Re-land "[dwarfdump] Add -lookup option"
authorJonas Devlieghere <jonas@devlieghere.com>
Wed, 25 Oct 2017 21:56:41 +0000 (21:56 +0000)
committerJonas Devlieghere <jonas@devlieghere.com>
Wed, 25 Oct 2017 21:56:41 +0000 (21:56 +0000)
Add the option to lookup an address in the debug information and print
out the file, function, block and line table details.

Differential revision: https://reviews.llvm.org/D38409

llvm-svn: 316619

llvm/include/llvm/DebugInfo/DIContext.h
llvm/include/llvm/DebugInfo/DWARF/DWARFContext.h
llvm/include/llvm/DebugInfo/DWARF/DWARFUnit.h
llvm/lib/DebugInfo/DWARF/DWARFContext.cpp
llvm/lib/DebugInfo/DWARF/DWARFUnit.cpp
llvm/test/tools/llvm-dwarfdump/X86/lookup.s [new file with mode: 0644]
llvm/test/tools/llvm-dwarfdump/cmdline.test
llvm/tools/llvm-dwarfdump/llvm-dwarfdump.cpp

index ce0a156..4a368be 100644 (file)
@@ -17,6 +17,7 @@
 
 #include "llvm/ADT/SmallVector.h"
 #include "llvm/Object/ObjectFile.h"
+#include "llvm/Support/raw_ostream.h"
 #include <cassert>
 #include <cstdint>
 #include <memory>
@@ -26,8 +27,6 @@
 
 namespace llvm {
 
-class raw_ostream;
-
 /// A format-neutral container for source line information.
 struct DILineInfo {
   std::string FileName;
@@ -46,15 +45,30 @@ struct DILineInfo {
            FileName == RHS.FileName && FunctionName == RHS.FunctionName &&
            StartLine == RHS.StartLine && Discriminator == RHS.Discriminator;
   }
+
   bool operator!=(const DILineInfo &RHS) const {
     return !(*this == RHS);
   }
+
   bool operator<(const DILineInfo &RHS) const {
     return std::tie(FileName, FunctionName, Line, Column, StartLine,
                     Discriminator) <
            std::tie(RHS.FileName, RHS.FunctionName, RHS.Line, RHS.Column,
                     RHS.StartLine, RHS.Discriminator);
   }
+
+  explicit operator bool() const { return *this != DILineInfo(); }
+
+  void dump(raw_ostream &OS) {
+    OS << "Line info: ";
+    if (FileName != "<invalid>")
+      OS << "file '" << FileName << "', ";
+    if (FunctionName != "<invalid>")
+      OS << "function '" << FunctionName << "', ";
+    OS << "line " << Line << ", ";
+    OS << "column " << Column << ", ";
+    OS << "start line " << StartLine << '\n';
+  }
 };
 
 using DILineInfoTable = SmallVector<std::pair<uint64_t, DILineInfo>, 16>;
index 4cecd22..2ddbc4b 100644 (file)
@@ -257,6 +257,18 @@ public:
   /// Get a pointer to a parsed line table corresponding to a compile unit.
   const DWARFDebugLine::LineTable *getLineTableForUnit(DWARFUnit *cu);
 
+  /// Wraps the returned DIEs for a given address.
+  struct DIEsForAddress {
+    DWARFCompileUnit *CompileUnit = nullptr;
+    DWARFDie FunctionDIE;
+    DWARFDie BlockDIE;
+    explicit operator bool() const { return CompileUnit != nullptr; }
+  };
+
+  /// Get the compilation unit, the function DIE and lexical block DIE for the
+  /// given address where applicable.
+  DIEsForAddress getDIEsForAddress(uint64_t Address);
+
   DILineInfo getLineInfoForAddress(uint64_t Address,
       DILineInfoSpecifier Specifier = DILineInfoSpecifier()) override;
   DILineInfoTable getLineInfoForAddressRange(uint64_t Address, uint64_t Size,
index 696299f..e9178e0 100644 (file)
@@ -338,6 +338,11 @@ public:
 
   void collectAddressRanges(DWARFAddressRangesVector &CURanges);
 
+  /// Returns subprogram DIE with address range encompassing the provided
+  /// address. The pointer is alive as long as parsed compile unit DIEs are not
+  /// cleared.
+  DWARFDie getSubroutineForAddress(uint64_t Address);
+
   /// getInlinedChainForAddress - fetches inlined chain for a given address.
   /// Returns empty chain if there is no subprogram containing address. The
   /// chain is valid as long as parsed compile unit DIEs are not cleared.
@@ -421,11 +426,6 @@ private:
   /// parseDWO - Parses .dwo file for current compile unit. Returns true if
   /// it was actually constructed.
   bool parseDWO();
-
-  /// getSubroutineForAddress - Returns subprogram DIE with address range
-  /// encompassing the provided address. The pointer is alive as long as parsed
-  /// compile unit DIEs are not cleared.
-  DWARFDie getSubroutineForAddress(uint64_t Address);
 };
 
 } // end namespace llvm
index bf0c4b0..24aa666 100644 (file)
@@ -48,7 +48,6 @@
 #include <cstdint>
 #include <map>
 #include <string>
-#include <tuple>
 #include <utility>
 #include <vector>
 
@@ -723,6 +722,35 @@ DWARFCompileUnit *DWARFContext::getCompileUnitForAddress(uint64_t Address) {
   return getCompileUnitForOffset(CUOffset);
 }
 
+DWARFContext::DIEsForAddress DWARFContext::getDIEsForAddress(uint64_t Address) {
+  DIEsForAddress Result;
+
+  DWARFCompileUnit *CU = getCompileUnitForAddress(Address);
+  if (!CU)
+    return Result;
+
+  Result.CompileUnit = CU;
+  Result.FunctionDIE = CU->getSubroutineForAddress(Address);
+
+  std::vector<DWARFDie> Worklist;
+  Worklist.push_back(Result.FunctionDIE);
+  while (!Worklist.empty()) {
+    DWARFDie DIE = Worklist.back();
+    Worklist.pop_back();
+
+    if (DIE.getTag() == DW_TAG_lexical_block &&
+        DIE.addressRangeContainsAddress(Address)) {
+      Result.BlockDIE = DIE;
+      break;
+    }
+
+    for (auto Child : DIE)
+      Worklist.push_back(Child);
+  }
+
+  return Result;
+}
+
 static bool getFunctionNameAndStartLineForAddress(DWARFCompileUnit *CU,
                                                   uint64_t Address,
                                                   FunctionNameKind Kind,
index 65ab594..c3d8ff2 100644 (file)
@@ -440,7 +440,7 @@ DWARFDie DWARFUnit::getSibling(const DWARFDebugInfoEntry *Die) {
   // NULL DIEs don't have siblings.
   if (Die->getAbbreviationDeclarationPtr() == nullptr)
     return DWARFDie();
-  
+
   // Find the next DIE whose depth is the same as the Die's depth.
   for (size_t I = getDIEIndex(Die) + 1, EndIdx = DieArray.size(); I < EndIdx;
        ++I) {
diff --git a/llvm/test/tools/llvm-dwarfdump/X86/lookup.s b/llvm/test/tools/llvm-dwarfdump/X86/lookup.s
new file mode 100644 (file)
index 0000000..47a9e55
--- /dev/null
@@ -0,0 +1,285 @@
+# RUN: llvm-mc %s -filetype obj -triple x86_64-apple-darwin -o - \
+# RUN:   | llvm-dwarfdump -lookup=0x7fffffff - | \
+# RUN: FileCheck %s --check-prefix=EMPTY --allow-empty
+# EMPTY: {{^$}}
+
+# RUN: llvm-mc %s -filetype obj -triple x86_64-apple-darwin -o - \
+# RUN:   | llvm-dwarfdump -lookup=0x4 - | \
+# RUN: FileCheck %s -check-prefixes=CHECK,LEX,A
+
+# RUN: llvm-mc %s -filetype obj -triple x86_64-apple-darwin -o - \
+# RUN:   | llvm-dwarfdump -lookup=0xb - | \
+# RUN: FileCheck %s -check-prefixes=CHECK,LEX,B
+
+# RUN: llvm-mc %s -filetype obj -triple x86_64-apple-darwin -o - \
+# RUN:   | llvm-dwarfdump -lookup=0x14 - | \
+# RUN: FileCheck %s -check-prefixes=CHECK,C
+
+# CHECK: Compile Unit: length = 0x00000060 version = 0x0004 abbr_offset = 0x0000 addr_size = 0x08 (next unit at 0x00000064)
+
+# CHECK: DW_TAG_compile_unit
+# CHECK:   DW_AT_name        ("foo.c")
+# CHECK:   DW_AT_stmt_list   (0x00000000)
+# CHECK:   DW_AT_low_pc      (0x0000000000000000)
+# CHECK:   DW_AT_high_pc     (0x00000016)
+
+# CHECK: DW_TAG_subprogram
+# CHECK:     DW_AT_low_pc    (0x0000000000000000)
+# CHECK:     DW_AT_high_pc   (0x00000016)
+# CHECK:     DW_AT_name      ("foo")
+
+# LEX: DW_TAG_lexical_block
+# LEX:       DW_AT_low_pc  (0x0000000000000004)
+# LEX:       DW_AT_high_pc (0x00000010)
+
+# A: Line info: file 'foo.c', line 3, column 9, start line 1
+# B: Line info: file 'foo.c', line 4, column 6, start line 1
+# C: Line info: file 'foo.c', line 6, column 1, start line 1
+
+       .section        __TEXT,__text,regular,pure_instructions
+       .macosx_version_min 10, 13
+       .globl  _foo                    ## -- Begin function foo
+       .p2align        4, 0x90
+_foo:                                   ## @foo
+Lfunc_begin0:
+       .file   1 "foo.c"
+       .loc    1 1 0                   ## foo.c:1:0
+       .cfi_startproc
+## BB#0:                                ## %entry
+       pushq   %rbp
+Lcfi0:
+       .cfi_def_cfa_offset 16
+Lcfi1:
+       .cfi_offset %rbp, -16
+       movq    %rsp, %rbp
+Lcfi2:
+       .cfi_def_cfa_register %rbp
+Ltmp0:
+       .loc    1 3 9 prologue_end      ## foo.c:3:9
+       movl    $1, -4(%rbp)
+       .loc    1 4 6                   ## foo.c:4:6
+       movl    -4(%rbp), %eax
+       addl    $1, %eax
+       movl    %eax, -4(%rbp)
+Ltmp1:
+       .loc    1 6 1                   ## foo.c:6:1
+       popq    %rbp
+       retq
+Ltmp2:
+Lfunc_end0:
+       .cfi_endproc
+                                        ## -- End function
+       .section        __DWARF,__debug_str,regular,debug
+Linfo_string:
+       .asciz  "clang version 6.0.0 (trunk 314509) (llvm/trunk 314517)" ## string offset=0
+       .asciz  "foo.c"                 ## string offset=55
+       .asciz  "/private/tmp"          ## string offset=61
+       .asciz  "foo"                   ## string offset=74
+       .asciz  "i"                     ## string offset=78
+       .asciz  "int"                   ## string offset=80
+       .section        __DWARF,__debug_abbrev,regular,debug
+Lsection_abbrev:
+       .byte   1                       ## Abbreviation Code
+       .byte   17                      ## DW_TAG_compile_unit
+       .byte   1                       ## DW_CHILDREN_yes
+       .byte   37                      ## DW_AT_producer
+       .byte   14                      ## DW_FORM_strp
+       .byte   19                      ## DW_AT_language
+       .byte   5                       ## DW_FORM_data2
+       .byte   3                       ## DW_AT_name
+       .byte   14                      ## DW_FORM_strp
+       .byte   16                      ## DW_AT_stmt_list
+       .byte   23                      ## DW_FORM_sec_offset
+       .byte   27                      ## DW_AT_comp_dir
+       .byte   14                      ## DW_FORM_strp
+       .byte   17                      ## DW_AT_low_pc
+       .byte   1                       ## DW_FORM_addr
+       .byte   18                      ## DW_AT_high_pc
+       .byte   6                       ## DW_FORM_data4
+       .byte   0                       ## EOM(1)
+       .byte   0                       ## EOM(2)
+       .byte   2                       ## Abbreviation Code
+       .byte   46                      ## DW_TAG_subprogram
+       .byte   1                       ## DW_CHILDREN_yes
+       .byte   17                      ## DW_AT_low_pc
+       .byte   1                       ## DW_FORM_addr
+       .byte   18                      ## DW_AT_high_pc
+       .byte   6                       ## DW_FORM_data4
+       .byte   64                      ## DW_AT_frame_base
+       .byte   24                      ## DW_FORM_exprloc
+       .byte   3                       ## DW_AT_name
+       .byte   14                      ## DW_FORM_strp
+       .byte   58                      ## DW_AT_decl_file
+       .byte   11                      ## DW_FORM_data1
+       .byte   59                      ## DW_AT_decl_line
+       .byte   11                      ## DW_FORM_data1
+       .byte   63                      ## DW_AT_external
+       .byte   25                      ## DW_FORM_flag_present
+       .byte   0                       ## EOM(1)
+       .byte   0                       ## EOM(2)
+       .byte   3                       ## Abbreviation Code
+       .byte   11                      ## DW_TAG_lexical_block
+       .byte   1                       ## DW_CHILDREN_yes
+       .byte   17                      ## DW_AT_low_pc
+       .byte   1                       ## DW_FORM_addr
+       .byte   18                      ## DW_AT_high_pc
+       .byte   6                       ## DW_FORM_data4
+       .byte   0                       ## EOM(1)
+       .byte   0                       ## EOM(2)
+       .byte   4                       ## Abbreviation Code
+       .byte   52                      ## DW_TAG_variable
+       .byte   0                       ## DW_CHILDREN_no
+       .byte   2                       ## DW_AT_location
+       .byte   24                      ## DW_FORM_exprloc
+       .byte   3                       ## DW_AT_name
+       .byte   14                      ## DW_FORM_strp
+       .byte   58                      ## DW_AT_decl_file
+       .byte   11                      ## DW_FORM_data1
+       .byte   59                      ## DW_AT_decl_line
+       .byte   11                      ## DW_FORM_data1
+       .byte   73                      ## DW_AT_type
+       .byte   19                      ## DW_FORM_ref4
+       .byte   0                       ## EOM(1)
+       .byte   0                       ## EOM(2)
+       .byte   5                       ## Abbreviation Code
+       .byte   36                      ## DW_TAG_base_type
+       .byte   0                       ## DW_CHILDREN_no
+       .byte   3                       ## DW_AT_name
+       .byte   14                      ## DW_FORM_strp
+       .byte   62                      ## DW_AT_encoding
+       .byte   11                      ## DW_FORM_data1
+       .byte   11                      ## DW_AT_byte_size
+       .byte   11                      ## DW_FORM_data1
+       .byte   0                       ## EOM(1)
+       .byte   0                       ## EOM(2)
+       .byte   0                       ## EOM(3)
+       .section        __DWARF,__debug_info,regular,debug
+Lsection_info:
+Lcu_begin0:
+       .long   96                      ## Length of Unit
+       .short  4                       ## DWARF version number
+Lset0 = Lsection_abbrev-Lsection_abbrev ## Offset Into Abbrev. Section
+       .long   Lset0
+       .byte   8                       ## Address Size (in bytes)
+       .byte   1                       ## Abbrev [1] 0xb:0x59 DW_TAG_compile_unit
+       .long   0                       ## DW_AT_producer
+       .short  12                      ## DW_AT_language
+       .long   55                      ## DW_AT_name
+Lset1 = Lline_table_start0-Lsection_line ## DW_AT_stmt_list
+       .long   Lset1
+       .long   61                      ## DW_AT_comp_dir
+       .quad   Lfunc_begin0            ## DW_AT_low_pc
+Lset2 = Lfunc_end0-Lfunc_begin0         ## DW_AT_high_pc
+       .long   Lset2
+       .byte   2                       ## Abbrev [2] 0x2a:0x32 DW_TAG_subprogram
+       .quad   Lfunc_begin0            ## DW_AT_low_pc
+Lset3 = Lfunc_end0-Lfunc_begin0         ## DW_AT_high_pc
+       .long   Lset3
+       .byte   1                       ## DW_AT_frame_base
+       .byte   86
+       .long   74                      ## DW_AT_name
+       .byte   1                       ## DW_AT_decl_file
+       .byte   1                       ## DW_AT_decl_line
+                                        ## DW_AT_external
+       .byte   3                       ## Abbrev [3] 0x3f:0x1c DW_TAG_lexical_block
+       .quad   Ltmp0                   ## DW_AT_low_pc
+Lset4 = Ltmp1-Ltmp0                     ## DW_AT_high_pc
+       .long   Lset4
+       .byte   4                       ## Abbrev [4] 0x4c:0xe DW_TAG_variable
+       .byte   2                       ## DW_AT_location
+       .byte   145
+       .byte   124
+       .long   78                      ## DW_AT_name
+       .byte   1                       ## DW_AT_decl_file
+       .byte   3                       ## DW_AT_decl_line
+       .long   92                      ## DW_AT_type
+       .byte   0                       ## End Of Children Mark
+       .byte   0                       ## End Of Children Mark
+       .byte   5                       ## Abbrev [5] 0x5c:0x7 DW_TAG_base_type
+       .long   80                      ## DW_AT_name
+       .byte   5                       ## DW_AT_encoding
+       .byte   4                       ## DW_AT_byte_size
+       .byte   0                       ## End Of Children Mark
+       .section        __DWARF,__debug_ranges,regular,debug
+Ldebug_range:
+       .section        __DWARF,__debug_macinfo,regular,debug
+Ldebug_macinfo:
+Lcu_macro_begin0:
+       .byte   0                       ## End Of Macro List Mark
+       .section        __DWARF,__apple_names,regular,debug
+Lnames_begin:
+       .long   1212240712              ## Header Magic
+       .short  1                       ## Header Version
+       .short  0                       ## Header Hash Function
+       .long   1                       ## Header Bucket Count
+       .long   1                       ## Header Hash Count
+       .long   12                      ## Header Data Length
+       .long   0                       ## HeaderData Die Offset Base
+       .long   1                       ## HeaderData Atom Count
+       .short  1                       ## DW_ATOM_die_offset
+       .short  6                       ## DW_FORM_data4
+       .long   0                       ## Bucket 0
+       .long   193491849               ## Hash in Bucket 0
+       .long   LNames0-Lnames_begin    ## Offset in Bucket 0
+LNames0:
+       .long   74                      ## foo
+       .long   1                       ## Num DIEs
+       .long   42
+       .long   0
+       .section        __DWARF,__apple_objc,regular,debug
+Lobjc_begin:
+       .long   1212240712              ## Header Magic
+       .short  1                       ## Header Version
+       .short  0                       ## Header Hash Function
+       .long   1                       ## Header Bucket Count
+       .long   0                       ## Header Hash Count
+       .long   12                      ## Header Data Length
+       .long   0                       ## HeaderData Die Offset Base
+       .long   1                       ## HeaderData Atom Count
+       .short  1                       ## DW_ATOM_die_offset
+       .short  6                       ## DW_FORM_data4
+       .long   -1                      ## Bucket 0
+       .section        __DWARF,__apple_namespac,regular,debug
+Lnamespac_begin:
+       .long   1212240712              ## Header Magic
+       .short  1                       ## Header Version
+       .short  0                       ## Header Hash Function
+       .long   1                       ## Header Bucket Count
+       .long   0                       ## Header Hash Count
+       .long   12                      ## Header Data Length
+       .long   0                       ## HeaderData Die Offset Base
+       .long   1                       ## HeaderData Atom Count
+       .short  1                       ## DW_ATOM_die_offset
+       .short  6                       ## DW_FORM_data4
+       .long   -1                      ## Bucket 0
+       .section        __DWARF,__apple_types,regular,debug
+Ltypes_begin:
+       .long   1212240712              ## Header Magic
+       .short  1                       ## Header Version
+       .short  0                       ## Header Hash Function
+       .long   1                       ## Header Bucket Count
+       .long   1                       ## Header Hash Count
+       .long   20                      ## Header Data Length
+       .long   0                       ## HeaderData Die Offset Base
+       .long   3                       ## HeaderData Atom Count
+       .short  1                       ## DW_ATOM_die_offset
+       .short  6                       ## DW_FORM_data4
+       .short  3                       ## DW_ATOM_die_tag
+       .short  5                       ## DW_FORM_data2
+       .short  4                       ## DW_ATOM_type_flags
+       .short  11                      ## DW_FORM_data1
+       .long   0                       ## Bucket 0
+       .long   193495088               ## Hash in Bucket 0
+       .long   Ltypes0-Ltypes_begin    ## Offset in Bucket 0
+Ltypes0:
+       .long   80                      ## int
+       .long   1                       ## Num DIEs
+       .long   92
+       .short  36
+       .byte   0
+       .long   0
+
+.subsections_via_symbols
+       .section        __DWARF,__debug_line,regular,debug
+Lsection_line:
+Lline_table_start0:
index de17cf9..fd94242 100644 (file)
@@ -8,6 +8,7 @@ HELP: -eh-frame
 HELP: Specific Options
 HELP: -find
 HELP: -ignore-case
+HELP: -lookup
 HELP: -name
 HELP: -recurse-depth=<N>
 HELP: -regex
index 9dad3b9..04371b7 100644 (file)
@@ -157,6 +157,11 @@ static list<std::string> Name(
          "the -regex option <pattern> is interpreted as a regular expression."),
     value_desc("pattern"), cat(DwarfDumpCategory));
 static alias NameAlias("n", desc("Alias for -name"), aliasopt(Name));
+static opt<unsigned>
+    Lookup("lookup",
+           desc("Lookup <address> in the debug information and print out any"
+                "available file, function, block and line table details."),
+           value_desc("address"), cat(DwarfDumpCategory));
 static opt<std::string>
     OutputFilename("out-file", cl::init(""),
                    cl::desc("Redirect output to the specified file."),
@@ -303,6 +308,30 @@ static void filterByName(const StringSet<> &Names,
           Die.dump(OS, 0, getDumpOpts());
       }
     }
+
+}
+
+/// Handle the --lookup option and dump the DIEs and line info for the given
+/// address.
+static bool lookup(DWARFContext &DICtx, uint64_t Address, raw_ostream &OS) {
+  auto DIEsForAddr = DICtx.getDIEsForAddress(Lookup);
+
+  if (!DIEsForAddr)
+    return false;
+
+  DIDumpOptions DumpOpts = getDumpOpts();
+  DumpOpts.RecurseDepth = 0;
+  DIEsForAddr.CompileUnit->dump(OS, DumpOpts);
+  if (DIEsForAddr.FunctionDIE) {
+    DIEsForAddr.FunctionDIE.dump(OS, 2, DumpOpts);
+    if (DIEsForAddr.BlockDIE)
+      DIEsForAddr.BlockDIE.dump(OS, 4, DumpOpts);
+  }
+
+  if (DILineInfo LineInfo = DICtx.getLineInfoForAddress(Lookup))
+    LineInfo.dump(OS);
+
+  return true;
 }
 
 bool collectStatsForObjectFile(ObjectFile &Obj, DWARFContext &DICtx,
@@ -312,11 +341,14 @@ static bool dumpObjectFile(ObjectFile &Obj, DWARFContext &DICtx, Twine Filename,
                            raw_ostream &OS) {
   logAllUnhandledErrors(DICtx.loadRegisterInfo(Obj), errs(),
                         Filename.str() + ": ");
-
   // The UUID dump already contains all the same information.
   if (!(DumpType & DIDT_UUID) || DumpType == DIDT_All)
     OS << Filename << ":\tfile format " << Obj.getFileFormatName() << '\n';
 
+  // Handle the --lookup option.
+  if (Lookup)
+    return lookup(DICtx, Lookup, OS);
+
   // Handle the --name option.
   if (!Name.empty()) {
     StringSet<> Names;