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 # 7. Due to MASM limitations [and certain general counter-intuitivity
  46 #    of ip-relative addressing] generation of position-independent
  47 #    code is assisted by synthetic directive, .picmeup, which puts
  48 #    address of the *next* instruction into target register.
  49 #
  50 #    Example 1:
  51 #               .picmeup        %rax
  52 #               lea             .Label-.(%rax),%rax
  53 #    Example 2:
  54 #               .picmeup        %rcx
  55 #       .Lpic_point:
  56 #               ...
  57 #               lea             .Label-.Lpic_point(%rcx),%rbp
  58
  59 my $output = shift;
  60 open STDOUT,">$output" || die "can't open $output: $!";
  61
  62 my $masm=1 if ($output =~ /\.asm/);
  63
  64 my $current_segment;
  65 my $current_function;
  66
  67 { package opcode;       # pick up opcodes
  68     sub re {
  69         my      $self = shift;  # single instance in enough...
  70         local   *line = shift;
  71         undef   $ret;
  72
  73         if ($line =~ /^([a-z]+)/i) {
  74             $self->{op} = $1;
  75             $ret = $self;
  76             $line = substr($line,@+[0]); $line =~ s/^\s+//;
  77
  78             undef $self->{sz};
  79             if ($self->{op} =~ /(movz)b.*/) {   # movz is pain...
  80                 $self->{op} = $1;
  81                 $self->{sz} = "b";
  82             } elsif ($self->{op} =~ /([a-z]{3,})([qlwb])/) {
  83                 $self->{op} = $1;
  84                 $self->{sz} = $2;
  85             }
  86         }
  87         $ret;
  88     }
  89     sub size {
  90         my $self = shift;
  91         my $sz   = shift;
  92         $self->{sz} = $sz if (defined($sz) && !defined($self->{sz}));
  93         $self->{sz};
  94     }
  95     sub out {
  96         my $self = shift;
  97         if (!$masm) {
  98             if ($self->{op} eq "movz") {        # movz in pain...
  99                 sprintf "%s%s%s",$self->{op},$self->{sz},shift;
 100             } elsif ($self->{op} eq "ret") {
 101                 ".byte  0xf3,0xc3";
 102             } else {
 103                 "$self->{op}$self->{sz}";
 104             }
 105         } else {
 106             $self->{op} =~ s/movz/movzx/;
 107             if ($self->{op} eq "ret") {
 108                 $self->{op} = "";
 109                 if ($current_function->{abi} eq "svr4") {
 110                     $self->{op} = "mov  rdi,QWORD PTR 8[rsp]\t;WIN64 epilogue\n\t".
 111                                   "mov  rsi,QWORD PTR 16[rsp]\n\t";
 112                 }
 113                 $self->{op} .= "DB\t0F3h,0C3h\t\t;repret";
 114             }
 115             $self->{op};
 116         }
 117     }
 118 }
 119 { package const;        # pick up constants, which start with $
 120     sub re {
 121         my      $self = shift;  # single instance in enough...
 122         local   *line = shift;
 123         undef   $ret;
 124
 125         if ($line =~ /^\$([^,]+)/) {
 126             $self->{value} = $1;
 127             $ret = $self;
 128             $line = substr($line,@+[0]); $line =~ s/^\s+//;
 129         }
 130         $ret;
 131     }
 132     sub out {
 133         my $self = shift;
 134
 135         if (!$masm) {
 136             sprintf "\$%s",$self->{value};
 137         } else {
 138             $self->{value} =~ s/0x([0-9a-f]+)/0$1h/ig;
 139             sprintf "%s",$self->{value};
 140         }
 141     }
 142 }
 143 { package ea;           # pick up effective addresses: expr(%reg,%reg,scale)
 144     sub re {
 145         my      $self = shift;  # single instance in enough...
 146         local   *line = shift;
 147         undef   $ret;
 148
 149         if ($line =~ /^([^\(,]*)\(([%\w,]+)\)/) {
 150             $self->{label} = $1;
 151             ($self->{base},$self->{index},$self->{scale})=split(/,/,$2);
 152             $self->{scale} = 1 if (!defined($self->{scale}));
 153             $ret = $self;
 154             $line = substr($line,@+[0]); $line =~ s/^\s+//;
 155
 156             $self->{base}  =~ s/^%//;
 157             $self->{index} =~ s/^%// if (defined($self->{index}));
 158         }
 159         $ret;
 160     }
 161     sub size {}
 162     sub out {
 163         my $self = shift;
 164         my $sz = shift;
 165
 166         if (!$masm) {
 167             # elder GNU assembler insists on 64-bit EAs:-(
 168             # on pros side, this results in more compact code:-)
 169             $self->{index} =~ s/^[er](.?[0-9xp])[d]?$/r\1/;
 170             $self->{base}  =~ s/^[er](.?[0-9xp])[d]?$/r\1/;
 171
 172             if (defined($self->{index})) {
 173                 sprintf "%s(%%%s,%%%s,%d)",
 174                                         $self->{label},$self->{base},
 175                                         $self->{index},$self->{scale};
 176             } else {
 177                 sprintf "%s(%%%s)",     $self->{label},$self->{base};
 178             }
 179         } else {
 180             %szmap = ( b=>"BYTE", w=>"WORD", l=>"DWORD", q=>"QWORD" );
 181
 182             $self->{label} =~ s/\./\$/g;
 183             $self->{label} =~ s/0x([0-9a-f]+)/0$1h/ig;
 184             $self->{label} = "($self->{label})" if ($self->{label} =~ /[\*\+\-\/]/);
 185
 186             if (defined($self->{index})) {
 187                 sprintf "%s PTR %s[%s*%d+%s]",$szmap{$sz},
 188                                         $self->{label},
 189                                         $self->{index},$self->{scale},
 190                                         $self->{base};
 191             } else {
 192                 sprintf "%s PTR %s[%s]",$szmap{$sz},
 193                                         $self->{label},$self->{base};
 194             }
 195         }
 196     }
 197 }
 198 { package register;     # pick up registers, which start with %.
 199     sub re {
 200         my      $class = shift; # muliple instances...
 201         my      $self = {};
 202         local   *line = shift;
 203         undef   $ret;
 204
 205         if ($line =~ /^%(\w+)/) {
 206             bless $self,$class;
 207             $self->{value} = $1;
 208             $ret = $self;
 209             $line = substr($line,@+[0]); $line =~ s/^\s+//;
 210         }
 211         $ret;
 212     }
 213     sub size {
 214         my      $self = shift;
 215         undef   $ret;
 216
 217         if    ($self->{value} =~ /^r[\d]+b$/i)  { $ret="b"; }
 218         elsif ($self->{value} =~ /^r[\d]+w$/i)  { $ret="w"; }
 219         elsif ($self->{value} =~ /^r[\d]+d$/i)  { $ret="l"; }
 220         elsif ($self->{value} =~ /^r[\w]+$/i)   { $ret="q"; }
 221         elsif ($self->{value} =~ /^[a-d][hl]$/i){ $ret="b"; }
 222         elsif ($self->{value} =~ /^[\w]{2}l$/i) { $ret="b"; }
 223         elsif ($self->{value} =~ /^[\w]{2}$/i)  { $ret="w"; }
 224         elsif ($self->{value} =~ /^e[a-z]{2}$/i){ $ret="l"; }
 225
 226         $ret;
 227     }
 228     sub out {
 229         my $self = shift;
 230         sprintf $masm?"%s":"%%%s",$self->{value};
 231     }
 232 }
 233 { package label;        # pick up labels, which end with :
 234     sub re {
 235         my      $self = shift;  # single instance is enough...
 236         local   *line = shift;
 237         undef   $ret;
 238
 239         if ($line =~ /(^[\.\w]+\:)/) {
 240             $self->{value} = $1;
 241             $ret = $self;
 242             $line = substr($line,@+[0]); $line =~ s/^\s+//;
 243
 244             $self->{value} =~ s/\.L/\$L/ if ($masm);
 245         }
 246         $ret;
 247     }
 248     sub out {
 249         my $self = shift;
 250
 251         if (!$masm) {
 252             $self->{value};
 253         } elsif ($self->{value} ne "$current_function->{name}:") {
 254             $self->{value};
 255         } elsif ($current_function->{abi} eq "svr4") {
 256             my $func =  "$current_function->{name}      PROC\n".
 257                         "       mov     QWORD PTR 8[rsp],rdi\t;WIN64 prologue\n".
 258                         "       mov     QWORD PTR 16[rsp],rsi\n";
 259             my $narg = $current_function->{narg};
 260             $narg=6 if (!defined($narg));
 261             $func .= "  mov     rdi,rcx\n" if ($narg>0);
 262             $func .= "  mov     rsi,rdx\n" if ($narg>1);
 263             $func .= "  mov     rdx,r8\n"  if ($narg>2);
 264             $func .= "  mov     rcx,r9\n"  if ($narg>3);
 265             $func .= "  mov     r8,QWORD PTR 40[rsp]\n" if ($narg>4);
 266             $func .= "  mov     r9,QWORD PTR 48[rsp]\n" if ($narg>5);
 267             $func .= "\n";
 268         } else {
 269            "$current_function->{name}   PROC";
 270         }
 271     }
 272 }
 273 { package expr;         # pick up expressioins
 274     sub re {
 275         my      $self = shift;  # single instance is enough...
 276         local   *line = shift;
 277         undef   $ret;
 278
 279         if ($line =~ /(^[^,]+)/) {
 280             $self->{value} = $1;
 281             $ret = $self;
 282             $line = substr($line,@+[0]); $line =~ s/^\s+//;
 283
 284             $self->{value} =~ s/\.L/\$L/g if ($masm);
 285         }
 286         $ret;
 287     }
 288     sub out {
 289         my $self = shift;
 290         $self->{value};
 291     }
 292 }
 293 { package directive;    # pick up directives, which start with .
 294     sub re {
 295         my      $self = shift;  # single instance is enough...
 296         local   *line = shift;
 297         undef   $ret;
 298         my      $dir;
 299         my      %opcode =       # lea 2f-1f(%rip),%dst; 1: nop; 2:
 300                 (       "%rax"=>0x01058d48,     "%rcx"=>0x010d8d48,
 301                         "%rdx"=>0x01158d48,     "%rbx"=>0x011d8d48,
 302                         "%rsp"=>0x01258d48,     "%rbp"=>0x012d8d48,
 303                         "%rsi"=>0x01358d48,     "%rdi"=>0x013d8d48,
 304                         "%r8" =>0x01058d4c,     "%r9" =>0x010d8d4c,
 305                         "%r10"=>0x01158d4c,     "%r11"=>0x011d8d4c,
 306                         "%r12"=>0x01258d4c,     "%r13"=>0x012d8d4c,
 307                         "%r14"=>0x01358d4c,     "%r15"=>0x013d8d4c      );
 308
 309         if ($line =~ /^\s*(\.\w+)/) {
 310             if (!$masm) {
 311                 $self->{value} = $1;
 312                 $line =~ s/\@abi\-omnipotent/\@function/;
 313                 $line =~ s/\@function.*/\@function/;
 314                 if ($line =~ /\.picmeup\s+(%r[\w]+)/i) {
 315                     $self->{value} = sprintf "\t.long\t0x%x,0x90000000",$opcode{$1};
 316                 } else {
 317                     $self->{value} = $line;
 318                 }
 319                 $line = "";
 320                 return $self;
 321             }
 322
 323             $dir = $1;
 324             $ret = $self;
 325             undef $self->{value};
 326             $line = substr($line,@+[0]); $line =~ s/^\s+//;
 327             SWITCH: for ($dir) {
 328                 /\.(text)/
 329                             && do { my $v=undef;
 330                                     $v="$current_segment\tENDS\n" if ($current_segment);
 331                                     $current_segment = "_$1\$";
 332                                     $current_segment =~ tr/[a-z]/[A-Z]/;
 333                                     $v.="$current_segment\tSEGMENT ALIGN(64) 'CODE'";
 334                                     $self->{value} = $v;
 335                                     last;
 336                                   };
 337                 /\.globl/   && do { $self->{value} = "PUBLIC\t".$line; last; };
 338                 /\.type/    && do { ($sym,$type,$narg) = split(',',$line);
 339                                     if ($type eq "\@function") {
 340                                         undef $current_function;
 341                                         $current_function->{name} = $sym;
 342                                         $current_function->{abi}  = "svr4";
 343                                         $current_function->{narg} = $narg;
 344                                     } elsif ($type eq "\@abi-omnipotent") {
 345                                         undef $current_function;
 346                                         $current_function->{name} = $sym;
 347                                     }
 348                                     last;
 349                                   };
 350                 /\.size/    && do { if (defined($current_function)) {
 351                                         $self->{value}="$current_function->{name}\tENDP";
 352                                         undef $current_function;
 353                                     }
 354                                     last;
 355                                   };
 356                 /\.align/   && do { $self->{value} = "ALIGN\t".$line; last; };
 357                 /\.(byte|value|long|quad)/
 358                             && do { my @arr = split(',',$line);
 359                                     my $sz  = substr($1,0,1);
 360                                     my $last = pop(@arr);
 361
 362                                     $sz =~ tr/bvlq/BWDQ/;
 363                                     $self->{value} = "\tD$sz\t";
 364                                     for (@arr) { $self->{value} .= sprintf"0%Xh,",oct; }
 365                                     $self->{value} .= sprintf"0%Xh",oct($last);
 366                                     last;
 367                                   };
 368                 /\.picmeup/ && do { $self->{value} = sprintf"\tDD\t 0%Xh,090000000h",$opcode{$line};
 369                                     last;
 370                                   };
 371             }
 372             $line = "";
 373         }
 374
 375         $ret;
 376     }
 377     sub out {
 378         my $self = shift;
 379         $self->{value};
 380     }
 381 }
 382
 383 while($line=<>) {
 384
 385     chomp($line);
 386
 387     $line =~ s|[#!].*$||;       # get rid of asm-style comments...
 388     $line =~ s|/\*.*\*/||;      # ... and C-style comments...
 389     $line =~ s|^\s+||;          # ... and skip white spaces in beginning
 390
 391     undef $label;
 392     undef $opcode;
 393     undef $dst;
 394     undef $src;
 395     undef $sz;
 396
 397     if ($label=label->re(\$line))       { print $label->out(); }
 398
 399     if (directive->re(\$line)) {
 400         printf "%s",directive->out();
 401     } elsif ($opcode=opcode->re(\$line)) { ARGUMENT: {
 402
 403         if ($src=register->re(\$line))  { opcode->size($src->size()); }
 404         elsif ($src=const->re(\$line))  { }
 405         elsif ($src=ea->re(\$line))     { }
 406         elsif ($src=expr->re(\$line))   { }
 407
 408         last ARGUMENT if ($line !~ /^,/);
 409
 410         $line = substr($line,1); $line =~ s/^\s+//;
 411
 412         if ($dst=register->re(\$line))  { opcode->size($dst->size()); }
 413         elsif ($dst=const->re(\$line))  { }
 414         elsif ($dst=ea->re(\$line))     { }
 415
 416         } # ARGUMENT:
 417
 418         $sz=opcode->size();
 419
 420         if (defined($dst)) {
 421             if (!$masm) {
 422                 printf "\t%s\t%s,%s",   $opcode->out($dst->size()),
 423                                         $src->out($sz),$dst->out($sz);
 424             } else {
 425                 printf "\t%s\t%s,%s",   $opcode->out(),
 426                                         $dst->out($sz),$src->out($sz);
 427             }
 428         } elsif (defined($src)) {
 429             printf "\t%s\t%s",$opcode->out(),$src->out($sz);
 430         } else {
 431             printf "\t%s",$opcode->out();
 432         }
 433     }
 434
 435     print $line,"\n";
 436 }
 437
 438 print "\n$current_segment\tENDS\nEND\n" if ($masm);
 439
 440 close STDOUT;
 441
 442 #################################################
 443 # Cross-reference x86_64 ABI "card"
 444 #
 445 #               Unix            Win64
 446 # %rax          *               *
 447 # %rbx          -               -
 448 # %rcx          #4              #1
 449 # %rdx          #3              #2
 450 # %rsi          #2              -
 451 # %rdi          #1              -
 452 # %rbp          -               -
 453 # %rsp          -               -
 454 # %r8           #5              #3
 455 # %r9           #6              #4
 456 # %r10          *               *
 457 # %r11          *               *
 458 # %r12          -               -
 459 # %r13          -               -
 460 # %r14          -               -
 461 # %r15          -               -
 462 #
 463 # (*)   volatile register
 464 # (-)   preserved by callee
 465 # (#)   Nth argument, volatile
 466 #
 467 # In Unix terms top of stack is argument transfer area for arguments
 468 # which could not be accomodated in registers. Or in other words 7th
 469 # [integer] argument resides at 8(%rsp) upon function entry point.
 470 # 128 bytes above %rsp constitute a "red zone" which is not touched
 471 # by signal handlers and can be used as temporal storage without
 472 # allocating a frame.
 473 #
 474 # In Win64 terms N*8 bytes on top of stack is argument transfer area,
 475 # which belongs to/can be overwritten by callee. N is the number of
 476 # arguments passed to callee, *but* not less than 4! This means that
 477 # upon function entry point 5th argument resides at 40(%rsp), as well
 478 # as that 32 bytes from 8(%rsp) can always be used as temporal
 479 # storage [without allocating a frame].
 480 #
 481 # All the above means that if assembler programmer adheres to Unix
 482 # register and stack layout, but disregards the "red zone" existense,
 483 # it's possible to use following prologue and epilogue to "gear" from
 484 # Unix to Win64 ABI in leaf functions with not more than 6 arguments.
 485 #
 486 # omnipotent_function:
 487 # ifdef WIN64
 488 #       movq    %rdi,8(%rsp)
 489 #       movq    %rsi,16(%rsp)
 490 #       movq    %rcx,%rdi       ; if 1st argument is actually present
 491 #       movq    %rdx,%rsi       ; if 2nd argument is actually ...
 492 #       movq    %r8,%rdx        ; if 3rd argument is ...
 493 #       movq    %r9,%rcx        ; if 4th argument ...
 494 #       movq    40(%rsp),%r8    ; if 5th ...
 495 #       movq    48(%rsp),%r9    ; if 6th ...
 496 # endif
 497 #       ...
 498 # ifdef WIN64
 499 #       movq    8(%rsp),%rdi
 500 #       movq    16(%rsp),%rsi
 501 # endif
 502 #       ret