Use udis on x86

basis/tools/disassembler/disassembler-docs.factor

@@ -3,11 +3,11 @@ USING: help.markup help.syntax sequences.private ;
 
 HELP: disassemble
 { $values { "obj" "a word or a pair of addresses" } }
-{ $description "Disassembles either a compiled word definition or an arbitrary memory range (in the case " { $snippet "obj" } " is a pair of integers) by attaching " { $snippet "gdb" } " to the Factor VM and capturing the output." }
+{ $description "Disassembles either a compiled word definition or an arbitrary memory range (in the case " { $snippet "obj" } " is a pair of integers)." }
-{ $notes "In some cases the Factor compiler emits data inline with code, which can confuse " { $snippet "gdb" } ". This occurs in words which call " { $link dispatch } ", where the jump table addresses are compiled inline. Also on the ARM architecture, various pointers are often compiled inline, and the preceeding instruction jumps over the inline pinter." } ;
+{ $notes "In some cases the Factor compiler emits data inline with code, which can confuse the disassembler. This occurs in words which call " { $link dispatch } ", where the jump table addresses are compiled inline." } ;
 
 ARTICLE: "tools.disassembler" "Disassembling words"
-"The " { $vocab-link "tools.disassembler" } " vocabulary integrates Factor with the GNU debugger (" { $snippet "gdb" } ") for viewing the assembly code generated by the compiler. It can be used on both Unix and Windows as long as a working copy of " { $snippet "gdb" } " is installed and available in the " { $snippet "PATH" } "."
+"The " { $vocab-link "tools.disassembler" } " vocabulary provides support for disassembling compiled word definitions. It uses the " { $snippet "libudis86" } " library on x86-32 and x86-64, and " { $snippet "gdb" } " on PowerPC."
 { $subsection disassemble } ;
 
 ABOUT: "tools.disassembler"

basis/tools/disassembler/disassembler.factor

@@ -1,43 +1,25 @@
-! Copyright (C) 2008 Slava Pestov, Jorge Acereda Macia.
+! Copyright (C) 2008 Slava Pestov.
 ! See http://factorcode.org/license.txt for BSD license.
-USING: io.files io words alien kernel math.parser alien.syntax
+USING: tr arrays sequences io words generic system combinators
-io.launcher system assocs arrays sequences namespaces make
+vocabs.loader ;
-qualified system math compiler.codegen.fixup
-io.encodings.ascii accessors generic tr ;
 IN: tools.disassembler
 
-: in-file ( -- path ) "gdb-in.txt" temp-file ;
+GENERIC: disassemble ( obj -- )
 
-: out-file ( -- path ) "gdb-out.txt" temp-file ;
+SYMBOL: disassembler-backend
 
-GENERIC: make-disassemble-cmd ( obj -- )
+HOOK: disassemble* disassembler-backend ( from to -- lines )
-
-M: word make-disassemble-cmd
-    word-xt code-format - 2array make-disassemble-cmd ;
-
-M: pair make-disassemble-cmd
-    in-file ascii [
-        "attach " write
-        current-process-handle number>string print
-        "disassemble " write
-        [ number>string write bl ] each
-    ] with-file-writer ;
-
-M: method-spec make-disassemble-cmd
-    first2 method make-disassemble-cmd ;
-
-: gdb-binary ( -- string ) "gdb" ;
-
-: run-gdb ( -- lines )
-    <process>
-        +closed+ >>stdin
-        out-file >>stdout
-        [ gdb-binary , "-x" , in-file , "-batch" , ] { } make >>command
-    try-process
-    out-file ascii file-lines ;
 
 TR: tabs>spaces "\t" "\s" ;
 
-: disassemble ( obj -- )
+M: pair disassemble first2 disassemble* [ tabs>spaces print ] each ;
-    make-disassemble-cmd run-gdb
+
-    [ tabs>spaces ] map [ print ] each ;
+M: word disassemble word-xt 2array disassemble ;
+
+M: method-spec disassemble first2 method disassemble ;
+
+cpu {
+    { x86.32 [ "tools.disassembler.udis" ] }
+    { x86.64 [ "tools.disassembler.udis" ] }
+    { ppc [ "tools.disassembler.gdb" ] }
+} case require

basis/tools/disassembler/gdb/gdb.factor

@@ -0,0 +1,36 @@
+! Copyright (C) 2008 Slava Pestov, Jorge Acereda Macia.
+! See http://factorcode.org/license.txt for BSD license.
+USING: io.files io words alien kernel math.parser alien.syntax
+io.launcher system assocs arrays sequences namespaces make
+qualified system math io.encodings.ascii accessors
+tools.disassembler ;
+IN: tools.disassembler.gdb
+
+SINGLETON: gdb-disassembler
+
+: in-file ( -- path ) "gdb-in.txt" temp-file ;
+
+: out-file ( -- path ) "gdb-out.txt" temp-file ;
+
+: make-disassemble-cmd ( from to -- )
+    in-file ascii [
+        "attach " write
+        current-process-handle number>string print
+        "disassemble " write
+        [ number>string write bl ] bi@
+    ] with-file-writer ;
+
+: gdb-binary ( -- string ) "gdb" ;
+
+: run-gdb ( -- lines )
+    <process>
+        +closed+ >>stdin
+        out-file >>stdout
+        [ gdb-binary , "-x" , in-file , "-batch" , ] { } make >>command
+    try-process
+    out-file ascii file-lines ;
+
+M: gdb-disassembler disassemble*
+    make-disassemble-cmd run-gdb ;
+
+gdb-disassembler disassembler-backend set-global

basis/tools/disassembler/gdb/tags.txt

@@ -0,0 +1 @@
+unportable

basis/tools/disassembler/udis/udis.factor

@@ -0,0 +1,91 @@
+! Copyright (C) 2008 Slava Pestov, Jorge Acereda Macia.
+! See http://factorcode.org/license.txt for BSD license.
+USING: tools.disassembler namespaces combinators
+alien alien.syntax alien.c-types lexer parser kernel
+sequences layouts math math.parser system make fry arrays ;
+IN: tools.disassembler.udis
+
+<< : & scan "c-library" get load-library dlsym parsed ; parsing >>
+
+<<
+"libudis86" {
+    { [ os macosx? ] [ "libudis86.0.dylib" ] }
+    { [ os unix? ] [ "libudis86.so.0" ] }
+    { [ os winnt? ] [ "libudis86.dll" ] }
+} cond "cdecl" add-library
+>>
+
+LIBRARY: libudis86
+
+TYPEDEF: char[592] ud
+
+FUNCTION: void ud_translate_intel ( ud* u ) ;
+FUNCTION: void ud_translate_att ( ud* u ) ;
+
+: UD_SYN_INTEL    & ud_translate_intel ; inline
+: UD_SYN_ATT      & ud_translate_att ; inline
+: UD_EOI          -1 ; inline
+: UD_INP_CACHE_SZ 32 ; inline
+: UD_VENDOR_AMD   0 ; inline
+: UD_VENDOR_INTEL 1 ; inline
+
+FUNCTION: void ud_init ( ud* u ) ;
+FUNCTION: void ud_set_mode ( ud* u, uint8_t mode ) ;
+FUNCTION: void ud_set_pc ( ud* u, ulonglong pc ) ;
+FUNCTION: void ud_set_input_buffer ( ud* u, uint8_t* offset, size_t size ) ;
+FUNCTION: void ud_set_vendor ( ud* u, uint vendor ) ;
+FUNCTION: void ud_set_syntax ( ud* u, void* syntax ) ;
+FUNCTION: void ud_input_skip ( ud* u, size_t size ) ;
+FUNCTION: int ud_input_end ( ud* u ) ;
+FUNCTION: uint ud_decode ( ud* u ) ;
+FUNCTION: uint ud_disassemble ( ud* u ) ;
+FUNCTION: char* ud_insn_asm ( ud* u ) ;
+FUNCTION: void* ud_insn_ptr ( ud* u ) ;
+FUNCTION: ulonglong ud_insn_off ( ud* u ) ;
+FUNCTION: char* ud_insn_hex ( ud* u ) ;
+FUNCTION: uint ud_insn_len ( ud* u ) ;
+FUNCTION: char* ud_lookup_mnemonic ( int c ) ;
+
+: <ud> ( -- ud )
+    "ud" <c-object>
+    dup ud_init
+    dup cell-bits ud_set_mode
+    dup UD_SYN_INTEL ud_set_syntax ;
+
+SINGLETON: udis-disassembler
+
+: buf/len ( from to -- buf len ) [ drop <alien> ] [ swap - ] 2bi ;
+
+: format-disassembly ( lines -- lines' )
+    dup [ second length ] map supremum
+    '[
+        [
+            [ first >hex cell 2 * CHAR: 0 pad-left % ": " % ]
+            [ second _ CHAR: \s pad-right % "  " % ]
+            [ third % ]
+            tri
+        ] "" make
+    ] map ;
+
+: (disassemble) ( ud -- lines )
+    [
+        dup '[
+            _ ud_disassemble 0 =
+            [ f ] [
+                _
+                [ ud_insn_off ]
+                [ ud_insn_hex ]
+                [ ud_insn_asm ]
+                tri 3array , t
+            ] if
+        ] loop
+    ] { } make ;
+
+M: udis-disassembler disassemble* ( from to -- buffer )
+    [ <ud> ] 2dip {
+        [ drop ud_set_pc ]
+        [ buf/len ud_set_input_buffer ]
+        [ 2drop (disassemble) format-disassembly ]
+    } 3cleave ;
+
+udis-disassembler disassembler-backend set-global