Debug::DWARF: more decoding optimizations

src/callstack.cr

@@ -250,7 +250,7 @@ struct CallStack
         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?
+            value = strings.try(&.decode(value.as(UInt32 | UInt64).to_u64)) if form.strp?
             name = value.as(String)
           when Debug::DWARF::AT::DW_AT_low_pc
             low_pc = value.as(LibC::SizeT)
@@ -279,14 +279,17 @@ 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.to_u64, sh.size.to_u64)
           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 true
+              pos = io.pos
+              offset = pos - sh.offset
+              break unless offset < sh.size
+
               info = Debug::DWARF::Info.new(io, offset)
 
               mach_o.read_section?("__debug_abbrev") do |sh, io|
@@ -381,14 +384,17 @@ 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 true
+              pos = io.pos
+              offset = pos - sh.offset
+              break if offset < sh.size
+
               info = Debug::DWARF::Info.new(io, offset)
 
               elf.read_section?(".debug_abbrev") do |sh, io|

src/debug/dwarf/info.cr

@@ -4,9 +4,9 @@ require "./abbrev"
 module Debug
   module DWARF
     struct Info
-      property unit_length : UInt32 | UInt64
+      property unit_length : UInt64
       property version : UInt16
-      property debug_abbrev_offset : UInt32 | UInt64
+      property debug_abbrev_offset : UInt64
       property address_size : UInt8
       property! abbreviations : Array(Abbrev)
 
@@ -17,7 +17,7 @@ module Debug
       def initialize(@io : IO::FileDescriptor, @offset)
         @ref_offset = offset
 
-        @unit_length = @io.read_bytes(UInt32)
+        @unit_length = @io.read_bytes(UInt32).to_u64
         if @unit_length == 0xffffffff
           @dwarf64 = true
           @unit_length = @io.read_bytes(UInt64)
@@ -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
@@ -133,7 +133,7 @@ module Debug
         if @dwarf64
           @io.read_bytes(UInt64)
         else
-          @io.read_bytes(UInt32)
+          @io.read_bytes(UInt32).to_u64
         end
       end
     end

src/debug/dwarf/line_numbers.cr

@@ -219,10 +219,13 @@ module Debug
           filename:  {} of String => Int32,
         }
 
-        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)
@@ -255,15 +258,15 @@ 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)

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 : UInt64, @size : UInt64)
+        # 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)
+      def decode(strp : UInt64)
+        # 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

@@ -117,13 +117,13 @@ module Debug
       property! name : UInt32
       property! type : Type
       property! flags : Flags
-      property! addr : UInt32 | UInt64
-      property! offset : UInt32 | UInt64
-      property! size : UInt32 | UInt64
+      property! addr : UInt64
+      property! offset : UInt64
+      property! size : UInt64
       property! link : UInt32
       property! info : UInt32
-      property! addralign : UInt32 | UInt64
-      property! entsize : UInt32 | UInt64
+      property! addralign : UInt64
+      property! entsize : UInt64
     end
 
     class Error < Exception
@@ -133,9 +133,9 @@ module Debug
     property! type : Type
     property! machine : Machine
     property! version : UInt32
-    property! entry : UInt32 | UInt64
-    property! phoff : UInt32 | UInt64
-    property! shoff : UInt32 | UInt64
+    property! entry : UInt64
+    property! phoff : UInt64
+    property! shoff : UInt64
     property! flags : UInt32
     property! ehsize : UInt16
     property! phentsize : UInt16
@@ -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
 
@@ -226,7 +226,7 @@ module Debug
 
     private def read_ulong
       case ident.klass
-      when Klass::ELF32 then @io.read_bytes(UInt32, endianness)
+      when Klass::ELF32 then @io.read_bytes(UInt32, endianness).to_u64
       when Klass::ELF64 then @io.read_bytes(UInt64, endianness)
       else                   raise Error.new("Unsupported")
       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