[llvm-pdbdump] Display padding bytes on record layout

When dumping classes, show where padding occurs, and at the end of the
class print statistics about how many bytes total of padding exist in a
class.

Since PDB doesn't specifically contain information about padding, we have
to mimic this by sort of reversing a small portion of the record layout
algorithm (e.g. looking at offsets and sizes and trying to determine
whether something is part of the same field or a new field).

Differential Revision: https://reviews.llvm.org/D31800

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@299869 91177308-0d34-0410-b5e6-96231b3b80d8

diff --git a/test/DebugInfo/PDB/DIA/pdbdump-symbol-format.test b/test/DebugInfo/PDB/DIA/pdbdump-symbol-format.test
index 697a723..60a1953 100644 (file)
--- a/test/DebugInfo/PDB/DIA/pdbdump-symbol-format.test
+++ b/test/DebugInfo/PDB/DIA/pdbdump-symbol-format.test
@@ -40,29 +40,31 @@
 
 ; TYPES_2: Classes
 ; TYPES_2: struct MemberTest {
-; TYPES_2: data +0x00 MemberTest::NestedEnum m_nested_enum
-; TYPES_2: data +0x04 int m_typedef
-; TYPES_2: data +0x08 bool m_bool
-; TYPES_2: data +0x09 char m_char
-; TYPES_2: data +0x0a wchar_t m_wchar_t
-; TYPES_2: data +0x0c int m_int
-; TYPES_2: data +0x10 unsigned int m_unsigned
-; TYPES_2: data +0x14 long m_long
-; TYPES_2: data +0x18 unsigned long m_unsigned_long
-; TYPES_2: data +0x20 __int64 m_int64
-; TYPES_2: data +0x28 unsigned __int64 m_unsigned_int64
-; TYPES_2: data +0x30 float m_float
-; TYPES_2: data +0x38 double m_double
-; TYPES_2: data +0x40 void (__cdecl * m_pfn_2_args)(int, double)
-; TYPES_2: data +0x44 int m_multidimensional_array[2][3]
+; TYPES_2: data +0x00 [sizeof=4] MemberTest::NestedEnum m_nested_enum
+; TYPES_2: data +0x04 [sizeof=4] int m_typedef
+; TYPES_2: data +0x08 [sizeof=1] bool m_bool
+; TYPES_2: data +0x09 [sizeof=1] char m_char
+; TYPES_2: data +0x0a [sizeof=2] wchar_t m_wchar_t
+; TYPES_2: data +0x0c [sizeof=4] int m_int
+; TYPES_2: data +0x10 [sizeof=4] unsigned int m_unsigned
+; TYPES_2: data +0x14 [sizeof=4] long m_long
+; TYPES_2: data +0x18 [sizeof=4] unsigned long m_unsigned_long
+; TYPES_2: <padding> (4 bytes)
+; TYPES_2: data +0x20 [sizeof=8] __int64 m_int64
+; TYPES_2: data +0x28 [sizeof=8] unsigned __int64 m_unsigned_int64
+; TYPES_2: data +0x30 [sizeof=4] float m_float
+; TYPES_2: <padding> (4 bytes)
+; TYPES_2: data +0x38 [sizeof=8] double m_double
+; TYPES_2: data +0x40 [sizeof=4] void  (__cdecl * m_pfn_2_args)(int, double)
+; TYPES_2: data +0x44 [sizeof=24] int m_multidimensional_array[2][3]
 ; TYPES_2: }
 
 ; GLOBALS: ---GLOBALS---
 ; GLOBALS-DAG: func [{{.*}}] (FPO) unsigned int __cdecl fpo_func(unsigned int n)
-; GLOBALS-DAG: data [{{.*}}] static void* g_global_pointer
-; GLOBALS-DAG: data [{{.*}}] static int g_global_int
-; GLOBALS-DAG: data [{{.*}}] static int g_array[3]
-; GLOBALS-DAG: data [{{.*}}] static int (* g_pointer_to_array)[3]
-; GLOBALS-DAG: data [{{.*}}] static const int* g_pointer_to_const_int
-; GLOBALS-DAG: data int* const g_const_pointer_to_int = 0
-; GLOBALS-DAG: data const int* const g_const_pointer_to_const_int = 0
+; GLOBALS-DAG: data [{{.*}}, sizeof=4] static void* g_global_pointer
+; GLOBALS-DAG: data [{{.*}}, sizeof=4] static int g_global_int
+; GLOBALS-DAG: data [{{.*}}, sizeof=12] static int g_array[3]
+; GLOBALS-DAG: data [{{.*}}, sizeof=4] static int (* g_pointer_to_array)[3]
+; GLOBALS-DAG: data [{{.*}}, sizeof=4] static const int* g_pointer_to_const_int
+; GLOBALS-DAG: data [sizeof=4] int* const g_const_pointer_to_int = 0
+; GLOBALS-DAG: data [sizeof=4] const int* const g_const_pointer_to_const_int = 0

diff --git a/test/tools/llvm-pdbdump/class-layout.test b/test/tools/llvm-pdbdump/class-layout.test
index 93127b1..e2921d2 100644 (file)
--- a/test/tools/llvm-pdbdump/class-layout.test
+++ b/test/tools/llvm-pdbdump/class-layout.test
@@ -46,9 +46,9 @@
 
 ; BITFIELD_TEST: ---TYPES---
 ; BITFIELD_TEST: struct BitFieldTest::A {
-; BITFIELD_TEST-NEXT: +0x00 int Bits1 : 1
-; BITFIELD_TEST-NEXT: +0x00 int Bits2 : 2
-; BITFIELD_TEST-NEXT: +0x00 int Bits3 : 3
-; BITFIELD_TEST-NEXT: +0x00 int Bits4 : 4
-; BITFIELD_TEST-NEXT: +0x00 int Bits22 : 22
-; BITFIELD_TEST-NEXT: +0x04 int Offset0x04
+; BITFIELD_TEST-NEXT: +0x00 [sizeof=4] int Bits1 : 1
+; BITFIELD_TEST-NEXT: +0x00 [sizeof=4] int Bits2 : 2
+; BITFIELD_TEST-NEXT: +0x00 [sizeof=4] int Bits3 : 3
+; BITFIELD_TEST-NEXT: +0x00 [sizeof=4] int Bits4 : 4
+; BITFIELD_TEST-NEXT: +0x00 [sizeof=4] int Bits22 : 22
+; BITFIELD_TEST-NEXT: +0x04 [sizeof=4] int Offset0x04

diff --git a/tools/llvm-pdbdump/LinePrinter.cpp b/tools/llvm-pdbdump/LinePrinter.cpp
index d376810..e5dd66f 100644 (file)
--- a/tools/llvm-pdbdump/LinePrinter.cpp
+++ b/tools/llvm-pdbdump/LinePrinter.cpp
@@ -99,6 +99,9 @@ void WithColor::applyColor(PDB_ColorItem C) {
   case PDB_ColorItem::None:
     OS.resetColor();
     return;
+  case PDB_ColorItem::Comment:
+    OS.changeColor(raw_ostream::GREEN, false);
+    return;
   case PDB_ColorItem::Address:
     OS.changeColor(raw_ostream::YELLOW, /*bold=*/true);
     return;
@@ -118,6 +121,7 @@ void WithColor::applyColor(PDB_ColorItem C) {
   case PDB_ColorItem::Path:
     OS.changeColor(raw_ostream::CYAN, false);
     return;
+  case PDB_ColorItem::Padding:
   case PDB_ColorItem::SectionHeader:
     OS.changeColor(raw_ostream::RED, true);
     return;

diff --git a/tools/llvm-pdbdump/LinePrinter.h b/tools/llvm-pdbdump/LinePrinter.h
index 87c51e3..8b3d875 100644 (file)
--- a/tools/llvm-pdbdump/LinePrinter.h
+++ b/tools/llvm-pdbdump/LinePrinter.h
@@ -70,6 +70,8 @@ enum class PDB_ColorItem {
   None,
   Address,
   Type,
+  Comment,
+  Padding,
   Keyword,
   Offset,
   Identifier,

diff --git a/tools/llvm-pdbdump/PrettyClassDefinitionDumper.cpp b/tools/llvm-pdbdump/PrettyClassDefinitionDumper.cpp
index 07e1662..b48ed23 100644 (file)
--- a/tools/llvm-pdbdump/PrettyClassDefinitionDumper.cpp
+++ b/tools/llvm-pdbdump/PrettyClassDefinitionDumper.cpp
@@ -16,6 +16,8 @@
 #include "PrettyVariableDumper.h"
 #include "llvm-pdbdump.h"
 
+#include "llvm/ADT/APFloat.h"
+#include "llvm/ADT/SmallString.h"
 #include "llvm/DebugInfo/PDB/IPDBSession.h"
 #include "llvm/DebugInfo/PDB/PDBExtras.h"
 #include "llvm/DebugInfo/PDB/PDBSymbolData.h"
@@ -26,6 +28,7 @@
 #include "llvm/DebugInfo/PDB/PDBSymbolTypeTypedef.h"
 #include "llvm/DebugInfo/PDB/PDBSymbolTypeUDT.h"
 #include "llvm/DebugInfo/PDB/PDBSymbolTypeVTable.h"
+#include "llvm/Support/Compiler.h"
 #include "llvm/Support/Format.h"
 
 using namespace llvm;
@@ -34,11 +37,56 @@ using namespace llvm::pdb;
 ClassDefinitionDumper::ClassDefinitionDumper(LinePrinter &P)
     : PDBSymDumper(true), Printer(P) {}
 
+static void analyzePadding(const PDBSymbolTypeUDT &Class, BitVector &Padding,
+                           uint32_t &FirstFieldOffset) {
+  Padding.resize(Class.getLength(), true);
+  auto Children = Class.findAllChildren<PDBSymbolData>();
+  bool IsFirst = true;
+  FirstFieldOffset = Class.getLength();
+
+  while (auto Data = Children->getNext()) {
+    // Ignore data members which are not relative to this.  Usually these are
+    // static data members or constexpr and occupy no space.  We also need to
+    // handle BitFields since the PDB doesn't consider them ThisRel, but they
+    // still occupy space in the record layout.
+    auto LocType = Data->getLocationType();
+    if (LocType != PDB_LocType::ThisRel && LocType != PDB_LocType::BitField)
+      continue;
+
+    uint64_t Start = Data->getOffset();
+    if (IsFirst) {
+      FirstFieldOffset = Start;
+      IsFirst = false;
+    }
+
+    auto VarType = Data->getType();
+    uint64_t Size = VarType->getRawSymbol().getLength();
+    Padding.reset(Start, Start + Size);
+  }
+
+  // Unmark anything that comes before the first field so it doesn't get
+  // counted as padding.  In reality this is going to be vptrs or base class
+  // members, but we don't correctly handle that yet.
+  // FIXME: Handle it.
+  Padding.reset(0, FirstFieldOffset);
+}
+
 void ClassDefinitionDumper::start(const PDBSymbolTypeUDT &Class) {
   assert(opts::pretty::ClassFormat !=
          opts::pretty::ClassDefinitionFormat::None);
 
-  std::string Name = Class.getName();
+  uint32_t Size = Class.getLength();
+  uint32_t FirstFieldOffset = 0;
+  BitVector Padding;
+  analyzePadding(Class, Padding, FirstFieldOffset);
+
+  if (opts::pretty::OnlyPaddingClasses && (Padding.count() == 0))
+    return;
+
+  Printer.NewLine();
+  WithColor(Printer, PDB_ColorItem::Comment).get() << "// sizeof = " << Size;
+  Printer.NewLine();
+
   WithColor(Printer, PDB_ColorItem::Keyword).get() << Class.getUdtKind() << " ";
   WithColor(Printer, PDB_ColorItem::Type).get() << Class.getName();
 
@@ -66,24 +114,23 @@ void ClassDefinitionDumper::start(const PDBSymbolTypeUDT &Class) {
   auto Children = Class.findAllChildren();
   Printer.Indent();
   int DumpedCount = 0;
-  while (auto Child = Children->getNext()) {
 
-    if (opts::pretty::ClassFormat ==
-        opts::pretty::ClassDefinitionFormat::LayoutOnly) {
-      if (auto Data = Child->cast<PDBSymbolData>()) {
-        switch (Data->getLocationType()) {
-        case PDB_LocType::ThisRel:
-        case PDB_LocType::BitField:
-          break;
-        default:
-          // All other types of data field do not occupy any storage (e.g. are
-          // const), so in layout mode we skip them.
-          continue;
+  int NextPaddingByte = Padding.find_first();
+  while (auto Child = Children->getNext()) {
+    if (auto Data = llvm::dyn_cast<PDBSymbolData>(Child.get())) {
+      if (Data->getDataKind() == PDB_DataKind::Member && NextPaddingByte >= 0) {
+        // If there are padding bytes remaining, see if this field is the first
+        // to cross a padding boundary, and print a padding field indicator if
+        // so.
+        int Off = Data->getOffset();
+        if (Off > NextPaddingByte) {
+          uint32_t Amount = Off - NextPaddingByte;
+          Printer.NewLine();
+          WithColor(Printer, PDB_ColorItem::Padding).get()
+              << "<padding> (" << Amount << " bytes)";
+          assert(Padding.find_next_unset(NextPaddingByte) == Off);
+          NextPaddingByte = Padding.find_next(Off);
         }
-      } else {
-        // Only data symbols affect record layout, so skip any non-data symbols
-        // if we're in record layout mode.
-        continue;
       }
     }
 
@@ -100,10 +147,27 @@ void ClassDefinitionDumper::start(const PDBSymbolTypeUDT &Class) {
     Child->dump(*this);
   }
 
+  if (NextPaddingByte >= 0) {
+    uint32_t Amount = Size - NextPaddingByte;
+    Printer.NewLine();
+    WithColor(Printer, PDB_ColorItem::Padding).get() << "<padding> (" << Amount
+                                                     << " bytes)";
+  }
   Printer.Unindent();
   if (DumpedCount > 0)
     Printer.NewLine();
   Printer << "}";
+  Printer.NewLine();
+  if (Padding.count() > 0) {
+    APFloat Pct(100.0 * (double)Padding.count() /
+                (double)(Size - FirstFieldOffset));
+    SmallString<8> PctStr;
+    Pct.toString(PctStr, 4);
+    WithColor(Printer, PDB_ColorItem::Padding).get()
+        << "Total padding " << Padding.count() << " bytes (" << PctStr
+        << "% of class size)";
+    Printer.NewLine();
+  }
 }
 
 void ClassDefinitionDumper::dump(const PDBSymbolTypeBaseClass &Symbol) {}

diff --git a/tools/llvm-pdbdump/PrettyClassDefinitionDumper.h b/tools/llvm-pdbdump/PrettyClassDefinitionDumper.h
index e145707..8f0c35c 100644 (file)
--- a/tools/llvm-pdbdump/PrettyClassDefinitionDumper.h
+++ b/tools/llvm-pdbdump/PrettyClassDefinitionDumper.h
@@ -10,6 +10,8 @@
 #ifndef LLVM_TOOLS_LLVMPDBDUMP_PRETTYCLASSDEFINITIONDUMPER_H
 #define LLVM_TOOLS_LLVMPDBDUMP_PRETTYCLASSDEFINITIONDUMPER_H
 
+#include "llvm/ADT/BitVector.h"
+
 #include "llvm/DebugInfo/PDB/PDBSymDumper.h"
 #include "llvm/DebugInfo/PDB/PDBSymbolData.h"
 #include "llvm/DebugInfo/PDB/PDBSymbolFunc.h"

diff --git a/tools/llvm-pdbdump/PrettyTypeDumper.cpp b/tools/llvm-pdbdump/PrettyTypeDumper.cpp
index 023806c..12a47d2 100644 (file)
--- a/tools/llvm-pdbdump/PrettyTypeDumper.cpp
+++ b/tools/llvm-pdbdump/PrettyTypeDumper.cpp
@@ -96,9 +96,8 @@ void TypeDumper::dump(const PDBSymbolTypeUDT &Symbol) {
   if (Printer.IsTypeExcluded(Symbol.getName()))
     return;
 
-  Printer.NewLine();
-
   if (opts::pretty::ClassFormat == opts::pretty::ClassDefinitionFormat::None) {
+    Printer.NewLine();
     WithColor(Printer, PDB_ColorItem::Keyword).get() << "class ";
     WithColor(Printer, PDB_ColorItem::Identifier).get() << Symbol.getName();
   } else {

diff --git a/tools/llvm-pdbdump/PrettyVariableDumper.cpp b/tools/llvm-pdbdump/PrettyVariableDumper.cpp
index 8b85f97..65f0139 100644 (file)
--- a/tools/llvm-pdbdump/PrettyVariableDumper.cpp
+++ b/tools/llvm-pdbdump/PrettyVariableDumper.cpp
@@ -43,13 +43,15 @@ void VariableDumper::start(const PDBSymbolData &Var) {
 
   auto VarType = Var.getType();
 
+  uint64_t Length = VarType->getRawSymbol().getLength();
+
   switch (auto LocType = Var.getLocationType()) {
   case PDB_LocType::Static:
     Printer.NewLine();
     Printer << "data [";
     WithColor(Printer, PDB_ColorItem::Address).get()
         << format_hex(Var.getVirtualAddress(), 10);
-    Printer << "] ";
+    Printer << ", sizeof=" << Length << "] ";
     WithColor(Printer, PDB_ColorItem::Keyword).get() << "static ";
     dumpSymbolTypeAndName(*VarType, Var.getName());
     break;
@@ -57,7 +59,7 @@ void VariableDumper::start(const PDBSymbolData &Var) {
     if (isa<PDBSymbolTypeEnum>(*VarType))
       break;
     Printer.NewLine();
-    Printer << "data ";
+    Printer << "data [sizeof=" << Length << "] ";
     dumpSymbolTypeAndName(*VarType, Var.getName());
     Printer << " = ";
     WithColor(Printer, PDB_ColorItem::LiteralValue).get() << Var.getValue();
@@ -66,21 +68,23 @@ void VariableDumper::start(const PDBSymbolData &Var) {
     Printer.NewLine();
     Printer << "data ";
     WithColor(Printer, PDB_ColorItem::Offset).get()
-        << "+" << format_hex(Var.getOffset(), 4) << " ";
+        << "+" << format_hex(Var.getOffset(), 4) << " [sizeof=" << Length
+        << "] ";
     dumpSymbolTypeAndName(*VarType, Var.getName());
     break;
   case PDB_LocType::BitField:
     Printer.NewLine();
     Printer << "data ";
     WithColor(Printer, PDB_ColorItem::Offset).get()
-        << "+" << format_hex(Var.getOffset(), 4) << " ";
+        << "+" << format_hex(Var.getOffset(), 4) << " [sizeof=" << Length
+        << "] ";
     dumpSymbolTypeAndName(*VarType, Var.getName());
     Printer << " : ";
     WithColor(Printer, PDB_ColorItem::LiteralValue).get() << Var.getLength();
     break;
   default:
     Printer.NewLine();
-    Printer << "data ";
+    Printer << "data [sizeof=" << Length << "] ";
     Printer << "unknown(" << LocType << ") ";
     WithColor(Printer, PDB_ColorItem::Identifier).get() << Var.getName();
     break;

diff --git a/tools/llvm-pdbdump/llvm-pdbdump.cpp b/tools/llvm-pdbdump/llvm-pdbdump.cpp
index 3d3d253..8a749ba 100644 (file)
--- a/tools/llvm-pdbdump/llvm-pdbdump.cpp
+++ b/tools/llvm-pdbdump/llvm-pdbdump.cpp
@@ -122,6 +122,14 @@ cl::opt<bool> Enums("enums", cl::desc("Display enum types"),
                     cl::cat(TypeCategory), cl::sub(PrettySubcommand));
 cl::opt<bool> Typedefs("typedefs", cl::desc("Display typedef types"),
                        cl::cat(TypeCategory), cl::sub(PrettySubcommand));
+cl::opt<ClassDefinitionFormat>
+    ClassFormat("class-definitions", cl::desc("Class definition format"),
+                cl::init(ClassDefinitionFormat::Standard),
+                cl::values(clEnumValN(ClassDefinitionFormat::Standard, "full",
+                                      "Display complete class definition"),
+                           clEnumValN(ClassDefinitionFormat::None, "none",
+                                      "Don't display class definitions")),
+                cl::cat(TypeCategory), cl::sub(PrettySubcommand));
 
 cl::opt<bool> Lines("lines", cl::desc("Line tables"), cl::cat(TypeCategory),
                     cl::sub(PrettySubcommand));
@@ -161,6 +169,10 @@ cl::list<std::string> IncludeCompilands(
     "include-compilands",
     cl::desc("Include only compilands those which match a regular expression"),
     cl::ZeroOrMore, cl::cat(FilterCategory), cl::sub(PrettySubcommand));
+cl::opt<bool> OnlyPaddingClasses(
+    "only-padding-classes", cl::desc("When dumping classes, only display those "
+                                     "with non-zero amounts of padding bytes"),
+    cl::ZeroOrMore, cl::cat(FilterCategory), cl::sub(PrettySubcommand));
 
 cl::opt<bool> ExcludeCompilerGenerated(
     "no-compiler-generated",
@@ -170,16 +182,6 @@ cl::opt<bool>
     ExcludeSystemLibraries("no-system-libs",
                            cl::desc("Don't show symbols from system libraries"),
                            cl::cat(FilterCategory), cl::sub(PrettySubcommand));
-cl::opt<ClassDefinitionFormat> ClassFormat(
-    "class-definitions", cl::desc("Class definition format"),
-    cl::init(ClassDefinitionFormat::Full),
-    cl::values(clEnumValN(ClassDefinitionFormat::Full, "full",
-                          "Display complete class definition"),
-               clEnumValN(ClassDefinitionFormat::LayoutOnly, "layout",
-                          "Display only members which affect record layout"),
-               clEnumValN(ClassDefinitionFormat::None, "none",
-                          "Don't display class definitions")),
-    cl::cat(FilterCategory), cl::sub(PrettySubcommand));
 
 cl::opt<bool> NoEnumDefs("no-enum-definitions",
                          cl::desc("Don't display full enum definitions"),

diff --git a/tools/llvm-pdbdump/llvm-pdbdump.h b/tools/llvm-pdbdump/llvm-pdbdump.h
index 0fdcab2..a335d30 100644 (file)
--- a/tools/llvm-pdbdump/llvm-pdbdump.h
+++ b/tools/llvm-pdbdump/llvm-pdbdump.h
@@ -17,7 +17,8 @@
 namespace opts {
 
 namespace pretty {
-enum class ClassDefinitionFormat { None, LayoutOnly, Full };
+
+enum class ClassDefinitionFormat { None, Standard };
 
 extern llvm::cl::opt<bool> Compilands;
 extern llvm::cl::opt<bool> Symbols;
@@ -28,7 +29,6 @@ extern llvm::cl::opt<bool> Typedefs;
 extern llvm::cl::opt<bool> All;
 extern llvm::cl::opt<bool> ExcludeCompilerGenerated;
 
-extern llvm::cl::opt<ClassDefinitionFormat> ClassFormat;
 extern llvm::cl::opt<bool> NoEnumDefs;
 extern llvm::cl::list<std::string> ExcludeTypes;
 extern llvm::cl::list<std::string> ExcludeSymbols;
@@ -36,6 +36,8 @@ extern llvm::cl::list<std::string> ExcludeCompilands;
 extern llvm::cl::list<std::string> IncludeTypes;
 extern llvm::cl::list<std::string> IncludeSymbols;
 extern llvm::cl::list<std::string> IncludeCompilands;
+extern llvm::cl::opt<bool> OnlyPaddingClasses;
+extern llvm::cl::opt<ClassDefinitionFormat> ClassFormat;
 }
 
 namespace raw {