X-Git-Url: http://git.shadowcat.co.uk/gitweb/gitweb.cgi?a=blobdiff_plain;f=t%2Flib%2FDBICTest%2FUtil%2FLeakTracer.pm;h=58790e4332e554294e1df777d7a00ee0808b03f9;hb=de4705b7393350ae7cde8f5409f204747c1a4a4e;hp=d0e63f2197da86af39530594efc1b3450d71ca3e;hpb=cf8fa2867da29e20c8371827a98c6e49efa744e3;p=dbsrgits%2FDBIx-Class.git

diff --git a/t/lib/DBICTest/Util/LeakTracer.pm b/t/lib/DBICTest/Util/LeakTracer.pm
index d0e63f2..58790e4 100644
--- a/t/lib/DBICTest/Util/LeakTracer.pm
+++ b/t/lib/DBICTest/Util/LeakTracer.pm
@@ -4,107 +4,290 @@ use warnings;
 use strict;
 
 use Carp;
-use Scalar::Util qw/isweak weaken blessed reftype refaddr/;
+use Scalar::Util qw(isweak weaken blessed reftype);
+use DBIx::Class::_Util qw(refcount hrefaddr);
+use DBIx::Class::Optional::Dependencies;
+use Data::Dumper::Concise;
 use DBICTest::Util 'stacktrace';
+use constant {
+  CV_TRACING => DBIx::Class::Optional::Dependencies->req_ok_for ('test_leaks_heavy'),
+  SKIP_SCALAR_REFS => ( $] > 5.017 ) ? 1 : 0,
+};
 
 use base 'Exporter';
-our @EXPORT_OK = qw/populate_weakregistry assert_empty_weakregistry/;
+our @EXPORT_OK = qw(populate_weakregistry assert_empty_weakregistry visit_refs);
 
 my $refs_traced = 0;
-my $leaks_found;
+my $leaks_found = 0;
 my %reg_of_regs;
 
+# so we don't trigger stringification
+sub _describe_ref {
+  sprintf '%s%s(%s)',
+    (defined blessed $_[0]) ? blessed($_[0]) . '=' : '',
+    reftype $_[0],
+    hrefaddr $_[0],
+  ;
+}
+
 sub populate_weakregistry {
-  my ($weak_registry, $target, $slot) = @_;
+  my ($weak_registry, $target, $note) = @_;
 
   croak 'Expecting a registry hashref' unless ref $weak_registry eq 'HASH';
   croak 'Target is not a reference' unless length ref $target;
 
-  $slot ||= (sprintf '%s%s(0x%x)', # so we don't trigger stringification
-    (defined blessed $target) ? blessed($target) . '=' : '',
-    reftype $target,
-    refaddr $target,
-  );
+  my $refaddr = hrefaddr $target;
 
-  if (defined $weak_registry->{$slot}{weakref}) {
-    if ( refaddr($weak_registry->{$slot}{weakref}) != (refaddr $target) ) {
-      print STDERR "Bail out! Weak Registry slot collision: $weak_registry->{$slot}{weakref} / $target\n";
-      exit 255;
-    }
+  # a registry could be fed to itself or another registry via recursive sweeps
+  return $target if $reg_of_regs{$refaddr};
+
+  weaken( $reg_of_regs{ hrefaddr($weak_registry) } = $weak_registry )
+    unless( $reg_of_regs{ hrefaddr($weak_registry) } );
+
+  # an explicit "garbage collection" pass every time we store a ref
+  # if we do not do this the registry will keep growing appearing
+  # as if the traced program is continuously slowly leaking memory
+  for my $reg (values %reg_of_regs) {
+    (defined $reg->{$_}{weakref}) or delete $reg->{$_}
+      for keys %$reg;
   }
-  else {
+
+  # FIXME/INVESTIGATE - something fishy is going on with refs to plain
+  # strings, perhaps something to do with the CoW work etc...
+  return $target if SKIP_SCALAR_REFS and reftype($target) eq 'SCALAR';
+
+  if (! defined $weak_registry->{$refaddr}{weakref}) {
+    $weak_registry->{$refaddr} = {
+      stacktrace => stacktrace(1),
+      weakref => $target,
+    };
+    weaken( $weak_registry->{$refaddr}{weakref} );
     $refs_traced++;
-    weaken( $weak_registry->{$slot}{weakref} = $target );
-    $weak_registry->{$slot}{stacktrace} = stacktrace(1);
-    $weak_registry->{$slot}{renumber} = 1 unless $_[2];
   }
 
-  weaken( $reg_of_regs{ refaddr($weak_registry) } = $weak_registry )
-    unless( $reg_of_regs{ refaddr($weak_registry) } );
+  my $desc = _describe_ref($target);
+  $weak_registry->{$refaddr}{slot_names}{$desc} = 1;
+  if ($note) {
+    $note =~ s/\s*\Q$desc\E\s*//g;
+    $weak_registry->{$refaddr}{slot_names}{$note} = 1;
+  }
 
   $target;
 }
 
-# Renumber everything we auto-named on a thread spawn
+# Regenerate the slots names on a thread spawn
 sub CLONE {
   my @individual_regs = grep { scalar keys %{$_||{}} } values %reg_of_regs;
   %reg_of_regs = ();
 
   for my $reg (@individual_regs) {
-    my @live_slots = grep { defined $reg->{$_}{weakref} } keys %$reg
+    my @live_slots = grep { defined $_->{weakref} } values %$reg
       or next;
 
-    my @live_instances = @{$reg}{@live_slots};
-
     $reg = {};  # get a fresh hashref in the new thread ctx
-    weaken( $reg_of_regs{refaddr($reg)} = $reg );
+    weaken( $reg_of_regs{hrefaddr($reg)} = $reg );
 
-    while (@live_slots) {
-      my $slot = shift @live_slots;
-      my $inst = shift @live_instances;
+    for my $slot_info (@live_slots) {
+      my $new_addr = hrefaddr $slot_info->{weakref};
 
-      $slot =~ s/0x[0-9A-F]+/'0x' . sprintf ('0x%x', refaddr($inst))/ieg
-        if $inst->{renumber};
+      # replace all slot names
+      $slot_info->{slot_names} = { map {
+        my $name = $_;
+        $name =~ s/\(0x[0-9A-F]+\)/sprintf ('(%s)', $new_addr)/ieg;
+        ($name => 1);
+      } keys %{$slot_info->{slot_names}} };
 
-      $reg->{$slot} = $inst;
+      $reg->{$new_addr} = $slot_info;
     }
   }
 }
 
+sub visit_refs {
+  my $args = { (ref $_[0]) ? %{$_[0]} : @_ };
+
+  $args->{seen_refs} ||= {};
+
+  my $visited_cnt = '0E0';
+  for my $i (0 .. $#{$args->{refs}} ) {
+    next if isweak($args->{refs}[$i]);
+
+    my $r = $args->{refs}[$i];
+
+    next unless length ref $r;
+
+    # no diving into weakregistries
+    next if $reg_of_regs{hrefaddr $r};
+
+    next if $args->{seen_refs}{my $dec_addr = Scalar::Util::refaddr($r)}++;
+
+    $visited_cnt++;
+    $args->{action}->($r) or next;
+
+    # This may end up being necessarry some day, but do not slow things
+    # down for now
+    #if ( defined( my $t = tied($r) ) ) {
+    #  $visited_cnt += visit_refs({ %$args, refs => [ $t ] });
+    #}
+
+    local $@;
+    eval {
+      my $type = reftype $r;
+      if ($type eq 'HASH') {
+        $visited_cnt += visit_refs({ %$args, refs => [ map {
+          ( !isweak($r->{$_}) ) ? $r->{$_} : ()
+        } keys %$r ] });
+      }
+      elsif ($type eq 'ARRAY') {
+        $visited_cnt += visit_refs({ %$args, refs => [ map {
+          ( !isweak($r->[$_]) ) ? $r->[$_] : ()
+        } 0..$#$r ] });
+      }
+      elsif ($type eq 'REF' and !isweak($$r)) {
+        $visited_cnt += visit_refs({ %$args, refs => [ $$r ] });
+      }
+      elsif (CV_TRACING and $type eq 'CODE') {
+        $visited_cnt += visit_refs({ %$args, refs => [ map {
+          ( !isweak($_) ) ? $_ : ()
+        } scalar PadWalker::closed_over($r) ] }); # scalar due to RT#92269
+      }
+      1;
+    } or warn "Could not descend into @{[ _describe_ref($r) ]}: $@\n";
+  }
+  $visited_cnt;
+}
+
 sub assert_empty_weakregistry {
   my ($weak_registry, $quiet) = @_;
 
+  # in case we hooked bless any extra object creation will wreak
+  # havoc during the assert phase
+  local *CORE::GLOBAL::bless;
+  *CORE::GLOBAL::bless = sub { CORE::bless( $_[0], (@_ > 1) ? $_[1] : caller() ) };
+
   croak 'Expecting a registry hashref' unless ref $weak_registry eq 'HASH';
 
+  defined $weak_registry->{$_}{weakref} or delete $weak_registry->{$_}
+    for keys %$weak_registry;
+
   return unless keys %$weak_registry;
 
   my $tb = eval { Test::Builder->new }
-    or croak 'Calling test_weakregistry without a loaded Test::Builder makes no sense';
+    or croak "Calling assert_empty_weakregistry in $0 without a loaded Test::Builder makes no sense";
+
+  for my $addr (keys %$weak_registry) {
+    $weak_registry->{$addr}{display_name} = join ' | ', (
+      sort
+        { length $a <=> length $b or $a cmp $b }
+        keys %{$weak_registry->{$addr}{slot_names}}
+    );
 
-  for my $slot (sort keys %$weak_registry) {
-    next if ! defined $weak_registry->{$slot}{weakref};
-    $tb->BAILOUT("!!!! WEAK REGISTRY SLOT $slot IS NOT A WEAKREF !!!!")
-      unless isweak( $weak_registry->{$slot}{weakref} );
+    $tb->BAILOUT("!!!! WEAK REGISTRY SLOT $weak_registry->{$addr}{display_name} IS NOT A WEAKREF !!!!")
+      if defined $weak_registry->{$addr}{weakref} and ! isweak( $weak_registry->{$addr}{weakref} );
   }
 
+  # compile a list of refs stored as globals (possibly even catching
+  # class data in the form of method closures), so we can skip them
+  # further on
+  my ($seen_refs, $classdata_refs) = ({}, undef);
+
+  # the walk is very expensive - if we are $quiet (running in an END block)
+  # we do not really need to be too thorough
+  unless ($quiet) {
+    my ($symwalker, $symcounts);
+    $symwalker = sub {
+      no strict 'refs';
+      my $pkg = shift || '::';
+
+      # any non-weak globals are "clasdata" in all possible sense
+      #
+      # the unless regex at the end skips some dangerous namespaces outright
+      # (but does not prevent descent)
+      $symcounts->{$pkg} += visit_refs (
+        seen_refs => $seen_refs,
+        action => sub { ++$classdata_refs->{hrefaddr $_[0]} },
+        refs => [ map { my $sym = $_;
+          # *{"$pkg$sym"}{CODE} won't simply work - MRO-cached CVs are invisible there
+          ( CV_TRACING ? Class::MethodCache::get_cv("${pkg}$sym") : () ),
+
+          ( defined *{"$pkg$sym"}{SCALAR} and length ref ${"$pkg$sym"} and ! isweak( ${"$pkg$sym"} ) )
+            ? ${"$pkg$sym"} : ()
+          ,
+          ( map {
+            ( defined *{"$pkg$sym"}{$_} and ! isweak(defined *{"$pkg$sym"}{$_}) )
+              ? *{"$pkg$sym"}{$_}
+              : ()
+          } qw(HASH ARRAY IO GLOB) ),
+        } keys %$pkg ],
+      ) unless $pkg =~ /^ :: (?:
+        DB | next | B | .+? ::::ISA (?: ::CACHE ) | Class::C3
+      ) :: $/x;
+
+      $symwalker->("${pkg}$_") for grep { $_ =~ /(?<!^main)::$/ } keys %$pkg;
+    };
+
+    $symwalker->();
 
-  for my $slot (sort keys %$weak_registry) {
-    ! defined $weak_registry->{$slot}{weakref} and next if $quiet;
+#    use Devel::Dwarn;
+#    Ddie [ map
+#      { { $_ => $symcounts->{$_} } }
+#      sort
+#        {$symcounts->{$a} <=> $symcounts->{$b} }
+#        keys %$symcounts
+#    ];
+  }
 
-    $tb->ok (! defined $weak_registry->{$slot}{weakref}, "No leaks of $slot") or do {
-      $leaks_found = 1;
+  delete $weak_registry->{$_} for keys %$classdata_refs;
 
-      my $diag = '';
+  for my $addr (sort { $weak_registry->{$a}{display_name} cmp $weak_registry->{$b}{display_name} } keys %$weak_registry) {
 
-      $diag .= Devel::FindRef::track ($weak_registry->{$slot}{weakref}, 20) . "\n"
-        if ( $ENV{TEST_VERBOSE} && eval { require Devel::FindRef });
+    next if ! defined $weak_registry->{$addr}{weakref};
 
-      if (my $stack = $weak_registry->{$slot}{stacktrace}) {
-        $diag .= "    Reference first seen$stack";
-      }
+    $leaks_found++ unless $tb->in_todo;
+    $tb->ok (0, "Leaked $weak_registry->{$addr}{display_name}");
 
-      $tb->diag($diag) if $diag;
+    my $diag = do {
+      local $Data::Dumper::Maxdepth = 1;
+      sprintf "\n%s (refcnt %d) => %s\n",
+        $weak_registry->{$addr}{display_name},
+        refcount($weak_registry->{$addr}{weakref}),
+        (
+          ref($weak_registry->{$addr}{weakref}) eq 'CODE'
+            and
+          B::svref_2object($weak_registry->{$addr}{weakref})->XSUB
+        ) ? '__XSUB__' : Dumper( $weak_registry->{$addr}{weakref} )
+      ;
     };
+
+    # FIXME - need to add a circular reference seeker based on the visitor
+    # (will need a bunch of modifications, punting with just a stub for now)
+
+    $diag .= Devel::FindRef::track ($weak_registry->{$addr}{weakref}, 50) . "\n"
+      if ( $ENV{TEST_VERBOSE} && eval { require Devel::FindRef });
+
+    $diag =~ s/^/    /mg;
+
+    if (my $stack = $weak_registry->{$addr}{stacktrace}) {
+      $diag .= "    Reference first seen$stack";
+    }
+
+    $tb->diag($diag);
+
+#    if ($leaks_found == 1) {
+#      # using the fh dumper due to intermittent buffering issues
+#      # in case we decide to exit soon after (possibly via _exit)
+#      require Devel::MAT::Dumper;
+#      local $Devel::MAT::Dumper::MAX_STRING = -1;
+#      open( my $fh, '>:raw', "leaked_${addr}_pid$$.pmat" ) or die $!;
+#      Devel::MAT::Dumper::dumpfh( $fh );
+#      close ($fh) or die $!;
+#
+#      use POSIX;
+#      POSIX::_exit(1);
+#    }
+  }
+
+  if (! $quiet and !$leaks_found and ! $tb->in_todo) {
+    $tb->ok(1, sprintf "No leaks found at %s line %d", (caller())[1,2] );
   }
 }