$T_hsc_cc_PAT = '\.ascii.*\)(hsc|cc) (.*)\\\\11"\n\t\.ascii\s+"(.*)\\\\0"';
$T_DOT_WORD = '\.quad';
- $T_DOT_GLOBAL = "\t\.globl";
+ $T_DOT_GLOBAL = '^\t\.globl';
$T_HDR_literal = "\.rdata\n\t\.align 3\n";
$T_HDR_misc = "\.text\n\t\.align 3\n";
$T_HDR_data = "\.data\n\t\.align 3\n";
$T_hsc_cc_PAT = '\.STRING.*\)(hsc|cc) (.*)\\\\x09(.*)\\\\x00';
$T_DOT_WORD = '\.word';
- $T_DOT_GLOBAL = '\s+\.EXPORT';
+ $T_DOT_GLOBAL = '^\s+\.EXPORT';
$T_HDR_literal = "\t.SPACE \$TEXT\$\n\t.SUBSPA \$LIT\$\n";
$T_HDR_misc = "\t.SPACE \$TEXT\$\n\t.SUBSPA \$CODE\$\n\t\.align 4\n";
$T_HDR_data = "\t.SPACE \$PRIVATE\$\n\t.SUBSPA \$DATA\$\n\t\.align 4\n";
$T_HDR_direct = "\t.SPACE \$TEXT\$\n\t.SUBSPA \$CODE\$\n\t\.align 4\n";
#--------------------------------------------------------#
- } elsif ( $TargetPlatform =~ /^i386-.*-(linuxaout|freebsd)/ ) {
+ } elsif ( $TargetPlatform =~ /^i386-.*-(linuxaout|freebsd|nextstep3|cygwin32)/ ) {
+ # NeXT added but not tested. CaS
$T_STABBY = 1; # 1 iff .stab things (usually if a.out format)
$T_US = '_'; # _ if symbols have an underscore on the front
$T_hsc_cc_PAT = '\.ascii.*\)(hsc|cc) (.*)\\\\11"\n\t\.ascii\s+"(.*)\\\\0"';
$T_DOT_WORD = '\.long';
$T_DOT_GLOBAL = '\.globl';
- $T_HDR_literal = "\.text\n\t\.align 2\n"; # .align 4 is 486-cache friendly
+ $T_HDR_literal = "\.text\n\t\.align 2\n";
$T_HDR_misc = "\.text\n\t\.align 2,0x90\n";
- $T_HDR_data = "\.data\n\t\.align 2\n"; # ToDo: change align??
+ $T_HDR_data = "\.data\n\t\.align 2\n";
$T_HDR_consist = "\.text\n";
- $T_HDR_closure = "\.data\n\t\.align 2\n"; # ToDo: change align?
+ $T_HDR_closure = "\.data\n\t\.align 2\n";
$T_HDR_info = "\.text\n\t\.align 2\n"; # NB: requires padding
$T_HDR_entry = "\.text\n"; # no .align so we're right next to _info (arguably wrong...?)
$T_HDR_fast = "\.text\n\t\.align 2,0x90\n";
$T_STABBY = 0; # 1 iff .stab things (usually if a.out format)
$T_US = ''; # _ if symbols have an underscore on the front
$T_DO_GC = 'PerformGC_wrapper';
- $T_PRE_APP = '/'; # regexp that says what comes before APP/NO_APP
+ $T_PRE_APP = # regexp that says what comes before APP/NO_APP
+ ($TargetPlatform =~ /-linux$/) ? '#' : '/' ;
$T_CONST_LBL = '^\.LC(\d+):$'; # regexp for what such a lbl looks like
$T_POST_LBL = ':';
$T_X86_PRE_LLBL_PAT = '\.L';
$T_CONST_LBL = '^LC(\d+):$';
$T_POST_LBL = ':';
- $T_MOVE_DIRVS = '(\s*(\.align\s+\d+|\.proc\s+\d+|\.const|\.cstring|\.globl\s+\S+|\.text|\.data|\.even|\.stab[^n].*)\n)';
+ $T_MOVE_DIRVS = '^(\s*(\.align\s+\d+|\.proc\s+\d+|\.const|\.cstring|\.globl\s+\S+|\.text|\.data|\.even|\.stab[^n].*)\n)';
$T_COPY_DIRVS = '\.(globl|proc|stab)';
$T_hsc_cc_PAT = '\.ascii.*\)(hsc|cc) (.*)\\\\11"\n\t\.ascii\s+"(.*)\\\\0"';
$T_CONST_LBL = '^\$LC(\d+):$'; # regexp for what such a lbl looks like
$T_POST_LBL = ':';
- $T_MOVE_DIRVS = '(\s*(\.align\s+\d+|\.(globl|ent)\s+\S+|\.text|\.r?data)\n)';
+ $T_MOVE_DIRVS = '^(\s*(\.align\s+\d+|\.(globl|ent)\s+\S+|\.text|\.r?data)\n)';
$T_COPY_DIRVS = '\.(globl|ent)';
$T_hsc_cc_PAT = 'I WAS TOO LAZY TO DO THIS BIT (WDP 95/05)';
$T_DOT_WORD = '\.word';
- $T_DOT_GLOBAL = '\t\.globl';
+ $T_DOT_GLOBAL = '^\t\.globl';
$T_HDR_literal = "\t\.rdata\n\t\.align 2\n";
$T_HDR_misc = "\t\.text\n\t\.align 2\n";
$T_HDR_data = "\t\.data\n\t\.align 2\n";
$T_HDR_direct = "\t\.text\n\t\.align 2\n";
#--------------------------------------------------------#
- } elsif ( $TargetPlatform =~ /^powerpc-.*/ ) {
+ } elsif ( $TargetPlatform =~ /^powerpc-.*|^rs6000-.*/ ) {
$T_STABBY = 0; # 1 iff .stab things (usually if a.out format)
- $T_US = '\.'; # _ if symbols have an underscore on the front
- $T_DO_GC = 'PerformGC_wrapper';
+ $T_US = ''; # _ if symbols have an underscore on the front
+ $T_DO_GC = '\.PerformGC_wrapper';
$T_PRE_APP = 'NOT APPLICABLE'; # regexp that says what comes before APP/NO_APP
- $T_CONST_LBL = '^LC\.\.(\d+):$'; # regexp for what such a lbl looks like
+ $T_CONST_LBL = 'NOT APPLICABLE'; # regexp for what such a lbl looks like
$T_POST_LBL = ':';
- $T_MOVE_DIRVS = '^(\s*(\.align\s+\d+(,0x90)?|\.globl\s+\S+|\.text|\.data|\.section\s+.*|\.type\s+.*|\.Lfe.*\n\t\.size\s+.*|\.size\s+.*|\.ident.*)\n)';
- $T_COPY_DIRVS = '\.(globl)';
+ $T_MOVE_DIRVS = '^(\s*(\.toc|\.align \d+|\.csect \S+|\t\.?l?globl \S+)\n)';
+ $T_COPY_DIRVS = '\.(l?globl)';
- $T_hsc_cc_PAT = '\.string.*\)(hsc|cc) (.*)\\\\t(.*)"';
+ $T_hsc_cc_PAT = '\.byte.*\)(hsc|cc) (.*)"\n\t\.byte \d+\n\t\.byte "(.*)"\n\t\.byte \d+';
$T_DOT_WORD = '\.long';
$T_DOT_GLOBAL = '\.globl';
- $T_HDR_literal = "\.section\t\.rodata\n"; # or just use .text??? (WDP 95/11)
- $T_HDR_misc = "\.text\n\t\.align 16\n";
- $T_HDR_data = "\.data\n\t\.align 4\n"; # ToDo: change align??
- $T_HDR_consist = "\.text\n";
- $T_HDR_closure = "\.data\n\t\.align 4\n"; # ToDo: change align?
- $T_HDR_info = "\.text\n\t\.align 16\n"; # NB: requires padding
- $T_HDR_entry = "\.text\n"; # no .align so we're right next to _info (arguably wrong...?)
- $T_HDR_fast = "\.text\n\t\.align 16\n";
- $T_HDR_vector = "\.text\n\t\.align 16\n"; # NB: requires padding
- $T_HDR_direct = "\.text\n\t\.align 16\n";
+ $T_HDR_toc = "\.toc\n";
+ $T_HDR_literal = "\.csect .data[RW]\n\t\.align 2\n"; #not RO!?
+ $T_HDR_misc = "# misc\n\.csect \.text[PR]\n\t\.align 2\n";
+ $T_HDR_data = "# data\n\.csect \.data[RW]\n\t\.align 2\n";
+ $T_HDR_consist = "# consist\n\.csect \.data[RW]\n\t\.align 2\n";
+ $T_HDR_closure = "# closure\n\.csect \.data[RW]\n\t\.align 2\n";
+ $T_HDR_info = "# info\n\.csect \.data[RW]\n\t\.align 2\n"; #not RO!?
+ $T_HDR_entry = "# entry\n\.csect \.text[PR]\n\t\.align 2\n";
+ $T_HDR_fast = "# fast\n\.csect \.text[PR]\n\t\.align 2\n";
+ $T_HDR_vector = "# vector\n\.csect \.data[RW]\n\t\.align 2\n"; #not RO!?
+ $T_HDR_direct = "# direct\n";
#--------------------------------------------------------#
} elsif ( $TargetPlatform =~ /^sparc-.*-solaris2/ ) {
$T_hsc_cc_PAT = '\.asciz.*\)(hsc|cc) (.*)\\\\t(.*)"';
$T_DOT_WORD = '\.word';
- $T_DOT_GLOBAL = '\.global';
+ $T_DOT_GLOBAL = '^\t\.global';
$T_HDR_literal = "\.text\n\t\.align 8\n";
$T_HDR_misc = "\.text\n\t\.align 4\n";
$T_HDR_data = "\.data\n\t\.align 8\n";
# multi-line regexp matching:
local($*) = 1;
local($i, $c);
+
+
&init_TARGET_STUFF();
&init_FUNNY_THINGS();
+ # perl4 on alphas SEGVs when give ${foo} substitutions in patterns.
+ # To avoid them we declare some locals that allows to avoid using curlies.
+ local($TUS) = ${T_US};
+ local($TPOSTLBL) = ${T_POST_LBL};
+ local($TMOVEDIRVS) = ${T_MOVE_DIRVS};
+ local($TPREAPP) = ${T_PRE_APP};
+ local($TCOPYDIRVS) = ${T_COPY_DIRVS};
+ local($TDOTWORD) = ${T_DOT_WORD};
+
open(INASM, "< $in_asmf")
|| &tidy_up_and_die(1,"$Pgm: failed to open `$in_asmf' (to read)\n");
open(OUTASM,"> $out_asmf")
$i = 0; $chkcat[0] = 'misc'; $chk[0] = '';
while (<INASM>) {
- next if $T_STABBY && /^\.stab.*${T_US}__stg_split_marker/o;
+ next if $T_STABBY && /^\.stab.*$TUS[@]?__stg_split_marker/o;
next if $T_STABBY && /^\.stab.*ghc.*c_ID/;
- next if /${T_PRE_APP}(NO_)?APP/o;
-
+ next if /$TPREAPP(NO_)?APP/o;
next if /^;/ && $TargetPlatform =~ /^hppa/;
- next if /(^$|^\t\.file\t|^ # )/ && $TargetPlatform =~ /^mips-/;
+ next if /(^$|^\t\.file\t|^ # )/ && $TargetPlatform =~ /^(mips|powerpc|rs6000)-/;
+
+ last if /^_section_\.text:$/ && $TargetPlatform =~ /^powerpc-|^rs6000-/;
if ( $TargetPlatform =~ /^mips-/
&& /^\t\.(globl \S+ \.text|comm\t)/ ) {
$chkcat[$i] = 'literal';
$chksymb[$i] = $1;
- } elsif ( /^${T_US}__stg_split_marker(\d+)${T_POST_LBL}$/o ) {
+ } elsif ( /^$TUS[@]?__stg_split_marker(\d+)$TPOSTLBL[@]?$/o ) {
$chk[++$i] = $_;
$chkcat[$i] = 'splitmarker';
$chksymb[$i] = $1;
- } elsif ( /^${T_US}([A-Za-z0-9_]+)_info${T_POST_LBL}$/o ) {
+ } elsif ( /^$TUS[@]?([A-Za-z0-9_]+)_info$TPOSTLBL[@]?$/o ) {
$symb = $1;
$chk[++$i] = $_;
$chkcat[$i] = 'infotbl';
$infochk{$symb} = $i;
- } elsif ( /^${T_US}([A-Za-z0-9_]+)_entry${T_POST_LBL}$/o ) {
+ } elsif ( /^$TUS[@]?([A-Za-z0-9_]+)_entry$TPOSTLBL[@]?$/o ) {
$chk[++$i] = $_;
$chkcat[$i] = 'slow';
$chksymb[$i] = $1;
$slowchk{$1} = $i;
- } elsif ( /^${T_US}([A-Za-z0-9_]+)_fast\d+${T_POST_LBL}$/o ) {
+ } elsif ( /^$TUS[@]?([A-Za-z0-9_]+)_fast\d+$TPOSTLBL[@]?$/o ) {
$chk[++$i] = $_;
$chkcat[$i] = 'fast';
$chksymb[$i] = $1;
$fastchk{$1} = $i;
- } elsif ( /^${T_US}([A-Za-z0-9_]+)_closure${T_POST_LBL}$/o ) {
+ } elsif ( /^$TUS[@]?([A-Za-z0-9_]+)_closure$TPOSTLBL[@]?$/o ) {
$chk[++$i] = $_;
$chkcat[$i] = 'closure';
$chksymb[$i] = $1;
$closurechk{$1} = $i;
- } elsif ( /^${T_US}ghc.*c_ID${T_POST_LBL}/o ) {
+ } elsif ( /^$TUS[@]?ghc.*c_ID$TPOSTLBL/o ) {
$chk[++$i] = $_;
$chkcat[$i] = 'consist';
- } elsif ( /^(${T_US}__gnu_compiled_c|gcc2_compiled\.)${T_POST_LBL}/o ) {
+ } elsif ( /^($TUS[@]?__gnu_compiled_c|gcc2_compiled\.)$TPOSTLBL/o ) {
; # toss it
- } elsif ( /^${T_US}ErrorIO_call_count${T_POST_LBL}$/o # HACK!!!!
- || /^${T_US}[A-Za-z0-9_]+\.\d+${T_POST_LBL}$/o
- || /^${T_US}.*_CAT${T_POST_LBL}$/o # PROF: _entryname_CAT
- || /^${T_US}CC_.*_struct${T_POST_LBL}$/o # PROF: _CC_ccident_struct
- || /^${T_US}.*_done${T_POST_LBL}$/o # PROF: _module_done
- || /^${T_US}_module_registered${T_POST_LBL}$/o # PROF: _module_registered
+ } elsif ( /^$TUS[@]?ErrorIO_call_count$TPOSTLBL[@]?$/o # HACK!!!!
+ || /^$TUS[A-Za-z0-9_]+\.\d+$TPOSTLBL[@]?$/o
+ || /^$TUS[@]?.*_CAT$TPOSTLBL[@]?$/o # PROF: _entryname_CAT
+ || /^$TUS[@]?CC_.*_struct$TPOSTLBL[@]?$/o # PROF: _CC_ccident_struct
+ || /^$TUS[@]?.*_done$TPOSTLBL[@]?$/o # PROF: _module_done
+ || /^$TUS[@]?_module_registered$TPOSTLBL[@]?$/o # PROF: _module_registered
) {
$chk[++$i] = $_;
$chkcat[$i] = 'data';
} elsif ( /^([A-Za-z0-9_]+)\s+\.comm/ && $TargetPlatform =~ /^hppa/ ) {
$chk[++$i] = $_;
$chkcat[$i] = 'bss';
- $chksymb[$i] = $1;
-
- } elsif ( /^${T_US}(ret_|djn_)/o ) {
+ $chksymb[$i] = '';
+
+ } elsif ( $TargetPlatform =~ /^powerpc-|^rs6000-/ && /^LC\.\.([0-9]+)/ ) {
+ $chk[++$i] = $_;
+ $chkcat[$i] = 'toc';
+ $chksymb[$i] = $1;
+
+ } elsif ( /^($TUS[@]?(ret_|djn_)[A-Za-z0-9_]+)/o ) {
$chk[++$i] = $_;
$chkcat[$i] = 'misc';
$chksymb[$i] = '';
+ $symbtmp = $1;
+ $chksymb[$i] = $symbtmp if ($TargetPlatform =~ /^powerpc-|^rs6000-/) ; #rm andre
- } elsif ( /^${T_US}vtbl_([A-Za-z0-9_]+)${T_POST_LBL}$/o ) {
+ } elsif ( /^$TUS[@]?vtbl_([A-Za-z0-9_]+)$TPOSTLBL[@]?$/o ) {
$chk[++$i] = $_;
$chkcat[$i] = 'vector';
$chksymb[$i] = $1;
$vectorchk{$1} = $i;
- } elsif ( /^${T_US}([A-Za-z0-9_]+)DirectReturn${T_POST_LBL}$/o ) {
+ } elsif ( /^$TUS[@]?([A-Za-z0-9_]+)DirectReturn$TPOSTLBL[@]?$/o ) {
$chk[++$i] = $_;
$chkcat[$i] = 'direct';
$chksymb[$i] = $1;
$directchk{$1} = $i;
- } elsif ( /^${T_US}[A-Za-z0-9_]+_upd${T_POST_LBL}$/o ) {
+ } elsif ( /^$TUS[@]?([A-Za-z0-9_]+)_upd$TPOSTLBL[@]?$/o ) {
$chk[++$i] = $_;
$chkcat[$i] = 'misc';
- $chksymb[$i] = '';
+ print STDERR "_upd!!!!! I guess this code is dead!!!!\n";
+ # I guess this is never entered, since all _upds are
+ # either vtbl_'s or ret_'s, caught above. - andre
+ $chksymb[$i] = '';
} elsif ( $TargetPlatform =~ /^i386-.*-solaris2/
&& /^(_uname|uname|stat|fstat):/ ) {
$chkcat[$i] = 'toss';
$chksymb[$i] = $1;
- } elsif ( /^${T_US}[A-Za-z0-9_]/o
+ } elsif ( /^$TUS[@]?[A-Za-z0-9_]/o
&& ( $TargetPlatform !~ /^hppa/ # need to avoid local labels in this case
- || /^L\$\d+$/ ) ) {
+ || ! /^L\$\d+$/ )
+ && ( $TargetPlatform !~ /^powerpc|^rs6000/ # ditto
+ || ! /^(L\.\.\d+|LT\.\..*):$/ ) ) {
local($thing);
chop($thing = $_);
print STDERR "Funny global thing?: $_"
unless $KNOWN_FUNNY_THING{$thing}
- || /^${T_US}_(PRIn|PRStart).*${T_POST_LBL}$/o # pointer reversal GC routines
- || /^${T_US}CC_.*${T_POST_LBL}$/o # PROF: _CC_ccident
- || /^${T_US}_reg.*${T_POST_LBL}$/o; # PROF: __reg<module>
+ || /^$TUS[@]?_(PRIn|PRStart).*$TPOSTLBL[@]?$/o # pointer reversal GC routines
+ || /^$TUS[@]?CC_.*$TPOSTLBL$/o # PROF: _CC_ccident ([@]? is a silly hack (see above))
+ || /^$TUS[@]?_reg.*$TPOSTLBL$/o; # PROF: __reg<module>
$chk[++$i] = $_;
$chkcat[$i] = 'misc';
- $chksymb[$i] = '';
+ if ($TargetPlatform =~ /^powerpc-|^rs6000-/)
+ { $chksymb[$i] = $thing; }
+ else { $chksymb[$i] = ''; };
} else { # simple line (duplicated at the top)
# HPPAs, MIPSen: also start medding at chunk 1
+ if ($TargetPlatform =~ /^powerpc|^rs6000/) {
+ print OUTASM $T_HDR_toc; # yes, we have to put a .toc
+ # in the beginning of every file!
+ %tocequiv = (); # maps toc symbol number to toc symbol
+ %revtocequiv = (); # maps toc symbol to toc symbol number
+ for ($i = 1; $i < $numchks; $i++) {
+ $chk[$i] =~ s/\[RW\]//g;
+ $chk[$i] =~ s/\[DS\]//g;
+ $chk[$i] =~ s/^\.csect .*\[DS\]$//g;
+
+ if ( $chkcat[$i] eq 'toc' && $chk[$i] !~ /\.byte/ )
+ { $chk[$i] =~ s/$T_MOVE_DIRVS//g;
+ $chk[$i] =~ s/\t\.tc (\S+)\[TC\],(\S+_fast\d+)/\t\.tc \1\[TC\],\.\2/;
+ $chk[$i] =~ s/\t\.tc (\S+)\[TC\],(\S+_entry)\n/\t\.tc \1\[TC\],\.\2\n/;
+ $chk[$i] =~ s/\t\.tc (\S+)\[TC\],(ret_\S+)/\t\.tc \1\[TC\],\.\2/;
+ $chk[$i] =~ s/\t\.tc (\S+)\[TC\],(djn_\S+)/\t\.tc \1\[TC\],\.\2/;
+ $chk[$i] =~ s/\t\.tc (\S+)\[TC\],(vtbl_\S+)/\t\.tc \1\[TC\],\.\2/;
+ $chk[$i] =~ s/\t\.tc (\S+)\[TC\],ErrorIO_innards/\t\.tc \1\[TC\],\.ErrorIO_innards/;
+ $chk[$i] =~ s/\t\.tc (\S+)\[TC\],startStgWorld/\t\.tc \1\[TC\],\.startStgWorld/;
+ $chk[$i] =~ s/\.tc UpdatePAP\[TC\],UpdatePAP/\.tc UpdatePAP\[TC\],\.UpdatePAP/;
+ $chk[$i] =~ s/\.tc IndUpdRetDir\[TC\],IndUpdRetDir/\.tc IndUpdRetDir\[TC\],\.IndUpdRetDir/;
+ $chk[$i] =~ s/\t\.tc (_PRStart_\S+)\[TC\],_PRStart_\S+/\t\.tc \1\[TC\],\.\1/;
+
+ $tocnumber = $chksymb[$i];
+ $tocsymb = $chk[$i];
+ $tocsymb =~ s/^LC\.\.\d+:\n//;
+ $tocsymb =~ s/^\t\.tc \S+,(\S+)\n/\1/;
+ $tocequiv{$tocnumber} = $tocsymb;
+
+ } elsif ( $chkcat[$i] eq 'toc' && $chk[$i] =~ /\.byte/ ) {
+ $chkcat[$i] = 'literal';
+ }
+ }
+ };
+
for ($i = $FIRST_MANGLABLE; $i < $numchks; $i++) {
$c = $chk[$i]; # convenience copy
$p =~ s/^\tpushl \%edi\n//;
$p =~ s/^\tpushl \%esi\n//;
$p =~ s/^\tsubl \$\d+,\%esp\n//;
+ $p =~ s/^\tmovl \$\d+,\%eax\n\tcall __alloca\n// if ($TargetPlatform =~ /^.*-cygwin32/);
} elsif ($TargetPlatform =~ /^m68k-/) {
$p =~ s/^\tlink a6,#-?\d.*\n//;
+ $p =~ s/^\tpea a6@\n\tmovel sp,a6\n//;
+ # The above showed up in the asm code,
+ # so I added it here.
+ # I hope it's correct.
+ # CaS
$p =~ s/^\tmovel d2,sp\@-\n//;
$p =~ s/^\tmovel d5,sp\@-\n//; # SMmark.* only?
$p =~ s/^\tmoveml \#0x[0-9a-f]+,sp\@-\n//; # SMmark.* only?
$p =~ s/^\tsw\t\$fp,\d+\(\$sp\)\n//;
$p =~ s/^\tsw\t\$28,\d+\(\$sp\)\n//;
$p =~ s/__FRAME__/$FRAME/;
+ } elsif ($TargetPlatform =~ /^powerpc-|^rs6000/) {
+ $p =~ s/^\tmflr 0\n//;
+ $p =~ s/^\tstm \d+,-\d+\(1\)\n//;
+ $p =~ s/^\tstw? 0,\d+\(1\)\n//;
+ $p =~ s/^\tstw?u 1,-\d+\(1\)\n//;
+ $p =~ s/^\tstw? \d+,-\d+\(1\)\n//g;
+ $p =~ s/^\tstfd \d+,-\d+\(1\)\n//g;
} else {
print STDERR "$Pgm: unknown prologue mangling? $TargetPlatform\n";
}
- die "Prologue junk?: $p\n" if $p =~ /^\t[^\.]/;
+ die "Prologue junk?: $p\n" if $p =~ /^\t[^\.]/
+ && $TargetPlatform !~ /^powerpc-/; #ToDo: remove test
# glue together what's left
$c = $p . $r;
+ $c =~ s/\n\t\n/\n/; # junk blank line
}
}
+ if ( $TargetPlatform =~ /^mips-/ ) {
+ # MIPS: first, this basic sequence may occur "--- END ---" or not
+ $c =~ s/^\tlw\t\$31,\d+\(\$sp\)\n\taddu\t\$sp,\$sp,\d+\n\tj\t\$31\n\t\.end/\t\.end/;
+ }
+
# toss all epilogue stuff; again, paranoidly
if ( $c =~ /--- END ---/ ) {
if (($r, $e) = split(/--- END ---/, $c)) {
$e =~ s/^\tlw\t\$fp,\d+\(\$sp\)\n//;
$e =~ s/^\taddu\t\$sp,\$sp,\d+\n//;
$e =~ s/^\tj\t\$31\n//;
+ } elsif ($TargetPlatform =~ /^powerpc-|^rs6000-/) {
+ $e =~ s/^\taddi 1,1,\d+\n//;
+ $e =~ s/^\tcal 1,\d+\(1\)\n//;
+ $e =~ s/^\tlw?z? \d+,\d+\(1\)\n//;
+ $e =~ s/^\tmtlr 0\n//;
+ $e =~ s/^\tbl?r\n//;
} else {
print STDERR "$Pgm: unknown epilogue mangling? $TargetPlatform\n";
}
- die "Epilogue junk?: $e\n" if $e =~ /^\t[^\.]/;
+ die "Epilogue junk?: $e\n" if $e =~ /^\t[^\.]/
+ && $TargetPlatform !~ /^powerpc-/; #ToDo: remove test
# glue together what's left
$c = $r . $e;
# On Alphas, the prologue mangling is done a little later (below)
# toss all calls to __DISCARD__
- $c =~ s/^\t(call|jbsr|jal) ${T_US}__DISCARD__\n//go;
+ $c =~ s/^\t(call|jbsr|jal)\s+$TUS[@]?__DISCARD__\n//go;
# MIPS: that may leave some gratuitous asm macros around
# (no harm done; but we get rid of them to be tidier)
# pin a funny end-thing on (for easier matching):
$c .= 'FUNNY#END#THING';
- while ( $c =~ /${T_MOVE_DIRVS}FUNNY#END#THING/o ) {
- $to_move = $1;
+ while ( $c =~ /$TMOVEDIRVS[@]?FUNNY#END#THING/o ) { # [@]? is a silly hack to avoid having to use curlies for T_PRE_APP
+ # (this SEGVs perl4 on alphas, you see)
+ $to_move = $1;
if ( $i < ($numchks - 1)
- && ( $to_move =~ /${T_COPY_DIRVS}/
+ && ( $to_move =~ /$TCOPYDIRVS/
|| ($TargetPlatform =~ /^hppa/ && $to_move =~ /align/ && $chkcat[$i+1] eq 'literal') )) {
$chk[$i + 1] = $to_move . $chk[$i + 1];
# otherwise they're tossed
}
- $c =~ s/${T_MOVE_DIRVS}FUNNY#END#THING/FUNNY#END#THING/o;
+ $c =~ s/$TMOVEDIRVS[@]?FUNNY#END#THING/FUNNY#END#THING/o; # [@]? is a hack (see above)
}
if ( $TargetPlatform =~ /^alpha-/ && $c =~ /^\t\.ent\s+(\S+)/ ) {
if ( $chkcat[$i] eq 'misc' ) {
if ($chk[$i] ne '') {
print OUTASM $T_HDR_misc;
- &print_doctored($chk[$i], 0);
+ if ($TargetPlatform =~ /^powerpc-|^rs6000/) {
+ $chksymb[$i] =~ s/://;
+#new if ($chksymb[$i] =~ /ret.*upd/ || $KNOWN_FUNNY_THING{$chksymb[$i]}
+#new || $chksymb[$i] =~ /^$.{T_US}_(PRIn|PRStart).*${T_POST_LBL}$/o )
+#new { print OUTASM "\t\.globl $chksymb[$i]\n"; }
+ if ($chksymb[$i] ne '')
+ { print OUTASM "\t\.globl \.$chksymb[$i]\n"; };
+ if ($chk[$i] =~ /TOC\[tc0\], 0\n/)
+ { ($p, $r) = split(/TOC\[tc0\], 0\n/, $chk[$i]); $printDS = 1;}
+ else { $r = $chk[$i]; $printDS = 0; };
+ $chk[$i] = &mangle_powerpc_tailjump($r);
+ };
+ &print_doctored($chk[$i], 0);
+ if ($TargetPlatform =~ /^powerpc-|^rs6000-/ && $printDS) {
+ print OUTASM "\.csect ${chksymb[$i]}[DS]\n";
+ print OUTASM "${p}TOC[tc0], 0\n";
+ }
}
} elsif ( $chkcat[$i] eq 'toss' ) {
$consist =~ s/\//./g;
$consist =~ s/-/_/g;
$consist =~ s/[^A-Za-z0-9_.]/ZZ/g; # ToDo: properly?
- print OUTASM $T_HDR_consist, "${consist}${T_POST_LBL}\n"
- if $TargetPlatform !~ /^mips-/; # we just don't try in that case
- } else {
+ print OUTASM $T_HDR_consist, "${consist}${T_POST_LBL}\n";
+
+ } elsif ( $TargetPlatform !~ /^(mips)-/ ) { # we just don't try in those case (ToDo)
+ # on mips: consistency string is just a v
+ # horrible bunch of .bytes,
+ # which I am too lazy to sort out (WDP 95/05)
+
print STDERR "Couldn't grok consistency: ", $chk[$i];
}
if ( defined($infochk{$symb}) ) {
print OUTASM $T_HDR_info;
+ if ($TargetPlatform =~ /^powerpc-|^rs6000-/) {
+ $chk[$infochk{$symb}] =~ s/\.long ([_A-Za-z]\S+_entry)/\.long \.\1/;
+ $chk[$infochk{$symb}] =~ s/\.long ([A-Za-z]\S+_upd)/\.long \.\1/;
+ $chk[$infochk{$symb}] =~ s/\.long (IndUpdRet\S+)/\.long \.\1/;
+ $chk[$infochk{$symb}] =~ s/\.long StdErrorCode/\.long \.StdErrorCode/;
+ $chk[$infochk{$symb}] =~ s/\.long UpdErr/\.long \.UpdErr/;
+ print OUTASM $chk[$infochk{$symb}];
+ } else {
print OUTASM &rev_tbl($symb, $chk[$infochk{$symb}], 1);
+ }
# entry code will be put here!
# paranoia
- if ( $chk[$infochk{$symb}] =~ /${T_DOT_WORD}\s+([A-Za-z0-9_]+_entry)$/o
+ if ( $chk[$infochk{$symb}] =~ /$TDOTWORD[@]?\s+([A-Za-z0-9_]+_entry)$/o
&& $1 ne "${T_US}${symb}_entry" ) {
print STDERR "!!! entry point???\n",$chk[$infochk{$symb}];
}
# teach it to drop through to the fast entry point:
$c = $chk[$slowchk{$symb}];
+ if ($TargetPlatform =~ /^powerpc-|^rs6000-/) {
+ ($p, $r) = split(/TOC\[tc0\], 0\n/, $c);
+ if ($symb =~ /^[_A-Z]/)
+ {
+ print OUTASM "\t\.globl \.${chksymb[$i]}_entry\n";
+ print OUTASM "\.csect ${symb}_entry[DS]\n";
+ print OUTASM "${p}TOC[tc0], 0\n";
+ };
+ $r =~ s/\.csect \.text\[PR\]\n//; # todo: properly - andre
+ $c = &mangle_powerpc_tailjump($r);
+ };
+
if ( defined($fastchk{$symb}) ) {
if ( $TargetPlatform =~ /^alpha-/ ) {
$c =~ s/^\tjmp \$31,\(\$27\),0\n\t\.align 4\n\t\.end/\t.align 4\n\t.end/;
} elsif ( $TargetPlatform =~ /^m68k-/ ) {
$c =~ s/^\tjmp ${T_US}${symb}_fast\d+.*\n\tnop\n//;
$c =~ s/^\tjmp ${T_US}${symb}_fast\d+.*\n//;
+ } elsif ( $TargetPlatform =~ /^powerpc-|^rs6000-/ ) {
+ $c =~ s/^\tb \.${T_US}${symb}_fast\d+\n//;
} elsif ( $TargetPlatform =~ /^sparc-/ ) {
$c =~ s/^\tcall ${T_US}${symb}_fast\d+,.*\n\tnop\n//;
$c =~ s/^\tcall ${T_US}${symb}_fast\d+,.*\n(\t[a-z].*\n)/$1/;
# references to fast-entry point.
# (questionable re hppa and mips...)
print STDERR "still has jump to fast entry point:\n$c"
- if $c =~ /${T_US}${symb}_fast/; # NB: paranoia
+ if $c =~ /$TUS[@]?$symb[@]?_fast/; # NB: paranoia
}
print OUTASM $T_HDR_entry;
# FAST ENTRY POINT
if ( defined($fastchk{$symb}) ) {
+ $c = $chk[$fastchk{$symb}];
if ( ! defined($slowchk{$symb})
# ToDo: the || clause can go once we're no longer
# concerned about producing exactly the same output as before
- || $TargetPlatform =~ /^(m68k|sparc|i386)-/
+#OLD: || $TargetPlatform =~ /^(m68k|sparc|i386)-/
) {
print OUTASM $T_HDR_fast;
}
- &print_doctored($chk[$fastchk{$symb}], 0);
+ if ($TargetPlatform =~ /^powerpc-|^rs6000-/) {
+ local(@lbls) = split(/:/, $c);
+ $fullname = $lbls[0];
+ $fullname =~ s/$T_MOVE_DIRVS//g;
+ if ( $fullname =~ /^[A-Z]/)
+ { print OUTASM "\t\.globl \.${fullname}\n";
+ } else {
+ print OUTASM "\t\.lglobl \.${fullname}\n"; #todo: rm - andre
+ };
+ $c =~ s/((.*\n)*)\t.long \S+, TOC\[tc0\], 0\n\.csect \.text\[PR\]\n((.*\n)*)/\1\3/;
+ $c = &mangle_powerpc_tailjump($c);
+ };
+ &print_doctored($c, 0);
$chkcat[$fastchk{$symb}] = 'DONE ALREADY';
}
# VECTOR TABLE
if ( defined($vectorchk{$symb}) ) {
print OUTASM $T_HDR_vector;
- print OUTASM &rev_tbl($symb, $chk[$vectorchk{$symb}], 0);
+ if ($TargetPlatform =~ /^powerpc-|^rs6000-/) {
+ if ( $symb =~ /^[A-Z]/) {
+ print OUTASM "\t\.globl \.vtbl_${symb}\n";
+ print OUTASM "\t\.globl vtbl_${symb}\n";
+ };
+ $chk[$vectorchk{$symb}] =~ s/\.long (\S+)/\.long \.\1/g;
+ print OUTASM ".vtbl_${symb}:\n";
+ print OUTASM $chk[$vectorchk{$symb}];
+ } else {
+ print OUTASM &rev_tbl($symb, $chk[$vectorchk{$symb}], 0);
+ }
# direct return code will be put here!
$chkcat[$vectorchk{$symb}] = 'DONE ALREADY';
}
# DIRECT RETURN
if ( defined($directchk{$symb}) ) {
print OUTASM $T_HDR_direct;
- &print_doctored($chk[$directchk{$symb}], 0);
+ if ($TargetPlatform =~ /^powerpc-|^rs6000-/) {
+ ($p, $r) = split(/TOC\[tc0\], 0\n/, $chk[$directchk{$symb}]);
+ &print_doctored($r, 0);
+ print OUTASM "\.csect ${symb}DirectReturn[DS]\n";
+ print OUTASM "${p}TOC[tc0], 0\n";
+ } else {
+ &print_doctored($chk[$directchk{$symb}], 0);
+ }
$chkcat[$directchk{$symb}] = 'DONE ALREADY';
} elsif ( $TargetPlatform =~ /^alpha-/ ) {
print OUTASM "\t# nop\n";
}
+ } elsif ( $chkcat[$i] eq 'toc' ) {
+ # silly optimisation to print tocs, since they come in groups...
+ print OUTASM $T_HDR_toc;
+ local($j) = $i;
+ while ($chkcat[$j] eq 'toc')
+ { print OUTASM $chk[$j];
+ $chkcat[$j] = 'DONE ALREADY';
+ $j++;
+ }
+
} else {
&tidy_up_and_die(1,"$Pgm: unknown chkcat (ghc-asm: $TargetPlatform)\n$chkcat[$i]\n$chk[$i]\n");
}
}
+
+ print OUTASM $EXTERN_DECLS if $TargetPlatform =~ /^mips-/;
+
+ if ($TargetPlatform =~ /^powerpc-|^rs6000-/) {
+ print OUTASM ".csect .text[PR]\n_section_.text:\n.csect .data[RW]\n\t.long _section_.text\n"
+ };
+
# finished
close(OUTASM) || &tidy_up_and_die(1,"Failed writing to $out_asmf\n");
close(INASM) || &tidy_up_and_die(1,"Failed reading from $in_asmf\n");
# final peephole fixes
s/^\tmovl \%eax,36\(\%ebx\)\n\tjmp \*36\(\%ebx\)\n/\tmovl \%eax,36\(\%ebx\)\n\tjmp \*\%eax\n/;
- s/^\tmovl \$_(.*),(\%e[abcd]x)\n\tjmp \*$2/\tjmp _$1/g;
+# the short form may tickle perl bug:
+# s/^\tmovl \$${T_US}(.*),(\%e[abcd]x)\n\tjmp \*$2/\tjmp $T_US$1/g;
+ s/^\tmovl \$${T_US}(.*),\%eax\n\tjmp \*\%eax/\tjmp $T_US$1/g;
+ s/^\tmovl \$${T_US}(.*),\%ebx\n\tjmp \*\%ebx/\tjmp $T_US$1/g;
+ s/^\tmovl \$${T_US}(.*),\%ecx\n\tjmp \*\%ecx/\tjmp $T_US$1/g;
+ s/^\tmovl \$${T_US}(.*),\%edx\n\tjmp \*\%edx/\tjmp $T_US$1/g;
# Hacks to eliminate some reloads of Hp. Worth about 5% code size.
# We could do much better than this, but at least it catches about
"${T_US}_PRMarking_MarkNextGA${T_POST_LBL}", 1,
"${T_US}_PRMarking_MarkNextRoot${T_POST_LBL}", 1,
"${T_US}_PRMarking_MarkNextSpark${T_POST_LBL}", 1,
+ "${T_US}_PRMarking_MarkNextEvent${T_POST_LBL}", 1,
+ "${T_US}_PRMarking_MarkNextClosureInFetchBuffer${T_POST_LBL}", 1,
"${T_US}_Scavenge_Forward_Ref${T_POST_LBL}", 1,
"${T_US}__std_entry_error__${T_POST_LBL}", 1,
"${T_US}_startMarkWorld${T_POST_LBL}", 1,
local(@words) = ();
local($after) = '';
local(@lines) = split(/\n/, $tbl);
- local($i, $extra, $words_to_pad, $j);
-
- for ($i = 0; $i <= $#lines && $lines[$i] !~ /^\t${T_DOT_WORD}\s+/o; $i++) {
+ local($i, $j); #local ($i, $extra, $words_to_pad, $j);
+
+ # see comment in mangleAsm as to why this silliness is needed.
+ local($TDOTWORD) = ${T_DOT_WORD};
+ local($TDOTGLOBAL) = ${T_DOT_GLOBAL};
+ local($TUS) = ${T_US};
+ local($TPOSTLBL) = ${T_POST_LBL};
+
+ for ($i = 0; $i <= $#lines && $lines[$i] !~ /^\t$TDOTWORD\s+/o; $i++) {
$label .= $lines[$i] . "\n",
- next if $lines[$i] =~ /^[A-Za-z0-9_]+_info${T_POST_LBL}$/o
- || $lines[$i] =~ /^${T_DOT_GLOBAL}/o
- || $lines[$i] =~ /^${T_US}vtbl_\S+${T_POST_LBL}$/o;
+ next if $lines[$i] =~ /^[A-Za-z0-9_]+_info$TPOSTLBL[@]?$/o
+ || $lines[$i] =~ /$TDOTGLOBAL/o
+ || $lines[$i] =~ /^$TUS[@]?vtbl_\S+$TPOSTLBL[@]?$/o;
$before .= $lines[$i] . "\n"; # otherwise...
}
if ( $TargetPlatform !~ /^hppa/ ) {
- for ( ; $i <= $#lines && $lines[$i] =~ /^\t${T_DOT_WORD}\s+/o; $i++) {
+ for ( ; $i <= $#lines && $lines[$i] =~ /^\t$TDOTWORD\s+/o; $i++) {
push(@words, $lines[$i]);
}
} else { # hppa weirdness
&init_TARGET_STUFF();
+ # see mangleAsm comment
+ local($TUS) = ${T_US};
+ local($TPOSTLBL)=${T_POST_LBL};
+
open(INASM, "< $in_asmf")
|| &tidy_up_and_die(1,"$Pgm: failed to open `$in_asmf' (to read)\n");
open(OUTASM,"> $out_asmf")
print OUTASM;
next unless
- /^${T_US}(PerformGC|StackOverflow|Yield|PerformReschedule)_wrapper${T_POST_LBL}\n/o;
+ /^$TUS[@]?(PerformGC|StackOverflow|Yield|PerformReschedule)_wrapper$TPOSTLBL\n/o;
print OUTASM "\tmovl \%esp, ${T_US}__temp_esp\n";
print OUTASM "\tmovl \%eax, ${T_US}__temp_eax\n";
}
close(INASM) || &tidy_up_and_die(1,"Failed reading from $in_asmf\n");
}
+\end{code}
+
+\begin{code}
+sub mini_mangle_asm_powerpc {
+ local($in_asmf, $out_asmf) = @_;
+
+ open(INASM, "< $in_asmf")
+ || &tidy_up_and_die(1,"$Pgm: failed to open `$in_asmf' (to read)\n");
+ open(OUTASM,"> $out_asmf")
+ || &tidy_up_and_die(1,"$Pgm: failed to open `$out_asmf' (to write)\n");
+
+ while (<INASM>) {
+ s/long _PRStart/long ._PRStart/;
+ s/long _PRIn_/long ._PRIn_/;
+ s/long _Dummy_(\S+)_entry/long ._Dummy_\1_entry/;
+ s/long _PRMarking_MarkNextRoot\[DS\]/long ._PRMarking_MarkNextRoot/;
+ s/long _PRMarking_MarkNextCAF\[DS\]/long ._PRMarking_MarkNextCAF/;
+ s/long _PRMarking_MarkNextAStack\[DS\]/long ._PRMarking_MarkNextAStack/;
+ s/long _PRMarking_MarkNextBStack\[DS\]/long ._PRMarking_MarkNextBStack/;
+ print OUTASM;
+ }
+
+ # finished:
+ close(OUTASM) || &tidy_up_and_die(1,"Failed writing to $out_asmf\n");
+ close(INASM) || &tidy_up_and_die(1,"Failed reading from $in_asmf\n");
+}
+
+sub mangle_powerpc_tailjump {
+ local($c) = @_;
+ local($maybe_more) = 1;
+ while (($c =~ /\tlw?z? \d+,LC\.\.\d+\(2\)\n\tmtctr \d+\n\tbctr\n/) && $maybe_more)
+ { $maybe_more = 0;
+ $lcsymb = $c;
+ $lcsymb =~ s/(.*\n)*\tlw?z? \d+,LC\.\.(\d+)\(2\)\n\tmtctr \d+\n\tbctr\n(.*\n)*/\2/;
+# the checks for r1 and r2 are mostly paranoia...
+ $r1 = $c;
+ $r1 =~ s/(.*\n)*\tlw?z? (\d+),LC\.\.\d+\(2\)\n\tmtctr \d+\n\tbctr\n(.*\n)*/\2/;
+ $r2 = $c;
+ $r2 =~ s/(.*\n)*\tlw?z? \d+,LC\.\.(\d+)\(2\)\n\tmtctr (\d+)\n\tbctr\n(.*\n)*/\3/;
+ if (r1 == r2)
+ { $maybe_more = 1;
+ $c =~ s/((.*\n)*)\tlw?z? \d+,LC\.\.\d+\(2\)\n\tmtctr \d+\n\tbctr\n((.*\n)*)/\1\tb $tocequiv{$lcsymb}\n\3/;
+ }
+ };
+ $c;
+}
+
# make "require"r happy...
1;
\end{code}