2017-10-06 22:24:34 +02:00
|
|
|
#include "llvm/ADT/DenseMap.h"
|
2018-09-21 23:59:34 +02:00
|
|
|
#include "llvm/ADT/StringExtras.h"
|
|
|
|
#include "llvm/ADT/StringSet.h"
|
2017-10-06 22:24:34 +02:00
|
|
|
#include "llvm/DebugInfo/DIContext.h"
|
|
|
|
#include "llvm/DebugInfo/DWARF/DWARFContext.h"
|
|
|
|
#include "llvm/DebugInfo/DWARF/DWARFDebugLoc.h"
|
|
|
|
#include "llvm/Object/ObjectFile.h"
|
2019-09-20 11:25:11 +02:00
|
|
|
#include "llvm/Support/JSON.h"
|
2017-10-06 22:24:34 +02:00
|
|
|
|
|
|
|
#define DEBUG_TYPE "dwarfdump"
|
|
|
|
using namespace llvm;
|
|
|
|
using namespace object;
|
|
|
|
|
2019-09-10 12:37:28 +02:00
|
|
|
/// This represents the number of categories of debug location coverage being
|
|
|
|
/// calculated. The first category is the number of variables with 0% location
|
|
|
|
/// coverage, but the last category is the number of variables with 100%
|
|
|
|
/// location coverage.
|
|
|
|
constexpr int NumOfCoverageCategories = 12;
|
|
|
|
|
2017-10-06 22:24:34 +02:00
|
|
|
/// Holds statistics for one function (or other entity that has a PC range and
|
|
|
|
/// contains variables, such as a compile unit).
|
|
|
|
struct PerFunctionStats {
|
|
|
|
/// Number of inlined instances of this function.
|
|
|
|
unsigned NumFnInlined = 0;
|
2019-03-02 00:51:54 +01:00
|
|
|
/// Number of inlined instances that have abstract origins.
|
|
|
|
unsigned NumAbstractOrigins = 0;
|
|
|
|
/// Number of variables and parameters with location across all inlined
|
|
|
|
/// instances.
|
2017-10-06 22:24:34 +02:00
|
|
|
unsigned TotalVarWithLoc = 0;
|
|
|
|
/// Number of constants with location across all inlined instances.
|
|
|
|
unsigned ConstantMembers = 0;
|
2019-03-02 00:51:54 +01:00
|
|
|
/// List of all Variables and parameters in this function.
|
2018-09-21 23:59:34 +02:00
|
|
|
StringSet<> VarsInFunction;
|
2017-10-06 22:24:34 +02:00
|
|
|
/// Compile units also cover a PC range, but have this flag set to false.
|
|
|
|
bool IsFunction = false;
|
2019-02-08 01:51:33 +01:00
|
|
|
/// Verify function definition has PC addresses (for detecting when
|
|
|
|
/// a function has been inlined everywhere).
|
|
|
|
bool HasPCAddresses = false;
|
2019-03-02 00:51:54 +01:00
|
|
|
/// Function has source location information.
|
|
|
|
bool HasSourceLocation = false;
|
|
|
|
/// Number of function parameters.
|
|
|
|
unsigned NumParams = 0;
|
|
|
|
/// Number of function parameters with source location.
|
|
|
|
unsigned NumParamSourceLocations = 0;
|
|
|
|
/// Number of function parameters with type.
|
|
|
|
unsigned NumParamTypes = 0;
|
|
|
|
/// Number of function parameters with a DW_AT_location.
|
|
|
|
unsigned NumParamLocations = 0;
|
|
|
|
/// Number of variables.
|
|
|
|
unsigned NumVars = 0;
|
|
|
|
/// Number of variables with source location.
|
|
|
|
unsigned NumVarSourceLocations = 0;
|
2019-09-04 11:44:09 +02:00
|
|
|
/// Number of variables with type.
|
2019-03-02 00:51:54 +01:00
|
|
|
unsigned NumVarTypes = 0;
|
2019-09-04 11:44:09 +02:00
|
|
|
/// Number of variables with DW_AT_location.
|
2019-03-02 00:51:54 +01:00
|
|
|
unsigned NumVarLocations = 0;
|
2017-10-06 22:24:34 +02:00
|
|
|
};
|
|
|
|
|
2018-11-09 19:10:02 +01:00
|
|
|
/// Holds accumulated global statistics about DIEs.
|
2017-10-06 22:24:34 +02:00
|
|
|
struct GlobalStats {
|
|
|
|
/// Total number of PC range bytes covered by DW_AT_locations.
|
|
|
|
unsigned ScopeBytesCovered = 0;
|
|
|
|
/// Total number of PC range bytes in each variable's enclosing scope,
|
|
|
|
/// starting from the first definition of the variable.
|
|
|
|
unsigned ScopeBytesFromFirstDefinition = 0;
|
2019-09-10 12:37:28 +02:00
|
|
|
/// Total number of PC range bytes covered by DW_AT_locations with
|
|
|
|
/// the debug entry values (DW_OP_entry_value).
|
|
|
|
unsigned ScopeEntryValueBytesCovered = 0;
|
|
|
|
/// Total number of PC range bytes covered by DW_AT_locations of
|
|
|
|
/// formal parameters.
|
|
|
|
unsigned ParamScopeBytesCovered = 0;
|
|
|
|
/// Total number of PC range bytes in each variable's enclosing scope,
|
|
|
|
/// starting from the first definition of the variable (only for parameters).
|
|
|
|
unsigned ParamScopeBytesFromFirstDefinition = 0;
|
|
|
|
/// Total number of PC range bytes covered by DW_AT_locations with
|
|
|
|
/// the debug entry values (DW_OP_entry_value) (only for parameters).
|
|
|
|
unsigned ParamScopeEntryValueBytesCovered = 0;
|
|
|
|
/// Total number of PC range bytes covered by DW_AT_locations (only for local
|
|
|
|
/// variables).
|
|
|
|
unsigned VarScopeBytesCovered = 0;
|
|
|
|
/// Total number of PC range bytes in each variable's enclosing scope,
|
|
|
|
/// starting from the first definition of the variable (only for local
|
|
|
|
/// variables).
|
|
|
|
unsigned VarScopeBytesFromFirstDefinition = 0;
|
|
|
|
/// Total number of PC range bytes covered by DW_AT_locations with
|
|
|
|
/// the debug entry values (DW_OP_entry_value) (only for local variables).
|
|
|
|
unsigned VarScopeEntryValueBytesCovered = 0;
|
2019-07-31 18:51:28 +02:00
|
|
|
/// Total number of call site entries (DW_AT_call_file & DW_AT_call_line).
|
2018-10-05 22:37:17 +02:00
|
|
|
unsigned CallSiteEntries = 0;
|
2019-07-31 18:51:28 +02:00
|
|
|
/// Total number of call site DIEs (DW_TAG_call_site).
|
|
|
|
unsigned CallSiteDIEs = 0;
|
|
|
|
/// Total number of call site parameter DIEs (DW_TAG_call_site_parameter).
|
|
|
|
unsigned CallSiteParamDIEs = 0;
|
2018-11-09 19:10:02 +01:00
|
|
|
/// Total byte size of concrete functions. This byte size includes
|
|
|
|
/// inline functions contained in the concrete functions.
|
2019-09-20 11:25:11 +02:00
|
|
|
unsigned FunctionSize = 0;
|
2018-11-09 19:10:02 +01:00
|
|
|
/// Total byte size of inlined functions. This is the total number of bytes
|
|
|
|
/// for the top inline functions within concrete functions. This can help
|
|
|
|
/// tune the inline settings when compiling to match user expectations.
|
2019-09-20 11:25:11 +02:00
|
|
|
unsigned InlineFunctionSize = 0;
|
2017-10-06 22:24:34 +02:00
|
|
|
};
|
|
|
|
|
2019-09-10 12:37:28 +02:00
|
|
|
/// Holds accumulated debug location statistics about local variables and
|
|
|
|
/// formal parameters.
|
|
|
|
struct LocationStats {
|
|
|
|
/// Map the scope coverage decile to the number of variables in the decile.
|
|
|
|
/// The first element of the array (at the index zero) represents the number
|
|
|
|
/// of variables with the no debug location at all, but the last element
|
|
|
|
/// in the vector represents the number of fully covered variables within
|
|
|
|
/// its scope.
|
|
|
|
std::vector<unsigned> VarParamLocStats{
|
|
|
|
std::vector<unsigned>(NumOfCoverageCategories, 0)};
|
|
|
|
/// Map non debug entry values coverage.
|
|
|
|
std::vector<unsigned> VarParamNonEntryValLocStats{
|
|
|
|
std::vector<unsigned>(NumOfCoverageCategories, 0)};
|
|
|
|
/// The debug location statistics for formal parameters.
|
|
|
|
std::vector<unsigned> ParamLocStats{
|
|
|
|
std::vector<unsigned>(NumOfCoverageCategories, 0)};
|
|
|
|
/// Map non debug entry values coverage for formal parameters.
|
|
|
|
std::vector<unsigned> ParamNonEntryValLocStats{
|
|
|
|
std::vector<unsigned>(NumOfCoverageCategories, 0)};
|
|
|
|
/// The debug location statistics for local variables.
|
|
|
|
std::vector<unsigned> VarLocStats{
|
|
|
|
std::vector<unsigned>(NumOfCoverageCategories, 0)};
|
|
|
|
/// Map non debug entry values coverage for local variables.
|
|
|
|
std::vector<unsigned> VarNonEntryValLocStats{
|
|
|
|
std::vector<unsigned>(NumOfCoverageCategories, 0)};
|
|
|
|
/// Total number of local variables and function parameters processed.
|
|
|
|
unsigned NumVarParam = 0;
|
|
|
|
/// Total number of formal parameters processed.
|
|
|
|
unsigned NumParam = 0;
|
|
|
|
/// Total number of local variables processed.
|
|
|
|
unsigned NumVar = 0;
|
|
|
|
};
|
|
|
|
|
2017-10-06 22:24:34 +02:00
|
|
|
/// Extract the low pc from a Die.
|
|
|
|
static uint64_t getLowPC(DWARFDie Die) {
|
2018-06-21 00:56:37 +02:00
|
|
|
auto RangesOrError = Die.getAddressRanges();
|
|
|
|
DWARFAddressRangesVector Ranges;
|
|
|
|
if (RangesOrError)
|
|
|
|
Ranges = RangesOrError.get();
|
|
|
|
else
|
|
|
|
llvm::consumeError(RangesOrError.takeError());
|
|
|
|
if (Ranges.size())
|
|
|
|
return Ranges[0].LowPC;
|
2017-10-06 22:24:34 +02:00
|
|
|
return dwarf::toAddress(Die.find(dwarf::DW_AT_low_pc), 0);
|
|
|
|
}
|
|
|
|
|
2019-09-10 12:37:28 +02:00
|
|
|
/// Collect debug location statistics for one DIE.
|
|
|
|
static void collectLocStats(uint64_t BytesCovered, uint64_t BytesInScope,
|
|
|
|
std::vector<unsigned> &VarParamLocStats,
|
|
|
|
std::vector<unsigned> &ParamLocStats,
|
|
|
|
std::vector<unsigned> &VarLocStats, bool IsParam,
|
|
|
|
bool IsLocalVar) {
|
|
|
|
auto getCoverageBucket = [BytesCovered, BytesInScope]() -> unsigned {
|
|
|
|
unsigned LocBucket = 100 * (double)BytesCovered / BytesInScope;
|
|
|
|
if (LocBucket == 0) {
|
|
|
|
// No debug location at all for the variable.
|
|
|
|
return 0;
|
|
|
|
} else if (LocBucket == 100 || BytesCovered > BytesInScope) {
|
|
|
|
// Fully covered variable within its scope.
|
|
|
|
return NumOfCoverageCategories - 1;
|
|
|
|
} else {
|
|
|
|
// Get covered range (e.g. 20%-29%).
|
|
|
|
LocBucket /= 10;
|
|
|
|
return LocBucket + 1;
|
|
|
|
}
|
|
|
|
};
|
|
|
|
|
|
|
|
unsigned CoverageBucket = getCoverageBucket();
|
|
|
|
VarParamLocStats[CoverageBucket]++;
|
|
|
|
if (IsParam)
|
|
|
|
ParamLocStats[CoverageBucket]++;
|
|
|
|
else if (IsLocalVar)
|
|
|
|
VarLocStats[CoverageBucket]++;
|
|
|
|
}
|
|
|
|
|
2017-10-06 22:24:34 +02:00
|
|
|
/// Collect debug info quality metrics for one DIE.
|
2019-08-31 03:05:46 +02:00
|
|
|
static void collectStatsForDie(DWARFDie Die, uint64_t UnitLowPC, std::string FnPrefix,
|
2018-09-21 23:59:34 +02:00
|
|
|
std::string VarPrefix, uint64_t ScopeLowPC,
|
2019-03-02 00:51:54 +01:00
|
|
|
uint64_t BytesInScope, uint32_t InlineDepth,
|
2017-10-06 22:24:34 +02:00
|
|
|
StringMap<PerFunctionStats> &FnStatMap,
|
2019-09-10 12:37:28 +02:00
|
|
|
GlobalStats &GlobalStats,
|
|
|
|
LocationStats &LocStats) {
|
2017-10-06 22:24:34 +02:00
|
|
|
bool HasLoc = false;
|
2019-03-02 00:51:54 +01:00
|
|
|
bool HasSrcLoc = false;
|
|
|
|
bool HasType = false;
|
|
|
|
bool IsArtificial = false;
|
2017-10-06 22:24:34 +02:00
|
|
|
uint64_t BytesCovered = 0;
|
2019-09-10 12:37:28 +02:00
|
|
|
uint64_t BytesEntryValuesCovered = 0;
|
2017-10-06 22:24:34 +02:00
|
|
|
uint64_t OffsetToFirstDefinition = 0;
|
2019-09-10 12:37:28 +02:00
|
|
|
auto &FnStats = FnStatMap[FnPrefix];
|
|
|
|
bool IsParam = Die.getTag() == dwarf::DW_TAG_formal_parameter;
|
|
|
|
bool IsLocalVar = Die.getTag() == dwarf::DW_TAG_variable;
|
2018-09-21 23:59:34 +02:00
|
|
|
|
2019-07-31 18:51:28 +02:00
|
|
|
if (Die.getTag() == dwarf::DW_TAG_call_site ||
|
|
|
|
Die.getTag() == dwarf::DW_TAG_GNU_call_site) {
|
|
|
|
GlobalStats.CallSiteDIEs++;
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (Die.getTag() == dwarf::DW_TAG_call_site_parameter ||
|
|
|
|
Die.getTag() == dwarf::DW_TAG_GNU_call_site_parameter) {
|
|
|
|
GlobalStats.CallSiteParamDIEs++;
|
2018-10-05 22:37:17 +02:00
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
2019-09-10 12:37:28 +02:00
|
|
|
if (!IsParam && !IsLocalVar && Die.getTag() != dwarf::DW_TAG_member) {
|
2018-09-21 23:59:34 +02:00
|
|
|
// Not a variable or constant member.
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
2019-03-02 00:51:54 +01:00
|
|
|
if (Die.findRecursively(dwarf::DW_AT_decl_file) &&
|
|
|
|
Die.findRecursively(dwarf::DW_AT_decl_line))
|
|
|
|
HasSrcLoc = true;
|
|
|
|
|
|
|
|
if (Die.findRecursively(dwarf::DW_AT_type))
|
|
|
|
HasType = true;
|
|
|
|
|
|
|
|
if (Die.find(dwarf::DW_AT_artificial))
|
|
|
|
IsArtificial = true;
|
|
|
|
|
2019-09-10 12:37:28 +02:00
|
|
|
auto IsEntryValue = [&](ArrayRef<uint8_t> D) -> bool {
|
|
|
|
DWARFUnit *U = Die.getDwarfUnit();
|
|
|
|
DataExtractor Data(toStringRef(D),
|
|
|
|
Die.getDwarfUnit()->getContext().isLittleEndian(), 0);
|
|
|
|
DWARFExpression Expression(Data, U->getVersion(), U->getAddressByteSize());
|
|
|
|
// Consider the expression containing the DW_OP_entry_value as
|
|
|
|
// an entry value.
|
|
|
|
return llvm::any_of(Expression, [](DWARFExpression::Operation &Op) {
|
|
|
|
return Op.getCode() == dwarf::DW_OP_entry_value ||
|
|
|
|
Op.getCode() == dwarf::DW_OP_GNU_entry_value;
|
|
|
|
});
|
|
|
|
};
|
|
|
|
|
2017-10-06 22:24:34 +02:00
|
|
|
if (Die.find(dwarf::DW_AT_const_value)) {
|
|
|
|
// This catches constant members *and* variables.
|
|
|
|
HasLoc = true;
|
|
|
|
BytesCovered = BytesInScope;
|
2018-09-21 23:59:34 +02:00
|
|
|
} else {
|
|
|
|
if (Die.getTag() == dwarf::DW_TAG_member) {
|
|
|
|
// Non-const member.
|
|
|
|
return;
|
|
|
|
}
|
2017-10-06 22:24:34 +02:00
|
|
|
// Handle variables and function arguments.
|
|
|
|
auto FormValue = Die.find(dwarf::DW_AT_location);
|
|
|
|
HasLoc = FormValue.hasValue();
|
|
|
|
if (HasLoc) {
|
|
|
|
// Get PC coverage.
|
|
|
|
if (auto DebugLocOffset = FormValue->getAsSectionOffset()) {
|
|
|
|
auto *DebugLoc = Die.getDwarfUnit()->getContext().getDebugLoc();
|
DWARFDebugLoc(v4): Add an incremental parsing function
Summary:
This adds a visitLocationList function to the DWARF v4 location lists,
similar to what already exists for DWARF v5. It follows the approach
outlined in previous patches (D69672), where the parsed form is always
stored in the DWARF v5 format, which makes it easier for generic code to
be built on top of that. v4 location lists are "upgraded" during
parsing, and then this upgrade is undone while dumping.
Both "inline" and section-based dumping is rewritten to reuse the
existing "generic" location list dumper. This means that the output
format is consistent for all location lists (the only thing one needs to
implement is the function which prints the "raw" form of a location
list), and that debug_loc dumping correctly processes base address
selection entries, etc.
The previous existing debug_loc functionality (e.g.,
parseOneLocationList) is rewritten on top of the new API, but it is not
removed as there is still code which uses them. This will be done in
follow-up patches, after I build the API to access the "interpreted"
location lists in a generic way (as that is what those users really
want).
Reviewers: dblaikie, probinson, JDevlieghere, aprantl, SouraVX
Subscribers: hiraditya, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D69847
2019-11-05 14:46:54 +01:00
|
|
|
// TODO: This code does not handle DWARF5 nor DWARF4 base address
|
|
|
|
// selection entries. This should use a higher-level API which abstracts
|
|
|
|
// these away.
|
2017-10-06 22:24:34 +02:00
|
|
|
if (auto List = DebugLoc->getLocationListAtOffset(*DebugLocOffset)) {
|
DWARFDebugLoc(v4): Add an incremental parsing function
Summary:
This adds a visitLocationList function to the DWARF v4 location lists,
similar to what already exists for DWARF v5. It follows the approach
outlined in previous patches (D69672), where the parsed form is always
stored in the DWARF v5 format, which makes it easier for generic code to
be built on top of that. v4 location lists are "upgraded" during
parsing, and then this upgrade is undone while dumping.
Both "inline" and section-based dumping is rewritten to reuse the
existing "generic" location list dumper. This means that the output
format is consistent for all location lists (the only thing one needs to
implement is the function which prints the "raw" form of a location
list), and that debug_loc dumping correctly processes base address
selection entries, etc.
The previous existing debug_loc functionality (e.g.,
parseOneLocationList) is rewritten on top of the new API, but it is not
removed as there is still code which uses them. This will be done in
follow-up patches, after I build the API to access the "interpreted"
location lists in a generic way (as that is what those users really
want).
Reviewers: dblaikie, probinson, JDevlieghere, aprantl, SouraVX
Subscribers: hiraditya, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D69847
2019-11-05 14:46:54 +01:00
|
|
|
ArrayRef<DWARFLocationEntry> Entries = List->Entries;
|
|
|
|
// Ignore end-of-list entries
|
|
|
|
Entries = Entries.drop_back();
|
|
|
|
|
|
|
|
for (auto Entry : Entries) {
|
|
|
|
uint64_t BytesEntryCovered = Entry.Value1 - Entry.Value0;
|
2019-09-10 12:37:28 +02:00
|
|
|
BytesCovered += BytesEntryCovered;
|
|
|
|
if (IsEntryValue(Entry.Loc))
|
|
|
|
BytesEntryValuesCovered += BytesEntryCovered;
|
|
|
|
}
|
DWARFDebugLoc(v4): Add an incremental parsing function
Summary:
This adds a visitLocationList function to the DWARF v4 location lists,
similar to what already exists for DWARF v5. It follows the approach
outlined in previous patches (D69672), where the parsed form is always
stored in the DWARF v5 format, which makes it easier for generic code to
be built on top of that. v4 location lists are "upgraded" during
parsing, and then this upgrade is undone while dumping.
Both "inline" and section-based dumping is rewritten to reuse the
existing "generic" location list dumper. This means that the output
format is consistent for all location lists (the only thing one needs to
implement is the function which prints the "raw" form of a location
list), and that debug_loc dumping correctly processes base address
selection entries, etc.
The previous existing debug_loc functionality (e.g.,
parseOneLocationList) is rewritten on top of the new API, but it is not
removed as there is still code which uses them. This will be done in
follow-up patches, after I build the API to access the "interpreted"
location lists in a generic way (as that is what those users really
want).
Reviewers: dblaikie, probinson, JDevlieghere, aprantl, SouraVX
Subscribers: hiraditya, llvm-commits
Tags: #llvm
Differential Revision: https://reviews.llvm.org/D69847
2019-11-05 14:46:54 +01:00
|
|
|
if (Entries.size()) {
|
|
|
|
uint64_t FirstDef = Entries[0].Value0;
|
2019-08-31 03:05:46 +02:00
|
|
|
uint64_t UnitOfs = UnitLowPC;
|
2017-10-06 22:24:34 +02:00
|
|
|
// Ranges sometimes start before the lexical scope.
|
|
|
|
if (UnitOfs + FirstDef >= ScopeLowPC)
|
|
|
|
OffsetToFirstDefinition = UnitOfs + FirstDef - ScopeLowPC;
|
|
|
|
// Or even after it. Count that as a failure.
|
|
|
|
if (OffsetToFirstDefinition > BytesInScope)
|
|
|
|
OffsetToFirstDefinition = 0;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
assert(BytesInScope);
|
|
|
|
} else {
|
|
|
|
// Assume the entire range is covered by a single location.
|
|
|
|
BytesCovered = BytesInScope;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2019-09-10 12:37:28 +02:00
|
|
|
// Calculate the debug location statistics.
|
|
|
|
if (BytesInScope) {
|
|
|
|
LocStats.NumVarParam++;
|
|
|
|
if (IsParam)
|
|
|
|
LocStats.NumParam++;
|
|
|
|
else if (IsLocalVar)
|
|
|
|
LocStats.NumVar++;
|
|
|
|
|
|
|
|
collectLocStats(BytesCovered, BytesInScope, LocStats.VarParamLocStats,
|
|
|
|
LocStats.ParamLocStats, LocStats.VarLocStats, IsParam,
|
|
|
|
IsLocalVar);
|
|
|
|
// Non debug entry values coverage statistics.
|
|
|
|
collectLocStats(BytesCovered - BytesEntryValuesCovered, BytesInScope,
|
|
|
|
LocStats.VarParamNonEntryValLocStats,
|
|
|
|
LocStats.ParamNonEntryValLocStats,
|
|
|
|
LocStats.VarNonEntryValLocStats, IsParam, IsLocalVar);
|
|
|
|
}
|
|
|
|
|
2017-10-06 22:24:34 +02:00
|
|
|
// Collect PC range coverage data.
|
|
|
|
if (DWARFDie D =
|
|
|
|
Die.getAttributeValueAsReferencedDie(dwarf::DW_AT_abstract_origin))
|
|
|
|
Die = D;
|
2018-09-21 23:59:34 +02:00
|
|
|
// By using the variable name + the path through the lexical block tree, the
|
|
|
|
// keys are consistent across duplicate abstract origins in different CUs.
|
|
|
|
std::string VarName = StringRef(Die.getName(DINameKind::ShortName));
|
2019-03-02 00:51:54 +01:00
|
|
|
FnStats.VarsInFunction.insert(VarPrefix + VarName);
|
2017-10-06 22:24:34 +02:00
|
|
|
if (BytesInScope) {
|
|
|
|
FnStats.TotalVarWithLoc += (unsigned)HasLoc;
|
|
|
|
// Adjust for the fact the variables often start their lifetime in the
|
|
|
|
// middle of the scope.
|
|
|
|
BytesInScope -= OffsetToFirstDefinition;
|
|
|
|
// Turns out we have a lot of ranges that extend past the lexical scope.
|
|
|
|
GlobalStats.ScopeBytesCovered += std::min(BytesInScope, BytesCovered);
|
|
|
|
GlobalStats.ScopeBytesFromFirstDefinition += BytesInScope;
|
2019-09-10 12:37:28 +02:00
|
|
|
GlobalStats.ScopeEntryValueBytesCovered += BytesEntryValuesCovered;
|
|
|
|
if (IsParam) {
|
|
|
|
GlobalStats.ParamScopeBytesCovered +=
|
|
|
|
std::min(BytesInScope, BytesCovered);
|
|
|
|
GlobalStats.ParamScopeBytesFromFirstDefinition += BytesInScope;
|
|
|
|
GlobalStats.ParamScopeEntryValueBytesCovered += BytesEntryValuesCovered;
|
|
|
|
} else if (IsLocalVar) {
|
|
|
|
GlobalStats.VarScopeBytesCovered += std::min(BytesInScope, BytesCovered);
|
|
|
|
GlobalStats.VarScopeBytesFromFirstDefinition += BytesInScope;
|
|
|
|
GlobalStats.VarScopeEntryValueBytesCovered += BytesEntryValuesCovered;
|
|
|
|
}
|
2017-10-06 22:24:34 +02:00
|
|
|
assert(GlobalStats.ScopeBytesCovered <=
|
|
|
|
GlobalStats.ScopeBytesFromFirstDefinition);
|
2019-02-08 01:51:33 +01:00
|
|
|
} else if (Die.getTag() == dwarf::DW_TAG_member) {
|
2017-10-06 22:24:34 +02:00
|
|
|
FnStats.ConstantMembers++;
|
2019-02-08 01:51:33 +01:00
|
|
|
} else {
|
|
|
|
FnStats.TotalVarWithLoc += (unsigned)HasLoc;
|
2017-10-06 22:24:34 +02:00
|
|
|
}
|
2019-03-02 00:51:54 +01:00
|
|
|
if (!IsArtificial) {
|
2019-09-10 12:37:28 +02:00
|
|
|
if (IsParam) {
|
2019-03-02 00:51:54 +01:00
|
|
|
FnStats.NumParams++;
|
|
|
|
if (HasType)
|
|
|
|
FnStats.NumParamTypes++;
|
|
|
|
if (HasSrcLoc)
|
|
|
|
FnStats.NumParamSourceLocations++;
|
|
|
|
if (HasLoc)
|
|
|
|
FnStats.NumParamLocations++;
|
2019-09-10 12:37:28 +02:00
|
|
|
} else if (IsLocalVar) {
|
2019-03-02 00:51:54 +01:00
|
|
|
FnStats.NumVars++;
|
|
|
|
if (HasType)
|
|
|
|
FnStats.NumVarTypes++;
|
|
|
|
if (HasSrcLoc)
|
|
|
|
FnStats.NumVarSourceLocations++;
|
|
|
|
if (HasLoc)
|
|
|
|
FnStats.NumVarLocations++;
|
|
|
|
}
|
|
|
|
}
|
2017-10-06 22:24:34 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
/// Recursively collect debug info quality metrics.
|
2019-08-31 03:05:46 +02:00
|
|
|
static void collectStatsRecursive(DWARFDie Die, uint64_t UnitLowPC, std::string FnPrefix,
|
2018-09-21 23:59:34 +02:00
|
|
|
std::string VarPrefix, uint64_t ScopeLowPC,
|
2019-03-02 00:51:54 +01:00
|
|
|
uint64_t BytesInScope, uint32_t InlineDepth,
|
2017-10-06 22:24:34 +02:00
|
|
|
StringMap<PerFunctionStats> &FnStatMap,
|
2019-09-10 12:37:28 +02:00
|
|
|
GlobalStats &GlobalStats,
|
|
|
|
LocationStats &LocStats) {
|
2017-10-06 22:24:34 +02:00
|
|
|
// Handle any kind of lexical scope.
|
2018-11-09 19:10:02 +01:00
|
|
|
const dwarf::Tag Tag = Die.getTag();
|
|
|
|
const bool IsFunction = Tag == dwarf::DW_TAG_subprogram;
|
|
|
|
const bool IsBlock = Tag == dwarf::DW_TAG_lexical_block;
|
|
|
|
const bool IsInlinedFunction = Tag == dwarf::DW_TAG_inlined_subroutine;
|
|
|
|
if (IsFunction || IsInlinedFunction || IsBlock) {
|
2018-09-21 23:59:34 +02:00
|
|
|
|
|
|
|
// Reset VarPrefix when entering a new function.
|
|
|
|
if (Die.getTag() == dwarf::DW_TAG_subprogram ||
|
|
|
|
Die.getTag() == dwarf::DW_TAG_inlined_subroutine)
|
|
|
|
VarPrefix = "v";
|
2018-11-09 19:10:02 +01:00
|
|
|
|
2017-10-06 22:24:34 +02:00
|
|
|
// Ignore forward declarations.
|
|
|
|
if (Die.find(dwarf::DW_AT_declaration))
|
|
|
|
return;
|
|
|
|
|
2019-03-02 00:51:54 +01:00
|
|
|
// Check for call sites.
|
|
|
|
if (Die.find(dwarf::DW_AT_call_file) && Die.find(dwarf::DW_AT_call_line))
|
|
|
|
GlobalStats.CallSiteEntries++;
|
|
|
|
|
2019-02-08 01:51:33 +01:00
|
|
|
// PC Ranges.
|
|
|
|
auto RangesOrError = Die.getAddressRanges();
|
|
|
|
if (!RangesOrError) {
|
|
|
|
llvm::consumeError(RangesOrError.takeError());
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
auto Ranges = RangesOrError.get();
|
|
|
|
uint64_t BytesInThisScope = 0;
|
|
|
|
for (auto Range : Ranges)
|
|
|
|
BytesInThisScope += Range.HighPC - Range.LowPC;
|
|
|
|
ScopeLowPC = getLowPC(Die);
|
|
|
|
|
2017-10-06 22:24:34 +02:00
|
|
|
// Count the function.
|
2018-11-09 19:10:02 +01:00
|
|
|
if (!IsBlock) {
|
2017-10-06 22:24:34 +02:00
|
|
|
StringRef Name = Die.getName(DINameKind::LinkageName);
|
|
|
|
if (Name.empty())
|
|
|
|
Name = Die.getName(DINameKind::ShortName);
|
2018-09-21 23:59:34 +02:00
|
|
|
FnPrefix = Name;
|
2017-10-06 22:24:34 +02:00
|
|
|
// Skip over abstract origins.
|
|
|
|
if (Die.find(dwarf::DW_AT_inline))
|
|
|
|
return;
|
|
|
|
// We've seen an (inlined) instance of this function.
|
|
|
|
auto &FnStats = FnStatMap[Name];
|
2019-03-02 00:51:54 +01:00
|
|
|
if (IsInlinedFunction) {
|
2019-02-08 01:51:33 +01:00
|
|
|
FnStats.NumFnInlined++;
|
2019-03-02 00:51:54 +01:00
|
|
|
if (Die.findRecursively(dwarf::DW_AT_abstract_origin))
|
|
|
|
FnStats.NumAbstractOrigins++;
|
|
|
|
}
|
2017-10-06 22:24:34 +02:00
|
|
|
FnStats.IsFunction = true;
|
2019-02-08 01:51:33 +01:00
|
|
|
if (BytesInThisScope && !IsInlinedFunction)
|
|
|
|
FnStats.HasPCAddresses = true;
|
2019-03-02 00:51:54 +01:00
|
|
|
std::string FnName = StringRef(Die.getName(DINameKind::ShortName));
|
|
|
|
if (Die.findRecursively(dwarf::DW_AT_decl_file) &&
|
|
|
|
Die.findRecursively(dwarf::DW_AT_decl_line))
|
|
|
|
FnStats.HasSourceLocation = true;
|
2017-10-06 22:24:34 +02:00
|
|
|
}
|
|
|
|
|
2018-11-09 19:10:02 +01:00
|
|
|
if (BytesInThisScope) {
|
2017-10-06 22:24:34 +02:00
|
|
|
BytesInScope = BytesInThisScope;
|
2018-11-09 19:10:02 +01:00
|
|
|
if (IsFunction)
|
|
|
|
GlobalStats.FunctionSize += BytesInThisScope;
|
|
|
|
else if (IsInlinedFunction && InlineDepth == 0)
|
|
|
|
GlobalStats.InlineFunctionSize += BytesInThisScope;
|
|
|
|
}
|
2017-10-06 22:24:34 +02:00
|
|
|
} else {
|
|
|
|
// Not a scope, visit the Die itself. It could be a variable.
|
2019-08-31 03:05:46 +02:00
|
|
|
collectStatsForDie(Die, UnitLowPC, FnPrefix, VarPrefix, ScopeLowPC, BytesInScope,
|
2019-09-10 12:37:28 +02:00
|
|
|
InlineDepth, FnStatMap, GlobalStats, LocStats);
|
2017-10-06 22:24:34 +02:00
|
|
|
}
|
|
|
|
|
2018-11-09 19:10:02 +01:00
|
|
|
// Set InlineDepth correctly for child recursion
|
|
|
|
if (IsFunction)
|
|
|
|
InlineDepth = 0;
|
|
|
|
else if (IsInlinedFunction)
|
|
|
|
++InlineDepth;
|
|
|
|
|
2017-10-06 22:24:34 +02:00
|
|
|
// Traverse children.
|
2018-09-21 23:59:34 +02:00
|
|
|
unsigned LexicalBlockIndex = 0;
|
2017-10-06 22:24:34 +02:00
|
|
|
DWARFDie Child = Die.getFirstChild();
|
|
|
|
while (Child) {
|
2018-09-21 23:59:34 +02:00
|
|
|
std::string ChildVarPrefix = VarPrefix;
|
|
|
|
if (Child.getTag() == dwarf::DW_TAG_lexical_block)
|
|
|
|
ChildVarPrefix += toHex(LexicalBlockIndex++) + '.';
|
|
|
|
|
2019-08-31 03:05:46 +02:00
|
|
|
collectStatsRecursive(Child, UnitLowPC, FnPrefix, ChildVarPrefix, ScopeLowPC,
|
2019-09-10 12:37:28 +02:00
|
|
|
BytesInScope, InlineDepth, FnStatMap, GlobalStats,
|
|
|
|
LocStats);
|
2017-10-06 22:24:34 +02:00
|
|
|
Child = Child.getSibling();
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
/// Print machine-readable output.
|
|
|
|
/// The machine-readable format is single-line JSON output.
|
|
|
|
/// \{
|
2019-09-20 11:25:11 +02:00
|
|
|
static void printDatum(raw_ostream &OS, const char *Key, json::Value Value) {
|
2017-10-06 22:24:34 +02:00
|
|
|
OS << ",\"" << Key << "\":" << Value;
|
2018-05-14 14:53:11 +02:00
|
|
|
LLVM_DEBUG(llvm::dbgs() << Key << ": " << Value << '\n');
|
2017-10-06 22:24:34 +02:00
|
|
|
}
|
2019-09-10 12:37:28 +02:00
|
|
|
static void printLocationStats(raw_ostream &OS,
|
|
|
|
const char *Key,
|
|
|
|
std::vector<unsigned> &LocationStats) {
|
|
|
|
OS << ",\"" << Key << " with 0% of its scope covered\":"
|
|
|
|
<< LocationStats[0];
|
|
|
|
LLVM_DEBUG(llvm::dbgs() << Key << " with 0% of its scope covered: "
|
|
|
|
<< LocationStats[0] << '\n');
|
|
|
|
OS << ",\"" << Key << " with 1-9% of its scope covered\":"
|
|
|
|
<< LocationStats[1];
|
|
|
|
LLVM_DEBUG(llvm::dbgs() << Key << " with 1-9% of its scope covered: "
|
|
|
|
<< LocationStats[1] << '\n');
|
|
|
|
for (unsigned i = 2; i < NumOfCoverageCategories - 1; ++i) {
|
|
|
|
OS << ",\"" << Key << " with " << (i - 1) * 10 << "-" << i * 10 - 1
|
|
|
|
<< "% of its scope covered\":" << LocationStats[i];
|
|
|
|
LLVM_DEBUG(llvm::dbgs()
|
|
|
|
<< Key << " with " << (i - 1) * 10 << "-" << i * 10 - 1
|
|
|
|
<< "% of its scope covered: " << LocationStats[i]);
|
|
|
|
}
|
|
|
|
OS << ",\"" << Key << " with 100% of its scope covered\":"
|
|
|
|
<< LocationStats[NumOfCoverageCategories - 1];
|
|
|
|
LLVM_DEBUG(llvm::dbgs() << Key << " with 100% of its scope covered: "
|
|
|
|
<< LocationStats[NumOfCoverageCategories - 1]);
|
|
|
|
}
|
2017-10-06 22:24:34 +02:00
|
|
|
/// \}
|
|
|
|
|
|
|
|
/// Collect debug info quality metrics for an entire DIContext.
|
|
|
|
///
|
|
|
|
/// Do the impossible and reduce the quality of the debug info down to a few
|
|
|
|
/// numbers. The idea is to condense the data into numbers that can be tracked
|
|
|
|
/// over time to identify trends in newer compiler versions and gauge the effect
|
|
|
|
/// of particular optimizations. The raw numbers themselves are not particularly
|
|
|
|
/// useful, only the delta between compiling the same program with different
|
|
|
|
/// compilers is.
|
|
|
|
bool collectStatsForObjectFile(ObjectFile &Obj, DWARFContext &DICtx,
|
|
|
|
Twine Filename, raw_ostream &OS) {
|
|
|
|
StringRef FormatName = Obj.getFileFormatName();
|
|
|
|
GlobalStats GlobalStats;
|
2019-09-10 12:37:28 +02:00
|
|
|
LocationStats LocStats;
|
2017-10-06 22:24:34 +02:00
|
|
|
StringMap<PerFunctionStats> Statistics;
|
|
|
|
for (const auto &CU : static_cast<DWARFContext *>(&DICtx)->compile_units())
|
2019-05-09 23:53:33 +02:00
|
|
|
if (DWARFDie CUDie = CU->getNonSkeletonUnitDIE(false))
|
2019-09-10 12:37:28 +02:00
|
|
|
collectStatsRecursive(CUDie, getLowPC(CUDie), "/", "g", 0, 0, 0,
|
|
|
|
Statistics, GlobalStats, LocStats);
|
2017-10-06 22:24:34 +02:00
|
|
|
|
|
|
|
/// The version number should be increased every time the algorithm is changed
|
|
|
|
/// (including bug fixes). New metrics may be added without increasing the
|
|
|
|
/// version.
|
2019-03-02 00:51:54 +01:00
|
|
|
unsigned Version = 3;
|
|
|
|
unsigned VarParamTotal = 0;
|
|
|
|
unsigned VarParamUnique = 0;
|
|
|
|
unsigned VarParamWithLoc = 0;
|
2017-10-06 22:24:34 +02:00
|
|
|
unsigned NumFunctions = 0;
|
|
|
|
unsigned NumInlinedFunctions = 0;
|
2019-03-02 00:51:54 +01:00
|
|
|
unsigned NumFuncsWithSrcLoc = 0;
|
|
|
|
unsigned NumAbstractOrigins = 0;
|
|
|
|
unsigned ParamTotal = 0;
|
|
|
|
unsigned ParamWithType = 0;
|
|
|
|
unsigned ParamWithLoc = 0;
|
|
|
|
unsigned ParamWithSrcLoc = 0;
|
|
|
|
unsigned VarTotal = 0;
|
|
|
|
unsigned VarWithType = 0;
|
|
|
|
unsigned VarWithSrcLoc = 0;
|
|
|
|
unsigned VarWithLoc = 0;
|
2017-10-06 22:24:34 +02:00
|
|
|
for (auto &Entry : Statistics) {
|
|
|
|
PerFunctionStats &Stats = Entry.getValue();
|
|
|
|
unsigned TotalVars = Stats.VarsInFunction.size() * Stats.NumFnInlined;
|
2019-02-08 01:51:33 +01:00
|
|
|
// Count variables in concrete out-of-line functions and in global scope.
|
|
|
|
if (Stats.HasPCAddresses || !Stats.IsFunction)
|
|
|
|
TotalVars += Stats.VarsInFunction.size();
|
2017-10-06 22:24:34 +02:00
|
|
|
unsigned Constants = Stats.ConstantMembers;
|
2019-03-02 00:51:54 +01:00
|
|
|
VarParamWithLoc += Stats.TotalVarWithLoc + Constants;
|
|
|
|
VarParamTotal += TotalVars;
|
|
|
|
VarParamUnique += Stats.VarsInFunction.size();
|
|
|
|
LLVM_DEBUG(for (auto &V
|
|
|
|
: Stats.VarsInFunction) llvm::dbgs()
|
2018-09-21 23:59:34 +02:00
|
|
|
<< Entry.getKey() << ": " << V.getKey() << "\n");
|
2017-10-06 22:24:34 +02:00
|
|
|
NumFunctions += Stats.IsFunction;
|
2019-03-02 00:51:54 +01:00
|
|
|
NumFuncsWithSrcLoc += Stats.HasSourceLocation;
|
2017-10-06 22:24:34 +02:00
|
|
|
NumInlinedFunctions += Stats.IsFunction * Stats.NumFnInlined;
|
2019-03-02 00:51:54 +01:00
|
|
|
NumAbstractOrigins += Stats.IsFunction * Stats.NumAbstractOrigins;
|
|
|
|
ParamTotal += Stats.NumParams;
|
|
|
|
ParamWithType += Stats.NumParamTypes;
|
|
|
|
ParamWithLoc += Stats.NumParamLocations;
|
|
|
|
ParamWithSrcLoc += Stats.NumParamSourceLocations;
|
|
|
|
VarTotal += Stats.NumVars;
|
|
|
|
VarWithType += Stats.NumVarTypes;
|
|
|
|
VarWithLoc += Stats.NumVarLocations;
|
|
|
|
VarWithSrcLoc += Stats.NumVarSourceLocations;
|
2017-10-06 22:24:34 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
// Print summary.
|
|
|
|
OS.SetBufferSize(1024);
|
2018-10-05 19:41:30 +02:00
|
|
|
OS << "{\"version\":" << Version;
|
2018-05-14 14:53:11 +02:00
|
|
|
LLVM_DEBUG(llvm::dbgs() << "Variable location quality metrics\n";
|
|
|
|
llvm::dbgs() << "---------------------------------\n");
|
2017-10-06 22:24:34 +02:00
|
|
|
printDatum(OS, "file", Filename.str());
|
|
|
|
printDatum(OS, "format", FormatName);
|
|
|
|
printDatum(OS, "source functions", NumFunctions);
|
2019-03-02 00:51:54 +01:00
|
|
|
printDatum(OS, "source functions with location", NumFuncsWithSrcLoc);
|
2017-10-06 22:24:34 +02:00
|
|
|
printDatum(OS, "inlined functions", NumInlinedFunctions);
|
2019-03-02 00:51:54 +01:00
|
|
|
printDatum(OS, "inlined funcs with abstract origins", NumAbstractOrigins);
|
|
|
|
printDatum(OS, "unique source variables", VarParamUnique);
|
|
|
|
printDatum(OS, "source variables", VarParamTotal);
|
|
|
|
printDatum(OS, "variables with location", VarParamWithLoc);
|
2018-10-05 22:37:17 +02:00
|
|
|
printDatum(OS, "call site entries", GlobalStats.CallSiteEntries);
|
2019-07-31 18:51:28 +02:00
|
|
|
printDatum(OS, "call site DIEs", GlobalStats.CallSiteDIEs);
|
|
|
|
printDatum(OS, "call site parameter DIEs", GlobalStats.CallSiteParamDIEs);
|
2017-10-06 22:24:34 +02:00
|
|
|
printDatum(OS, "scope bytes total",
|
|
|
|
GlobalStats.ScopeBytesFromFirstDefinition);
|
|
|
|
printDatum(OS, "scope bytes covered", GlobalStats.ScopeBytesCovered);
|
2019-09-10 12:37:28 +02:00
|
|
|
printDatum(OS, "entry value scope bytes covered",
|
|
|
|
GlobalStats.ScopeEntryValueBytesCovered);
|
|
|
|
printDatum(OS, "formal params scope bytes total",
|
|
|
|
GlobalStats.ParamScopeBytesFromFirstDefinition);
|
|
|
|
printDatum(OS, "formal params scope bytes covered",
|
|
|
|
GlobalStats.ParamScopeBytesCovered);
|
|
|
|
printDatum(OS, "formal params entry value scope bytes covered",
|
|
|
|
GlobalStats.ParamScopeEntryValueBytesCovered);
|
|
|
|
printDatum(OS, "vars scope bytes total",
|
|
|
|
GlobalStats.VarScopeBytesFromFirstDefinition);
|
|
|
|
printDatum(OS, "vars scope bytes covered", GlobalStats.VarScopeBytesCovered);
|
|
|
|
printDatum(OS, "vars entry value scope bytes covered",
|
|
|
|
GlobalStats.VarScopeEntryValueBytesCovered);
|
2018-11-09 19:10:02 +01:00
|
|
|
printDatum(OS, "total function size", GlobalStats.FunctionSize);
|
|
|
|
printDatum(OS, "total inlined function size", GlobalStats.InlineFunctionSize);
|
2019-03-02 00:51:54 +01:00
|
|
|
printDatum(OS, "total formal params", ParamTotal);
|
|
|
|
printDatum(OS, "formal params with source location", ParamWithSrcLoc);
|
|
|
|
printDatum(OS, "formal params with type", ParamWithType);
|
|
|
|
printDatum(OS, "formal params with binary location", ParamWithLoc);
|
|
|
|
printDatum(OS, "total vars", VarTotal);
|
|
|
|
printDatum(OS, "vars with source location", VarWithSrcLoc);
|
|
|
|
printDatum(OS, "vars with type", VarWithType);
|
|
|
|
printDatum(OS, "vars with binary location", VarWithLoc);
|
2019-09-10 12:37:28 +02:00
|
|
|
printDatum(OS, "total variables procesed by location statistics",
|
|
|
|
LocStats.NumVarParam);
|
|
|
|
printLocationStats(OS, "variables", LocStats.VarParamLocStats);
|
|
|
|
printLocationStats(OS, "variables (excluding the debug entry values)",
|
|
|
|
LocStats.VarParamNonEntryValLocStats);
|
|
|
|
printDatum(OS, "total params procesed by location statistics",
|
|
|
|
LocStats.NumParam);
|
|
|
|
printLocationStats(OS, "params", LocStats.ParamLocStats);
|
|
|
|
printLocationStats(OS, "params (excluding the debug entry values)",
|
|
|
|
LocStats.ParamNonEntryValLocStats);
|
|
|
|
printDatum(OS, "total vars procesed by location statistics", LocStats.NumVar);
|
|
|
|
printLocationStats(OS, "vars", LocStats.VarLocStats);
|
|
|
|
printLocationStats(OS, "vars (excluding the debug entry values)",
|
2019-10-02 15:24:45 +02:00
|
|
|
LocStats.VarNonEntryValLocStats);
|
2017-10-06 22:24:34 +02:00
|
|
|
OS << "}\n";
|
2018-05-14 14:53:11 +02:00
|
|
|
LLVM_DEBUG(
|
2017-10-06 22:24:34 +02:00
|
|
|
llvm::dbgs() << "Total Availability: "
|
2019-03-02 00:51:54 +01:00
|
|
|
<< (int)std::round((VarParamWithLoc * 100.0) / VarParamTotal)
|
|
|
|
<< "%\n";
|
2017-10-06 22:24:34 +02:00
|
|
|
llvm::dbgs() << "PC Ranges covered: "
|
|
|
|
<< (int)std::round((GlobalStats.ScopeBytesCovered * 100.0) /
|
|
|
|
GlobalStats.ScopeBytesFromFirstDefinition)
|
|
|
|
<< "%\n");
|
|
|
|
return true;
|
|
|
|
}
|