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         $ret;
 118     }
 119     sub out {
 120         my $self = shift;
 121         sprintf $masm?"%s":"\$%s",$self->{value};
 122     }
 123 }
 124 { package ea;           # pick up effective addresses: expr(%reg,%reg,scale)
 125     sub re {
 126         my      $self = shift;  # single instance in enough...
 127         local   *line = shift;
 128         undef   $ret;
 129
 130         if ($line =~ /^([^\(,]*)\(([%\w,]+)\)/) {
 131             $self->{label} = $1;
 132             ($self->{base},$self->{index},$self->{scale})=split(/,/,$2);
 133             $self->{scale} = 1 if (!defined($self->{scale}));
 134             $ret = $self;
 135             $line = substr($line,@+[0]); $line =~ s/^\s+//;
 136
 137             $self->{label} =~ s/\.L/\$L/g;
 138             $self->{base}  =~ s/^%//;
 139             $self->{index} =~ s/^%// if (defined($self->{index}));
 140         }
 141         $ret;
 142     }
 143     sub size {}
 144     sub out {
 145         my $self = shift;
 146         my $sz = shift;
 147
 148         if (!$masm) {
 149             # elder GNU assembler insists on 64-bit EAs:-(
 150             # on pros side, this results in more compact code:-)
 151             $self->{index} =~ s/^[er](.?[0-9xp])[d]?$/r\1/;
 152             $self->{base}  =~ s/^[er](.?[0-9xp])[d]?$/r\1/;
 153
 154             if (defined($self->{index})) {
 155                 sprintf "%s(%%%s,%%%s,%d)",     $self->{label},$self->{base},
 156                                         $self->{index},$self->{scale};
 157             }
 158             else {
 159                 sprintf "%s(%%%s)",     $self->{label},$self->{base};
 160             }
 161         } else {
 162             %szmap = ( b=>"BYTE", w=>"WORD", l=>"DWORD", q=>"QWORD" );
 163
 164             if (defined($self->{index})) {
 165                 sprintf "%s PTR %s[%s*%d+%s]",$szmap{$sz},
 166                                         $self->{label},
 167                                         $self->{index},$self->{scale},
 168                                         $self->{base};
 169             }
 170             else {
 171                 sprintf "%s PTR %s[%s]",$szmap{$sz},
 172                                         $self->{label},$self->{base};
 173             }
 174         }
 175     }
 176 }
 177 { package register;     # pick up registers, which start with %.
 178     sub re {
 179         my      $class = shift; # muliple instances...
 180         my      $self = {};
 181         local   *line = shift;
 182         undef   $ret;
 183
 184         if ($line =~ /^%(\w+)/) {
 185             bless $self,$class;
 186             $self->{value} = $1;
 187             $ret = $self;
 188             $line = substr($line,@+[0]); $line =~ s/^\s+//;
 189         }
 190         $ret;
 191     }
 192     sub size {
 193         my      $self = shift;
 194         undef   $ret;
 195
 196         if    ($self->{value} =~ /^r[\d]+b$/i)  { $ret="b"; }
 197         elsif ($self->{value} =~ /^r[\d]+w$/i)  { $ret="w"; }
 198         elsif ($self->{value} =~ /^r[\d]+d$/i)  { $ret="l"; }
 199         elsif ($self->{value} =~ /^r[\w]+$/i)   { $ret="q"; }
 200         elsif ($self->{value} =~ /^[a-d][hl]$/i){ $ret="b"; }
 201         elsif ($self->{value} =~ /^[\w]{2}l$/i) { $ret="b"; }
 202         elsif ($self->{value} =~ /^[\w]{2}$/i)  { $ret="w"; }
 203         elsif ($self->{value} =~ /^e[a-z]{2}$/i){ $ret="l"; }
 204
 205         $ret;
 206     }
 207     sub out {
 208         my $self = shift;
 209         sprintf $masm?"%s":"%%%s",$self->{value};
 210     }
 211 }
 212 { package label;        # pick up labels, which end with :
 213     sub re {
 214         my      $self = shift;  # single instance is enough...
 215         local   *line = shift;
 216         undef   $ret;
 217
 218         if ($line =~ /(^[\.\w]+\:)/) {
 219             $self->{value} = $1;
 220             $ret = $self;
 221             $line = substr($line,@+[0]); $line =~ s/^\s+//;
 222
 223             $self->{value} =~ s/\.L/\$L/ if ($masm);
 224         }
 225         $ret;
 226     }
 227     sub out {
 228         my $self = shift;
 229
 230         if (!$masm) {
 231             $self->{value};
 232         } elsif ($self->{value} ne "$current_function->{name}:") {
 233             $self->{value};
 234         } elsif ($current_function->{abi} eq "svr4") {
 235             my $func =  "$current_function->{name}      PROC\n".
 236                         "       mov     QWORD PTR 8[rsp],rdi\t;WIN64 prologue\n".
 237                         "       mov     QWORD PTR 16[rsp],rsi\n";
 238             my $narg = $current_function->{narg};
 239             $narg=6 if (!defined($narg));
 240             $func .= "  mov     rdi,rcx\n" if ($narg>0);
 241             $func .= "  mov     rsi,rdx\n" if ($narg>1);
 242             $func .= "  mov     rdx,r8\n"  if ($narg>2);
 243             $func .= "  mov     rcx,r9\n"  if ($narg>3);
 244             $func .= "  mov     r8,QWORD PTR 40[rsp]\n" if ($narg>4);
 245             $func .= "  mov     r9,QWORD PTR 48[rsp]\n" if ($narg>5);
 246             $func .= "\n";
 247         } else {
 248            "$current_function->{name}   PROC";
 249         }
 250     }
 251 }
 252 { package expr;         # pick up expressioins
 253     sub re {
 254         my      $self = shift;  # single instance is enough...
 255         local   *line = shift;
 256         undef   $ret;
 257
 258         if ($line =~ /(^[^,]+)/) {
 259             $self->{value} = $1;
 260             $ret = $self;
 261             $line = substr($line,@+[0]); $line =~ s/^\s+//;
 262
 263             $self->{value} =~ s/\.L/\$L/g if ($masm);
 264         }
 265         $ret;
 266     }
 267     sub out {
 268         my $self = shift;
 269         $self->{value};
 270     }
 271 }
 272 { package directive;    # pick up directives, which start with .
 273     sub re {
 274         my      $self = shift;  # single instance is enough...
 275         local   *line = shift;
 276         undef   $ret;
 277         my      $dir;
 278
 279         if ($line =~ /^\s*(\.\w+)/) {
 280             if (!$masm) {
 281                 $self->{value} = $1;
 282                 $line =~ s/\@abi\-omnipotent/\@function/;
 283                 $line =~ s/\@function.*/\@function/;
 284                 $self->{value} = $line;
 285                 $line = "";
 286                 return $self;
 287             }
 288
 289             $dir = $1;
 290             $ret = $self;
 291             undef $self->{value};
 292             $line = substr($line,@+[0]); $line =~ s/^\s+//;
 293             SWITCH: for ($dir) {
 294                 /\.(text|data)/
 295                             && do { my $v=undef;
 296                                     $v="$current_segment\tENDS\n" if ($current_segment);
 297                                     $current_segment = "_$1";
 298                                     $current_segment =~ tr/[a-z]/[A-Z]/;
 299                                     $v.="$current_segment\tSEGMENT PARA";
 300                                     $self->{value} = $v;
 301                                     last;
 302                                   };
 303                 /\.globl/   && do { $self->{value} = "PUBLIC\t".$line; last; };
 304                 /\.type/    && do { ($sym,$type,$narg) = split(',',$line);
 305                                     if ($type eq "\@function")
 306                                     {   undef $current_function;
 307                                         $current_function->{name} = $sym;
 308                                         $current_function->{abi}  = "svr4";
 309                                         $current_function->{narg} = $narg;
 310                                     }
 311                                     elsif ($type eq "\@abi-omnipotent")
 312                                     {   undef $current_function;
 313                                         $current_function->{name} = $sym;
 314                                     }
 315                                     last;
 316                                   };
 317                 /\.size/    && do { if (defined($current_function))
 318                                     {   $self->{value}="$current_function->{name}\tENDP";
 319                                         undef $current_function;
 320                                     }
 321                                     last;
 322                                   };
 323                 /\.align/   && do { $self->{value} = "ALIGN\t".$line; last; };
 324                 /\.(byte|value|long|quad)/
 325                             && do { my @arr = split(',',$line);
 326                                     my $sz  = substr($1,0,1);
 327                                     my $last = pop(@arr);
 328
 329                                     $sz =~ tr/bvlq/BWDQ/;
 330                                     $self->{value} = "\tD$sz\t";
 331                                     for (@arr) { $self->{value} .= sprintf"0%Xh,",oct; }
 332                                     $self->{value} .= sprintf"0%Xh",oct($last);
 333                                     last;
 334                                   };
 335             }
 336             $line = "";
 337         }
 338
 339         $ret;
 340     }
 341     sub out {
 342         my $self = shift;
 343         $self->{value};
 344     }
 345 }
 346
 347 while($line=<>) {
 348
 349     chomp($line);
 350
 351     $line =~ s|[#!].*$||;       # get rid of asm-style comments...
 352     $line =~ s|/\*.*\*/||;      # ... and C-style comments...
 353     $line =~ s|^\s+||;          # ... and skip white spaces in beginning
 354
 355     undef $label;
 356     undef $opcode;
 357     undef $dst;
 358     undef $src;
 359     undef $sz;
 360
 361     if ($label=label->re(\$line))       { print $label->out(); }
 362
 363     if (directive->re(\$line)) {
 364         printf "%s",directive->out();
 365     } elsif ($opcode=opcode->re(\$line)) { ARGUMENT: {
 366
 367         if ($src=register->re(\$line))  { opcode->size($src->size()); }
 368         elsif ($src=const->re(\$line))  { }
 369         elsif ($src=ea->re(\$line))     { }
 370         elsif ($src=expr->re(\$line))   { }
 371
 372         last ARGUMENT if ($line !~ /^,/);
 373
 374         $line = substr($line,1); $line =~ s/^\s+//;
 375
 376         if ($dst=register->re(\$line))  { opcode->size($dst->size()); }
 377         elsif ($dst=const->re(\$line))  { }
 378         elsif ($dst=ea->re(\$line))     { }
 379
 380         } # ARGUMENT:
 381
 382         $sz=opcode->size();
 383
 384         if (defined($dst)) {
 385             if (!$masm) {
 386                 printf "\t%s\t%s,%s",   $opcode->out($dst->size()),
 387                                         $src->out($sz),$dst->out($sz);
 388             }
 389             else {
 390                 printf "\t%s\t%s,%s",   $opcode->out(),
 391                                         $dst->out($sz),$src->out($sz);
 392             }
 393         }
 394         elsif (defined($src)) {
 395             printf "\t%s\t%s",$opcode->out(),$src->out($sz);
 396         } else {
 397             printf "\t%s",$opcode->out();
 398         }
 399     }
 400
 401     print $line,"\n";
 402 }
 403
 404 print "\n$current_segment\tENDS\nEND\n" if ($masm);
 405
 406 close STDOUT;
 407
 408 #################################################
 409 # Cross-reference x86_64 ABI "card"
 410 #
 411 #               Unix            Win64
 412 # %rax          *               *
 413 # %rbx          -               -
 414 # %rcx          #4              #1
 415 # %rdx          #3              #2
 416 # %rsi          #2              -
 417 # %rdi          #1              -
 418 # %rbp          -               -
 419 # %rsp          -               -
 420 # %r8           #5              #3
 421 # %r9           #6              #4
 422 # %r10          *               *
 423 # %r11          *               *
 424 # %r12          -               -
 425 # %r13          -               -
 426 # %r14          -               -
 427 # %r15          -               -
 428 #
 429 # (*)   volatile register
 430 # (-)   preserved by callee
 431 # (#)   Nth argument, volatile
 432 #
 433 # In Unix terms top of stack is argument transfer area for arguments
 434 # which could not be accomodated in registers. Or in other words 7th
 435 # [integer] argument resides at 8(%rsp) upon function entry point.
 436 # 128 bytes above %rsp constitute a "red zone" which is not touched
 437 # by signal handlers and can be used as temporal storage without
 438 # allocating a frame.
 439 #
 440 # In Win64 terms N*8 bytes on top of stack is argument transfer area,
 441 # which belongs to/can be overwritten by callee. N is the number of
 442 # arguments passed to callee, *but* not less than 4! This means that
 443 # upon function entry point 5th argument resides at 40(%rsp), as well
 444 # as that 32 bytes from 8(%rsp) can always be used as temporal
 445 # storage [without allocating a frame].
 446 #
 447 # All the above means that if assembler programmer adheres to Unix
 448 # register and stack layout, but disregards the "red zone" existense,
 449 # it's possible to use following prologue and epilogue to "gear" from
 450 # Unix to Win64 ABI in leaf functions with not more than 6 arguments.
 451 #
 452 # omnipotent_function:
 453 # ifdef WIN64
 454 #       movq    %rdi,8(%rsp)
 455 #       movq    %rsi,16(%rsp)
 456 #       movq    %rcx,%rdi       ; if 1st argument is actually present
 457 #       movq    %rdx,%rsi       ; if 2nd argument is actually ...
 458 #       movq    %r8,%rdx        ; if 3rd argument is ...
 459 #       movq    %r9,%rcx        ; if 4th argument ...
 460 #       movq    40(%rsp),%r8    ; if 5th ...
 461 #       movq    48(%rsp),%r9    ; if 6th ...
 462 # endif
 463 #       ...
 464 # ifdef WIN64
 465 #       movq    8(%rsp),%rdi
 466 #       movq    16(%rsp),%rsi
 467 # endif
 468 #       ret