF in .o files. So the fixes in the checkin for this makes these changes: - Add a new setting called "target.inline-breakpoint-strategy" which can be set to "never", "always", or "headers". "never" will never try and set any inlined breakpoints (fastest). "always" always looks for inlined breakpoint locations (slowest, but most accurate). "headers", which is the default setting, will only look for inlined breakpoint locations if the breakpoint is set in what are consudered to be header files, which is realy defined as "not in an implementation source file". - modify the breakpoint setting by file and line to check the current "target.inline-breakpoint-strategy" setting and act accordingly - Modify compile units to be able to get their language and other info lazily. This allows us to create compile units from the debug map and not have to fill all of the details in, and then lazily discover this information as we go on debuggging. This is needed to avoid parsing all .o files when setting breakpoints in implementation only files (no inlines). Otherwise we would need to parse the .o file, the object file (mach-o in our case) and the symbol file (DWARF in the object file) just to see what the compile unit was. - modify the "SymbolFileDWARFDebugMap" to subclass lldb_private::Module so that the virtual "GetObjectFile()" and "GetSymbolVendor()" functions can be intercepted when the .o file contenst are later lazilly needed. Prior to this fix, when we first instantiated the "SymbolFileDWARFDebugMap" class, we would also make modules, object files and symbol files for every .o file in the debug map because we needed to fix up the sections in the .o files with information that is in the executable debug map. Now we lazily do this in the DebugMapModule::GetObjectFile() Cleaned up header includes a bit as well. llvm-svn: 162860
2012-08-29 21:13:06 +00:00
if (!resolve) {
FileSpec so_file_spec;
if (GetFileSpecForSO(i, so_file_spec)) {
// Match the full path if the incoming file_spec has a directory (not
// just a basename)
const bool full_match = (bool)file_spec.GetDirectory();
<rdar://problem/11757916> Make breakpoint setting by file and line much more efficient by only looking for inlined breakpoint locations if we are setting a breakpoint in anything but a source implementation file. Implementing this complex for a many reasons. Turns out that parsing compile units lazily had some issues with respect to how we need to do things with DWARF in .o files. So the fixes in the checkin for this makes these changes: - Add a new setting called "target.inline-breakpoint-strategy" which can be set to "never", "always", or "headers". "never" will never try and set any inlined breakpoints (fastest). "always" always looks for inlined breakpoint locations (slowest, but most accurate). "headers", which is the default setting, will only look for inlined breakpoint locations if the breakpoint is set in what are consudered to be header files, which is realy defined as "not in an implementation source file". - modify the breakpoint setting by file and line to check the current "target.inline-breakpoint-strategy" setting and act accordingly - Modify compile units to be able to get their language and other info lazily. This allows us to create compile units from the debug map and not have to fill all of the details in, and then lazily discover this information as we go on debuggging. This is needed to avoid parsing all .o files when setting breakpoints in implementation only files (no inlines). Otherwise we would need to parse the .o file, the object file (mach-o in our case) and the symbol file (DWARF in the object file) just to see what the compile unit was. - modify the "SymbolFileDWARFDebugMap" to subclass lldb_private::Module so that the virtual "GetObjectFile()" and "GetSymbolVendor()" functions can be intercepted when the .o file contenst are later lazilly needed. Prior to this fix, when we first instantiated the "SymbolFileDWARFDebugMap" class, we would also make modules, object files and symbol files for every .o file in the debug map because we needed to fix up the sections in the .o files with information that is in the executable debug map. Now we lazily do this in the DebugMapModule::GetObjectFile() Cleaned up header includes a bit as well. llvm-svn: 162860
2012-08-29 21:13:06 +00:00
resolve = FileSpec::Equal(file_spec, so_file_spec, full_match);
}
}
if (resolve) {
<rdar://problem/11757916> Make breakpoint setting by file and line much more efficient by only looking for inlined breakpoint locations if we are setting a breakpoint in anything but a source implementation file. Implementing this complex for a many reasons. Turns out that parsing compile units lazily had some issues with respect to how we need to do things with DWARF in .o files. So the fixes in the checkin for this makes these changes: - Add a new setting called "target.inline-breakpoint-strategy" which can be set to "never", "always", or "headers". "never" will never try and set any inlined breakpoints (fastest). "always" always looks for inlined breakpoint locations (slowest, but most accurate). "headers", which is the default setting, will only look for inlined breakpoint locations if the breakpoint is set in what are consudered to be header files, which is realy defined as "not in an implementation source file". - modify the breakpoint setting by file and line to check the current "target.inline-breakpoint-strategy" setting and act accordingly - Modify compile units to be able to get their language and other info lazily. This allows us to create compile units from the debug map and not have to fill all of the details in, and then lazily discover this information as we go on debuggging. This is needed to avoid parsing all .o files when setting breakpoints in implementation only files (no inlines). Otherwise we would need to parse the .o file, the object file (mach-o in our case) and the symbol file (DWARF in the object file) just to see what the compile unit was. - modify the "SymbolFileDWARFDebugMap" to subclass lldb_private::Module so that the virtual "GetObjectFile()" and "GetSymbolVendor()" functions can be intercepted when the .o file contenst are later lazilly needed. Prior to this fix, when we first instantiated the "SymbolFileDWARFDebugMap" class, we would also make modules, object files and symbol files for every .o file in the debug map because we needed to fix up the sections in the .o files with information that is in the executable debug map. Now we lazily do this in the DebugMapModule::GetObjectFile() Cleaned up header includes a bit as well. llvm-svn: 162860
2012-08-29 21:13:06 +00:00
SymbolFileDWARF *oso_dwarf = GetSymbolFileByOSOIndex(i);
if (oso_dwarf)
oso_dwarf->ResolveSymbolContext(file_spec, line, check_inlines,
resolve_scope, sc_list);
}
}
return sc_list.GetSize() - initial;
}
void SymbolFileDWARFDebugMap::PrivateFindGlobalVariables(
ConstString name, const CompilerDeclContext *parent_decl_ctx,
const std::vector<uint32_t>
&indexes, // Indexes into the symbol table that match "name"
uint32_t max_matches, VariableList &variables) {
const size_t match_count = indexes.size();
for (size_t i = 0; i < match_count; ++i) {
uint32_t oso_idx;
CompileUnitInfo *comp_unit_info =
GetCompileUnitInfoForSymbolWithIndex(indexes[i], &oso_idx);
if (comp_unit_info) {
SymbolFileDWARF *oso_dwarf = GetSymbolFileByOSOIndex(oso_idx);
if (oso_dwarf) {
oso_dwarf->FindGlobalVariables(name, parent_decl_ctx, max_matches,
variables);
if (variables.GetSize() > max_matches)
break;
}
}
}
}
void SymbolFileDWARFDebugMap::FindGlobalVariables(
ConstString name, const CompilerDeclContext *parent_decl_ctx,
uint32_t max_matches, VariableList &variables) {
std::lock_guard<std::recursive_mutex> guard(GetModuleMutex());
uint32_t total_matches = 0;
ForEachSymbolFile([&](SymbolFileDWARF *oso_dwarf) -> bool {
const uint32_t old_size = variables.GetSize();
oso_dwarf->FindGlobalVariables(name, parent_decl_ctx, max_matches,
variables);
const uint32_t oso_matches = variables.GetSize() - old_size;
if (oso_matches > 0) {
total_matches += oso_matches;
// Are we getting all matches?
if (max_matches == UINT32_MAX)
return false; // Yep, continue getting everything
// If we have found enough matches, lets get out
Final bit of type system cleanup that abstracts declaration contexts into lldb_private::CompilerDeclContext and renames ClangType to CompilerType in many accessors and functions. Create a new "lldb_private::CompilerDeclContext" class that will replace all direct uses of "clang::DeclContext" when used in compiler agnostic code, yet still allow for conversion to clang::DeclContext subclasses by clang specific code. This completes the abstraction of type parsing by removing all "clang::" references from the SymbolFileDWARF. The new "lldb_private::CompilerDeclContext" class abstracts decl contexts found in compiler type systems so they can be used in internal API calls. The TypeSystem is required to support CompilerDeclContexts with new pure virtual functions that start with "DeclContext" in the member function names. Converted all code that used lldb_private::ClangNamespaceDecl over to use the new CompilerDeclContext class and removed the ClangNamespaceDecl.cpp and ClangNamespaceDecl.h files. Removed direct use of clang APIs from SBType and now use the abstract type systems to correctly explore types. Bulk renames for things that used to return a ClangASTType which is now CompilerType: "Type::GetClangFullType()" to "Type::GetFullCompilerType()" "Type::GetClangLayoutType()" to "Type::GetLayoutCompilerType()" "Type::GetClangForwardType()" to "Type::GetForwardCompilerType()" "Value::GetClangType()" to "Value::GetCompilerType()" "Value::SetClangType (const CompilerType &)" to "Value::SetCompilerType (const CompilerType &)" "ValueObject::GetClangType ()" to "ValueObject::GetCompilerType()" many more renames that are similar. llvm-svn: 245905
2015-08-24 23:46:31 +00:00
if (max_matches >= total_matches)
return true;
// Update the max matches for any subsequent calls to find globals in any
// other object files with DWARF
max_matches -= oso_matches;
}
return false;
});
}
void SymbolFileDWARFDebugMap::FindGlobalVariables(
const RegularExpression &regex, uint32_t max_matches,
VariableList &variables) {
std::lock_guard<std::recursive_mutex> guard(GetModuleMutex());
uint32_t total_matches = 0;
ForEachSymbolFile([&](SymbolFileDWARF *oso_dwarf) -> bool {
const uint32_t old_size = variables.GetSize();
oso_dwarf->FindGlobalVariables(regex, max_matches, variables);
const uint32_t oso_matches = variables.GetSize() - old_size;
if (oso_matches > 0) {
total_matches += oso_matches;
// Are we getting all matches?
if (max_matches == UINT32_MAX)
return false; // Yep, continue getting everything
// If we have found enough matches, lets get out
if (max_matches >= total_matches)
return true;
// Update the max matches for any subsequent calls to find globals in any
// other object files with DWARF
max_matches -= oso_matches;
}
return false;
});
}
int SymbolFileDWARFDebugMap::SymbolContainsSymbolWithIndex(
uint32_t *symbol_idx_ptr, const CompileUnitInfo *comp_unit_info) {
const uint32_t symbol_idx = *symbol_idx_ptr;
if (symbol_idx < comp_unit_info->first_symbol_index)
return -1;
if (symbol_idx <= comp_unit_info->last_symbol_index)
return 0;
return 1;
}
int SymbolFileDWARFDebugMap::SymbolContainsSymbolWithID(
user_id_t *symbol_idx_ptr, const CompileUnitInfo *comp_unit_info) {
const user_id_t symbol_id = *symbol_idx_ptr;
<rdar://problem/11757916> Make breakpoint setting by file and line much more efficient by only looking for inlined breakpoint locations if we are setting a breakpoint in anything but a source implementation file. Implementing this complex for a many reasons. Turns out that parsing compile units lazily had some issues with respect to how we need to do things with DWARF in .o files. So the fixes in the checkin for this makes these changes: - Add a new setting called "target.inline-breakpoint-strategy" which can be set to "never", "always", or "headers". "never" will never try and set any inlined breakpoints (fastest). "always" always looks for inlined breakpoint locations (slowest, but most accurate). "headers", which is the default setting, will only look for inlined breakpoint locations if the breakpoint is set in what are consudered to be header files, which is realy defined as "not in an implementation source file". - modify the breakpoint setting by file and line to check the current "target.inline-breakpoint-strategy" setting and act accordingly - Modify compile units to be able to get their language and other info lazily. This allows us to create compile units from the debug map and not have to fill all of the details in, and then lazily discover this information as we go on debuggging. This is needed to avoid parsing all .o files when setting breakpoints in implementation only files (no inlines). Otherwise we would need to parse the .o file, the object file (mach-o in our case) and the symbol file (DWARF in the object file) just to see what the compile unit was. - modify the "SymbolFileDWARFDebugMap" to subclass lldb_private::Module so that the virtual "GetObjectFile()" and "GetSymbolVendor()" functions can be intercepted when the .o file contenst are later lazilly needed. Prior to this fix, when we first instantiated the "SymbolFileDWARFDebugMap" class, we would also make modules, object files and symbol files for every .o file in the debug map because we needed to fix up the sections in the .o files with information that is in the executable debug map. Now we lazily do this in the DebugMapModule::GetObjectFile() Cleaned up header includes a bit as well. llvm-svn: 162860
2012-08-29 21:13:06 +00:00
if (symbol_id < comp_unit_info->first_symbol_id)
return -1;
<rdar://problem/11757916> Make breakpoint setting by file and line much more efficient by only looking for inlined breakpoint locations if we are setting a breakpoint in anything but a source implementation file. Implementing this complex for a many reasons. Turns out that parsing compile units lazily had some issues with respect to how we need to do things with DWARF in .o files. So the fixes in the checkin for this makes these changes: - Add a new setting called "target.inline-breakpoint-strategy" which can be set to "never", "always", or "headers". "never" will never try and set any inlined breakpoints (fastest). "always" always looks for inlined breakpoint locations (slowest, but most accurate). "headers", which is the default setting, will only look for inlined breakpoint locations if the breakpoint is set in what are consudered to be header files, which is realy defined as "not in an implementation source file". - modify the breakpoint setting by file and line to check the current "target.inline-breakpoint-strategy" setting and act accordingly - Modify compile units to be able to get their language and other info lazily. This allows us to create compile units from the debug map and not have to fill all of the details in, and then lazily discover this information as we go on debuggging. This is needed to avoid parsing all .o files when setting breakpoints in implementation only files (no inlines). Otherwise we would need to parse the .o file, the object file (mach-o in our case) and the symbol file (DWARF in the object file) just to see what the compile unit was. - modify the "SymbolFileDWARFDebugMap" to subclass lldb_private::Module so that the virtual "GetObjectFile()" and "GetSymbolVendor()" functions can be intercepted when the .o file contenst are later lazilly needed. Prior to this fix, when we first instantiated the "SymbolFileDWARFDebugMap" class, we would also make modules, object files and symbol files for every .o file in the debug map because we needed to fix up the sections in the .o files with information that is in the executable debug map. Now we lazily do this in the DebugMapModule::GetObjectFile() Cleaned up header includes a bit as well. llvm-svn: 162860
2012-08-29 21:13:06 +00:00
if (symbol_id <= comp_unit_info->last_symbol_id)
return 0;
return 1;
}
SymbolFileDWARFDebugMap::CompileUnitInfo *
SymbolFileDWARFDebugMap::GetCompileUnitInfoForSymbolWithIndex(
uint32_t symbol_idx, uint32_t *oso_idx_ptr) {
const uint32_t oso_index_count = m_compile_unit_infos.size();
CompileUnitInfo *comp_unit_info = nullptr;
if (oso_index_count) {
comp_unit_info = (CompileUnitInfo *)bsearch(
&symbol_idx, &m_compile_unit_infos[0], m_compile_unit_infos.size(),
sizeof(CompileUnitInfo),
(ComparisonFunction)SymbolContainsSymbolWithIndex);
}
if (oso_idx_ptr) {
if (comp_unit_info != nullptr)
*oso_idx_ptr = comp_unit_info - &m_compile_unit_infos[0];
else
*oso_idx_ptr = UINT32_MAX;
}
return comp_unit_info;
}
SymbolFileDWARFDebugMap::CompileUnitInfo *
SymbolFileDWARFDebugMap::GetCompileUnitInfoForSymbolWithID(
user_id_t symbol_id, uint32_t *oso_idx_ptr) {
const uint32_t oso_index_count = m_compile_unit_infos.size();
CompileUnitInfo *comp_unit_info = nullptr;
if (oso_index_count) {
comp_unit_info = (CompileUnitInfo *)::bsearch(
&symbol_id, &m_compile_unit_infos[0], m_compile_unit_infos.size(),
sizeof(CompileUnitInfo),
(ComparisonFunction)SymbolContainsSymbolWithID);
}
if (oso_idx_ptr) {
if (comp_unit_info != nullptr)
*oso_idx_ptr = comp_unit_info - &m_compile_unit_infos[0];
else
*oso_idx_ptr = UINT32_MAX;
}
return comp_unit_info;
}
static void RemoveFunctionsWithModuleNotEqualTo(const ModuleSP &module_sp,
SymbolContextList &sc_list,
uint32_t start_idx) {
// We found functions in .o files. Not all functions in the .o files will
// have made it into the final output file. The ones that did make it into
// the final output file will have a section whose module matches the module
// from the ObjectFile for this SymbolFile. When the modules don't match,
// then we have something that was in a .o file, but doesn't map to anything
// in the final executable.
uint32_t i = start_idx;
while (i < sc_list.GetSize()) {
SymbolContext sc;
sc_list.GetContextAtIndex(i, sc);
if (sc.function) {
const SectionSP section_sp(
sc.function->GetAddressRange().GetBaseAddress().GetSection());
if (section_sp->GetModule() != module_sp) {
sc_list.RemoveContextAtIndex(i);
continue;
}
}
++i;
}
}
void SymbolFileDWARFDebugMap::FindFunctions(
ConstString name, const CompilerDeclContext *parent_decl_ctx,
FunctionNameType name_type_mask, bool include_inlines,
SymbolContextList &sc_list) {
std::lock_guard<std::recursive_mutex> guard(GetModuleMutex());
static Timer::Category func_cat(LLVM_PRETTY_FUNCTION);
Timer scoped_timer(func_cat,
"SymbolFileDWARFDebugMap::FindFunctions (name = %s)",
name.GetCString());
ForEachSymbolFile([&](SymbolFileDWARF *oso_dwarf) -> bool {
uint32_t sc_idx = sc_list.GetSize();
oso_dwarf->FindFunctions(name, parent_decl_ctx, name_type_mask,
include_inlines, sc_list);
if (!sc_list.IsEmpty()) {
RemoveFunctionsWithModuleNotEqualTo(m_objfile_sp->GetModule(), sc_list,
sc_idx);
}
return false;
});
}
void SymbolFileDWARFDebugMap::FindFunctions(const RegularExpression &regex,
bool include_inlines,
SymbolContextList &sc_list) {
std::lock_guard<std::recursive_mutex> guard(GetModuleMutex());
static Timer::Category func_cat(LLVM_PRETTY_FUNCTION);
Timer scoped_timer(func_cat,
"SymbolFileDWARFDebugMap::FindFunctions (regex = '%s')",
regex.GetText().str().c_str());
ForEachSymbolFile([&](SymbolFileDWARF *oso_dwarf) -> bool {
uint32_t sc_idx = sc_list.GetSize();
oso_dwarf->FindFunctions(regex, include_inlines, sc_list);
if (!sc_list.IsEmpty()) {
RemoveFunctionsWithModuleNotEqualTo(m_objfile_sp->GetModule(), sc_list,
sc_idx);
}
return false;
});
}
void SymbolFileDWARFDebugMap::GetTypes(SymbolContextScope *sc_scope,
lldb::TypeClass type_mask,
TypeList &type_list) {
std::lock_guard<std::recursive_mutex> guard(GetModuleMutex());
static Timer::Category func_cat(LLVM_PRETTY_FUNCTION);
Timer scoped_timer(func_cat,
"SymbolFileDWARFDebugMap::GetTypes (type_mask = 0x%8.8x)",
type_mask);
SymbolFileDWARF *oso_dwarf = nullptr;
Added the ability to get a list of types from a SBModule or SBCompileUnit. Sebastien Metrot wanted this, and sent a hollowed out patch. I filled in the blanks and did the low level implementation. The new functions are: //------------------------------------------------------------------ /// Get all types matching \a type_mask from debug info in this /// module. /// /// @param[in] type_mask /// A bitfield that consists of one or more bits logically OR'ed /// together from the lldb::TypeClass enumeration. This allows /// you to request only structure types, or only class, struct /// and union types. Passing in lldb::eTypeClassAny will return /// all types found in the debug information for this module. /// /// @return /// A list of types in this module that match \a type_mask //------------------------------------------------------------------ lldb::SBTypeList SBModule::GetTypes (uint32_t type_mask) //------------------------------------------------------------------ /// Get all types matching \a type_mask from debug info in this /// compile unit. /// /// @param[in] type_mask /// A bitfield that consists of one or more bits logically OR'ed /// together from the lldb::TypeClass enumeration. This allows /// you to request only structure types, or only class, struct /// and union types. Passing in lldb::eTypeClassAny will return /// all types found in the debug information for this compile /// unit. /// /// @return /// A list of types in this compile unit that match \a type_mask //------------------------------------------------------------------ lldb::SBTypeList SBCompileUnit::GetTypes (uint32_t type_mask = lldb::eTypeClassAny); This lets you request types by filling out a mask that contains one or more bits from the lldb::TypeClass enumerations, so you can only get the types you really want. llvm-svn: 184251
2013-06-18 22:51:05 +00:00
if (sc_scope) {
SymbolContext sc;
Added the ability to get a list of types from a SBModule or SBCompileUnit. Sebastien Metrot wanted this, and sent a hollowed out patch. I filled in the blanks and did the low level implementation. The new functions are: //------------------------------------------------------------------ /// Get all types matching \a type_mask from debug info in this /// module. /// /// @param[in] type_mask /// A bitfield that consists of one or more bits logically OR'ed /// together from the lldb::TypeClass enumeration. This allows /// you to request only structure types, or only class, struct /// and union types. Passing in lldb::eTypeClassAny will return /// all types found in the debug information for this module. /// /// @return /// A list of types in this module that match \a type_mask //------------------------------------------------------------------ lldb::SBTypeList SBModule::GetTypes (uint32_t type_mask) //------------------------------------------------------------------ /// Get all types matching \a type_mask from debug info in this /// compile unit. /// /// @param[in] type_mask /// A bitfield that consists of one or more bits logically OR'ed /// together from the lldb::TypeClass enumeration. This allows /// you to request only structure types, or only class, struct /// and union types. Passing in lldb::eTypeClassAny will return /// all types found in the debug information for this compile /// unit. /// /// @return /// A list of types in this compile unit that match \a type_mask //------------------------------------------------------------------ lldb::SBTypeList SBCompileUnit::GetTypes (uint32_t type_mask = lldb::eTypeClassAny); This lets you request types by filling out a mask that contains one or more bits from the lldb::TypeClass enumerations, so you can only get the types you really want. llvm-svn: 184251
2013-06-18 22:51:05 +00:00
sc_scope->CalculateSymbolContext(&sc);
Added the ability to get a list of types from a SBModule or SBCompileUnit. Sebastien Metrot wanted this, and sent a hollowed out patch. I filled in the blanks and did the low level implementation. The new functions are: //------------------------------------------------------------------ /// Get all types matching \a type_mask from debug info in this /// module. /// /// @param[in] type_mask /// A bitfield that consists of one or more bits logically OR'ed /// together from the lldb::TypeClass enumeration. This allows /// you to request only structure types, or only class, struct /// and union types. Passing in lldb::eTypeClassAny will return /// all types found in the debug information for this module. /// /// @return /// A list of types in this module that match \a type_mask //------------------------------------------------------------------ lldb::SBTypeList SBModule::GetTypes (uint32_t type_mask) //------------------------------------------------------------------ /// Get all types matching \a type_mask from debug info in this /// compile unit. /// /// @param[in] type_mask /// A bitfield that consists of one or more bits logically OR'ed /// together from the lldb::TypeClass enumeration. This allows /// you to request only structure types, or only class, struct /// and union types. Passing in lldb::eTypeClassAny will return /// all types found in the debug information for this compile /// unit. /// /// @return /// A list of types in this compile unit that match \a type_mask //------------------------------------------------------------------ lldb::SBTypeList SBCompileUnit::GetTypes (uint32_t type_mask = lldb::eTypeClassAny); This lets you request types by filling out a mask that contains one or more bits from the lldb::TypeClass enumerations, so you can only get the types you really want. llvm-svn: 184251
2013-06-18 22:51:05 +00:00
CompileUnitInfo *cu_info = GetCompUnitInfo(sc);
if (cu_info) {
oso_dwarf = GetSymbolFileByCompUnitInfo(cu_info);
if (oso_dwarf)
oso_dwarf->GetTypes(sc_scope, type_mask, type_list);
}
} else {
ForEachSymbolFile([&](SymbolFileDWARF *oso_dwarf) -> bool {
oso_dwarf->GetTypes(sc_scope, type_mask, type_list);
return false;
});
}
}
std::vector<lldb_private::CallEdge>
SymbolFileDWARFDebugMap::ParseCallEdgesInFunction(UserID func_id) {
uint32_t oso_idx = GetOSOIndexFromUserID(func_id.GetID());
SymbolFileDWARF *oso_dwarf = GetSymbolFileByOSOIndex(oso_idx);
if (oso_dwarf)
return oso_dwarf->ParseCallEdgesInFunction(func_id);
return {};
}
Added the ability to get a list of types from a SBModule or SBCompileUnit. Sebastien Metrot wanted this, and sent a hollowed out patch. I filled in the blanks and did the low level implementation. The new functions are: //------------------------------------------------------------------ /// Get all types matching \a type_mask from debug info in this /// module. /// /// @param[in] type_mask /// A bitfield that consists of one or more bits logically OR'ed /// together from the lldb::TypeClass enumeration. This allows /// you to request only structure types, or only class, struct /// and union types. Passing in lldb::eTypeClassAny will return /// all types found in the debug information for this module. /// /// @return /// A list of types in this module that match \a type_mask //------------------------------------------------------------------ lldb::SBTypeList SBModule::GetTypes (uint32_t type_mask) //------------------------------------------------------------------ /// Get all types matching \a type_mask from debug info in this /// compile unit. /// /// @param[in] type_mask /// A bitfield that consists of one or more bits logically OR'ed /// together from the lldb::TypeClass enumeration. This allows /// you to request only structure types, or only class, struct /// and union types. Passing in lldb::eTypeClassAny will return /// all types found in the debug information for this compile /// unit. /// /// @return /// A list of types in this compile unit that match \a type_mask //------------------------------------------------------------------ lldb::SBTypeList SBCompileUnit::GetTypes (uint32_t type_mask = lldb::eTypeClassAny); This lets you request types by filling out a mask that contains one or more bits from the lldb::TypeClass enumerations, so you can only get the types you really want. llvm-svn: 184251
2013-06-18 22:51:05 +00:00
TypeSP SymbolFileDWARFDebugMap::FindDefinitionTypeForDWARFDeclContext(
const DWARFDeclContext &die_decl_ctx) {
TypeSP type_sp;
ForEachSymbolFile([&](SymbolFileDWARF *oso_dwarf) -> bool {
type_sp = oso_dwarf->FindDefinitionTypeForDWARFDeclContext(die_decl_ctx);
return ((bool)type_sp);
});
Added the ability to get a list of types from a SBModule or SBCompileUnit. Sebastien Metrot wanted this, and sent a hollowed out patch. I filled in the blanks and did the low level implementation. The new functions are: //------------------------------------------------------------------ /// Get all types matching \a type_mask from debug info in this /// module. /// /// @param[in] type_mask /// A bitfield that consists of one or more bits logically OR'ed /// together from the lldb::TypeClass enumeration. This allows /// you to request only structure types, or only class, struct /// and union types. Passing in lldb::eTypeClassAny will return /// all types found in the debug information for this module. /// /// @return /// A list of types in this module that match \a type_mask //------------------------------------------------------------------ lldb::SBTypeList SBModule::GetTypes (uint32_t type_mask) //------------------------------------------------------------------ /// Get all types matching \a type_mask from debug info in this /// compile unit. /// /// @param[in] type_mask /// A bitfield that consists of one or more bits logically OR'ed /// together from the lldb::TypeClass enumeration. This allows /// you to request only structure types, or only class, struct /// and union types. Passing in lldb::eTypeClassAny will return /// all types found in the debug information for this compile /// unit. /// /// @return /// A list of types in this compile unit that match \a type_mask //------------------------------------------------------------------ lldb::SBTypeList SBCompileUnit::GetTypes (uint32_t type_mask = lldb::eTypeClassAny); This lets you request types by filling out a mask that contains one or more bits from the lldb::TypeClass enumerations, so you can only get the types you really want. llvm-svn: 184251
2013-06-18 22:51:05 +00:00
return type_sp;
}
bool SymbolFileDWARFDebugMap::Supports_DW_AT_APPLE_objc_complete_type(
SymbolFileDWARF *skip_dwarf_oso) {
if (m_supports_DW_AT_APPLE_objc_complete_type == eLazyBoolCalculate) {
m_supports_DW_AT_APPLE_objc_complete_type = eLazyBoolNo;
ForEachSymbolFile([&](SymbolFileDWARF *oso_dwarf) -> bool {
if (skip_dwarf_oso != oso_dwarf &&
oso_dwarf->Supports_DW_AT_APPLE_objc_complete_type(nullptr)) {
m_supports_DW_AT_APPLE_objc_complete_type = eLazyBoolYes;
return true;
}
return false;
});
}
return m_supports_DW_AT_APPLE_objc_complete_type == eLazyBoolYes;
}
TypeSP SymbolFileDWARFDebugMap::FindCompleteObjCDefinitionTypeForDIE(
const DWARFDIE &die, ConstString type_name,
bool must_be_implementation) {
// If we have a debug map, we will have an Objective-C symbol whose name is
// the type name and whose type is eSymbolTypeObjCClass. If we can find that
// symbol and find its containing parent, we can locate the .o file that will
// contain the implementation definition since it will be scoped inside the
// N_SO and we can then locate the SymbolFileDWARF that corresponds to that
// N_SO.
SymbolFileDWARF *oso_dwarf = nullptr;
TypeSP type_sp;
ObjectFile *module_objfile = m_objfile_sp->GetModule()->GetObjectFile();
if (module_objfile) {
Symtab *symtab = module_objfile->GetSymtab();
if (symtab) {
Symbol *objc_class_symbol = symtab->FindFirstSymbolWithNameAndType(
type_name, eSymbolTypeObjCClass, Symtab::eDebugAny,
Symtab::eVisibilityAny);
if (objc_class_symbol) {
// Get the N_SO symbol that contains the objective C class symbol as
// this should be the .o file that contains the real definition...
const Symbol *source_file_symbol = symtab->GetParent(objc_class_symbol);
if (source_file_symbol &&
source_file_symbol->GetType() == eSymbolTypeSourceFile) {
const uint32_t source_file_symbol_idx =
symtab->GetIndexForSymbol(source_file_symbol);
if (source_file_symbol_idx != UINT32_MAX) {
CompileUnitInfo *compile_unit_info =
GetCompileUnitInfoForSymbolWithIndex(source_file_symbol_idx,
nullptr);
if (compile_unit_info) {
oso_dwarf = GetSymbolFileByCompUnitInfo(compile_unit_info);
if (oso_dwarf) {
TypeSP type_sp(oso_dwarf->FindCompleteObjCDefinitionTypeForDIE(
die, type_name, must_be_implementation));
if (type_sp) {
return type_sp;
}
}
}
}
}
}
}
}
// Only search all .o files for the definition if we don't need the
// implementation because otherwise, with a valid debug map we should have
// the ObjC class symbol and the code above should have found it.
if (!must_be_implementation) {
TypeSP type_sp;
ForEachSymbolFile([&](SymbolFileDWARF *oso_dwarf) -> bool {
type_sp = oso_dwarf->FindCompleteObjCDefinitionTypeForDIE(
die, type_name, must_be_implementation);
return (bool)type_sp;
});
return type_sp;
}
return TypeSP();
}
void SymbolFileDWARFDebugMap::FindTypes(
ConstString name, const CompilerDeclContext *parent_decl_ctx,
uint32_t max_matches,
llvm::DenseSet<lldb_private::SymbolFile *> &searched_symbol_files,
TypeMap &types) {
std::lock_guard<std::recursive_mutex> guard(GetModuleMutex());
ForEachSymbolFile([&](SymbolFileDWARF *oso_dwarf) -> bool {
oso_dwarf->FindTypes(name, parent_decl_ctx, max_matches,
searched_symbol_files, types);
return types.GetSize() >= max_matches;
});
}
//
// uint32_t
// SymbolFileDWARFDebugMap::FindTypes (const SymbolContext& sc, const
// RegularExpression& regex, bool append, uint32_t max_matches, Type::Encoding
// encoding, lldb::user_id_t udt_uid, TypeList& types)
//{
// SymbolFileDWARF *oso_dwarf = GetSymbolFile (sc);
// if (oso_dwarf)
// return oso_dwarf->FindTypes (sc, regex, append, max_matches, encoding,
// udt_uid, types);
// return 0;
//}
CompilerDeclContext SymbolFileDWARFDebugMap::FindNamespace(
lldb_private::ConstString name,
Final bit of type system cleanup that abstracts declaration contexts into lldb_private::CompilerDeclContext and renames ClangType to CompilerType in many accessors and functions. Create a new "lldb_private::CompilerDeclContext" class that will replace all direct uses of "clang::DeclContext" when used in compiler agnostic code, yet still allow for conversion to clang::DeclContext subclasses by clang specific code. This completes the abstraction of type parsing by removing all "clang::" references from the SymbolFileDWARF. The new "lldb_private::CompilerDeclContext" class abstracts decl contexts found in compiler type systems so they can be used in internal API calls. The TypeSystem is required to support CompilerDeclContexts with new pure virtual functions that start with "DeclContext" in the member function names. Converted all code that used lldb_private::ClangNamespaceDecl over to use the new CompilerDeclContext class and removed the ClangNamespaceDecl.cpp and ClangNamespaceDecl.h files. Removed direct use of clang APIs from SBType and now use the abstract type systems to correctly explore types. Bulk renames for things that used to return a ClangASTType which is now CompilerType: "Type::GetClangFullType()" to "Type::GetFullCompilerType()" "Type::GetClangLayoutType()" to "Type::GetLayoutCompilerType()" "Type::GetClangForwardType()" to "Type::GetForwardCompilerType()" "Value::GetClangType()" to "Value::GetCompilerType()" "Value::SetClangType (const CompilerType &)" to "Value::SetCompilerType (const CompilerType &)" "ValueObject::GetClangType ()" to "ValueObject::GetCompilerType()" many more renames that are similar. llvm-svn: 245905
2015-08-24 23:46:31 +00:00
const CompilerDeclContext *parent_decl_ctx) {
std::lock_guard<std::recursive_mutex> guard(GetModuleMutex());
Final bit of type system cleanup that abstracts declaration contexts into lldb_private::CompilerDeclContext and renames ClangType to CompilerType in many accessors and functions. Create a new "lldb_private::CompilerDeclContext" class that will replace all direct uses of "clang::DeclContext" when used in compiler agnostic code, yet still allow for conversion to clang::DeclContext subclasses by clang specific code. This completes the abstraction of type parsing by removing all "clang::" references from the SymbolFileDWARF. The new "lldb_private::CompilerDeclContext" class abstracts decl contexts found in compiler type systems so they can be used in internal API calls. The TypeSystem is required to support CompilerDeclContexts with new pure virtual functions that start with "DeclContext" in the member function names. Converted all code that used lldb_private::ClangNamespaceDecl over to use the new CompilerDeclContext class and removed the ClangNamespaceDecl.cpp and ClangNamespaceDecl.h files. Removed direct use of clang APIs from SBType and now use the abstract type systems to correctly explore types. Bulk renames for things that used to return a ClangASTType which is now CompilerType: "Type::GetClangFullType()" to "Type::GetFullCompilerType()" "Type::GetClangLayoutType()" to "Type::GetLayoutCompilerType()" "Type::GetClangForwardType()" to "Type::GetForwardCompilerType()" "Value::GetClangType()" to "Value::GetCompilerType()" "Value::SetClangType (const CompilerType &)" to "Value::SetCompilerType (const CompilerType &)" "ValueObject::GetClangType ()" to "ValueObject::GetCompilerType()" many more renames that are similar. llvm-svn: 245905
2015-08-24 23:46:31 +00:00
CompilerDeclContext matching_namespace;
ForEachSymbolFile([&](SymbolFileDWARF *oso_dwarf) -> bool {
matching_namespace = oso_dwarf->FindNamespace(name, parent_decl_ctx);
return (bool)matching_namespace;
});
return matching_namespace;
}
void SymbolFileDWARFDebugMap::DumpClangAST(Stream &s) {
ForEachSymbolFile([&s](SymbolFileDWARF *oso_dwarf) -> bool {
oso_dwarf->DumpClangAST(s);
return true;
});
}
// PluginInterface protocol
lldb_private::ConstString SymbolFileDWARFDebugMap::GetPluginName() {
return GetPluginNameStatic();
}
uint32_t SymbolFileDWARFDebugMap::GetPluginVersion() { return 1; }
<rdar://problem/11757916> Make breakpoint setting by file and line much more efficient by only looking for inlined breakpoint locations if we are setting a breakpoint in anything but a source implementation file. Implementing this complex for a many reasons. Turns out that parsing compile units lazily had some issues with respect to how we need to do things with DWARF in .o files. So the fixes in the checkin for this makes these changes: - Add a new setting called "target.inline-breakpoint-strategy" which can be set to "never", "always", or "headers". "never" will never try and set any inlined breakpoints (fastest). "always" always looks for inlined breakpoint locations (slowest, but most accurate). "headers", which is the default setting, will only look for inlined breakpoint locations if the breakpoint is set in what are consudered to be header files, which is realy defined as "not in an implementation source file". - modify the breakpoint setting by file and line to check the current "target.inline-breakpoint-strategy" setting and act accordingly - Modify compile units to be able to get their language and other info lazily. This allows us to create compile units from the debug map and not have to fill all of the details in, and then lazily discover this information as we go on debuggging. This is needed to avoid parsing all .o files when setting breakpoints in implementation only files (no inlines). Otherwise we would need to parse the .o file, the object file (mach-o in our case) and the symbol file (DWARF in the object file) just to see what the compile unit was. - modify the "SymbolFileDWARFDebugMap" to subclass lldb_private::Module so that the virtual "GetObjectFile()" and "GetSymbolVendor()" functions can be intercepted when the .o file contenst are later lazilly needed. Prior to this fix, when we first instantiated the "SymbolFileDWARFDebugMap" class, we would also make modules, object files and symbol files for every .o file in the debug map because we needed to fix up the sections in the .o files with information that is in the executable debug map. Now we lazily do this in the DebugMapModule::GetObjectFile() Cleaned up header includes a bit as well. llvm-svn: 162860
2012-08-29 21:13:06 +00:00
lldb::CompUnitSP
SymbolFileDWARFDebugMap::GetCompileUnit(SymbolFileDWARF *oso_dwarf) {
if (oso_dwarf) {
const uint32_t cu_count = GetNumCompileUnits();
for (uint32_t cu_idx = 0; cu_idx < cu_count; ++cu_idx) {
SymbolFileDWARF *oso_symfile =
GetSymbolFileByCompUnitInfo(&m_compile_unit_infos[cu_idx]);
if (oso_symfile == oso_dwarf) {
if (!m_compile_unit_infos[cu_idx].compile_unit_sp)
m_compile_unit_infos[cu_idx].compile_unit_sp =
ParseCompileUnitAtIndex(cu_idx);
return m_compile_unit_infos[cu_idx].compile_unit_sp;
<rdar://problem/11757916> Make breakpoint setting by file and line much more efficient by only looking for inlined breakpoint locations if we are setting a breakpoint in anything but a source implementation file. Implementing this complex for a many reasons. Turns out that parsing compile units lazily had some issues with respect to how we need to do things with DWARF in .o files. So the fixes in the checkin for this makes these changes: - Add a new setting called "target.inline-breakpoint-strategy" which can be set to "never", "always", or "headers". "never" will never try and set any inlined breakpoints (fastest). "always" always looks for inlined breakpoint locations (slowest, but most accurate). "headers", which is the default setting, will only look for inlined breakpoint locations if the breakpoint is set in what are consudered to be header files, which is realy defined as "not in an implementation source file". - modify the breakpoint setting by file and line to check the current "target.inline-breakpoint-strategy" setting and act accordingly - Modify compile units to be able to get their language and other info lazily. This allows us to create compile units from the debug map and not have to fill all of the details in, and then lazily discover this information as we go on debuggging. This is needed to avoid parsing all .o files when setting breakpoints in implementation only files (no inlines). Otherwise we would need to parse the .o file, the object file (mach-o in our case) and the symbol file (DWARF in the object file) just to see what the compile unit was. - modify the "SymbolFileDWARFDebugMap" to subclass lldb_private::Module so that the virtual "GetObjectFile()" and "GetSymbolVendor()" functions can be intercepted when the .o file contenst are later lazilly needed. Prior to this fix, when we first instantiated the "SymbolFileDWARFDebugMap" class, we would also make modules, object files and symbol files for every .o file in the debug map because we needed to fix up the sections in the .o files with information that is in the executable debug map. Now we lazily do this in the DebugMapModule::GetObjectFile() Cleaned up header includes a bit as well. llvm-svn: 162860
2012-08-29 21:13:06 +00:00
}
}
}
llvm_unreachable("this shouldn't happen");
<rdar://problem/11757916> Make breakpoint setting by file and line much more efficient by only looking for inlined breakpoint locations if we are setting a breakpoint in anything but a source implementation file. Implementing this complex for a many reasons. Turns out that parsing compile units lazily had some issues with respect to how we need to do things with DWARF in .o files. So the fixes in the checkin for this makes these changes: - Add a new setting called "target.inline-breakpoint-strategy" which can be set to "never", "always", or "headers". "never" will never try and set any inlined breakpoints (fastest). "always" always looks for inlined breakpoint locations (slowest, but most accurate). "headers", which is the default setting, will only look for inlined breakpoint locations if the breakpoint is set in what are consudered to be header files, which is realy defined as "not in an implementation source file". - modify the breakpoint setting by file and line to check the current "target.inline-breakpoint-strategy" setting and act accordingly - Modify compile units to be able to get their language and other info lazily. This allows us to create compile units from the debug map and not have to fill all of the details in, and then lazily discover this information as we go on debuggging. This is needed to avoid parsing all .o files when setting breakpoints in implementation only files (no inlines). Otherwise we would need to parse the .o file, the object file (mach-o in our case) and the symbol file (DWARF in the object file) just to see what the compile unit was. - modify the "SymbolFileDWARFDebugMap" to subclass lldb_private::Module so that the virtual "GetObjectFile()" and "GetSymbolVendor()" functions can be intercepted when the .o file contenst are later lazilly needed. Prior to this fix, when we first instantiated the "SymbolFileDWARFDebugMap" class, we would also make modules, object files and symbol files for every .o file in the debug map because we needed to fix up the sections in the .o files with information that is in the executable debug map. Now we lazily do this in the DebugMapModule::GetObjectFile() Cleaned up header includes a bit as well. llvm-svn: 162860
2012-08-29 21:13:06 +00:00
}
SymbolFileDWARFDebugMap::CompileUnitInfo *
SymbolFileDWARFDebugMap::GetCompileUnitInfo(SymbolFileDWARF *oso_dwarf) {
if (oso_dwarf) {
const uint32_t cu_count = GetNumCompileUnits();
for (uint32_t cu_idx = 0; cu_idx < cu_count; ++cu_idx) {
SymbolFileDWARF *oso_symfile =
GetSymbolFileByCompUnitInfo(&m_compile_unit_infos[cu_idx]);
if (oso_symfile == oso_dwarf) {
return &m_compile_unit_infos[cu_idx];
}
}
}
return nullptr;
}
<rdar://problem/11757916> Make breakpoint setting by file and line much more efficient by only looking for inlined breakpoint locations if we are setting a breakpoint in anything but a source implementation file. Implementing this complex for a many reasons. Turns out that parsing compile units lazily had some issues with respect to how we need to do things with DWARF in .o files. So the fixes in the checkin for this makes these changes: - Add a new setting called "target.inline-breakpoint-strategy" which can be set to "never", "always", or "headers". "never" will never try and set any inlined breakpoints (fastest). "always" always looks for inlined breakpoint locations (slowest, but most accurate). "headers", which is the default setting, will only look for inlined breakpoint locations if the breakpoint is set in what are consudered to be header files, which is realy defined as "not in an implementation source file". - modify the breakpoint setting by file and line to check the current "target.inline-breakpoint-strategy" setting and act accordingly - Modify compile units to be able to get their language and other info lazily. This allows us to create compile units from the debug map and not have to fill all of the details in, and then lazily discover this information as we go on debuggging. This is needed to avoid parsing all .o files when setting breakpoints in implementation only files (no inlines). Otherwise we would need to parse the .o file, the object file (mach-o in our case) and the symbol file (DWARF in the object file) just to see what the compile unit was. - modify the "SymbolFileDWARFDebugMap" to subclass lldb_private::Module so that the virtual "GetObjectFile()" and "GetSymbolVendor()" functions can be intercepted when the .o file contenst are later lazilly needed. Prior to this fix, when we first instantiated the "SymbolFileDWARFDebugMap" class, we would also make modules, object files and symbol files for every .o file in the debug map because we needed to fix up the sections in the .o files with information that is in the executable debug map. Now we lazily do this in the DebugMapModule::GetObjectFile() Cleaned up header includes a bit as well. llvm-svn: 162860
2012-08-29 21:13:06 +00:00
void SymbolFileDWARFDebugMap::SetCompileUnit(SymbolFileDWARF *oso_dwarf,
const CompUnitSP &cu_sp) {
if (oso_dwarf) {
const uint32_t cu_count = GetNumCompileUnits();
for (uint32_t cu_idx = 0; cu_idx < cu_count; ++cu_idx) {
SymbolFileDWARF *oso_symfile =
GetSymbolFileByCompUnitInfo(&m_compile_unit_infos[cu_idx]);
if (oso_symfile == oso_dwarf) {
if (m_compile_unit_infos[cu_idx].compile_unit_sp) {
assert(m_compile_unit_infos[cu_idx].compile_unit_sp.get() ==
cu_sp.get());
<rdar://problem/11757916> Make breakpoint setting by file and line much more efficient by only looking for inlined breakpoint locations if we are setting a breakpoint in anything but a source implementation file. Implementing this complex for a many reasons. Turns out that parsing compile units lazily had some issues with respect to how we need to do things with DWARF in .o files. So the fixes in the checkin for this makes these changes: - Add a new setting called "target.inline-breakpoint-strategy" which can be set to "never", "always", or "headers". "never" will never try and set any inlined breakpoints (fastest). "always" always looks for inlined breakpoint locations (slowest, but most accurate). "headers", which is the default setting, will only look for inlined breakpoint locations if the breakpoint is set in what are consudered to be header files, which is realy defined as "not in an implementation source file". - modify the breakpoint setting by file and line to check the current "target.inline-breakpoint-strategy" setting and act accordingly - Modify compile units to be able to get their language and other info lazily. This allows us to create compile units from the debug map and not have to fill all of the details in, and then lazily discover this information as we go on debuggging. This is needed to avoid parsing all .o files when setting breakpoints in implementation only files (no inlines). Otherwise we would need to parse the .o file, the object file (mach-o in our case) and the symbol file (DWARF in the object file) just to see what the compile unit was. - modify the "SymbolFileDWARFDebugMap" to subclass lldb_private::Module so that the virtual "GetObjectFile()" and "GetSymbolVendor()" functions can be intercepted when the .o file contenst are later lazilly needed. Prior to this fix, when we first instantiated the "SymbolFileDWARFDebugMap" class, we would also make modules, object files and symbol files for every .o file in the debug map because we needed to fix up the sections in the .o files with information that is in the executable debug map. Now we lazily do this in the DebugMapModule::GetObjectFile() Cleaned up header includes a bit as well. llvm-svn: 162860
2012-08-29 21:13:06 +00:00
} else {
m_compile_unit_infos[cu_idx].compile_unit_sp = cu_sp;
2019-07-23 09:24:02 +00:00
SetCompileUnitAtIndex(cu_idx, cu_sp);
}
}
}
}
}
Final bit of type system cleanup that abstracts declaration contexts into lldb_private::CompilerDeclContext and renames ClangType to CompilerType in many accessors and functions. Create a new "lldb_private::CompilerDeclContext" class that will replace all direct uses of "clang::DeclContext" when used in compiler agnostic code, yet still allow for conversion to clang::DeclContext subclasses by clang specific code. This completes the abstraction of type parsing by removing all "clang::" references from the SymbolFileDWARF. The new "lldb_private::CompilerDeclContext" class abstracts decl contexts found in compiler type systems so they can be used in internal API calls. The TypeSystem is required to support CompilerDeclContexts with new pure virtual functions that start with "DeclContext" in the member function names. Converted all code that used lldb_private::ClangNamespaceDecl over to use the new CompilerDeclContext class and removed the ClangNamespaceDecl.cpp and ClangNamespaceDecl.h files. Removed direct use of clang APIs from SBType and now use the abstract type systems to correctly explore types. Bulk renames for things that used to return a ClangASTType which is now CompilerType: "Type::GetClangFullType()" to "Type::GetFullCompilerType()" "Type::GetClangLayoutType()" to "Type::GetLayoutCompilerType()" "Type::GetClangForwardType()" to "Type::GetForwardCompilerType()" "Value::GetClangType()" to "Value::GetCompilerType()" "Value::SetClangType (const CompilerType &)" to "Value::SetCompilerType (const CompilerType &)" "ValueObject::GetClangType ()" to "ValueObject::GetCompilerType()" many more renames that are similar. llvm-svn: 245905
2015-08-24 23:46:31 +00:00
CompilerDeclContext
SymbolFileDWARFDebugMap::GetDeclContextForUID(lldb::user_id_t type_uid) {
const uint64_t oso_idx = GetOSOIndexFromUserID(type_uid);
SymbolFileDWARF *oso_dwarf = GetSymbolFileByOSOIndex(oso_idx);
if (oso_dwarf)
Final bit of type system cleanup that abstracts declaration contexts into lldb_private::CompilerDeclContext and renames ClangType to CompilerType in many accessors and functions. Create a new "lldb_private::CompilerDeclContext" class that will replace all direct uses of "clang::DeclContext" when used in compiler agnostic code, yet still allow for conversion to clang::DeclContext subclasses by clang specific code. This completes the abstraction of type parsing by removing all "clang::" references from the SymbolFileDWARF. The new "lldb_private::CompilerDeclContext" class abstracts decl contexts found in compiler type systems so they can be used in internal API calls. The TypeSystem is required to support CompilerDeclContexts with new pure virtual functions that start with "DeclContext" in the member function names. Converted all code that used lldb_private::ClangNamespaceDecl over to use the new CompilerDeclContext class and removed the ClangNamespaceDecl.cpp and ClangNamespaceDecl.h files. Removed direct use of clang APIs from SBType and now use the abstract type systems to correctly explore types. Bulk renames for things that used to return a ClangASTType which is now CompilerType: "Type::GetClangFullType()" to "Type::GetFullCompilerType()" "Type::GetClangLayoutType()" to "Type::GetLayoutCompilerType()" "Type::GetClangForwardType()" to "Type::GetForwardCompilerType()" "Value::GetClangType()" to "Value::GetCompilerType()" "Value::SetClangType (const CompilerType &)" to "Value::SetCompilerType (const CompilerType &)" "ValueObject::GetClangType ()" to "ValueObject::GetCompilerType()" many more renames that are similar. llvm-svn: 245905
2015-08-24 23:46:31 +00:00
return oso_dwarf->GetDeclContextForUID(type_uid);
return CompilerDeclContext();
}
Final bit of type system cleanup that abstracts declaration contexts into lldb_private::CompilerDeclContext and renames ClangType to CompilerType in many accessors and functions. Create a new "lldb_private::CompilerDeclContext" class that will replace all direct uses of "clang::DeclContext" when used in compiler agnostic code, yet still allow for conversion to clang::DeclContext subclasses by clang specific code. This completes the abstraction of type parsing by removing all "clang::" references from the SymbolFileDWARF. The new "lldb_private::CompilerDeclContext" class abstracts decl contexts found in compiler type systems so they can be used in internal API calls. The TypeSystem is required to support CompilerDeclContexts with new pure virtual functions that start with "DeclContext" in the member function names. Converted all code that used lldb_private::ClangNamespaceDecl over to use the new CompilerDeclContext class and removed the ClangNamespaceDecl.cpp and ClangNamespaceDecl.h files. Removed direct use of clang APIs from SBType and now use the abstract type systems to correctly explore types. Bulk renames for things that used to return a ClangASTType which is now CompilerType: "Type::GetClangFullType()" to "Type::GetFullCompilerType()" "Type::GetClangLayoutType()" to "Type::GetLayoutCompilerType()" "Type::GetClangForwardType()" to "Type::GetForwardCompilerType()" "Value::GetClangType()" to "Value::GetCompilerType()" "Value::SetClangType (const CompilerType &)" to "Value::SetCompilerType (const CompilerType &)" "ValueObject::GetClangType ()" to "ValueObject::GetCompilerType()" many more renames that are similar. llvm-svn: 245905
2015-08-24 23:46:31 +00:00
CompilerDeclContext
SymbolFileDWARFDebugMap::GetDeclContextContainingUID(lldb::user_id_t type_uid) {
const uint64_t oso_idx = GetOSOIndexFromUserID(type_uid);
SymbolFileDWARF *oso_dwarf = GetSymbolFileByOSOIndex(oso_idx);
if (oso_dwarf)
Final bit of type system cleanup that abstracts declaration contexts into lldb_private::CompilerDeclContext and renames ClangType to CompilerType in many accessors and functions. Create a new "lldb_private::CompilerDeclContext" class that will replace all direct uses of "clang::DeclContext" when used in compiler agnostic code, yet still allow for conversion to clang::DeclContext subclasses by clang specific code. This completes the abstraction of type parsing by removing all "clang::" references from the SymbolFileDWARF. The new "lldb_private::CompilerDeclContext" class abstracts decl contexts found in compiler type systems so they can be used in internal API calls. The TypeSystem is required to support CompilerDeclContexts with new pure virtual functions that start with "DeclContext" in the member function names. Converted all code that used lldb_private::ClangNamespaceDecl over to use the new CompilerDeclContext class and removed the ClangNamespaceDecl.cpp and ClangNamespaceDecl.h files. Removed direct use of clang APIs from SBType and now use the abstract type systems to correctly explore types. Bulk renames for things that used to return a ClangASTType which is now CompilerType: "Type::GetClangFullType()" to "Type::GetFullCompilerType()" "Type::GetClangLayoutType()" to "Type::GetLayoutCompilerType()" "Type::GetClangForwardType()" to "Type::GetForwardCompilerType()" "Value::GetClangType()" to "Value::GetCompilerType()" "Value::SetClangType (const CompilerType &)" to "Value::SetCompilerType (const CompilerType &)" "ValueObject::GetClangType ()" to "ValueObject::GetCompilerType()" many more renames that are similar. llvm-svn: 245905
2015-08-24 23:46:31 +00:00
return oso_dwarf->GetDeclContextContainingUID(type_uid);
return CompilerDeclContext();
}
void SymbolFileDWARFDebugMap::ParseDeclsForContext(
lldb_private::CompilerDeclContext decl_ctx) {
ForEachSymbolFile([&](SymbolFileDWARF *oso_dwarf) -> bool {
oso_dwarf->ParseDeclsForContext(decl_ctx);
return true; // Keep iterating
});
}
bool SymbolFileDWARFDebugMap::AddOSOFileRange(CompileUnitInfo *cu_info,
lldb::addr_t exe_file_addr,
lldb::addr_t exe_byte_size,
lldb::addr_t oso_file_addr,
lldb::addr_t oso_byte_size) {
const uint32_t debug_map_idx =
m_debug_map.FindEntryIndexThatContains(exe_file_addr);
if (debug_map_idx != UINT32_MAX) {
DebugMap::Entry *debug_map_entry =
m_debug_map.FindEntryThatContains(exe_file_addr);
debug_map_entry->data.SetOSOFileAddress(oso_file_addr);
addr_t range_size = std::min<addr_t>(exe_byte_size, oso_byte_size);
if (range_size == 0) {
range_size = std::max<addr_t>(exe_byte_size, oso_byte_size);
if (range_size == 0)
range_size = 1;
}
cu_info->file_range_map.Append(
FileRangeMap::Entry(oso_file_addr, range_size, exe_file_addr));
return true;
}
return false;
}
void SymbolFileDWARFDebugMap::FinalizeOSOFileRanges(CompileUnitInfo *cu_info) {
cu_info->file_range_map.Sort();
#if defined(DEBUG_OSO_DMAP)
const FileRangeMap &oso_file_range_map = cu_info->GetFileRangeMap(this);
const size_t n = oso_file_range_map.GetSize();
printf("SymbolFileDWARFDebugMap::FinalizeOSOFileRanges (cu_info = %p) %s\n",
cu_info, cu_info->oso_sp->module_sp->GetFileSpec().GetPath().c_str());
for (size_t i = 0; i < n; ++i) {
const FileRangeMap::Entry &entry = oso_file_range_map.GetEntryRef(i);
printf("oso [0x%16.16" PRIx64 " - 0x%16.16" PRIx64
") ==> exe [0x%16.16" PRIx64 " - 0x%16.16" PRIx64 ")\n",
entry.GetRangeBase(), entry.GetRangeEnd(), entry.data,
entry.data + entry.GetByteSize());
}
#endif
}
lldb::addr_t
SymbolFileDWARFDebugMap::LinkOSOFileAddress(SymbolFileDWARF *oso_symfile,
lldb::addr_t oso_file_addr) {
CompileUnitInfo *cu_info = GetCompileUnitInfo(oso_symfile);
if (cu_info) {
const FileRangeMap::Entry *oso_range_entry =
cu_info->GetFileRangeMap(this).FindEntryThatContains(oso_file_addr);
if (oso_range_entry) {
const DebugMap::Entry *debug_map_entry =
m_debug_map.FindEntryThatContains(oso_range_entry->data);
if (debug_map_entry) {
const lldb::addr_t offset =
oso_file_addr - oso_range_entry->GetRangeBase();
const lldb::addr_t exe_file_addr =
debug_map_entry->GetRangeBase() + offset;
return exe_file_addr;
}
}
}
return LLDB_INVALID_ADDRESS;
}
bool SymbolFileDWARFDebugMap::LinkOSOAddress(Address &addr) {
// Make sure this address hasn't been fixed already
Module *exe_module = GetObjectFile()->GetModule().get();
Module *addr_module = addr.GetModule().get();
if (addr_module == exe_module)
return true; // Address is already in terms of the main executable module
CompileUnitInfo *cu_info = GetCompileUnitInfo(
GetSymbolFileAsSymbolFileDWARF(addr_module->GetSymbolFile()));
if (cu_info) {
const lldb::addr_t oso_file_addr = addr.GetFileAddress();
const FileRangeMap::Entry *oso_range_entry =
cu_info->GetFileRangeMap(this).FindEntryThatContains(oso_file_addr);
if (oso_range_entry) {
const DebugMap::Entry *debug_map_entry =
m_debug_map.FindEntryThatContains(oso_range_entry->data);
if (debug_map_entry) {
const lldb::addr_t offset =
oso_file_addr - oso_range_entry->GetRangeBase();
const lldb::addr_t exe_file_addr =
debug_map_entry->GetRangeBase() + offset;
return exe_module->ResolveFileAddress(exe_file_addr, addr);
}
}
}
return true;
}
LineTable *SymbolFileDWARFDebugMap::LinkOSOLineTable(SymbolFileDWARF *oso_dwarf,
LineTable *line_table) {
CompileUnitInfo *cu_info = GetCompileUnitInfo(oso_dwarf);
if (cu_info)
return line_table->LinkLineTable(cu_info->GetFileRangeMap(this));
return nullptr;
}
size_t
SymbolFileDWARFDebugMap::AddOSOARanges(SymbolFileDWARF *dwarf2Data,
DWARFDebugAranges *debug_aranges) {
size_t num_line_entries_added = 0;
if (debug_aranges && dwarf2Data) {
CompileUnitInfo *compile_unit_info = GetCompileUnitInfo(dwarf2Data);
if (compile_unit_info) {
const FileRangeMap &file_range_map =
compile_unit_info->GetFileRangeMap(this);
for (size_t idx = 0; idx < file_range_map.GetSize(); idx++) {
const FileRangeMap::Entry *entry = file_range_map.GetEntryAtIndex(idx);
if (entry) {
debug_aranges->AppendRange(dwarf2Data->GetID(), entry->GetRangeBase(),
entry->GetRangeEnd());
num_line_entries_added++;
}
}
}
}
return num_line_entries_added;
}