crypto/perlasm/x86_64-xlate.pl

   1 #!/usr/bin/env perl
   2
   3 # Ascetic x86_64 AT&T to MASM assembler translator by <appro>.
   4 #
   5 # Why AT&T to MASM and not vice versa? Several reasons. Because AT&T
   6 # format is way easier to parse. Because it's simpler to "gear" from
   7 # Unix ABI to Windows one [see cross-reference "card" at the end of
   8 # file]. Because Linux targets were available first...
   9 #
  10 # In addition the script also "distills" code suitable for GNU
  11 # assembler, so that it can be compiled with more rigid assemblers,
  12 # such as Solaris /usr/ccs/bin/as.
  13 #
  14 # This translator is not designed to convert *arbitrary* assembler
  15 # code from AT&T format to MASM one. It's designed to convert just
  16 # enough to provide for dual-ABI OpenSSL modules development...
  17 # There *are* limitations and you might have to modify your assembler
  18 # code or this script to achieve the desired result...
  19 #
  20 # Currently recognized limitations:
  21 #
  22 # - can't use multiple ops per line;
  23 # - indirect calls and jumps are not supported;
  24 #
  25 # Dual-ABI styling rules.
  26 #
  27 # 1. Adhere to Unix register and stack layout [see the end for
  28 #    explanation].
  29 # 2. Forget about "red zone," stick to more traditional blended
  30 #    stack frame allocation. If volatile storage is actually required
  31 #    that is. If not, just leave the stack as is.
  32 # 3. Functions tagged with ".type name,@function" get crafted with
  33 #    unified Win64 prologue and epilogue automatically. If you want
  34 #    to take care of ABI differences yourself, tag functions as
  35 #    ".type name,@abi-omnipotent" instead.
  36 # 4. To optimize the Win64 prologue you can specify number of input
  37 #    arguments as ".type name,@function,N." Keep in mind that if N is
  38 #    larger than 6, then you *have to* write "abi-omnipotent" code,
  39 #    because >6 cases can't be addressed with unified prologue.
  40 # 5. Name local labels as .L*, do *not* use dynamic labels such as 1:
  41 #    (sorry about latter).
  42 # 6. Don't use [or hand-code with .byte] "rep ret." "ret" mnemonic is
  43 #    required to identify the spots, where to inject Win64 epilogue!
  44 #    But on the pros, it's then prefixed with rep automatically:-)
  45
  46 my $output = shift;
  47 open STDOUT,">$output" || die "can't open $output: $!";
  48
  49 my $masm=1 if ($output =~ /\.asm/);
  50
  51 my $current_segment;
  52 my $current_function;
  53
  54 { package opcode;       # pick up opcodes
  55     sub re {
  56         my      $self = shift;  # single instance in enough...
  57         local   *line = shift;
  58         undef   $ret;
  59
  60         if ($line =~ /^([a-z]+)/i) {
  61             $self->{op} = $1;
  62             $ret = $self;
  63             $line = substr($line,@+[0]); $line =~ s/^\s+//;
  64
  65             undef $self->{sz};
  66             if ($self->{op} =~ /(movz)b.*/) {   # movz is pain...
  67                 $self->{op} = $1;
  68                 $self->{sz} = "b";
  69             } elsif ($self->{op} =~ /([a-z]{3,})([qlwb])/) {
  70                 $self->{op} = $1;
  71                 $self->{sz} = $2;
  72             }
  73         }
  74         $ret;
  75     }
  76     sub size {
  77         my $self = shift;
  78         my $sz   = shift;
  79         $self->{sz} = $sz if (defined($sz) && !defined($self->{sz}));
  80         $self->{sz};
  81     }
  82     sub out {
  83         my $self = shift;
  84         if (!$masm) {
  85             if ($self->{op} eq "movz") {        # movz in pain...
  86                 sprintf "%s%s%s",$self->{op},$self->{sz},shift;
  87             } elsif ($self->{op} eq "ret") {
  88                 ".byte  0xf3,0xc3";
  89             } else {
  90                 "$self->{op}$self->{sz}";
  91             }
  92         } else {
  93             $self->{op} =~ s/movz/movzx/;
  94             if ($self->{op} eq "ret") {
  95                 $self->{op} = "";
  96                 if ($current_function->{abi} eq "svr4") {
  97                     $self->{op} = "mov  rdi,QWORD PTR 8[rsp]\t;WIN64 epilogue\n\t".
  98                                   "mov  rsi,QWORD PTR 16[rsp]\n\t";
  99                 }
 100                 $self->{op} .= "DB\t0F3h,0C3h\t\t;repret";
 101             }
 102             $self->{op};
 103         }
 104     }
 105 }
 106 { package const;        # pick up constants, which start with $
 107     sub re {
 108         my      $self = shift;  # single instance in enough...
 109         local   *line = shift;
 110         undef   $ret;
 111
 112         if ($line =~ /^\$([^,]+)/) {
 113             $self->{value} = $1;
 114             $ret = $self;
 115             $line = substr($line,@+[0]); $line =~ s/^\s+//;
 116
 117             $self->{value} = oct($self->{value}) if ($self->{value} =~ /^0/);
 118         }
 119         $ret;
 120     }
 121     sub out {
 122         my $self = shift;
 123         sprintf $masm?"0%xh":"\$0x%x",$self->{value};
 124     }
 125 }
 126 { package ea;           # pick up effective addresses: expr(%reg,%reg,scale)
 127     sub re {
 128         my      $self = shift;  # single instance in enough...
 129         local   *line = shift;
 130         undef   $ret;
 131
 132         if ($line =~ /^([^\(,]*)\(([%\w,]+)\)/) {
 133             $self->{label} = $1;
 134             ($self->{base},$self->{index},$self->{scale})=split(/,/,$2);
 135             $self->{scale} = 1 if (!defined($self->{scale}));
 136             $ret = $self;
 137             $line = substr($line,@+[0]); $line =~ s/^\s+//;
 138
 139             $self->{base}  =~ s/^%//;
 140             $self->{index} =~ s/^%// if (defined($self->{index}));
 141         }
 142         $ret;
 143     }
 144     sub size {}
 145     sub out {
 146         my $self = shift;
 147         my $sz = shift;
 148
 149         if (!$masm) {
 150             # elder GNU assembler insists on 64-bit EAs:-(
 151             # on pros side, this results in more compact code:-)
 152             $self->{index} =~ s/^[er](.?[0-9xp])[d]?$/r\1/;
 153             $self->{base}  =~ s/^[er](.?[0-9xp])[d]?$/r\1/;
 154
 155             if (defined($self->{index})) {
 156                 sprintf "%s(%%%s,%%%s,%d)",
 157                                         $self->{label},$self->{base},
 158                                         $self->{index},$self->{scale};
 159             }
 160             else {
 161                 sprintf "%s(%%%s)",     $self->{label},$self->{base};
 162             }
 163         } else {
 164             %szmap = ( b=>"BYTE", w=>"WORD", l=>"DWORD", q=>"QWORD" );
 165
 166             $self->{label} =~ s/\./\$/g;
 167             $self->{label} =~ s/0x([0-9a-f]+)/0$1h/ig;
 168             $self->{label} = "($self->{label})" if ($self->{label} =~ /[\*\+\-\/]/);
 169
 170             if (defined($self->{index})) {
 171                 sprintf "%s PTR %s[%s*%d+%s]",$szmap{$sz},
 172                                         $self->{label},
 173                                         $self->{index},$self->{scale},
 174                                         $self->{base};
 175             }
 176             else {
 177                 sprintf "%s PTR %s[%s]",$szmap{$sz},
 178                                         $self->{label},$self->{base};
 179             }
 180         }
 181     }
 182 }
 183 { package register;     # pick up registers, which start with %.
 184     sub re {
 185         my      $class = shift; # muliple instances...
 186         my      $self = {};
 187         local   *line = shift;
 188         undef   $ret;
 189
 190         if ($line =~ /^%(\w+)/) {
 191             bless $self,$class;
 192             $self->{value} = $1;
 193             $ret = $self;
 194             $line = substr($line,@+[0]); $line =~ s/^\s+//;
 195         }
 196         $ret;
 197     }
 198     sub size {
 199         my      $self = shift;
 200         undef   $ret;
 201
 202         if    ($self->{value} =~ /^r[\d]+b$/i)  { $ret="b"; }
 203         elsif ($self->{value} =~ /^r[\d]+w$/i)  { $ret="w"; }
 204         elsif ($self->{value} =~ /^r[\d]+d$/i)  { $ret="l"; }
 205         elsif ($self->{value} =~ /^r[\w]+$/i)   { $ret="q"; }
 206         elsif ($self->{value} =~ /^[a-d][hl]$/i){ $ret="b"; }
 207         elsif ($self->{value} =~ /^[\w]{2}l$/i) { $ret="b"; }
 208         elsif ($self->{value} =~ /^[\w]{2}$/i)  { $ret="w"; }
 209         elsif ($self->{value} =~ /^e[a-z]{2}$/i){ $ret="l"; }
 210
 211         $ret;
 212     }
 213     sub out {
 214         my $self = shift;
 215         sprintf $masm?"%s":"%%%s",$self->{value};
 216     }
 217 }
 218 { package label;        # pick up labels, which end with :
 219     sub re {
 220         my      $self = shift;  # single instance is enough...
 221         local   *line = shift;
 222         undef   $ret;
 223
 224         if ($line =~ /(^[\.\w]+\:)/) {
 225             $self->{value} = $1;
 226             $ret = $self;
 227             $line = substr($line,@+[0]); $line =~ s/^\s+//;
 228
 229             $self->{value} =~ s/\.L/\$L/ if ($masm);
 230         }
 231         $ret;
 232     }
 233     sub out {
 234         my $self = shift;
 235
 236         if (!$masm) {
 237             $self->{value};
 238         } elsif ($self->{value} ne "$current_function->{name}:") {
 239             $self->{value};
 240         } elsif ($current_function->{abi} eq "svr4") {
 241             my $func =  "$current_function->{name}      PROC\n".
 242                         "       mov     QWORD PTR 8[rsp],rdi\t;WIN64 prologue\n".
 243                         "       mov     QWORD PTR 16[rsp],rsi\n";
 244             my $narg = $current_function->{narg};
 245             $narg=6 if (!defined($narg));
 246             $func .= "  mov     rdi,rcx\n" if ($narg>0);
 247             $func .= "  mov     rsi,rdx\n" if ($narg>1);
 248             $func .= "  mov     rdx,r8\n"  if ($narg>2);
 249             $func .= "  mov     rcx,r9\n"  if ($narg>3);
 250             $func .= "  mov     r8,QWORD PTR 40[rsp]\n" if ($narg>4);
 251             $func .= "  mov     r9,QWORD PTR 48[rsp]\n" if ($narg>5);
 252             $func .= "\n";
 253         } else {
 254            "$current_function->{name}   PROC";
 255         }
 256     }
 257 }
 258 { package expr;         # pick up expressioins
 259     sub re {
 260         my      $self = shift;  # single instance is enough...
 261         local   *line = shift;
 262         undef   $ret;
 263
 264         if ($line =~ /(^[^,]+)/) {
 265             $self->{value} = $1;
 266             $ret = $self;
 267             $line = substr($line,@+[0]); $line =~ s/^\s+//;
 268
 269             $self->{value} =~ s/\.L/\$L/g if ($masm);
 270         }
 271         $ret;
 272     }
 273     sub out {
 274         my $self = shift;
 275         $self->{value};
 276     }
 277 }
 278 { package directive;    # pick up directives, which start with .
 279     sub re {
 280         my      $self = shift;  # single instance is enough...
 281         local   *line = shift;
 282         undef   $ret;
 283         my      $dir;
 284
 285         if ($line =~ /^\s*(\.\w+)/) {
 286             if (!$masm) {
 287                 $self->{value} = $1;
 288                 $line =~ s/\@abi\-omnipotent/\@function/;
 289                 $line =~ s/\@function.*/\@function/;
 290                 $self->{value} = $line;
 291                 $line = "";
 292                 return $self;
 293             }
 294
 295             $dir = $1;
 296             $ret = $self;
 297             undef $self->{value};
 298             $line = substr($line,@+[0]); $line =~ s/^\s+//;
 299             SWITCH: for ($dir) {
 300                 /\.(text|data)/
 301                             && do { my $v=undef;
 302                                     $v="$current_segment\tENDS\n" if ($current_segment);
 303                                     $current_segment = "_$1";
 304                                     $current_segment =~ tr/[a-z]/[A-Z]/;
 305                                     $v.="$current_segment\tSEGMENT PARA";
 306                                     $self->{value} = $v;
 307                                     last;
 308                                   };
 309                 /\.globl/   && do { $self->{value} = "PUBLIC\t".$line; last; };
 310                 /\.type/    && do { ($sym,$type,$narg) = split(',',$line);
 311                                     if ($type eq "\@function")
 312                                     {   undef $current_function;
 313                                         $current_function->{name} = $sym;
 314                                         $current_function->{abi}  = "svr4";
 315                                         $current_function->{narg} = $narg;
 316                                     }
 317                                     elsif ($type eq "\@abi-omnipotent")
 318                                     {   undef $current_function;
 319                                         $current_function->{name} = $sym;
 320                                     }
 321                                     last;
 322                                   };
 323                 /\.size/    && do { if (defined($current_function))
 324                                     {   $self->{value}="$current_function->{name}\tENDP";
 325                                         undef $current_function;
 326                                     }
 327                                     last;
 328                                   };
 329                 /\.align/   && do { $self->{value} = "ALIGN\t".$line; last; };
 330                 /\.(byte|value|long|quad)/
 331                             && do { my @arr = split(',',$line);
 332                                     my $sz  = substr($1,0,1);
 333                                     my $last = pop(@arr);
 334
 335                                     $sz =~ tr/bvlq/BWDQ/;
 336                                     $self->{value} = "\tD$sz\t";
 337                                     for (@arr) { $self->{value} .= sprintf"0%Xh,",oct; }
 338                                     $self->{value} .= sprintf"0%Xh",oct($last);
 339                                     last;
 340                                   };
 341             }
 342             $line = "";
 343         }
 344
 345         $ret;
 346     }
 347     sub out {
 348         my $self = shift;
 349         $self->{value};
 350     }
 351 }
 352
 353 while($line=<>) {
 354
 355     chomp($line);
 356
 357     $line =~ s|[#!].*$||;       # get rid of asm-style comments...
 358     $line =~ s|/\*.*\*/||;      # ... and C-style comments...
 359     $line =~ s|^\s+||;          # ... and skip white spaces in beginning
 360
 361     undef $label;
 362     undef $opcode;
 363     undef $dst;
 364     undef $src;
 365     undef $sz;
 366
 367     if ($label=label->re(\$line))       { print $label->out(); }
 368
 369     if (directive->re(\$line)) {
 370         printf "%s",directive->out();
 371     } elsif ($opcode=opcode->re(\$line)) { ARGUMENT: {
 372
 373         if ($src=register->re(\$line))  { opcode->size($src->size()); }
 374         elsif ($src=const->re(\$line))  { }
 375         elsif ($src=ea->re(\$line))     { }
 376         elsif ($src=expr->re(\$line))   { }
 377
 378         last ARGUMENT if ($line !~ /^,/);
 379
 380         $line = substr($line,1); $line =~ s/^\s+//;
 381
 382         if ($dst=register->re(\$line))  { opcode->size($dst->size()); }
 383         elsif ($dst=const->re(\$line))  { }
 384         elsif ($dst=ea->re(\$line))     { }
 385
 386         } # ARGUMENT:
 387
 388         $sz=opcode->size();
 389
 390         if (defined($dst)) {
 391             if (!$masm) {
 392                 printf "\t%s\t%s,%s",   $opcode->out($dst->size()),
 393                                         $src->out($sz),$dst->out($sz);
 394             }
 395             else {
 396                 printf "\t%s\t%s,%s",   $opcode->out(),
 397                                         $dst->out($sz),$src->out($sz);
 398             }
 399         }
 400         elsif (defined($src)) {
 401             printf "\t%s\t%s",$opcode->out(),$src->out($sz);
 402         } else {
 403             printf "\t%s",$opcode->out();
 404         }
 405     }
 406
 407     print $line,"\n";
 408 }
 409
 410 print "\n$current_segment\tENDS\nEND\n" if ($masm);
 411
 412 close STDOUT;
 413
 414 #################################################
 415 # Cross-reference x86_64 ABI "card"
 416 #
 417 #               Unix            Win64
 418 # %rax          *               *
 419 # %rbx          -               -
 420 # %rcx          #4              #1
 421 # %rdx          #3              #2
 422 # %rsi          #2              -
 423 # %rdi          #1              -
 424 # %rbp          -               -
 425 # %rsp          -               -
 426 # %r8           #5              #3
 427 # %r9           #6              #4
 428 # %r10          *               *
 429 # %r11          *               *
 430 # %r12          -               -
 431 # %r13          -               -
 432 # %r14          -               -
 433 # %r15          -               -
 434 #
 435 # (*)   volatile register
 436 # (-)   preserved by callee
 437 # (#)   Nth argument, volatile
 438 #
 439 # In Unix terms top of stack is argument transfer area for arguments
 440 # which could not be accomodated in registers. Or in other words 7th
 441 # [integer] argument resides at 8(%rsp) upon function entry point.
 442 # 128 bytes above %rsp constitute a "red zone" which is not touched
 443 # by signal handlers and can be used as temporal storage without
 444 # allocating a frame.
 445 #
 446 # In Win64 terms N*8 bytes on top of stack is argument transfer area,
 447 # which belongs to/can be overwritten by callee. N is the number of
 448 # arguments passed to callee, *but* not less than 4! This means that
 449 # upon function entry point 5th argument resides at 40(%rsp), as well
 450 # as that 32 bytes from 8(%rsp) can always be used as temporal
 451 # storage [without allocating a frame].
 452 #
 453 # All the above means that if assembler programmer adheres to Unix
 454 # register and stack layout, but disregards the "red zone" existense,
 455 # it's possible to use following prologue and epilogue to "gear" from
 456 # Unix to Win64 ABI in leaf functions with not more than 6 arguments.
 457 #
 458 # omnipotent_function:
 459 # ifdef WIN64
 460 #       movq    %rdi,8(%rsp)
 461 #       movq    %rsi,16(%rsp)
 462 #       movq    %rcx,%rdi       ; if 1st argument is actually present
 463 #       movq    %rdx,%rsi       ; if 2nd argument is actually ...
 464 #       movq    %r8,%rdx        ; if 3rd argument is ...
 465 #       movq    %r9,%rcx        ; if 4th argument ...
 466 #       movq    40(%rsp),%r8    ; if 5th ...
 467 #       movq    48(%rsp),%r9    ; if 6th ...
 468 # endif
 469 #       ...
 470 # ifdef WIN64
 471 #       movq    8(%rsp),%rdi
 472 #       movq    16(%rsp),%rsi
 473 # endif
 474 #       ret