crypto/openssl/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)/egi;
 194             $self->{label} =~ s/([0-9]+\s*[\*\/\%]\s*[0-9]+)/eval($1)/eg;
 195             $self->{label} =~ s/([0-9]+)/$1<<32>>32/eg;
 196
 197             if (defined($self->{index})) {
 198                 sprintf "%s(%%%s,%%%s,%d)",
 199                                         $self->{label},$self->{base},
 200                                         $self->{index},$self->{scale};
 201             } else {
 202                 sprintf "%s(%%%s)",     $self->{label},$self->{base};
 203             }
 204         } else {
 205             %szmap = ( b=>"BYTE", w=>"WORD", l=>"DWORD", q=>"QWORD" );
 206
 207             $self->{label} =~ s/\./\$/g;
 208             $self->{label} =~ s/0x([0-9a-f]+)/0$1h/ig;
 209             $self->{label} = "($self->{label})" if ($self->{label} =~ /[\*\+\-\/]/);
 210
 211             if (defined($self->{index})) {
 212                 sprintf "%s PTR %s[%s*%d+%s]",$szmap{$sz},
 213                                         $self->{label},
 214                                         $self->{index},$self->{scale},
 215                                         $self->{base};
 216             } elsif ($self->{base} eq "rip") {
 217                 sprintf "%s PTR %s",$szmap{$sz},$self->{label};
 218             } else {
 219                 sprintf "%s PTR %s[%s]",$szmap{$sz},
 220                                         $self->{label},$self->{base};
 221             }
 222         }
 223     }
 224 }
 225 { package register;     # pick up registers, which start with %.
 226     sub re {
 227         my      $class = shift; # muliple instances...
 228         my      $self = {};
 229         local   *line = shift;
 230         undef   $ret;
 231
 232         if ($line =~ /^%(\w+)/) {
 233             bless $self,$class;
 234             $self->{value} = $1;
 235             $ret = $self;
 236             $line = substr($line,@+[0]); $line =~ s/^\s+//;
 237         }
 238         $ret;
 239     }
 240     sub size {
 241         my      $self = shift;
 242         undef   $ret;
 243
 244         if    ($self->{value} =~ /^r[\d]+b$/i)  { $ret="b"; }
 245         elsif ($self->{value} =~ /^r[\d]+w$/i)  { $ret="w"; }
 246         elsif ($self->{value} =~ /^r[\d]+d$/i)  { $ret="l"; }
 247         elsif ($self->{value} =~ /^r[\w]+$/i)   { $ret="q"; }
 248         elsif ($self->{value} =~ /^[a-d][hl]$/i){ $ret="b"; }
 249         elsif ($self->{value} =~ /^[\w]{2}l$/i) { $ret="b"; }
 250         elsif ($self->{value} =~ /^[\w]{2}$/i)  { $ret="w"; }
 251         elsif ($self->{value} =~ /^e[a-z]{2}$/i){ $ret="l"; }
 252
 253         $ret;
 254     }
 255     sub out {
 256         my $self = shift;
 257         sprintf $masm?"%s":"%%%s",$self->{value};
 258     }
 259 }
 260 { package label;        # pick up labels, which end with :
 261     sub re {
 262         my      $self = shift;  # single instance is enough...
 263         local   *line = shift;
 264         undef   $ret;
 265
 266         if ($line =~ /(^[\.\w]+\:)/) {
 267             $self->{value} = $1;
 268             $ret = $self;
 269             $line = substr($line,@+[0]); $line =~ s/^\s+//;
 270
 271             $self->{value} =~ s/\.L/\$L/ if ($masm);
 272         }
 273         $ret;
 274     }
 275     sub out {
 276         my $self = shift;
 277
 278         if (!$masm) {
 279             $self->{value};
 280         } elsif ($self->{value} ne "$current_function->{name}:") {
 281             $self->{value};
 282         } elsif ($current_function->{abi} eq "svr4") {
 283             my $func =  "$current_function->{name}      PROC\n".
 284                         "       mov     QWORD PTR 8[rsp],rdi\t;WIN64 prologue\n".
 285                         "       mov     QWORD PTR 16[rsp],rsi\n";
 286             my $narg = $current_function->{narg};
 287             $narg=6 if (!defined($narg));
 288             $func .= "  mov     rdi,rcx\n" if ($narg>0);
 289             $func .= "  mov     rsi,rdx\n" if ($narg>1);
 290             $func .= "  mov     rdx,r8\n"  if ($narg>2);
 291             $func .= "  mov     rcx,r9\n"  if ($narg>3);
 292             $func .= "  mov     r8,QWORD PTR 40[rsp]\n" if ($narg>4);
 293             $func .= "  mov     r9,QWORD PTR 48[rsp]\n" if ($narg>5);
 294             $func .= "\n";
 295         } else {
 296            "$current_function->{name}   PROC";
 297         }
 298     }
 299 }
 300 { package expr;         # pick up expressioins
 301     sub re {
 302         my      $self = shift;  # single instance is enough...
 303         local   *line = shift;
 304         undef   $ret;
 305
 306         if ($line =~ /(^[^,]+)/) {
 307             $self->{value} = $1;
 308             $ret = $self;
 309             $line = substr($line,@+[0]); $line =~ s/^\s+//;
 310
 311             $self->{value} =~ s/\.L/\$L/g if ($masm);
 312         }
 313         $ret;
 314     }
 315     sub out {
 316         my $self = shift;
 317         $self->{value};
 318     }
 319 }
 320 { package directive;    # pick up directives, which start with .
 321     sub re {
 322         my      $self = shift;  # single instance is enough...
 323         local   *line = shift;
 324         undef   $ret;
 325         my      $dir;
 326         my      %opcode =       # lea 2f-1f(%rip),%dst; 1: nop; 2:
 327                 (       "%rax"=>0x01058d48,     "%rcx"=>0x010d8d48,
 328                         "%rdx"=>0x01158d48,     "%rbx"=>0x011d8d48,
 329                         "%rsp"=>0x01258d48,     "%rbp"=>0x012d8d48,
 330                         "%rsi"=>0x01358d48,     "%rdi"=>0x013d8d48,
 331                         "%r8" =>0x01058d4c,     "%r9" =>0x010d8d4c,
 332                         "%r10"=>0x01158d4c,     "%r11"=>0x011d8d4c,
 333                         "%r12"=>0x01258d4c,     "%r13"=>0x012d8d4c,
 334                         "%r14"=>0x01358d4c,     "%r15"=>0x013d8d4c      );
 335
 336         if ($line =~ /^\s*(\.\w+)/) {
 337             if (!$masm) {
 338                 $self->{value} = $1;
 339                 $line =~ s/\@abi\-omnipotent/\@function/;
 340                 $line =~ s/\@function.*/\@function/;
 341                 if ($line =~ /\.picmeup\s+(%r[\w]+)/i) {
 342                     $self->{value} = sprintf "\t.long\t0x%x,0x90000000",$opcode{$1};
 343                 } elsif ($line =~ /\.asciz\s+"(.*)"$/) {
 344                     $self->{value} = ".byte\t".join(",",unpack("C*",$1),0);
 345                 } elsif ($line =~ /\.extern/) {
 346                     $self->{value} = ""; # swallow extern
 347                 } else {
 348                     $self->{value} = $line;
 349                 }
 350                 $line = "";
 351                 return $self;
 352             }
 353
 354             $dir = $1;
 355             $ret = $self;
 356             undef $self->{value};
 357             $line = substr($line,@+[0]); $line =~ s/^\s+//;
 358             SWITCH: for ($dir) {
 359                 /\.(text)/
 360                             && do { my $v=undef;
 361                                     $v="$current_segment\tENDS\n" if ($current_segment);
 362                                     $current_segment = "_$1\$";
 363                                     $current_segment =~ tr/[a-z]/[A-Z]/;
 364                                     $v.="$current_segment\tSEGMENT ";
 365                                     $v.=$masm>=$masmref ? "ALIGN(64)" : "PAGE";
 366                                     $v.=" 'CODE'";
 367                                     $self->{value} = $v;
 368                                     last;
 369                                   };
 370                 /\.extern/  && do { $self->{value} = "EXTRN\t".$line.":BYTE"; last;  };
 371                 /\.globl/   && do { $self->{value} = "PUBLIC\t".$line; last; };
 372                 /\.type/    && do { ($sym,$type,$narg) = split(',',$line);
 373                                     if ($type eq "\@function") {
 374                                         undef $current_function;
 375                                         $current_function->{name} = $sym;
 376                                         $current_function->{abi}  = "svr4";
 377                                         $current_function->{narg} = $narg;
 378                                     } elsif ($type eq "\@abi-omnipotent") {
 379                                         undef $current_function;
 380                                         $current_function->{name} = $sym;
 381                                     }
 382                                     last;
 383                                   };
 384                 /\.size/    && do { if (defined($current_function)) {
 385                                         $self->{value}="$current_function->{name}\tENDP";
 386                                         undef $current_function;
 387                                     }
 388                                     last;
 389                                   };
 390                 /\.align/   && do { $self->{value} = "ALIGN\t".$line; last; };
 391                 /\.(byte|value|long|quad)/
 392                             && do { my @arr = split(',',$line);
 393                                     my $sz  = substr($1,0,1);
 394                                     my $last = pop(@arr);
 395                                     my $conv = sub  {   my $var=shift;
 396                                                         if ($var=~s/0x([0-9a-f]+)/0$1h/i) { $var; }
 397                                                         else { sprintf"0%Xh",$var; }
 398                                                     };
 399
 400                                     $sz =~ tr/bvlq/BWDQ/;
 401                                     $self->{value} = "\tD$sz\t";
 402                                     for (@arr) { $self->{value} .= &$conv($_).","; }
 403                                     $self->{value} .= &$conv($last);
 404                                     last;
 405                                   };
 406                 /\.picmeup/ && do { $self->{value} = sprintf"\tDD\t 0%Xh,090000000h",$opcode{$line};
 407                                     last;
 408                                   };
 409                 /\.asciz/   && do { if ($line =~ /^"(.*)"$/) {
 410                                         my @str=unpack("C*",$1);
 411                                         push @str,0;
 412                                         while ($#str>15) {
 413                                             $self->{value}.="DB\t"
 414                                                 .join(",",@str[0..15])."\n";
 415                                             foreach (0..15) { shift @str; }
 416                                         }
 417                                         $self->{value}.="DB\t"
 418                                                 .join(",",@str) if (@str);
 419                                     }
 420                                     last;
 421                                   };
 422             }
 423             $line = "";
 424         }
 425
 426         $ret;
 427     }
 428     sub out {
 429         my $self = shift;
 430         $self->{value};
 431     }
 432 }
 433
 434 while($line=<>) {
 435
 436     chomp($line);
 437
 438     $line =~ s|[#!].*$||;       # get rid of asm-style comments...
 439     $line =~ s|/\*.*\*/||;      # ... and C-style comments...
 440     $line =~ s|^\s+||;          # ... and skip white spaces in beginning
 441
 442     undef $label;
 443     undef $opcode;
 444     undef $dst;
 445     undef $src;
 446     undef $sz;
 447
 448     if ($label=label->re(\$line))       { print $label->out(); }
 449
 450     if (directive->re(\$line)) {
 451         printf "%s",directive->out();
 452     } elsif ($opcode=opcode->re(\$line)) { ARGUMENT: {
 453
 454         if ($src=register->re(\$line))  { opcode->size($src->size()); }
 455         elsif ($src=const->re(\$line))  { }
 456         elsif ($src=ea->re(\$line))     { }
 457         elsif ($src=expr->re(\$line))   { }
 458
 459         last ARGUMENT if ($line !~ /^,/);
 460
 461         $line = substr($line,1); $line =~ s/^\s+//;
 462
 463         if ($dst=register->re(\$line))  { opcode->size($dst->size()); }
 464         elsif ($dst=const->re(\$line))  { }
 465         elsif ($dst=ea->re(\$line))     { }
 466
 467         } # ARGUMENT:
 468
 469         $sz=opcode->size();
 470
 471         if (defined($dst)) {
 472             if (!$masm) {
 473                 printf "\t%s\t%s,%s",   $opcode->out($dst->size()),
 474                                         $src->out($sz),$dst->out($sz);
 475             } else {
 476                 printf "\t%s\t%s,%s",   $opcode->out(),
 477                                         $dst->out($sz),$src->out($sz);
 478             }
 479         } elsif (defined($src)) {
 480             printf "\t%s\t%s",$opcode->out(),$src->out($sz);
 481         } else {
 482             printf "\t%s",$opcode->out();
 483         }
 484     }
 485
 486     print $line,"\n";
 487 }
 488
 489 print "\n$current_segment\tENDS\nEND\n" if ($masm);
 490
 491 close STDOUT;
 492
 493 #################################################
 494 # Cross-reference x86_64 ABI "card"
 495 #
 496 #               Unix            Win64
 497 # %rax          *               *
 498 # %rbx          -               -
 499 # %rcx          #4              #1
 500 # %rdx          #3              #2
 501 # %rsi          #2              -
 502 # %rdi          #1              -
 503 # %rbp          -               -
 504 # %rsp          -               -
 505 # %r8           #5              #3
 506 # %r9           #6              #4
 507 # %r10          *               *
 508 # %r11          *               *
 509 # %r12          -               -
 510 # %r13          -               -
 511 # %r14          -               -
 512 # %r15          -               -
 513 #
 514 # (*)   volatile register
 515 # (-)   preserved by callee
 516 # (#)   Nth argument, volatile
 517 #
 518 # In Unix terms top of stack is argument transfer area for arguments
 519 # which could not be accomodated in registers. Or in other words 7th
 520 # [integer] argument resides at 8(%rsp) upon function entry point.
 521 # 128 bytes above %rsp constitute a "red zone" which is not touched
 522 # by signal handlers and can be used as temporal storage without
 523 # allocating a frame.
 524 #
 525 # In Win64 terms N*8 bytes on top of stack is argument transfer area,
 526 # which belongs to/can be overwritten by callee. N is the number of
 527 # arguments passed to callee, *but* not less than 4! This means that
 528 # upon function entry point 5th argument resides at 40(%rsp), as well
 529 # as that 32 bytes from 8(%rsp) can always be used as temporal
 530 # storage [without allocating a frame]. One can actually argue that
 531 # one can assume a "red zone" above stack pointer under Win64 as well.
 532 # Point is that at apparently no occasion Windows kernel would alter
 533 # the area above user stack pointer in true asynchronous manner...
 534 #
 535 # All the above means that if assembler programmer adheres to Unix
 536 # register and stack layout, but disregards the "red zone" existense,
 537 # it's possible to use following prologue and epilogue to "gear" from
 538 # Unix to Win64 ABI in leaf functions with not more than 6 arguments.
 539 #
 540 # omnipotent_function:
 541 # ifdef WIN64
 542 #       movq    %rdi,8(%rsp)
 543 #       movq    %rsi,16(%rsp)
 544 #       movq    %rcx,%rdi       ; if 1st argument is actually present
 545 #       movq    %rdx,%rsi       ; if 2nd argument is actually ...
 546 #       movq    %r8,%rdx        ; if 3rd argument is ...
 547 #       movq    %r9,%rcx        ; if 4th argument ...
 548 #       movq    40(%rsp),%r8    ; if 5th ...
 549 #       movq    48(%rsp),%r9    ; if 6th ...
 550 # endif
 551 #       ...
 552 # ifdef WIN64
 553 #       movq    8(%rsp),%rdi
 554 #       movq    16(%rsp),%rsi
 555 # endif
 556 #       ret