DebugInfo: Restrict DW_AT_high_pc encoding as data4 offset to DWARF 4 as per spec
[oota-llvm.git] / lib / CodeGen / AsmPrinter / DIEHash.cpp
index 34d9fffb1ae85cee784e08d1d3b683300fc47d5d..cff8f2180a9a6a58e62cf10c75b2bb68d3cf3ddc 100644 (file)
 
 #define DEBUG_TYPE "dwarfdebug"
 
-#include "DIE.h"
 #include "DIEHash.h"
-#include "DwarfCompileUnit.h"
+#include "DIE.h"
 #include "llvm/ADT/ArrayRef.h"
 #include "llvm/ADT/StringRef.h"
+#include "llvm/CodeGen/AsmPrinter.h"
 #include "llvm/Support/Debug.h"
 #include "llvm/Support/Dwarf.h"
 #include "llvm/Support/Endian.h"
@@ -74,7 +74,7 @@ void DIEHash::addSLEB128(int64_t Value) {
   do {
     uint8_t Byte = Value & 0x7f;
     Value >>= 7;
-    More = !((((Value == 0 ) && ((Byte & 0x40) == 0)) ||
+    More = !((((Value == 0) && ((Byte & 0x40) == 0)) ||
               ((Value == -1) && ((Byte & 0x40) != 0))));
     if (More)
       Byte |= 0x80; // Mark this byte to show that more bytes will follow.
@@ -83,33 +83,36 @@ void DIEHash::addSLEB128(int64_t Value) {
 }
 
 /// \brief Including \p Parent adds the context of Parent to the hash..
-void DIEHash::addParentContext(DIE *Parent) {
+void DIEHash::addParentContext(const DIE &Parent) {
 
   DEBUG(dbgs() << "Adding parent context to hash...\n");
 
   // [7.27.2] For each surrounding type or namespace beginning with the
   // outermost such construct...
-  SmallVector<DIE *, 1> Parents;
-  while (Parent->getTag() != dwarf::DW_TAG_compile_unit) {
-    Parents.push_back(Parent);
-    Parent = Parent->getParent();
+  SmallVector<const DIE *, 1> Parents;
+  const DIE *Cur = &Parent;
+  while (Cur->getParent()) {
+    Parents.push_back(Cur);
+    Cur = Cur->getParent();
   }
+  assert(Cur->getTag() == dwarf::DW_TAG_compile_unit ||
+         Cur->getTag() == dwarf::DW_TAG_type_unit);
 
   // Reverse iterate over our list to go from the outermost construct to the
   // innermost.
-  for (SmallVectorImpl<DIE *>::reverse_iterator I = Parents.rbegin(),
-                                                E = Parents.rend();
+  for (SmallVectorImpl<const DIE *>::reverse_iterator I = Parents.rbegin(),
+                                                      E = Parents.rend();
        I != E; ++I) {
-    DIE *Die = *I;
+    const DIE &Die = **I;
 
     // ... Append the letter "C" to the sequence...
     addULEB128('C');
 
     // ... Followed by the DWARF tag of the construct...
-    addULEB128(Die->getTag());
+    addULEB128(Die.getTag());
 
     // ... Then the name, taken from the DW_AT_name attribute.
-    StringRef Name = getDIEStringAttr(*Die, dwarf::DW_AT_name);
+    StringRef Name = getDIEStringAttr(Die, dwarf::DW_AT_name);
     DEBUG(dbgs() << "... adding context: " << Name << "\n");
     if (!Name.empty())
       addString(Name);
@@ -117,9 +120,9 @@ void DIEHash::addParentContext(DIE *Parent) {
 }
 
 // Collect all of the attributes for a particular DIE in single structure.
-void DIEHash::collectAttributes(DIE *Die, DIEAttrs &Attrs) {
-  const SmallVectorImpl<DIEValue *> &Values = Die->getValues();
-  const DIEAbbrev &Abbrevs = Die->getAbbrev();
+void DIEHash::collectAttributes(const DIE &Die, DIEAttrs &Attrs) {
+  const SmallVectorImpl<DIEValue *> &Values = Die.getValues();
+  const DIEAbbrev &Abbrevs = Die.getAbbrev();
 
 #define COLLECT_ATTR(NAME)                                                     \
   case dwarf::NAME:                                                            \
@@ -132,55 +135,55 @@ void DIEHash::collectAttributes(DIE *Die, DIEAttrs &Attrs) {
                  << dwarf::AttributeString(Abbrevs.getData()[i].getAttribute())
                  << " added.\n");
     switch (Abbrevs.getData()[i].getAttribute()) {
-    COLLECT_ATTR(DW_AT_name);
-    COLLECT_ATTR(DW_AT_accessibility);
-    COLLECT_ATTR(DW_AT_address_class);
-    COLLECT_ATTR(DW_AT_allocated);
-    COLLECT_ATTR(DW_AT_artificial);
-    COLLECT_ATTR(DW_AT_associated);
-    COLLECT_ATTR(DW_AT_binary_scale);
-    COLLECT_ATTR(DW_AT_bit_offset);
-    COLLECT_ATTR(DW_AT_bit_size);
-    COLLECT_ATTR(DW_AT_bit_stride);
-    COLLECT_ATTR(DW_AT_byte_size);
-    COLLECT_ATTR(DW_AT_byte_stride);
-    COLLECT_ATTR(DW_AT_const_expr);
-    COLLECT_ATTR(DW_AT_const_value);
-    COLLECT_ATTR(DW_AT_containing_type);
-    COLLECT_ATTR(DW_AT_count);
-    COLLECT_ATTR(DW_AT_data_bit_offset);
-    COLLECT_ATTR(DW_AT_data_location);
-    COLLECT_ATTR(DW_AT_data_member_location);
-    COLLECT_ATTR(DW_AT_decimal_scale);
-    COLLECT_ATTR(DW_AT_decimal_sign);
-    COLLECT_ATTR(DW_AT_default_value);
-    COLLECT_ATTR(DW_AT_digit_count);
-    COLLECT_ATTR(DW_AT_discr);
-    COLLECT_ATTR(DW_AT_discr_list);
-    COLLECT_ATTR(DW_AT_discr_value);
-    COLLECT_ATTR(DW_AT_encoding);
-    COLLECT_ATTR(DW_AT_enum_class);
-    COLLECT_ATTR(DW_AT_endianity);
-    COLLECT_ATTR(DW_AT_explicit);
-    COLLECT_ATTR(DW_AT_is_optional);
-    COLLECT_ATTR(DW_AT_location);
-    COLLECT_ATTR(DW_AT_lower_bound);
-    COLLECT_ATTR(DW_AT_mutable);
-    COLLECT_ATTR(DW_AT_ordering);
-    COLLECT_ATTR(DW_AT_picture_string);
-    COLLECT_ATTR(DW_AT_prototyped);
-    COLLECT_ATTR(DW_AT_small);
-    COLLECT_ATTR(DW_AT_segment);
-    COLLECT_ATTR(DW_AT_string_length);
-    COLLECT_ATTR(DW_AT_threads_scaled);
-    COLLECT_ATTR(DW_AT_upper_bound);
-    COLLECT_ATTR(DW_AT_use_location);
-    COLLECT_ATTR(DW_AT_use_UTF8);
-    COLLECT_ATTR(DW_AT_variable_parameter);
-    COLLECT_ATTR(DW_AT_virtuality);
-    COLLECT_ATTR(DW_AT_visibility);
-    COLLECT_ATTR(DW_AT_vtable_elem_location);
-    COLLECT_ATTR(DW_AT_type);
+      COLLECT_ATTR(DW_AT_name);
+      COLLECT_ATTR(DW_AT_accessibility);
+      COLLECT_ATTR(DW_AT_address_class);
+      COLLECT_ATTR(DW_AT_allocated);
+      COLLECT_ATTR(DW_AT_artificial);
+      COLLECT_ATTR(DW_AT_associated);
+      COLLECT_ATTR(DW_AT_binary_scale);
+      COLLECT_ATTR(DW_AT_bit_offset);
+      COLLECT_ATTR(DW_AT_bit_size);
+      COLLECT_ATTR(DW_AT_bit_stride);
+      COLLECT_ATTR(DW_AT_byte_size);
+      COLLECT_ATTR(DW_AT_byte_stride);
+      COLLECT_ATTR(DW_AT_const_expr);
+      COLLECT_ATTR(DW_AT_const_value);
+      COLLECT_ATTR(DW_AT_containing_type);
+      COLLECT_ATTR(DW_AT_count);
+      COLLECT_ATTR(DW_AT_data_bit_offset);
+      COLLECT_ATTR(DW_AT_data_location);
+      COLLECT_ATTR(DW_AT_data_member_location);
+      COLLECT_ATTR(DW_AT_decimal_scale);
+      COLLECT_ATTR(DW_AT_decimal_sign);
+      COLLECT_ATTR(DW_AT_default_value);
+      COLLECT_ATTR(DW_AT_digit_count);
+      COLLECT_ATTR(DW_AT_discr);
+      COLLECT_ATTR(DW_AT_discr_list);
+      COLLECT_ATTR(DW_AT_discr_value);
+      COLLECT_ATTR(DW_AT_encoding);
+      COLLECT_ATTR(DW_AT_enum_class);
+      COLLECT_ATTR(DW_AT_endianity);
+      COLLECT_ATTR(DW_AT_explicit);
+      COLLECT_ATTR(DW_AT_is_optional);
+      COLLECT_ATTR(DW_AT_location);
+      COLLECT_ATTR(DW_AT_lower_bound);
+      COLLECT_ATTR(DW_AT_mutable);
+      COLLECT_ATTR(DW_AT_ordering);
+      COLLECT_ATTR(DW_AT_picture_string);
+      COLLECT_ATTR(DW_AT_prototyped);
+      COLLECT_ATTR(DW_AT_small);
+      COLLECT_ATTR(DW_AT_segment);
+      COLLECT_ATTR(DW_AT_string_length);
+      COLLECT_ATTR(DW_AT_threads_scaled);
+      COLLECT_ATTR(DW_AT_upper_bound);
+      COLLECT_ATTR(DW_AT_use_location);
+      COLLECT_ATTR(DW_AT_use_UTF8);
+      COLLECT_ATTR(DW_AT_variable_parameter);
+      COLLECT_ATTR(DW_AT_virtuality);
+      COLLECT_ATTR(DW_AT_visibility);
+      COLLECT_ATTR(DW_AT_vtable_elem_location);
+      COLLECT_ATTR(DW_AT_type);
     default:
       break;
     }
@@ -196,8 +199,8 @@ void DIEHash::hashShallowTypeReference(dwarf::Attribute Attribute,
   addULEB128(Attribute);
 
   // the context of the tag,
-  if (DIE *Parent = Entry.getParent())
-    addParentContext(Parent);
+  if (const DIE *Parent = Entry.getParent())
+    addParentContext(*Parent);
 
   // the letter 'E',
   addULEB128('E');
@@ -227,7 +230,7 @@ void DIEHash::hashRepeatedTypeReference(dwarf::Attribute Attribute,
 }
 
 void DIEHash::hashDIEEntry(dwarf::Attribute Attribute, dwarf::Tag Tag,
-                           DIE &Entry) {
+                           const DIE &Entry) {
   assert(Tag != dwarf::DW_TAG_friend && "No current LLVM clients emit friend "
                                         "tags. Add support here when there's "
                                         "a use case");
@@ -242,10 +245,12 @@ void DIEHash::hashDIEEntry(dwarf::Attribute Attribute, dwarf::Tag Tag,
       // there's a decl/def difference in the containing type of a
       // ptr_to_member_type, but it's what DWARF says, for some reason.
       Attribute == dwarf::DW_AT_type) {
-    // [FIXME] ... has a DW_AT_name attribute,
-    hashShallowTypeReference(Attribute, Entry,
-                             getDIEStringAttr(Entry, dwarf::DW_AT_name));
-    return;
+    // ... has a DW_AT_name attribute,
+    StringRef Name = getDIEStringAttr(Entry, dwarf::DW_AT_name);
+    if (!Name.empty()) {
+      hashShallowTypeReference(Attribute, Entry, Name);
+      return;
+    }
   }
 
   unsigned &DieNumber = Numbering[&Entry];
@@ -262,7 +267,16 @@ void DIEHash::hashDIEEntry(dwarf::Attribute Attribute, dwarf::Tag Tag,
   // ... process the type T recursively by performing Steps 2 through 7, and
   // use the result as the attribute value.
   DieNumber = Numbering.size();
-  computeHash(&Entry);
+  computeHash(Entry);
+}
+
+// Hash all of the values in a block like set of values. This assumes that
+// all of the data is going to be added as integers.
+void DIEHash::hashBlockData(const SmallVectorImpl<DIEValue *> &Values) {
+  for (SmallVectorImpl<DIEValue *>::const_iterator I = Values.begin(),
+                                                   E = Values.end();
+       I != E; ++I)
+    Hash.update((uint64_t)cast<DIEInteger>(*I)->getValue());
 }
 
 // Hash an individual attribute \param Attr based on the type of attribute and
@@ -272,43 +286,72 @@ void DIEHash::hashAttribute(AttrEntry Attr, dwarf::Tag Tag) {
   const DIEAbbrevData *Desc = Attr.Desc;
   dwarf::Attribute Attribute = Desc->getAttribute();
 
-  // 7.27 Step 3
-  // ... An attribute that refers to another type entry T is processed as
-  // follows:
-  if (const DIEEntry *EntryAttr = dyn_cast<DIEEntry>(Value)) {
-    hashDIEEntry(Attribute, Tag, *EntryAttr->getEntry());
-    return;
+  // Other attribute values use the letter 'A' as the marker, and the value
+  // consists of the form code (encoded as an unsigned LEB128 value) followed by
+  // the encoding of the value according to the form code. To ensure
+  // reproducibility of the signature, the set of forms used in the signature
+  // computation is limited to the following: DW_FORM_sdata, DW_FORM_flag,
+  // DW_FORM_string, and DW_FORM_block.
+
+  switch (Value->getType()) {
+    // 7.27 Step 3
+    // ... An attribute that refers to another type entry T is processed as
+    // follows:
+  case DIEValue::isEntry:
+    hashDIEEntry(Attribute, Tag, *cast<DIEEntry>(Value)->getEntry());
+    break;
+  case DIEValue::isInteger: {
+    addULEB128('A');
+    addULEB128(Attribute);
+    switch (Desc->getForm()) {
+    case dwarf::DW_FORM_data1:
+    case dwarf::DW_FORM_data2:
+    case dwarf::DW_FORM_data4:
+    case dwarf::DW_FORM_data8:
+    case dwarf::DW_FORM_udata:
+    case dwarf::DW_FORM_sdata:
+      addULEB128(dwarf::DW_FORM_sdata);
+      addSLEB128((int64_t)cast<DIEInteger>(Value)->getValue());
+      break;
+    // DW_FORM_flag_present is just flag with a value of one. We still give it a
+    // value so just use the value.
+    case dwarf::DW_FORM_flag_present:
+    case dwarf::DW_FORM_flag:
+      addULEB128(dwarf::DW_FORM_flag);
+      addULEB128((int64_t)cast<DIEInteger>(Value)->getValue());
+      break;
+    default:
+      llvm_unreachable("Unknown integer form!");
+    }
+    break;
   }
-
-  // Other attribute values use the letter 'A' as the marker, ...
-  addULEB128('A');
-
-  addULEB128(Attribute);
-
-  // ... and the value consists of the form code (encoded as an unsigned LEB128
-  // value) followed by the encoding of the value according to the form code. To
-  // ensure reproducibility of the signature, the set of forms used in the
-  // signature computation is limited to the following: DW_FORM_sdata,
-  // DW_FORM_flag, DW_FORM_string, and DW_FORM_block.
-  switch (Desc->getForm()) {
-  case dwarf::DW_FORM_string:
-    llvm_unreachable(
-        "Add support for DW_FORM_string if we ever start emitting them again");
-  case dwarf::DW_FORM_GNU_str_index:
-  case dwarf::DW_FORM_strp:
+  case DIEValue::isString:
+    addULEB128('A');
+    addULEB128(Attribute);
     addULEB128(dwarf::DW_FORM_string);
     addString(cast<DIEString>(Value)->getString());
     break;
-  case dwarf::DW_FORM_data1:
-  case dwarf::DW_FORM_data2:
-  case dwarf::DW_FORM_data4:
-  case dwarf::DW_FORM_data8:
-  case dwarf::DW_FORM_udata:
-    addULEB128(dwarf::DW_FORM_sdata);
-    addSLEB128((int64_t)cast<DIEInteger>(Value)->getValue());
+  case DIEValue::isBlock:
+  case DIEValue::isLoc:
+    addULEB128('A');
+    addULEB128(Attribute);
+    addULEB128(dwarf::DW_FORM_block);
+    if (isa<DIEBlock>(Value)) {
+      addULEB128(cast<DIEBlock>(Value)->ComputeSize(AP));
+      hashBlockData(cast<DIEBlock>(Value)->getValues());
+    } else {
+      addULEB128(cast<DIELoc>(Value)->ComputeSize(AP));
+      hashBlockData(cast<DIELoc>(Value)->getValues());
+    }
     break;
-  default:
-    llvm_unreachable("Add support for additional forms");
+    // FIXME: Handle loclistptr.
+  case DIEValue::isLocList:
+    // FIXME: It's uncertain whether or not we should handle this at the moment.
+  case DIEValue::isExpr:
+  case DIEValue::isLabel:
+  case DIEValue::isDelta:
+  case DIEValue::isTypeSignature:
+    llvm_unreachable("Add support for additional value types.");
   }
 }
 
@@ -375,28 +418,51 @@ void DIEHash::hashAttributes(const DIEAttrs &Attrs, dwarf::Tag Tag) {
 }
 
 // Add all of the attributes for \param Die to the hash.
-void DIEHash::addAttributes(DIE *Die) {
+void DIEHash::addAttributes(const DIE &Die) {
   DIEAttrs Attrs = {};
   collectAttributes(Die, Attrs);
-  hashAttributes(Attrs, Die->getTag());
+  hashAttributes(Attrs, Die.getTag());
+}
+
+void DIEHash::hashNestedType(const DIE &Die, StringRef Name) {
+  // 7.27 Step 7
+  // ... append the letter 'S',
+  addULEB128('S');
+
+  // the tag of C,
+  addULEB128(Die.getTag());
+
+  // and the name.
+  addString(Name);
 }
 
 // Compute the hash of a DIE. This is based on the type signature computation
 // given in section 7.27 of the DWARF4 standard. It is the md5 hash of a
 // flattened description of the DIE.
-void DIEHash::computeHash(DIE *Die) {
+void DIEHash::computeHash(const DIE &Die) {
   // Append the letter 'D', followed by the DWARF tag of the DIE.
   addULEB128('D');
-  addULEB128(Die->getTag());
+  addULEB128(Die.getTag());
 
   // Add each of the attributes of the DIE.
   addAttributes(Die);
 
   // Then hash each of the children of the DIE.
-  for (std::vector<DIE *>::const_iterator I = Die->getChildren().begin(),
-                                          E = Die->getChildren().end();
-       I != E; ++I)
-    computeHash(*I);
+  for (std::vector<DIE *>::const_iterator I = Die.getChildren().begin(),
+                                          E = Die.getChildren().end();
+       I != E; ++I) {
+    // 7.27 Step 7
+    // If C is a nested type entry or a member function entry, ...
+    if (isType((*I)->getTag()) || (*I)->getTag() == dwarf::DW_TAG_subprogram) {
+      StringRef Name = getDIEStringAttr(**I, dwarf::DW_AT_name);
+      // ... and has a DW_AT_name attribute
+      if (!Name.empty()) {
+        hashNestedType(**I, Name);
+        continue;
+      }
+    }
+    computeHash(**I);
+  }
 
   // Following the last (or if there are no children), append a zero byte.
   Hash.update(makeArrayRef((uint8_t)'\0'));
@@ -406,24 +472,23 @@ void DIEHash::computeHash(DIE *Die) {
 /// DWARF4 standard. It is the md5 hash of a flattened description of the DIE
 /// with the exception that we are hashing only the context and the name of the
 /// type.
-uint64_t DIEHash::computeDIEODRSignature(DIE *Die) {
+uint64_t DIEHash::computeDIEODRSignature(const DIE &Die) {
 
   // Add the contexts to the hash. We won't be computing the ODR hash for
   // function local types so it's safe to use the generic context hashing
   // algorithm here.
   // FIXME: If we figure out how to account for linkage in some way we could
   // actually do this with a slight modification to the parent hash algorithm.
-  DIE *Parent = Die->getParent();
-  if (Parent)
-    addParentContext(Parent);
+  if (const DIE *Parent = Die.getParent())
+    addParentContext(*Parent);
 
   // Add the current DIE information.
 
   // Add the DWARF tag of the DIE.
-  addULEB128(Die->getTag());
+  addULEB128(Die.getTag());
 
   // Add the name of the type to the hash.
-  addString(getDIEStringAttr(*Die, dwarf::DW_AT_name));
+  addString(getDIEStringAttr(Die, dwarf::DW_AT_name));
 
   // Now get the result.
   MD5::MD5Result Result;
@@ -439,9 +504,9 @@ uint64_t DIEHash::computeDIEODRSignature(DIE *Die) {
 /// DWARF4 standard. It is an md5 hash of the flattened description of the DIE
 /// with the inclusion of the full CU and all top level CU entities.
 // TODO: Initialize the type chain at 0 instead of 1 for CU signatures.
-uint64_t DIEHash::computeCUSignature(DIE *Die) {
+uint64_t DIEHash::computeCUSignature(const DIE &Die) {
   Numbering.clear();
-  Numbering[Die] = 1;
+  Numbering[&Die] = 1;
 
   // Hash the DIE.
   computeHash(Die);
@@ -460,12 +525,12 @@ uint64_t DIEHash::computeCUSignature(DIE *Die) {
 /// DWARF4 standard. It is an md5 hash of the flattened description of the DIE
 /// with the inclusion of additional forms not specifically called out in the
 /// standard.
-uint64_t DIEHash::computeTypeSignature(DIE *Die) {
+uint64_t DIEHash::computeTypeSignature(const DIE &Die) {
   Numbering.clear();
-  Numbering[Die] = 1;
+  Numbering[&Die] = 1;
 
-  if (DIE *Parent = Die->getParent())
-    addParentContext(Parent);
+  if (const DIE *Parent = Die.getParent())
+    addParentContext(*Parent);
 
   // Hash the DIE.
   computeHash(Die);