Backtrace: get function name from DWARF sections

spec/std/callstack_spec.cr

@@ -17,9 +17,9 @@ describe "Backtrace" do
     output = `#{tempfile.path}`
 
     # resolved file line:column
-    output.should match(/#{sample} 3:10/) # callee1
-    output.should match(/#{sample} 15:3/) # callee3
-    output.should match(/#{sample} 17:1/) # ???
+    output.should match(/callee1 at #{sample} 3:10/)
+    output.should match(/callee3 at #{sample} 15:3/)
+    output.should match(/__crystal_main at #{sample} 17:1/)
 
     # skipped internal details
     output.should_not match(/src\/callstack\.cr/)

src/callstack.cr

@@ -146,15 +146,19 @@ struct CallStack
 
   private def decode_backtrace
     @callstack.compact_map do |ip|
-      file, line, column = CallStack.decode_line_number(ip)
+      pc = CallStack.decode_address(ip)
+
+      file, line, column = CallStack.decode_line_number(pc)
       if file == "??"
         file_line_column = "??"
       else
         next if @@skip.includes?(file)
         file_line_column = "#{file} #{line}:#{column}"
       end
 
-      if frame = CallStack.decode_frame(ip)
+      if name = CallStack.decode_function_name(pc)
+        function = name
+      elsif frame = CallStack.decode_frame(ip)
         _, sname = frame
         function = String.new(sname)
       else
@@ -167,10 +171,12 @@ struct CallStack
 
   {% if flag?(:darwin) || flag?(:freebsd) || flag?(:linux) || flag?(:openbsd) %}
     @@dwarf_line_numbers : Debug::DWARF::LineNumbers?
+    @@dwarf_function_names : Array(Tuple(LibC::SizeT, LibC::SizeT, String))?
 
-    protected def self.decode_line_number(ip)
-      if ln = dwarf_line_numbers
-        if row = ln.find(decode_address(ip))
+    protected def self.decode_line_number(pc)
+      read_dwarf_sections unless @@dwarf_line_numbers
+      if ln = @@dwarf_line_numbers
+        if row = ln.find(pc)
           path = ln.files[row.file]?
           if dirname = ln.directories[row.directory]?
             path = "#{dirname}/#{path}"
@@ -181,13 +187,72 @@ struct CallStack
       {"??", 0, 0}
     end
 
+    protected def self.decode_function_name(pc)
+      read_dwarf_sections unless @@dwarf_function_names
+      if fn = @@dwarf_function_names
+        fn.each do |(low_pc, high_pc, function_name)|
+          return function_name if low_pc <= pc <= high_pc
+        end
+      end
+    end
+
+    protected def self.parse_function_names_from_dwarf(info, strings)
+      info.each do |code, abbrev, attributes|
+        next unless abbrev && abbrev.tag.subprogram?
+        name = low_pc = high_pc = nil
+
+        attributes.each do |(at, form, value)|
+          case at
+          when Debug::DWARF::AT::DW_AT_name
+            value = strings.try(&.decode(value.as(UInt32 | UInt64)))  if form.strp?
+            name = value.as(String)
+          when Debug::DWARF::AT::DW_AT_low_pc
+            low_pc = value.as(LibC::SizeT)
+          when Debug::DWARF::AT::DW_AT_high_pc
+            if form.addr?
+              high_pc = value.as(LibC::SizeT)
+            elsif value.responds_to?(:to_i)
+              high_pc = low_pc.as(LibC::SizeT) + value.to_i
+            end
+          end
+        end
+
+        if low_pc && high_pc && name
+          yield low_pc, high_pc, name
+        end
+      end
+    end
+
     {% if flag?(:darwin) %}
       @@image_slide : LibC::Long?
 
-      protected def self.dwarf_line_numbers
-        @@dwarf_line_numbers ||= locate_dsym_bundle do |mach_o|
+      protected def self.read_dwarf_sections
+        locate_dsym_bundle do |mach_o|
           mach_o.read_section?("__debug_line") do |sh, io|
-            Debug::DWARF::LineNumbers.new(io, sh.size)
+            @@dwarf_line_numbers = Debug::DWARF::LineNumbers.new(io, sh.size)
+          end
+
+          strings = mach_o.read_section?("__debug_str") do |sh, io|
+            Debug::DWARF::Strings.new(io, sh.offset)
+          end
+
+          mach_o.read_section?("__debug_info") do |sh, io|
+            names = [] of {LibC::SizeT, LibC::SizeT, String}
+
+            while io.tell - sh.offset < sh.size
+              offset = io.tell - sh.offset
+              info = Debug::DWARF::Info.new(io, offset)
+
+              mach_o.read_section?("__debug_abbrev") do |sh, io|
+                info.read_abbreviations(io)
+              end
+
+              parse_function_names_from_dwarf(info, strings) do |name, low_pc, high_pc|
+                names << {name, low_pc, high_pc}
+              end
+            end
+
+            @@dwarf_function_names = names
           end
         end
       end
@@ -259,14 +324,37 @@ struct CallStack
     {% else %}
       @@base_address : UInt64|UInt32|Nil
 
-      protected def self.dwarf_line_numbers
-        @@dwarf_line_numbers ||= Debug::ELF.open(PROGRAM_NAME) do |elf|
+      protected def self.read_dwarf_sections
+        Debug::ELF.open(PROGRAM_NAME) do |elf|
           elf.read_section?(".text") do |sh, _|
             @@base_address = sh.addr - sh.offset
           end
 
           elf.read_section?(".debug_line") do |sh, io|
-            Debug::DWARF::LineNumbers.new(io, sh.size)
+            @@dwarf_line_numbers = Debug::DWARF::LineNumbers.new(io, sh.size)
+          end
+
+          strings = elf.read_section?(".debug_str") do |sh, io|
+            Debug::DWARF::Strings.new(io, sh.offset)
+          end
+
+          elf.read_section?(".debug_info") do |sh, io|
+            names = [] of {LibC::SizeT, LibC::SizeT, String}
+
+            while io.tell - sh.offset < sh.size
+              offset = io.tell - sh.offset
+              info = Debug::DWARF::Info.new(io, offset)
+
+              elf.read_section?(".debug_abbrev") do |sh, io|
+                info.read_abbreviations(io)
+              end
+
+              parse_function_names_from_dwarf(info, strings) do |name, low_pc, high_pc|
+                names << {name, low_pc, high_pc}
+              end
+            end
+
+            @@dwarf_function_names = names
           end
         end
       end
@@ -287,9 +375,17 @@ struct CallStack
       end
     {% end %}
   {% else %}
-    def self.decode_line_number(ip)
+    def self.decode_address(ip)
+      ip
+    end
+
+    def self.decode_line_number(pc)
       {"??", 0, 0}
     end
+
+    def self.decode_function_name(pc)
+      nil
+    end
   {% end %}
 
   protected def self.decode_frame(ip, original_ip = ip)

src/debug/dwarf/info.cr

@@ -11,11 +11,11 @@ module Debug
       property! abbreviations : Array(Abbrev)
 
       property dwarf64 : Bool
-      @offset : Int64
-      @ref_offset : UInt64
+      @offset : LibC::OffT
+      @ref_offset : LibC::OffT
 
       def initialize(@io : IO::FileDescriptor, @offset)
-        @ref_offset = offset.to_u64
+        @ref_offset = offset
 
         @unit_length = @io.read_bytes(UInt32)
         if @unit_length == 0xffffffff
@@ -31,7 +31,7 @@ module Debug
         @address_size = @io.read_byte.not_nil!
       end
 
-      alias Value = Bool | Int32 | Slice(UInt8) | String | UInt16 | UInt32 | UInt64 | UInt8
+      alias Value = Bool | Int32 | Int64 | Slice(UInt8) | String | UInt16 | UInt32 | UInt64 | UInt8
 
       def read_abbreviations(io)
         @abbreviations = Abbrev.read(io, debug_abbrev_offset)