]> git.xonotic.org Git - xonotic/xonotic.git/blobdiff - misc/tools/progs-analyzer.pl
fix the last analyzing step
[xonotic/xonotic.git] / misc / tools / progs-analyzer.pl
index c46e98e567f89bba0ecee35dce5d62d1740a7849..0a0593b6c0d8c6e7977602cf24a404321dc27dd3 100644 (file)
@@ -1,5 +1,6 @@
 use strict;
 use warnings;
+use Digest::SHA;
 
 sub id()
 {
@@ -117,7 +118,7 @@ sub checkop($)
        }
        if($op =~ /^DONE$|^RETURN$/)
        {
-               return { a => 'inglobal', isreturn => 1 };
+               return { a => 'inglobalvec', isreturn => 1 };
        }
        return { a => 'inglobal', b => 'inglobal', c => 'outglobal' };
 }
@@ -226,10 +227,19 @@ sub parse_section($$$$$)
        return $out[0];
 }
 
+sub nfa_default_state_checker()
+{
+       my %seen;
+       return sub
+       {
+               my ($ip, $state) = @_;
+               return $seen{"$ip $state"}++;
+       };
+}
+
 sub run_nfa($$$$$$)
 {
-       my ($progs, $ip, $state, $copy_handler, $state_hasher, $instruction_handler) = @_;
-       my %seen = ();
+       my ($progs, $ip, $state, $copy_handler, $state_checker, $instruction_handler) = @_;
 
        my $statements = $progs->{statements};
 
@@ -239,17 +249,21 @@ sub run_nfa($$$$$$)
                no warnings 'recursion';
 
                my ($ip, $state) = @_;
+               my $ret = 0;
 
                for(;;)
                {
-                       my $statestr = $state_hasher->($state);
-                       return
-                               if $seen{"$ip:$statestr"}++;
+                       return $ret
+                               if $state_checker->($ip, $state);
 
                        my $s = $statements->[$ip];
                        my $c = checkop $s->{op};
 
-                       $instruction_handler->($ip, $state, $s, $c);
+                       if(($ret = $instruction_handler->($ip, $state, $s, $c)))
+                       {
+                               # abort execution
+                               last;
+                       }
 
                        if($c->{isreturn})
                        {
@@ -259,8 +273,19 @@ sub run_nfa($$$$$$)
                        {
                                if($c->{isconditional})
                                {
-                                       $nfa->($ip+1, $copy_handler->($state));
-                                       $ip += $s->{$c->{isjump}};
+                                       if(rand 2)
+                                       {
+                                               if(($ret = $nfa->($ip+$s->{$c->{isjump}}, $copy_handler->($state))) < 0)
+                                               {
+                                                       last;
+                                               }
+                                               $ip += 1;
+                                       }
+                                       else
+                                       {
+                                               $nfa->($ip+1, $copy_handler->($state));
+                                               $ip += $s->{$c->{isjump}};
+                                       }
                                }
                                else
                                {
@@ -272,15 +297,41 @@ sub run_nfa($$$$$$)
                                $ip += 1;
                        }
                }
+
+               return $ret;
        };
 
        $nfa->($ip, $copy_handler->($state));
 }
 
-use constant PRE_MARK_STATEMENT => "\e[1m";
-use constant POST_MARK_STATEMENT => "\e[m";
-use constant PRE_MARK_OPERAND => "\e[41m";
-use constant POST_MARK_OPERAND => "\e[49m";
+sub get_constant($$)
+{
+       my ($progs, $g) = @_;
+       if($g->{int} == 0)
+       {
+               return undef;
+       }
+       elsif($g->{int} > 0 && $g->{int} < 16777216)
+       {
+               if($g->{int} < length $progs->{strings} && $g->{int} > 0)
+               {
+                       return str($progs->{getstring}->($g->{int}));
+               }
+               else
+               {
+                       return $g->{int} . "i";
+               }
+       }
+       else
+       {
+               return $g->{float};
+       }
+}
+
+use constant PRE_MARK_STATEMENT => "";
+use constant POST_MARK_STATEMENT => "";
+use constant PRE_MARK_OPERAND => "*** ";
+use constant POST_MARK_OPERAND => " ***";
 
 use constant INSTRUCTION_FORMAT => "%8s %3s | %-12s ";
 use constant OPERAND_FORMAT => "%s";
@@ -303,22 +354,9 @@ sub disassemble_function($$;$)
        my $initializer = sub
        {
                my ($ofs) = @_;
-               my $g = $progs->{globals}[$ofs]{v};
-               if($g->{int} == 0)
-               {
-               }
-               elsif($g->{int} < 16777216)
-               {
-                       print " = $g->{int}%";
-                       if($g->{int} < length $progs->{strings} && $g->{int} > 0)
-                       {
-                               print " " . str($progs->{getstring}->($g->{int}));
-                       }
-               }
-               else
-               {
-                       print " = $g->{float}!";
-               }
+               my $g = get_constant($progs, $progs->{globals}[$ofs]{v});
+               print " = $g"
+                       if defined $g;
        };
 
        printf INSTRUCTION_FORMAT, '', '', '.PARM_START';
@@ -333,17 +371,14 @@ sub disassemble_function($$;$)
        my $p = $func->{parm_start};
        for(0..($func->{numparms}-1))
        {
-               if($func->{parm_size}[$_] <= 1)
-               {
-                       $override_locals{$p} //= "argv[$_]";
-               }
+               $override_locals{$p} //= "argv_$_";
                for my $comp(0..($func->{parm_size}[$_]-1))
                {
-                       $override_locals{$p} //= "argv[$_][$comp]";
+                       $override_locals{$p} //= "argv_$_\[$comp]";
                        ++$p;
                }
                printf INSTRUCTION_FORMAT, '', '', '.ARG';
-               printf OPERAND_FORMAT, "argv[$_]";
+               printf OPERAND_FORMAT, "argv_$_";
                print OPERAND_SEPARATOR;
                printf OPERAND_FORMAT, $func->{parm_size}[$_];
                print INSTRUCTION_SEPARATOR;
@@ -352,10 +387,10 @@ sub disassemble_function($$;$)
        {
                next
                        if exists $override_locals{$_};
-               $override_locals{$_} = "<local>\@$_";
+               $override_locals{$_} = "local_$_";
 
                printf INSTRUCTION_FORMAT, '', '', '.LOCAL';
-               printf OPERAND_FORMAT, "<local>\@$_";
+               printf OPERAND_FORMAT, "local_$_";
                $initializer->($_);
                print INSTRUCTION_SEPARATOR;
        }
@@ -363,9 +398,11 @@ sub disassemble_function($$;$)
        my $getname = sub
        {
                my ($ofs) = @_;
+               $ofs &= 0xFFFF;
                return $override_locals{$ofs}
                        if exists $override_locals{$ofs};
-               return $progs->{globaldef_byoffset}->($ofs)->{debugname};
+               my $def = $progs->{globaldef_byoffset}->($ofs);
+               return $def->{debugname};
        };
 
        my $operand = sub
@@ -408,7 +445,7 @@ sub disassemble_function($$;$)
 
        my %statements = ();
        my %come_from = ();
-       run_nfa $progs, $func->{first_statement}, "", id, id,
+       run_nfa $progs, $func->{first_statement}, "", id, nfa_default_state_checker,
                sub
                {
                        my ($ip, $state, $s, $c) = @_;
@@ -419,6 +456,8 @@ sub disassemble_function($$;$)
                                my $t = $ip + $s->{$j};
                                $come_from{$t}{$ip} = $c->{isconditional};
                        }
+
+                       return 0;
                };
 
        my $ipprev = undef;
@@ -483,6 +522,7 @@ sub find_uninitialized_locals($$)
 {
        my ($progs, $func) = @_;
 
+
        return
                if $func->{first_statement} < 0; # builtin
 
@@ -498,15 +538,16 @@ sub find_uninitialized_locals($$)
        use constant WATCHME_W => 2;
        use constant WATCHME_X => 4;
        use constant WATCHME_T => 8;
-       my %watchme = map { $_ => WATCHME_X } ($p .. ($func->{parm_start} + $func->{locals} - 1));
+       my %watchme = map { $_ => WATCHME_X } ($func->{parm_start} .. ($func->{parm_start} + $func->{locals} - 1));
 
-       for($progs->{temps})
+       for(keys %{$progs->{temps}})
        {
                $watchme{$_} = WATCHME_T | WATCHME_X
                        if not exists $watchme{$_};
        }
 
-       run_nfa $progs, $func->{first_statement}, "", id, id,
+       my %write_places = ();
+       run_nfa $progs, $func->{first_statement}, "", id, nfa_default_state_checker,
                sub
                {
                        my ($ip, $state, $s, $c) = @_;
@@ -530,14 +571,21 @@ sub find_uninitialized_locals($$)
                                elsif($type eq 'outglobal')
                                {
                                        $watchme{$ofs} |= WATCHME_W;
+                                       $write_places{$ip}{$_} = [$ofs]
+                                               if $watchme{$ofs} & WATCHME_X;
                                }
                                elsif($type eq 'outglobalvec')
                                {
                                        $watchme{$ofs} |= WATCHME_W;
                                        $watchme{$ofs+1} |= WATCHME_W;
                                        $watchme{$ofs+2} |= WATCHME_W;
+                                       my @l = grep { $watchme{$_} & WATCHME_X } $ofs .. ($ofs+2);
+                                       $write_places{$ip}{$_} = \@l
+                                               if @l;
                                }
                        }
+
+                       return 0;
                };
 
        for(keys %watchme)
@@ -551,22 +599,90 @@ sub find_uninitialized_locals($$)
 
        for(keys %watchme)
        {
-               $watchme{$_} = { flags => $watchme{$_}, valid => 0 };
+               $watchme{$_} = {
+                       flags => $watchme{$_},
+                       valid => [0, undef, undef]
+               };
+       }
+
+       # mark parameters as initialized
+       for($func->{parm_start} .. ($p-1))
+       {
+               $watchme{$_}{valid} = [1, undef, undef]
+                       if defined $watchme{$_};
+       }
+       # an initial run of STORE instruction is for receiving extra parameters
+       # (beyond 8). Only possible if the function is declared as having 8 params.
+       # Extra parameters behave otherwise like temps, but are initialized at
+       # startup.
+       for($func->{first_statement} .. (@{$progs->{statements}}-1))
+       {
+               my $s = $progs->{statements}[$_];
+               if($s->{op} eq 'STORE_V')
+               {
+                       $watchme{$s->{a}}{valid} = [1, undef, undef]
+                               if defined $watchme{$s->{a}};
+                       $watchme{$s->{a}+1}{valid} = [1, undef, undef]
+                               if defined $watchme{$s->{a}+1};
+                       $watchme{$s->{a}+2}{valid} = [1, undef, undef]
+                               if defined $watchme{$s->{a}+2};
+               }
+               elsif($s->{op} =~ /^STORE_/)
+               {
+                       $watchme{$s->{a}}{valid} = [1, undef, undef]
+                               if defined $watchme{$s->{a}};
+               }
+               else
+               {
+                       last;
+               }
        }
 
        my %warned = ();
+       my %ip_seen = ();
        run_nfa $progs, $func->{first_statement}, \%watchme,
                sub {
                        my ($h) = @_;
                        return { map { $_ => { %{$h->{$_}} } } keys %$h };
                },
                sub {
-                       my ($h) = @_;
-                       return join ' ', map { $h->{$_}->{valid}; } sort keys %$h;
+                       my ($ip, $state) = @_;
+
+                       my $s = $ip_seen{$ip};
+                       if($s)
+                       {
+                               # if $state is stronger or equal to $s, return 1
+
+                               for(keys %$state)
+                               {
+                                       if($state->{$_}{valid}[0] < $s->{$_})
+                                       {
+                                               # The current state is LESS valid than the previously run one. We NEED to run this.
+                                               # The saved state can safely become the intersection [citation needed].
+                                               for(keys %$state)
+                                               {
+                                                       $s->{$_} = $state->{$_}{valid}[0]
+                                                               if $state->{$_}{valid}[0] < $s->{$_};
+                                               }
+                                               return 0;
+                                       }
+                               }
+                               # if we get here, $state is stronger or equal. No need to try it.
+                               return 1;
+                       }
+                       else
+                       {
+                               # Never seen this IP yet.
+                               $ip_seen{$ip} = { map { ($_ => $state->{$_}{valid}[0]); } keys %$state };
+                               return 0;
+                       }
                },
                sub {
                        my ($ip, $state, $s, $c) = @_;
                        my $op = $s->{op};
+
+                       my $return_hack = $c->{isreturn} // 0;
+
                        for(qw(a b c))
                        {
                                my $type = $c->{$_};
@@ -578,42 +694,52 @@ sub find_uninitialized_locals($$)
                                my $read = sub
                                {
                                        my ($ofs) = @_;
+                                       ++$return_hack
+                                               if $return_hack;
                                        return
                                                if not exists $state->{$ofs};
                                        my $valid = $state->{$ofs}{valid};
-                                       if($valid == 0)
+                                       if($valid->[0] == 0)
+                                       {
+                                               if($return_hack <= 2 and ($op ne 'OR' && $op ne 'AND' || $_ ne 'b')) # fteqcc logicops cause this
+                                               {
+                                                       print "; Use of uninitialized value $ofs in $func->{debugname} at $ip.$_\n";
+                                                       ++$warned{$ip}{$_};
+                                               }
+                                       }
+                                       elsif($valid->[0] < 0)
                                        {
-                                               print "; Use of uninitialized value $ofs in $func->{debugname} at $ip.$_\n";
-                                               ++$warned{$ip}{$_};
+                                               if($return_hack <= 2 and ($op ne 'OR' && $op ne 'AND' || $_ ne 'b')) # fteqcc logicops cause this
+                                               {
+                                                       print "; Use of temporary $ofs across CALL in $func->{debugname} at $ip.$_\n";
+                                                       ++$warned{$ip}{$_};
+                                               }
                                        }
-                                       elsif($valid < 0)
+                                       else
                                        {
-                                               print "; Use of temporary $ofs across CALL in $func->{debugname} at $ip.$_\n";
-                                               ++$warned{$ip}{$_};
+                                               # it's VALID
+                                               if(defined $valid->[1])
+                                               {
+                                                       delete $write_places{$valid->[1]}{$valid->[2]};
+                                               }
                                        }
                                };
                                my $write = sub
                                {
                                        my ($ofs) = @_;
-                                       $state->{$ofs}{valid} = 1
+                                       $state->{$ofs}{valid} = [1, $ip, $_]
                                                if exists $state->{$ofs};
                                };
 
                                if($type eq 'inglobal' || $type eq 'inglobalfunc')
                                {
-                                       if($op ne 'OR' && $op ne 'AND') # fteqcc logicops cause this
-                                       {
-                                               $read->($ofs);
-                                       }
+                                       $read->($ofs);
                                }
                                elsif($type eq 'inglobalvec')
                                {
-                                       if($op ne 'OR' && $op ne 'AND') # fteqcc logicops cause this
-                                       {
-                                               $read->($ofs);
-                                               $read->($ofs+1);
-                                               $read->($ofs+2);
-                                       }
+                                       $read->($ofs);
+                                       $read->($ofs+1);
+                                       $read->($ofs+2);
                                }
                                elsif($type eq 'outglobal')
                                {
@@ -628,50 +754,136 @@ sub find_uninitialized_locals($$)
                        }
                        if($c->{iscall})
                        {
-                               # invalidate temps
-                               for(values %$state)
+                               # builtin calls may clobber stuff
+                               my $func = $s->{a};
+                               my $funcid = $progs->{globals}[$func]{v}{int};
+                               my $funcobj = $progs->{functions}[$funcid];
+                               if(!$funcobj || $funcobj->{first_statement} >= 0)
                                {
-                                       if($_->{flags} & WATCHME_T)
+                                       # invalidate temps
+                                       for(values %$state)
                                        {
-                                               $_->{valid} = -1;
+                                               if($_->{flags} & WATCHME_T)
+                                               {
+                                                       $_->{valid} = [-1, undef, undef];
+                                               }
                                        }
                                }
+                               else # builtin
+                               {
+                                       my $def = $progs->{globaldef_byoffset}->($func);
+                                       return 1
+                                               if $def->{debugname} eq 'error';
+                               }
                        }
+
+                       return 0;
                };
+
+       for my $ip(keys %write_places)
+       {
+               for my $operand(keys %{$write_places{$ip}})
+               {
+                       # TODO verify it
+                       my %left = map { $_ => 1 } @{$write_places{$ip}{$operand}};
+                       my $isread = 0;
+
+                       my %writeplace_seen = ();
+                       run_nfa $progs, $ip+1, \%left,
+                               sub
+                               {
+                                       return { %{$_[0]} };
+                               },
+                               sub
+                               {
+                                       my ($ip, $state) = @_;
+                                       return $writeplace_seen{"$ip " . join " ", sort keys %$state}++;
+                               },
+                               sub
+                               {
+                                       my ($ip, $state, $s, $c) = @_;
+                                       for(qw(a b c))
+                                       {
+                                               my $type = $c->{$_};
+                                               next
+                                                       unless defined $type;
+
+                                               my $ofs = $s->{$_};
+                                               if($type eq 'inglobal' || $type eq 'inglobalfunc')
+                                               {
+                                                       if($state->{$ofs})
+                                                       {
+                                                               $isread = 1;
+                                                               return -1; # exit TOTALLY
+                                                       }
+                                               }
+                                               elsif($type eq 'inglobalvec')
+                                               {
+                                                       if($state->{$ofs} || $state->{$ofs+1} || $state->{$ofs+2})
+                                                       {
+                                                               $isread = 1;
+                                                               return -1; # exit TOTALLY
+                                                       }
+                                               }
+                                               elsif($type eq 'outglobal')
+                                               {
+                                                       delete $state->{$ofs};
+                                                       return 1
+                                                               if !%$state;
+                                               }
+                                               elsif($type eq 'outglobalvec')
+                                               {
+                                                       delete $state->{$ofs};
+                                                       delete $state->{$ofs+1};
+                                                       delete $state->{$ofs+2};
+                                                       return 1
+                                                               if !%$state;
+                                               }
+                                       }
+                                       return 0;
+                               };
+
+                       if(!$isread)
+                       {
+                               print "; Value is never used in $func->{debugname} at $ip.$operand\n";
+                               ++$warned{$ip}{$operand};
+                       }
+               }
+       }
        
        disassemble_function($progs, $func, \%warned)
                if keys %warned;
 }
 
 use constant DEFAULTGLOBALS => [
-       "<OFS_NULL>",
-       "<OFS_RETURN>",
-       "<OFS_RETURN>[1]",
-       "<OFS_RETURN>[2]",
-       "<OFS_PARM0>",
-       "<OFS_PARM0>[1]",
-       "<OFS_PARM0>[2]",
-       "<OFS_PARM1>",
-       "<OFS_PARM1>[1]",
-       "<OFS_PARM1>[2]",
-       "<OFS_PARM2>",
-       "<OFS_PARM2>[1]",
-       "<OFS_PARM2>[2]",
-       "<OFS_PARM3>",
-       "<OFS_PARM3>[1]",
-       "<OFS_PARM3>[2]",
-       "<OFS_PARM4>",
-       "<OFS_PARM4>[1]",
-       "<OFS_PARM4>[2]",
-       "<OFS_PARM5>",
-       "<OFS_PARM5>[1]",
-       "<OFS_PARM5>[2]",
-       "<OFS_PARM6>",
-       "<OFS_PARM6>[1]",
-       "<OFS_PARM6>[2]",
-       "<OFS_PARM7>",
-       "<OFS_PARM7>[1]",
-       "<OFS_PARM7>[2]"
+       "OFS_NULL",
+       "OFS_RETURN",
+       "OFS_RETURN[1]",
+       "OFS_RETURN[2]",
+       "OFS_PARM0",
+       "OFS_PARM0[1]",
+       "OFS_PARM0[2]",
+       "OFS_PARM1",
+       "OFS_PARM1[1]",
+       "OFS_PARM1[2]",
+       "OFS_PARM2",
+       "OFS_PARM2[1]",
+       "OFS_PARM2[2]",
+       "OFS_PARM3",
+       "OFS_PARM3[1]",
+       "OFS_PARM3[2]",
+       "OFS_PARM4",
+       "OFS_PARM4[1]",
+       "OFS_PARM4[2]",
+       "OFS_PARM5",
+       "OFS_PARM5[1]",
+       "OFS_PARM5[2]",
+       "OFS_PARM6",
+       "OFS_PARM6[1]",
+       "OFS_PARM6[2]",
+       "OFS_PARM7",
+       "OFS_PARM7[1]",
+       "OFS_PARM7[2]"
 ];
 
 sub defaultglobal($)
@@ -721,23 +933,36 @@ sub parse_progs($)
        my %offsets_saved = ();
        for(@{$p{globaldefs}})
        {
+               my $type = $_->{type};
+               my $name = $p{getstring}->($_->{s_name});
                next
-                       unless $_->{type}{save};
-               next
-                       unless $p{getstring}->($_->{s_name}) eq "";
+                       unless $type->{save} or $name ne "";
                for my $i(0..(typesize($_->{type}{type})-1))
                {
                        ++$offsets_saved{$_->{ofs}+$i};
                }
        }
+       my %offsets_initialized = ();
+       for(0..(@{$p{globals}}-1))
+       {
+               if($p{globals}[$_]{v}{int})
+               {
+                       ++$offsets_initialized{$_};
+               }
+       }
        my %istemp = ();
+       my %isconst = ();
        for(0..(@{$p{globals}}-1))
        {
                next
-                       if $offsets_saved{$_};
-               $istemp{$_} = 1;
+                       if $_ < @{(DEFAULTGLOBALS)};
+               ++$isconst{$_}
+                       if !$offsets_saved{$_} and $offsets_initialized{$_};
+               ++$istemp{$_}
+                       if !$offsets_saved{$_} and !$offsets_initialized{$_};
        }
-       $p{temps} = [keys %istemp];
+       $p{temps} = \%istemp;
+       $p{consts} = \%isconst;
 
        print STDERR "Naming...\n";
 
@@ -745,12 +970,14 @@ sub parse_progs($)
        my @globaldefs = ();
        for(@{$p{globaldefs}})
        {
-               $_->{debugname} = $p{getstring}->($_->{s_name});
+               my $s = $p{getstring}->($_->{s_name});
+               $_->{debugname} //= "_$s"
+                       if length $s;
        }
        for(@{$p{globaldefs}})
        {
                $globaldefs[$_->{ofs}] //= $_
-                       if $_->{debugname} ne "";
+                       if defined $_->{debugname};
        }
        for(@{$p{globaldefs}})
        {
@@ -758,29 +985,49 @@ sub parse_progs($)
        }
        for(0..(@{$p{globals}}-1))
        {
-               $globaldefs[$_] //= { ofs => $_, s_name => undef, debugname => ($istemp{$_} ? "<temp>" : "<nodef>") . "\@$_" }, 
+               $globaldefs[$_] //= {
+                       ofs => $_,
+                       s_name => undef,
+                       debugname => undef
+               };
        }
        my %globaldefs = ();
-       for(@{$p{globaldefs}})
+       for(@globaldefs)
        {
-               $_->{debugname} = "<anon>\@$_->{ofs}"
-                       if $_->{debugname} eq "";
+               if(!defined $_->{debugname})
+               {
+                       if($istemp{$_->{ofs}})
+                       {
+                               $_->{debugname} = "temp_$_->{ofs}";
+                       }
+                       elsif($isconst{$_->{ofs}})
+                       {
+                               $_->{debugname} = "(" . get_constant(\%p, $p{globals}[$_->{ofs}]{v}) . ")";
+                       }
+                       else
+                       {
+                               $_->{debugname} = "global_$_->{ofs}";
+                       }
+               }
                ++$globaldefs{$_->{debugname}};
        }
-       for(@{$p{globaldefs}})
+       for(@globaldefs)
        {
                next
                        if $globaldefs{$_->{debugname}} <= 1;
+               print "Not unique: $_->{debugname} at $_->{ofs}\n";
                $_->{debugname} .= "\@$_->{ofs}";
        }
        $p{globaldef_byoffset} = sub
        {
                my ($ofs) = @_;
-               if($ofs < @{(DEFAULTGLOBALS)})
+               $ofs &= 0xFFFF;
+               if($ofs >= 0 && $ofs < @{(DEFAULTGLOBALS)})
                {
                        return { ofs => $ofs, s_name => undef, debugname => DEFAULTGLOBALS->[$ofs], type => undef };
                }
                my $def = $globaldefs[$ofs];
+               return $def;
        };
 
        # functions
@@ -802,6 +1049,7 @@ sub parse_progs($)
 
        # what do we want to do?
        my $checkfunc = \&find_uninitialized_locals;
+       #my $checkfunc = \&disassemble_function;
        for(sort { $a->{debugname} cmp $b->{debugname} } @{$p{functions}})
        {
                $checkfunc->(\%p, $_);