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 $output = shift;
  60
  61 { my ($stddev,$stdino,@junk)=stat(STDOUT);
  62   my ($outdev,$outino,@junk)=stat($output);
  63
  64     open STDOUT,">$output" || die "can't open $output: $!"
  65         if ($stddev!=$outdev || $stdino!=$outino);
  66 }
  67
  68 my $masmref=8 + 50727*2**-32;   # 8.00.50727 shipped with VS2005
  69 my $masm=$masmref if ($output =~ /\.asm/);
  70 if ($masm && `ml64 2>&1` =~ m/Version ([0-9]+)\.([0-9]+)(\.([0-9]+))?/)
  71 {   $masm=$1 + $2*2**-16 + $4*2**-32;   }
  72
  73 my $current_segment;
  74 my $current_function;
  75
  76 { package opcode;       # pick up opcodes
  77     sub re {
  78         my      $self = shift;  # single instance in enough...
  79         local   *line = shift;
  80         undef   $ret;
  81
  82         if ($line =~ /^([a-z][a-z0-9]*)/i) {
  83             $self->{op} = $1;
  84             $ret = $self;
  85             $line = substr($line,@+[0]); $line =~ s/^\s+//;
  86
  87             undef $self->{sz};
  88             if ($self->{op} =~ /^(movz)b.*/) {  # movz is pain...
  89                 $self->{op} = $1;
  90                 $self->{sz} = "b";
  91             } elsif ($self->{op} =~ /call/) {
  92                 $self->{sz} = ""
  93             } elsif ($self->{op} =~ /([a-z]{3,})([qlwb])$/) {
  94                 $self->{op} = $1;
  95                 $self->{sz} = $2;
  96             }
  97         }
  98         $ret;
  99     }
 100     sub size {
 101         my $self = shift;
 102         my $sz   = shift;
 103         $self->{sz} = $sz if (defined($sz) && !defined($self->{sz}));
 104         $self->{sz};
 105     }
 106     sub out {
 107         my $self = shift;
 108         if (!$masm) {
 109             if ($self->{op} eq "movz") {        # movz is pain...
 110                 sprintf "%s%s%s",$self->{op},$self->{sz},shift;
 111             } elsif ($self->{op} =~ /^set/) {
 112                 "$self->{op}";
 113             } elsif ($self->{op} eq "ret") {
 114                 ".byte  0xf3,0xc3";
 115             } else {
 116                 "$self->{op}$self->{sz}";
 117             }
 118         } else {
 119             $self->{op} =~ s/^movz/movzx/;
 120             if ($self->{op} eq "ret") {
 121                 $self->{op} = "";
 122                 if ($current_function->{abi} eq "svr4") {
 123                     $self->{op} = "mov  rdi,QWORD PTR 8[rsp]\t;WIN64 epilogue\n\t".
 124                                   "mov  rsi,QWORD PTR 16[rsp]\n\t";
 125                 }
 126                 $self->{op} .= "DB\t0F3h,0C3h\t\t;repret";
 127             }
 128             $self->{op};
 129         }
 130     }
 131 }
 132 { package const;        # pick up constants, which start with $
 133     sub re {
 134         my      $self = shift;  # single instance in enough...
 135         local   *line = shift;
 136         undef   $ret;
 137
 138         if ($line =~ /^\$([^,]+)/) {
 139             $self->{value} = $1;
 140             $ret = $self;
 141             $line = substr($line,@+[0]); $line =~ s/^\s+//;
 142         }
 143         $ret;
 144     }
 145     sub out {
 146         my $self = shift;
 147
 148         if (!$masm) {
 149             # Solaris /usr/ccs/bin/as can't handle multiplications
 150             # in $self->{value}
 151             $self->{value} =~ s/(?<![0-9a-f])(0[x0-9a-f]+)/oct($1)/egi;
 152             $self->{value} =~ s/([0-9]+\s*[\*\/\%]\s*[0-9]+)/eval($1)/eg;
 153             sprintf "\$%s",$self->{value};
 154         } else {
 155             $self->{value} =~ s/0x([0-9a-f]+)/0$1h/ig;
 156             sprintf "%s",$self->{value};
 157         }
 158     }
 159 }
 160 { package ea;           # pick up effective addresses: expr(%reg,%reg,scale)
 161     sub re {
 162         my      $self = shift;  # single instance in enough...
 163         local   *line = shift;
 164         undef   $ret;
 165
 166         if ($line =~ /^([^\(,]*)\(([%\w,]+)\)/) {
 167             $self->{label} = $1;
 168             ($self->{base},$self->{index},$self->{scale})=split(/,/,$2);
 169             $self->{scale} = 1 if (!defined($self->{scale}));
 170             $ret = $self;
 171             $line = substr($line,@+[0]); $line =~ s/^\s+//;
 172
 173             $self->{base}  =~ s/^%//;
 174             $self->{index} =~ s/^%// if (defined($self->{index}));
 175         }
 176         $ret;
 177     }
 178     sub size {}
 179     sub out {
 180         my $self = shift;
 181         my $sz = shift;
 182
 183         # Silently convert all EAs to 64-bit. This is required for
 184         # elder GNU assembler and results in more compact code,
 185         # *but* most importantly AES module depends on this feature!
 186         $self->{index} =~ s/^[er](.?[0-9xpi])[d]?$/r\1/;
 187         $self->{base}  =~ s/^[er](.?[0-9xpi])[d]?$/r\1/;
 188
 189         if (!$masm) {
 190             # Solaris /usr/ccs/bin/as can't handle multiplications
 191             # in $self->{label}
 192             use integer;
 193             $self->{label} =~ s/(?<![0-9a-f])(0[x0-9a-f]+)/oct($1)<<32>>32/egi;
 194             $self->{label} =~ s/([0-9]+\s*[\*\/\%]\s*[0-9]+)/eval($1)/eg;
 195
 196             if (defined($self->{index})) {
 197                 sprintf "%s(%%%s,%%%s,%d)",
 198                                         $self->{label},$self->{base},
 199                                         $self->{index},$self->{scale};
 200             } else {
 201                 sprintf "%s(%%%s)",     $self->{label},$self->{base};
 202             }
 203         } else {
 204             %szmap = ( b=>"BYTE", w=>"WORD", l=>"DWORD", q=>"QWORD" );
 205
 206             $self->{label} =~ s/\./\$/g;
 207             $self->{label} =~ s/0x([0-9a-f]+)/0$1h/ig;
 208             $self->{label} = "($self->{label})" if ($self->{label} =~ /[\*\+\-\/]/);
 209
 210             if (defined($self->{index})) {
 211                 sprintf "%s PTR %s[%s*%d+%s]",$szmap{$sz},
 212                                         $self->{label},
 213                                         $self->{index},$self->{scale},
 214                                         $self->{base};
 215             } elsif ($self->{base} eq "rip") {
 216                 sprintf "%s PTR %s",$szmap{$sz},$self->{label};
 217             } else {
 218                 sprintf "%s PTR %s[%s]",$szmap{$sz},
 219                                         $self->{label},$self->{base};
 220             }
 221         }
 222     }
 223 }
 224 { package register;     # pick up registers, which start with %.
 225     sub re {
 226         my      $class = shift; # muliple instances...
 227         my      $self = {};
 228         local   *line = shift;
 229         undef   $ret;
 230
 231         if ($line =~ /^%(\w+)/) {
 232             bless $self,$class;
 233             $self->{value} = $1;
 234             $ret = $self;
 235             $line = substr($line,@+[0]); $line =~ s/^\s+//;
 236         }
 237         $ret;
 238     }
 239     sub size {
 240         my      $self = shift;
 241         undef   $ret;
 242
 243         if    ($self->{value} =~ /^r[\d]+b$/i)  { $ret="b"; }
 244         elsif ($self->{value} =~ /^r[\d]+w$/i)  { $ret="w"; }
 245         elsif ($self->{value} =~ /^r[\d]+d$/i)  { $ret="l"; }
 246         elsif ($self->{value} =~ /^r[\w]+$/i)   { $ret="q"; }
 247         elsif ($self->{value} =~ /^[a-d][hl]$/i){ $ret="b"; }
 248         elsif ($self->{value} =~ /^[\w]{2}l$/i) { $ret="b"; }
 249         elsif ($self->{value} =~ /^[\w]{2}$/i)  { $ret="w"; }
 250         elsif ($self->{value} =~ /^e[a-z]{2}$/i){ $ret="l"; }
 251
 252         $ret;
 253     }
 254     sub out {
 255         my $self = shift;
 256         sprintf $masm?"%s":"%%%s",$self->{value};
 257     }
 258 }
 259 { package label;        # pick up labels, which end with :
 260     sub re {
 261         my      $self = shift;  # single instance is enough...
 262         local   *line = shift;
 263         undef   $ret;
 264
 265         if ($line =~ /(^[\.\w]+\:)/) {
 266             $self->{value} = $1;
 267             $ret = $self;
 268             $line = substr($line,@+[0]); $line =~ s/^\s+//;
 269
 270             $self->{value} =~ s/\.L/\$L/ if ($masm);
 271         }
 272         $ret;
 273     }
 274     sub out {
 275         my $self = shift;
 276
 277         if (!$masm) {
 278             $self->{value};
 279         } elsif ($self->{value} ne "$current_function->{name}:") {
 280             $self->{value};
 281         } elsif ($current_function->{abi} eq "svr4") {
 282             my $func =  "$current_function->{name}      PROC\n".
 283                         "       mov     QWORD PTR 8[rsp],rdi\t;WIN64 prologue\n".
 284                         "       mov     QWORD PTR 16[rsp],rsi\n";
 285             my $narg = $current_function->{narg};
 286             $narg=6 if (!defined($narg));
 287             $func .= "  mov     rdi,rcx\n" if ($narg>0);
 288             $func .= "  mov     rsi,rdx\n" if ($narg>1);
 289             $func .= "  mov     rdx,r8\n"  if ($narg>2);
 290             $func .= "  mov     rcx,r9\n"  if ($narg>3);
 291             $func .= "  mov     r8,QWORD PTR 40[rsp]\n" if ($narg>4);
 292             $func .= "  mov     r9,QWORD PTR 48[rsp]\n" if ($narg>5);
 293             $func .= "\n";
 294         } else {
 295            "$current_function->{name}   PROC";
 296         }
 297     }
 298 }
 299 { package expr;         # pick up expressioins
 300     sub re {
 301         my      $self = shift;  # single instance is enough...
 302         local   *line = shift;
 303         undef   $ret;
 304
 305         if ($line =~ /(^[^,]+)/) {
 306             $self->{value} = $1;
 307             $ret = $self;
 308             $line = substr($line,@+[0]); $line =~ s/^\s+//;
 309
 310             $self->{value} =~ s/\.L/\$L/g if ($masm);
 311         }
 312         $ret;
 313     }
 314     sub out {
 315         my $self = shift;
 316         $self->{value};
 317     }
 318 }
 319 { package directive;    # pick up directives, which start with .
 320     sub re {
 321         my      $self = shift;  # single instance is enough...
 322         local   *line = shift;
 323         undef   $ret;
 324         my      $dir;
 325         my      %opcode =       # lea 2f-1f(%rip),%dst; 1: nop; 2:
 326                 (       "%rax"=>0x01058d48,     "%rcx"=>0x010d8d48,
 327                         "%rdx"=>0x01158d48,     "%rbx"=>0x011d8d48,
 328                         "%rsp"=>0x01258d48,     "%rbp"=>0x012d8d48,
 329                         "%rsi"=>0x01358d48,     "%rdi"=>0x013d8d48,
 330                         "%r8" =>0x01058d4c,     "%r9" =>0x010d8d4c,
 331                         "%r10"=>0x01158d4c,     "%r11"=>0x011d8d4c,
 332                         "%r12"=>0x01258d4c,     "%r13"=>0x012d8d4c,
 333                         "%r14"=>0x01358d4c,     "%r15"=>0x013d8d4c      );
 334
 335         if ($line =~ /^\s*(\.\w+)/) {
 336             if (!$masm) {
 337                 $self->{value} = $1;
 338                 $line =~ s/\@abi\-omnipotent/\@function/;
 339                 $line =~ s/\@function.*/\@function/;
 340                 if ($line =~ /\.picmeup\s+(%r[\w]+)/i) {
 341                     $self->{value} = sprintf "\t.long\t0x%x,0x90000000",$opcode{$1};
 342                 } elsif ($line =~ /\.asciz\s+"(.*)"$/) {
 343                     $self->{value} = ".byte\t".join(",",unpack("C*",$1),0);
 344                 } elsif ($line =~ /\.extern/) {
 345                     $self->{value} = ""; # swallow extern
 346                 } else {
 347                     $self->{value} = $line;
 348                 }
 349                 $line = "";
 350                 return $self;
 351             }
 352
 353             $dir = $1;
 354             $ret = $self;
 355             undef $self->{value};
 356             $line = substr($line,@+[0]); $line =~ s/^\s+//;
 357             SWITCH: for ($dir) {
 358                 /\.(text)/
 359                             && do { my $v=undef;
 360                                     $v="$current_segment\tENDS\n" if ($current_segment);
 361                                     $current_segment = "_$1\$";
 362                                     $current_segment =~ tr/[a-z]/[A-Z]/;
 363                                     $v.="$current_segment\tSEGMENT ";
 364                                     $v.=$masm>=$masmref ? "ALIGN(64)" : "PAGE";
 365                                     $v.=" 'CODE'";
 366                                     $self->{value} = $v;
 367                                     last;
 368                                   };
 369                 /\.extern/  && do { $self->{value} = "EXTRN\t".$line.":BYTE"; last;  };
 370                 /\.globl/   && do { $self->{value} = "PUBLIC\t".$line; last; };
 371                 /\.type/    && do { ($sym,$type,$narg) = split(',',$line);
 372                                     if ($type eq "\@function") {
 373                                         undef $current_function;
 374                                         $current_function->{name} = $sym;
 375                                         $current_function->{abi}  = "svr4";
 376                                         $current_function->{narg} = $narg;
 377                                     } elsif ($type eq "\@abi-omnipotent") {
 378                                         undef $current_function;
 379                                         $current_function->{name} = $sym;
 380                                     }
 381                                     last;
 382                                   };
 383                 /\.size/    && do { if (defined($current_function)) {
 384                                         $self->{value}="$current_function->{name}\tENDP";
 385                                         undef $current_function;
 386                                     }
 387                                     last;
 388                                   };
 389                 /\.align/   && do { $self->{value} = "ALIGN\t".$line; last; };
 390                 /\.(byte|value|long|quad)/
 391                             && do { my @arr = split(',',$line);
 392                                     my $sz  = substr($1,0,1);
 393                                     my $last = pop(@arr);
 394                                     my $conv = sub  {   my $var=shift;
 395                                                         if ($var=~s/0x([0-9a-f]+)/0$1h/i) { $var; }
 396                                                         else { sprintf"0%Xh",$var; }
 397                                                     };
 398
 399                                     $sz =~ tr/bvlq/BWDQ/;
 400                                     $self->{value} = "\tD$sz\t";
 401                                     for (@arr) { $self->{value} .= &$conv($_).","; }
 402                                     $self->{value} .= &$conv($last);
 403                                     last;
 404                                   };
 405                 /\.picmeup/ && do { $self->{value} = sprintf"\tDD\t 0%Xh,090000000h",$opcode{$line};
 406                                     last;
 407                                   };
 408                 /\.asciz/   && do { if ($line =~ /^"(.*)"$/) {
 409                                         my @str=unpack("C*",$1);
 410                                         push @str,0;
 411                                         while ($#str>15) {
 412                                             $self->{value}.="DB\t"
 413                                                 .join(",",@str[0..15])."\n";
 414                                             foreach (0..15) { shift @str; }
 415                                         }
 416                                         $self->{value}.="DB\t"
 417                                                 .join(",",@str) if (@str);
 418                                     }
 419                                     last;
 420                                   };
 421             }
 422             $line = "";
 423         }
 424
 425         $ret;
 426     }
 427     sub out {
 428         my $self = shift;
 429         $self->{value};
 430     }
 431 }
 432
 433 while($line=<>) {
 434
 435     chomp($line);
 436
 437     $line =~ s|[#!].*$||;       # get rid of asm-style comments...
 438     $line =~ s|/\*.*\*/||;      # ... and C-style comments...
 439     $line =~ s|^\s+||;          # ... and skip white spaces in beginning
 440
 441     undef $label;
 442     undef $opcode;
 443     undef $dst;
 444     undef $src;
 445     undef $sz;
 446
 447     if ($label=label->re(\$line))       { print $label->out(); }
 448
 449     if (directive->re(\$line)) {
 450         printf "%s",directive->out();
 451     } elsif ($opcode=opcode->re(\$line)) { ARGUMENT: {
 452
 453         if ($src=register->re(\$line))  { opcode->size($src->size()); }
 454         elsif ($src=const->re(\$line))  { }
 455         elsif ($src=ea->re(\$line))     { }
 456         elsif ($src=expr->re(\$line))   { }
 457
 458         last ARGUMENT if ($line !~ /^,/);
 459
 460         $line = substr($line,1); $line =~ s/^\s+//;
 461
 462         if ($dst=register->re(\$line))  { opcode->size($dst->size()); }
 463         elsif ($dst=const->re(\$line))  { }
 464         elsif ($dst=ea->re(\$line))     { }
 465
 466         } # ARGUMENT:
 467
 468         $sz=opcode->size();
 469
 470         if (defined($dst)) {
 471             if (!$masm) {
 472                 printf "\t%s\t%s,%s",   $opcode->out($dst->size()),
 473                                         $src->out($sz),$dst->out($sz);
 474             } else {
 475                 printf "\t%s\t%s,%s",   $opcode->out(),
 476                                         $dst->out($sz),$src->out($sz);
 477             }
 478         } elsif (defined($src)) {
 479             printf "\t%s\t%s",$opcode->out(),$src->out($sz);
 480         } else {
 481             printf "\t%s",$opcode->out();
 482         }
 483     }
 484
 485     print $line,"\n";
 486 }
 487
 488 print "\n$current_segment\tENDS\nEND\n" if ($masm);
 489
 490 close STDOUT;
 491
 492 #################################################
 493 # Cross-reference x86_64 ABI "card"
 494 #
 495 #               Unix            Win64
 496 # %rax          *               *
 497 # %rbx          -               -
 498 # %rcx          #4              #1
 499 # %rdx          #3              #2
 500 # %rsi          #2              -
 501 # %rdi          #1              -
 502 # %rbp          -               -
 503 # %rsp          -               -
 504 # %r8           #5              #3
 505 # %r9           #6              #4
 506 # %r10          *               *
 507 # %r11          *               *
 508 # %r12          -               -
 509 # %r13          -               -
 510 # %r14          -               -
 511 # %r15          -               -
 512 #
 513 # (*)   volatile register
 514 # (-)   preserved by callee
 515 # (#)   Nth argument, volatile
 516 #
 517 # In Unix terms top of stack is argument transfer area for arguments
 518 # which could not be accomodated in registers. Or in other words 7th
 519 # [integer] argument resides at 8(%rsp) upon function entry point.
 520 # 128 bytes above %rsp constitute a "red zone" which is not touched
 521 # by signal handlers and can be used as temporal storage without
 522 # allocating a frame.
 523 #
 524 # In Win64 terms N*8 bytes on top of stack is argument transfer area,
 525 # which belongs to/can be overwritten by callee. N is the number of
 526 # arguments passed to callee, *but* not less than 4! This means that
 527 # upon function entry point 5th argument resides at 40(%rsp), as well
 528 # as that 32 bytes from 8(%rsp) can always be used as temporal
 529 # storage [without allocating a frame]. One can actually argue that
 530 # one can assume a "red zone" above stack pointer under Win64 as well.
 531 # Point is that at apparently no occasion Windows kernel would alter
 532 # the area above user stack pointer in true asynchronous manner...
 533 #
 534 # All the above means that if assembler programmer adheres to Unix
 535 # register and stack layout, but disregards the "red zone" existense,
 536 # it's possible to use following prologue and epilogue to "gear" from
 537 # Unix to Win64 ABI in leaf functions with not more than 6 arguments.
 538 #
 539 # omnipotent_function:
 540 # ifdef WIN64
 541 #       movq    %rdi,8(%rsp)
 542 #       movq    %rsi,16(%rsp)
 543 #       movq    %rcx,%rdi       ; if 1st argument is actually present
 544 #       movq    %rdx,%rsi       ; if 2nd argument is actually ...
 545 #       movq    %r8,%rdx        ; if 3rd argument is ...
 546 #       movq    %r9,%rcx        ; if 4th argument ...
 547 #       movq    40(%rsp),%r8    ; if 5th ...
 548 #       movq    48(%rsp),%r9    ; if 6th ...
 549 # endif
 550 #       ...
 551 # ifdef WIN64
 552 #       movq    8(%rsp),%rdi
 553 #       movq    16(%rsp),%rsi
 554 # endif
 555 #       ret