Merge pull request #7413 from asterite/opt/dwarf-decoding

Optimize dwarf line numbers decoding

src/callstack.cr

@@ -223,10 +223,7 @@ struct CallStack
       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}"
-          end
+          path = "#{row.directory}/#{row.file}"
           return {path, row.line, row.column}
         end
       end
@@ -279,14 +276,13 @@ struct CallStack
           end
 
           strings = mach_o.read_section?("__debug_str") do |sh, io|
-            Debug::DWARF::Strings.new(io, sh.offset)
+            Debug::DWARF::Strings.new(io, sh.offset, sh.size)
           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
+            while (offset = io.pos - sh.offset) < sh.size
               info = Debug::DWARF::Info.new(io, offset)
 
               mach_o.read_section?("__debug_abbrev") do |sh, io|
@@ -381,14 +377,13 @@ struct CallStack
           end
 
           strings = elf.read_section?(".debug_str") do |sh, io|
-            Debug::DWARF::Strings.new(io, sh.offset)
+            Debug::DWARF::Strings.new(io, sh.offset, sh.size)
           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
+            while (offset = io.pos - sh.offset) < sh.size
               info = Debug::DWARF::Info.new(io, offset)
 
               elf.read_section?(".debug_abbrev") do |sh, io|

src/debug/dwarf/info.cr

@@ -118,7 +118,7 @@ module Debug
         when FORM::RefSig8
           @io.read_bytes(UInt64)
         when FORM::String
-          @io.gets('\0').to_s.chomp('\0')
+          @io.gets('\0', chomp: true).to_s
         when FORM::Strp
           read_ulong
         when FORM::Indirect

src/debug/dwarf/line_numbers.cr

@@ -115,8 +115,8 @@ module Debug
       record Row,
         address : UInt64,
         op_index : UInt32,
-        directory : Int32,
-        file : Int32,
+        directory : String,
+        file : String,
         line : Int32,
         column : Int32,
         end_sequence : Bool
@@ -168,19 +168,11 @@ module Debug
       # reduce the memory usage of repeating a String many times.
       getter matrix : Array(Array(Row))
 
-      # The array of indexed directory paths.
-      getter directories : Array(String)
-
-      # The array of indexed file names.
-      getter files : Array(String)
-
       @offset : LibC::OffT
 
       def initialize(@io : IO::FileDescriptor, size)
         @offset = @io.tell
         @matrix = Array(Array(Row)).new
-        @directories = [] of String
-        @files = [] of String
         decode_sequences(size)
       end
 
@@ -213,10 +205,13 @@ module Debug
 
       # Decodes the compressed matrix of addresses to line numbers.
       private def decode_sequences(size)
-        while (@io.tell - @offset) < size
-          sequence = Sequence.new
+        while true
+          pos = @io.tell
+          offset = pos - @offset
+          break unless offset < size
 
-          sequence.offset = @io.tell - @offset
+          sequence = Sequence.new
+          sequence.offset = offset
           sequence.unit_length = @io.read_bytes(UInt32)
           sequence.version = @io.read_bytes(UInt16)
           sequence.header_length = @io.read_bytes(UInt32)
@@ -249,15 +244,15 @@ module Debug
 
       private def read_directory_table(sequence)
         loop do
-          name = @io.gets('\0').to_s.chomp('\0')
+          name = @io.gets('\0', chomp: true).to_s
           break if name.empty?
           sequence.include_directories << name
         end
       end
 
       private def read_filename_table(sequence)
         loop do
-          name = @io.gets('\0').to_s.chomp('\0')
+          name = @io.gets('\0', chomp: true).to_s
           break if name.empty?
           dir = DWARF.read_unsigned_leb128(@io)
           time = DWARF.read_unsigned_leb128(@io)
@@ -370,8 +365,8 @@ module Debug
         row = Row.new(
           registers.address,
           registers.op_index,
-          register_directory(path),
-          register_filename(file[0]),
+          path,
+          file[0],
           registers.line.to_i,
           registers.column.to_i,
           registers.end_sequence
@@ -388,22 +383,6 @@ module Debug
           @current_sequence_matrix = nil
         end
       end
-
-      private def register_filename(name)
-        if index = @files.index(name)
-          return index
-        end
-        @files << name
-        @files.size - 1
-      end
-
-      private def register_directory(name)
-        if index = @directories.index(name)
-          return index
-        end
-        @directories << name
-        @directories.size - 1
-      end
     end
   end
 end

src/debug/dwarf/strings.cr

@@ -1,12 +1,25 @@
 module Debug
   module DWARF
     struct Strings
-      def initialize(@io : IO::FileDescriptor, @offset : UInt32 | UInt64)
+      def initialize(@io : IO::FileDescriptor, @offset : UInt32 | UInt64, size)
+        # Read a good chunk of bytes to decode strings faster
+        # (avoid seeking/reading the IO too many times)
+        @buffer = Bytes.new(Math.max(16384, size))
+        pos = @io.pos
+        @io.read_fully(@buffer)
+        @io.pos = pos
       end
 
       def decode(strp)
+        # See if we can read it from the buffer
+        if strp < @buffer.size
+          index = @buffer.index('\0'.ord, offset: strp)
+          return String.new(@buffer[strp, index - strp]) if index
+        end
+
+        # If not, try directly from the IO
         @io.seek(@offset + strp) do
-          @io.gets('\0').to_s.chomp('\0')
+          @io.gets('\0', chomp: true).to_s
         end
       end
     end

src/debug/elf.cr

@@ -201,7 +201,7 @@ module Debug
     def sh_name(index)
       sh = section_headers[shstrndx]
       @io.seek(sh.offset + index) do
-        @io.gets('\0').to_s.chomp('\0')
+        @io.gets('\0', chomp: true).to_s
       end
     end
 

src/debug/mach_o.cr

@@ -458,7 +458,7 @@ module Debug
 
           if nlist.strx > 0
             @io.seek(symtab.stroff + nlist.strx) do
-              nlist.name = @io.gets('\0').to_s.chomp('\0')
+              nlist.name = @io.gets('\0', chomp: true).to_s
             end
           end