Major overhaul: MRI C AST -> YAML function defs -> probes

bin/parse

@@ -0,0 +1,7 @@
+#!/usr/bin/env ruby
+
+$LOAD_PATH.unshift File.dirname(__FILE__) + '/../lib'
+
+require 'instrumentation'
+
+Mri::Instrumentation::Parser.probes!( ARGV.first || '/Users/lourens/projects/matzruby' )

config/18/files.yml

@@ -0,0 +1,20 @@
+--- 
+- 'array.c'
+- 'io.c'
+- 'st.c'
+- 'lex.c'
+- 'util.c'
+- 'process.c'
+- 're.c'
+- 'time.c'
+- 'compar.c'
+- 'random.c'
+- 'bignum.c'
+- 'dir.c'
+- 'prec.c'
+- 'struct.c'
+- 'math.c'
+- 'range.c'
+- 'file.c'
+- 'numeric.c'
+- 'signal.c'

config/18/ignores.yml

@@ -0,0 +1,16 @@
+---
+- rb_class_of
+- rb_type
+- rb_special_const_p
+- memfill
+- rb_ary_modify_check
+- rb_ary_elt
+- bigfixize
+- bracket
+- do_stat
+- do_opendir
+- has_magic
+- find_dirsep
+- remove_backslashes
+- glob_make_pattern
+- rb_glob2

config/18/macros.yml

@@ -0,0 +1 @@
+--- {}

config/18/types.yml

@@ -0,0 +1,142 @@
+--- 
+- 'FILE64'
+- 'fd_set'
+- 'st_table_entry'
+- 'rb_atomic_t'
+- 'pointer'     
+- 'hdr'
+- 'C_block'
+- 'sFILE'
+- 'stack_node'
+- 'U'
+- 'Bigint'
+- 'unsigned_time_t'
+- 'VALUE'
+- 'ID'
+- 'RUBY_DATA_FUNC'
+- 'ruby_glob_func'
+- 'rb_nativethread_t'
+- 'mbc_startpos_func_t'
+- 'conftest_type'
+- 'Regexp'
+- 'pid_t'
+- 'sigset_t'
+- 'regex_t'
+- 're_pattern_buffer'
+- 'regoff_t'
+- 'regmatch_t'
+- 'Real'
+- 'uint8_t'
+- 'uint32_t'               
+- 'uint64_t'
+- 'rb_digest_hash_init_func_t'
+- 'rb_digest_hash_update_funct_t'
+- 'rb_digest_hash_finish_funct_t'
+- 'MD5_CTX'
+- 'RMD160_CTX'
+- 'CHAR64LONG16'
+- 'SHA1_CTX'
+- 'sha2_byte'
+- 'sha2_word32'
+- 'sha2_word64'
+- 'SHA256_CTX'
+- 'SHA512_CTX'
+- 'SHA384_CTX'
+- 's_voidp'
+- 's_short'
+- 's_int'
+- 's_long'
+- 's_float'
+- 's_double'
+- 'freefunc_t'
+- 'ANY_TYPE'
+- 'callback1_t'
+- 'nkf_char'
+- 'bfk_nfchar'
+- 'conftest_type'
+- 'd2i_of_void'               
+- 'ossl_asn1_info_t'
+- 'DBM'
+- 'FILE'
+- 'datum'
+- 'YYSTYPE'
+- 'yysigned_char'
+- 'bytestring_t'
+- 'SyckNode'
+- 'SyckParser'
+- 'SyckIoFile'
+- 'SyckIoStr'
+- 'SyckLevel'
+- 'SyckNodeHandler'
+- 'SyckErrorHandler'
+- 'SyckBadAnchorHandler'
+- 'SyckIoFileRead'
+- 'SyckIoStrRead'
+- 'SyckEmitter'
+- 'SyckEmitterNode'
+- 'SyckOutputHandler'
+- 'SyckEmitterHandler'
+- 'bytestring_t'
+- 'yamlbyte_utf8_t'
+- 'yamlbyte_utf16_t'           
+- 'yamlbyte_char_t'
+- 'yamlbyte_consumer_t'
+- 'yamlbyte_producer_t'
+- 'yamlbyte_result_t'
+- 'yamlbyte_buff_t'
+- 'yamlbyte_inst_t'
+- 'yamlbyte_pull_t'
+- 'yamlbyte_pushbuff_t'
+- 'yamlbyte_pullbuff_t'
+- 'Entry'
+- 'List'
+- 'Mutexv'
+- 'ConditionVariable'
+- 'legacy_wait_args'
+- 'Queue'
+- 'DL_HANDLE'
+- 'CallFrame'
+- 'DummyInterp'
+- 'time_t'
+- '_fsize_t'
+- 'intptr_t'
+- 'SIGHANDLER'
+- 'sighandler_t'
+- 'dev_t'
+- 'ino_t'
+- 'mode_t'
+- 'clock_t'
+- 'ptrdiff_t'
+- 'off_t'
+- 'u_char'
+- 'u_short'
+- 'caddr_t'
+- 'transitionTime'
+- 'va_list'
+- 'mbc_startpos_func_t'
+- 'DIR'
+- 'sighandler_t'                  
+- 'NODE'
+- 'rb_event_t'
+- 'rb_event_hook_func_t'
+- 'rb_jmpbuf_t'
+- 'rb_thread_t'                  
+- 'event_hook'
+- 'foobazzz'               
+- 'DWORD'
+- 'size_t'
+- 'USTR'
+- 'RVALUE'
+- '__uint16_t'
+- '__int32_t'
+- '__uint32_t'
+- 'uid_t'
+- 'gid_t'
+- 'blkcnt_t'
+- 'blksize_t'
+- '__int64_t'
+- 'nlink_t'
+- 'stack_type'        
+- 'st_data_t'
+- 'st_table' 
+- 'BDIGITv

lib/instrumentation.rb

@@ -9,12 +9,19 @@ module Instrumentation
 
     MRI_VERSION = RUBY_VERSION.gsub('.','')[0..1]
 
-    PROBES = File.join( File.dirname( __FILE__ ), '..', 'probes', MRI_VERSION ).freeze
+    DIRECTORY = File.expand_path( File.join( File.dirname( __FILE__ ), '..' ) )
+
+    PROBES = File.join( DIRECTORY, 'probes', MRI_VERSION ).freeze
+
+    CONFIG = File.join( DIRECTORY, 'config', MRI_VERSION ).freeze
+
+    EXTENSIONS = File.join( DIRECTORY, 'lib', 'mri', 'instrumentation', 'extensions' ).freeze
 
     autoload :Definition, 'mri/instrumentation/definition'        
     autoload :Probe, 'mri/instrumentation/probe'
     autoload :Argument, 'mri/instrumentation/argument'
     autoload :ProbeCollection, 'mri/instrumentation/probe_collection'
+    autoload :Parser, 'mri/instrumentation/parser'
 
     module Strategy
 
@@ -26,6 +33,7 @@ module Strategy
       autoload :Calls, 'mri/instrumentation/strategy/calls'
       autoload :Speculate, 'mri/instrumentation/strategy/speculate'
       autoload :Peek, 'mri/instrumentation/strategy/peek'
+      autoload :Stack, 'mri/instrumentation/strategy/stack'
 
     end
 

lib/mri/instrumentation/argument.rb

@@ -2,6 +2,9 @@ module Mri
   module Instrumentation
     class Argument < Struct.new(:type, :description, :order)
 
+      class TypeNotDefined < StandardError
+      end
+
       class << self
 
         # Yields the default probe argument
@@ -16,22 +19,76 @@ def probe
       # and integers.
       #
 
-      TYPES = { :char => ['copyinstr( %s )', '%s'],
-                :pointer => ['%s', '%#p'],
-                :int => ['stringof( %s )', '%s'],
-                :probe => ['%s', '%s'] }
+      TYPES = { 'VALUE' => ['%s', 'stringof(%d)'],
+                'VALUE *' => ['%s', '%s'],
+                'long int' => ['%s', 'stringof(%d)'],
+                'void' => ['%s', '%s'],
+                'int' => ['%s', 'stringof(%d)'],
+                'struct inspect_arg *' => ['%s', '%s'],
+                'VALUE (*)()' => ['%s', '%s'],
+                'struct ary_sort_data *' => ['%s', '%s'],
+                'VALUE (*)(VALUE, long int)' => ['%s', '%s'],
+                'void *' => ['%s', '%s'],
+                'struct st_hash_type *' => ['%s', '%s'],
+                'st_table' => ['%s', '%s'],
+                'st_table *' => ['%s', '%s'],
+                'st_data_t' => ['%s', '%s'],
+                'st_data_t *' => ['%s', '%s'],
+                'int (*)()' => ['%s', 'stringof(%d)'],
+                'const char *' => ['%s', '%s'],
+                'unsigned long int' => ['%s', 'stringof(%d)'],
+                'char *' => ['%s', '%s'],
+                'unsigned long long int' => ['%s', 'stringof(%d)'],
+                'long long int' => ['%s', 'stringof(%d)'],
+                'double' => ['%s', 'stringof(%d)'],
+                'double *' => ['%s', 'stringof(%d)'],
+                'const char **' => ['%s', '%s'],
+                'struct dir_data *' => ['%s', '%s'],
+                'struct chdir_data *' => ['%s', '%s'],
+                'struct stat *' => ['%s', '%s'],
+                'DIR *' => ['%s', '%s'],
+                'struct glob_pattern *' => ['%s', '%s'],
+                'enum answer' => ['%s', '%s'],
+                'ruby_glob_func *' => ['%s', '%s'],
+                'void (*)(const char *, VALUE)' => ['%s', '%s'],              
+                'gid_t' => ['%s', '%s'],
+                'OpenFile *' => ['%s', '%s'],
+                'FILE *' => ['%s', '%s'],
+                'ID' => ['%s', 'stringof(%d)'],
+                'struct foreach_arg *' => ['%s', '%s'],
+                'struct kwtable *' => ['%s', '%s'],
+                :probe => ['%s', '%s']
+              }  
+
+      # Hash representation
+      # 
+      def to_hash
+        { self.type => self.description }
+      end
+
+      # YAML representation
+      #
+      def to_yaml( opts = {} )
+        to_hash.to_yaml
+      end      
+
+      # Type lookup accessor
+      #
+      def lookup
+        TYPES[self.type] || raise( TypeNotDefined, "type not defined #{self.type.inspect}" )
+      end
 
       # Format helper for D's printf
       #
       def to_format
-        @format ||= TYPES[self.type.to_sym].last  
+        @format ||= lookup.last  
       end      
 
       # Massage into a format D can easily work with.
       # Standardized on everything being a string, for the time being.
       #
       def massage
-        @massage ||= TYPES[self.type.to_sym].first % to_arg  
+        @massage ||= lookup.first % to_var( '_copy' )
       end      
 
       # String representation as description
@@ -42,14 +99,21 @@ def to_s
 
       # Function variable
       #
-      def to_var
-        @to_var ||= probe? ? "this->type" : "this->#{to_arg}"
+      def to_var( suffix = '' )
+        @to_var ||= {}
+        @to_var[suffix] ||= probe? ? "self->type#{suffix}" : "self->#{to_arg}#{suffix}"
       end
 
       # Function variable assignment
       #
       def to_assignment
-        @to_assignment ||= "#{to_var} = #{massage};"
+        @to_assignment ||= default_assignment #init_assignment
+      end
+
+      # Copy argument to a thread local ( pointers )
+      #
+      def to_copy
+        @to_copy ||= "#{to_var( '_copy' )} = #{to_arg};"
       end
 
       # Argument representation.
@@ -64,6 +128,12 @@ def probe?
         self.order == -1
       end
 
+      private
+
+        def default_assignment
+          "#{to_var} = #{massage};"
+        end
+
     end
   end
 end

lib/mri/instrumentation/definition.rb

@@ -33,12 +33,15 @@ def setup_group
         # Setup all probes
         #    
         def setup_probes
-          read.each_pair do |probe, definition|
-            @probes << setup_probe( probe, 
-                                    safe_definition( definition, "desc" ),
-                                    safe_definition( definition, "args" ),
-                                    safe_definition( definition, "return" ) )
-          end  
+          read.each do |probe|
+            probe.each_pair do |probe, definition|
+              @probes << setup_probe( probe, 
+                                      safe_definition( definition, "desc" ),
+                                      safe_definition( definition, "arguments" ),
+                                      safe_definition( definition, "return" ),
+                                      safe_definition( definition, "storage" ) )
+            end  
+          end
         end
 
         # Don't assume all argument and return definitions is given
@@ -49,8 +52,8 @@ def safe_definition( definition, key )
 
         # Setup a single probe
         #
-        def setup_probe( probe, description, arguments, returns )
-          Mri::Instrumentation::Probe.new( group, probe, description, setup_arguments( arguments ), returns )
+        def setup_probe( probe, description, arguments, returns, storage )
+          Mri::Instrumentation::Probe.new( group, probe, description, setup_arguments( arguments ), returns, storage )
         end  
 
         # Setup arguments for a single probe