From a1aac4278696d303598679561cc8f399e1610cb7 Mon Sep 17 00:00:00 2001
From: Slava Pestov <slava@slava-pestovs-macbook-pro.local>
Date: Tue, 27 Oct 2009 21:31:28 -0500
Subject: [PATCH] vm: clean up gc events, remove -verbosegc switch, fix
 compaction bug

---
 basis/tools/memory/memory.factor | 74 +++++++++++---------------------
 basis/vm/vm.factor               | 42 ++++++++----------
 vm/byte_arrays.hpp               |  6 +--
 vm/code_heap.cpp                 |  8 +++-
 vm/compaction.cpp                |  8 ++--
 vm/data_heap.cpp                 |  8 +++-
 vm/factor.cpp                    |  4 --
 vm/free_list.cpp                 |  6 +--
 vm/free_list_allocator.hpp       | 15 +++++++
 vm/gc.cpp                        | 48 +++------------------
 vm/gc.hpp                        | 16 ++-----
 vm/image.hpp                     |  1 -
 vm/mark_bits.hpp                 | 14 ++++++
 vm/master.hpp                    |  2 +-
 vm/tenured_space.hpp             | 18 +-------
 vm/vm.hpp                        |  5 +--
 16 files changed, 110 insertions(+), 165 deletions(-)

diff --git a/basis/tools/memory/memory.factor b/basis/tools/memory/memory.factor
index 446f047150..6d0c4b7c86 100644
--- a/basis/tools/memory/memory.factor
+++ b/basis/tools/memory/memory.factor
@@ -10,10 +10,15 @@ IN: tools.memory
 
 <PRIVATE
 
+: commas ( n -- str )
+    number>string
+    reverse 3 group "," join reverse ;
+
 : kilobytes ( n -- str )
-    1024 /i number>string
-    dup length 4 > [ 3 cut* "," glue ] when
-    " KB" append ;
+    1024 /i commas " KB" append ;
+
+: micros>string ( n -- str )
+    commas " microseconds" append ;
 
 : fancy-table. ( obj alist -- )
     [ [ nip first ] [ second call( obj -- str ) ] 2bi 2array ] with map
@@ -97,50 +102,23 @@ PRIVATE>
     enable-gc-events [ ] [ disable-gc-events drop ] cleanup
     disable-gc-events byte-array>gc-event-array ; inline
 
-: generation-sizes-before ( events -- sizes )
-    [
-        {
-            [ start-time>> ]
-            [ nursery-size-before>> ]
-            [ aging-size-before>> ]
-            [ tenured-size-before>> ]
-            [ code-size-before>> ]
-        } cleave 5 narray
-    ] { } map-as ;
+: gc-op-string ( op -- string )
+    {
+        { collect-nursery-op      [ "copying from nursery" ] }
+        { collect-aging-op        [ "copying from aging"   ] }
+        { collect-to-tenured-op   [ "copying to tenured"   ] }
+        { collect-full-op         [ "mark and sweep"       ] }
+        { collect-compact-op      [ "mark and compact"     ] }
+        { collect-growing-heap-op [ "grow heap"            ] }
+    } case ;
 
-: generation-sizes-after ( events -- sizes )
-    [
-        {
-            [ start-time>> ]
-            [ nursery-size-after>> ]
-            [ aging-size-after>> ]
-            [ tenured-size-after>> ]
-            [ code-size-after>> ]
-        } cleave 5 narray
-    ] { } map-as ;
+: space-reclaimed ( event -- bytes )
+    [ data-heap-before>> ] [ data-heap-after>> ] bi
+    [ [ nursery>> ] [ aging>> ] [ tenured>> ] tri [ occupied>> ] tri@ + + ] bi@ - ;
 
-: reclaimed-space ( events -- sizes )
-    [
-        [ start-time>> ] [
-            {
-                [ [ nursery-size-before>> ] [ nursery-size-after>> ] bi - ]
-                [ [   aging-size-before>> ] [   aging-size-after>> ] bi - ]
-                [ [ tenured-size-before>> ] [ tenured-size-after>> ] bi - ]
-                [ [    code-size-before>> ] [    code-size-after>> ] bi - ]
-            } cleave
-            + + +
-        ] bi 2array
-    ] { } map-as ;
-
-: allocated-space ( events -- sizes )
-    2 <sliced-clumps> [
-        [ second start-time>> ] [
-            {
-                [ [ second nursery-size-before>> ] [ first nursery-size-after>> ] bi - ]
-                [ [ second   aging-size-before>> ] [ first   aging-size-after>> ] bi - ]
-                [ [ second tenured-size-before>> ] [ first tenured-size-after>> ] bi - ]
-                [ [ second    code-size-before>> ] [ first    code-size-after>> ] bi - ]
-            } cleave
-            + + +
-        ] bi 2array
-    ] { } map-as ;
+: gc-event. ( event -- )
+    {
+        { "Event type:" [ op>> gc-op-string ] }
+        { "Total time:" [ total-time>> micros>string ] }
+        { "Space reclaimed:" [ space-reclaimed kilobytes ] }
+    } fancy-table. ;
diff --git a/basis/vm/vm.factor b/basis/vm/vm.factor
index 1351ed10a3..6bcdd2862a 100644
--- a/basis/vm/vm.factor
+++ b/basis/vm/vm.factor
@@ -29,31 +29,6 @@ collect-full-op
 collect-compact-op
 collect-growing-heap-op ;
 
-STRUCT: gc-event
-{ op uint }
-{ nursery-size-before cell }
-{ aging-size-before cell }
-{ tenured-size-before cell }
-{ tenured-free-block-count-before cell }
-{ code-size-before cell }
-{ code-free-block-count-before cell }
-{ nursery-size-after cell }
-{ aging-size-after cell }
-{ tenured-size-after cell }
-{ tenured-free-block-count-after cell }
-{ code-size-after cell }
-{ code-free-block-count-after cell }
-{ cards-scanned cell }
-{ decks-scanned cell }
-{ code-blocks-scanned cell }
-{ start-time ulonglong }
-{ total-time cell }
-{ card-scan-time cell }
-{ code-scan-time cell }
-{ data-sweep-time cell }
-{ code-sweep-time cell }
-{ compaction-time cell } ;
-
 STRUCT: copying-sizes
 { size cell }
 { occupied cell }
@@ -73,3 +48,20 @@ STRUCT: data-heap-room
 { cards cell }
 { decks cell }
 { mark-stack cell } ;
+
+STRUCT: gc-event
+{ op uint }
+{ data-heap-before data-heap-room }
+{ code-heap-before mark-sweep-sizes }
+{ data-heap-after data-heap-room }
+{ code-heap-after mark-sweep-sizes }
+{ cards-scanned cell }
+{ decks-scanned cell }
+{ code-blocks-scanned cell }
+{ start-time ulonglong }
+{ total-time cell }
+{ card-scan-time cell }
+{ code-scan-time cell }
+{ data-sweep-time cell }
+{ code-sweep-time cell }
+{ compaction-time cell } ;
diff --git a/vm/byte_arrays.hpp b/vm/byte_arrays.hpp
index 93cb775b1e..0817f7dcdc 100755
--- a/vm/byte_arrays.hpp
+++ b/vm/byte_arrays.hpp
@@ -13,14 +13,14 @@ struct growable_byte_array {
 	void trim();
 };
 
-template<typename T> byte_array *factor_vm::byte_array_from_value(T *value)
+template<typename Type> byte_array *factor_vm::byte_array_from_value(Type *value)
 {
 	return byte_array_from_values(value,1);
 }
 
-template<typename T> byte_array *factor_vm::byte_array_from_values(T *values, cell len)
+template<typename Type> byte_array *factor_vm::byte_array_from_values(Type *values, cell len)
 {
-	cell size = sizeof(T) * len;
+	cell size = sizeof(Type) * len;
 	byte_array *data = allot_uninitialized_array<byte_array>(size);
 	memcpy(data->data<char>(),values,size);
 	return data;
diff --git a/vm/code_heap.cpp b/vm/code_heap.cpp
index ac6669ca49..77ae5180c8 100755
--- a/vm/code_heap.cpp
+++ b/vm/code_heap.cpp
@@ -196,7 +196,7 @@ void factor_vm::primitive_modify_code_heap()
 	update_code_heap_words();
 }
 
-void factor_vm::primitive_code_room()
+code_heap_room factor_vm::code_room()
 {
 	code_heap_room room;
 
@@ -206,6 +206,12 @@ void factor_vm::primitive_code_room()
 	room.contiguous_free  = code->allocator->free_blocks.largest_free_block();
 	room.free_block_count = code->allocator->free_blocks.free_block_count;
 
+	return room;
+}
+
+void factor_vm::primitive_code_room()
+{
+	code_heap_room room = code_room();
 	dpush(tag<byte_array>(byte_array_from_value(&room)));
 }
 
diff --git a/vm/compaction.cpp b/vm/compaction.cpp
index 4540560ac4..bda8035858 100644
--- a/vm/compaction.cpp
+++ b/vm/compaction.cpp
@@ -55,10 +55,12 @@ struct compaction_sizer {
 
 	cell operator()(object *obj)
 	{
-		if(obj->free_p() || obj->h.hi_tag() != TUPLE_TYPE)
-			return obj->size();
-		else
+		if(!forwarding_map->marked_p(obj))
+			return forwarding_map->unmarked_space_starting_at(obj);
+		else if(obj->h.hi_tag() == TUPLE_TYPE)
 			return align(tuple_size_with_forwarding(forwarding_map,obj),data_alignment);
+		else
+			return obj->size();
 	}
 };
 
diff --git a/vm/data_heap.cpp b/vm/data_heap.cpp
index 8f8db699f6..b38674fe1f 100755
--- a/vm/data_heap.cpp
+++ b/vm/data_heap.cpp
@@ -201,7 +201,7 @@ void factor_vm::primitive_size()
 	box_unsigned_cell(object_size(dpop()));
 }
 
-void factor_vm::primitive_data_room()
+data_heap_room factor_vm::data_room()
 {
 	data_heap_room room;
 
@@ -220,6 +220,12 @@ void factor_vm::primitive_data_room()
 	room.decks                    = data->decks_end - data->decks;
 	room.mark_stack               = data->tenured->mark_stack.capacity();
 
+	return room;
+}
+
+void factor_vm::primitive_data_room()
+{
+	data_heap_room room = data_room();
 	dpush(tag<byte_array>(byte_array_from_value(&room)));
 }
 
diff --git a/vm/factor.cpp b/vm/factor.cpp
index 6a235ec88d..510900a036 100755
--- a/vm/factor.cpp
+++ b/vm/factor.cpp
@@ -38,7 +38,6 @@ void factor_vm::default_parameters(vm_parameters *p)
 	p->max_pic_size = 3;
 
 	p->fep = false;
-	p->verbose_gc = false;
 	p->signals = true;
 
 #ifdef WINDOWS
@@ -87,7 +86,6 @@ void factor_vm::init_parameters_from_args(vm_parameters *p, int argc, vm_char **
 		else if(factor_arg(arg,STRING_LITERAL("-callbacks=%d"),&p->callback_size));
 		else if(STRCMP(arg,STRING_LITERAL("-fep")) == 0) p->fep = true;
 		else if(STRCMP(arg,STRING_LITERAL("-nosignals")) == 0) p->signals = false;
-		else if(STRCMP(arg,STRING_LITERAL("-verbosegc")) == 0) p->verbose_gc = true;
 		else if(STRNCMP(arg,STRING_LITERAL("-i="),3) == 0) p->image_path = arg + 3;
 		else if(STRCMP(arg,STRING_LITERAL("-console")) == 0) p->console = true;
 	}
@@ -142,8 +140,6 @@ void factor_vm::init_factor(vm_parameters *p)
 	if(p->signals)
 		init_signals();
 
-	verbose_gc = p->verbose_gc;
-
 	if(p->console)
 		open_console();
 
diff --git a/vm/free_list.cpp b/vm/free_list.cpp
index 24ffab9f2c..644f62d870 100644
--- a/vm/free_list.cpp
+++ b/vm/free_list.cpp
@@ -108,10 +108,8 @@ cell free_list::largest_free_block()
 {
 	if(large_blocks.size())
 	{
-		large_block_set::iterator end = large_blocks.end();
-		free_heap_block *block = *end;
-		large_blocks.erase(end);
-		return block->size();
+		large_block_set::reverse_iterator last = large_blocks.rbegin();
+		return (*last)->size();
 	}
 	else
 	{
diff --git a/vm/free_list_allocator.hpp b/vm/free_list_allocator.hpp
index d74dd5be9a..dad54b644d 100644
--- a/vm/free_list_allocator.hpp
+++ b/vm/free_list_allocator.hpp
@@ -14,6 +14,7 @@ template<typename Block> struct free_list_allocator {
 	Block *first_block();
 	Block *last_block();
 	Block *next_block_after(Block *block);
+	Block *next_allocated_block_after(Block *block);
 	bool can_allot_p(cell size);
 	Block *allot(cell size);
 	void free(Block *block);
@@ -58,6 +59,20 @@ template<typename Block> Block *free_list_allocator<Block>::next_block_after(Blo
 	return (Block *)((cell)block + block->size());
 }
 
+template<typename Block> Block *free_list_allocator<Block>::next_allocated_block_after(Block *block)
+{
+	while(block != this->last_block() && block->free_p())
+	{
+		free_heap_block *free_block = (free_heap_block *)block;
+		block = (object *)((cell)free_block + free_block->size());
+	}
+
+	if(block == this->last_block())
+		return NULL;
+	else
+		return block;
+}
+
 template<typename Block> bool free_list_allocator<Block>::can_allot_p(cell size)
 {
 	return free_blocks.can_allot_p(size);
diff --git a/vm/gc.cpp b/vm/gc.cpp
index 0c8750ad92..cce8298921 100755
--- a/vm/gc.cpp
+++ b/vm/gc.cpp
@@ -15,12 +15,8 @@ gc_event::gc_event(gc_op op_, factor_vm *parent) :
 	code_sweep_time(0),
 	compaction_time(0)
 {
-	nursery_size_before = parent->nursery.occupied_space();
-	aging_size_before = parent->data->aging->occupied_space();
-	tenured_size_before = parent->data->tenured->occupied_space();
-	tenured_free_block_count_before = parent->data->tenured->free_blocks.free_block_count;
-	code_size_before = parent->code->allocator->occupied_space();
-	code_free_block_count_before = parent->code->allocator->free_blocks.free_block_count;
+	data_heap_before = parent->data_room();
+	code_heap_before = parent->code_room();
 	start_time = current_micros();
 }
 
@@ -79,44 +75,11 @@ void gc_event::ended_compaction()
 
 void gc_event::ended_gc(factor_vm *parent)
 {
-	nursery_size_after = parent->nursery.occupied_space();
-	aging_size_after = parent->data->aging->occupied_space();
-	tenured_size_after = parent->data->tenured->occupied_space();
-	tenured_free_block_count_after = parent->data->tenured->free_blocks.free_block_count;
-	code_size_after = parent->code->allocator->occupied_space();
-	code_free_block_count_after = parent->code->allocator->free_blocks.free_block_count;
+	data_heap_after = parent->data_room();
+	code_heap_after = parent->code_room();
 	total_time = current_micros() - start_time;
 }
 
-std::ostream &operator<<(std::ostream &out, const gc_event *event)
-{
-	out << "<event\n"
-	    << " op                              = '" << event->op                              << "'\n"
-	    << " nursery_size_before             = '" << event->nursery_size_before             << "'\n"
-	    << " aging_size_before               = '" << event->aging_size_before               << "'\n"
-	    << " tenured_size_before             = '" << event->tenured_size_before             << "'\n"
-	    << " tenured_free_block_count_before = '" << event->tenured_free_block_count_before << "'\n"
-	    << " code_size_before                = '" << event->code_size_before                << "'\n"
-	    << " code_free_block_count_before    = '" << event->code_free_block_count_before    << "'\n"
-	    << " nursery_size_after              = '" << event->nursery_size_after              << "'\n"
-	    << " aging_size_after                = '" << event->aging_size_after                << "'\n"
-	    << " tenured_size_after              = '" << event->tenured_size_after              << "'\n"
-	    << " tenured_free_block_count_after  = '" << event->tenured_free_block_count_after  << "'\n"
-	    << " code_size_after                 = '" << event->code_size_after                 << "'\n"
-	    << " code_free_block_count_after     = '" << event->code_free_block_count_after     << "'\n"
-	    << " cards_scanned                   = '" << event->cards_scanned                   << "'\n"
-	    << " decks_scanned                   = '" << event->decks_scanned                   << "'\n"
-	    << " code_blocks_scanned             = '" << event->code_blocks_scanned             << "'\n"
-	    << " start_time                      = '" << event->start_time                      << "'\n"
-	    << " total_time                      = '" << event->total_time                      << "'\n"
-	    << " card_scan_time                  = '" << event->card_scan_time                  << "'\n"
-	    << " code_scan_time                  = '" << event->code_scan_time                  << "'\n"
-	    << " data_sweep_time                 = '" << event->data_sweep_time                 << "'\n"
-	    << " code_sweep_time                 = '" << event->code_sweep_time                 << "'\n"
-	    << " compaction_time                 = '" << event->compaction_time                 << "' />";
-	return out;
-}
-
 gc_state::gc_state(gc_op op_, factor_vm *parent) : op(op_), start_time(current_micros())
 {
 	event = new gc_event(op,parent);
@@ -131,7 +94,6 @@ gc_state::~gc_state()
 void factor_vm::end_gc()
 {
 	current_gc->event->ended_gc(this);
-	if(verbose_gc) std::cout << current_gc->event << std::endl;
 	if(gc_events) gc_events->push_back(*current_gc->event);
 	delete current_gc->event;
 	current_gc->event = NULL;
@@ -354,7 +316,7 @@ void factor_vm::primitive_disable_gc_events()
 {
 	if(gc_events)
 	{
-		byte_array *data = byte_array_from_values(&gc_events->front(),gc_events->size());
+		byte_array *data = byte_array_from_values(&gc_events->first(),gc_events->size());
 		dpush(tag<byte_array>(data));
 
 		delete gc_events;
diff --git a/vm/gc.hpp b/vm/gc.hpp
index fe19f93ff1..90b5ce6032 100755
--- a/vm/gc.hpp
+++ b/vm/gc.hpp
@@ -12,18 +12,10 @@ enum gc_op {
 
 struct gc_event {
 	gc_op op;
-	cell nursery_size_before;
-	cell aging_size_before;
-	cell tenured_size_before;
-	cell tenured_free_block_count_before;
-	cell code_size_before;
-	cell code_free_block_count_before;
-	cell nursery_size_after;
-	cell aging_size_after;
-	cell tenured_size_after;
-	cell tenured_free_block_count_after;
-	cell code_size_after;
-	cell code_free_block_count_after;
+	data_heap_room data_heap_before;
+	code_heap_room code_heap_before;
+	data_heap_room data_heap_after;
+	code_heap_room code_heap_after;
 	cell cards_scanned;
 	cell decks_scanned;
 	cell code_blocks_scanned;
diff --git a/vm/image.hpp b/vm/image.hpp
index f740b65f68..cca0e9e378 100755
--- a/vm/image.hpp
+++ b/vm/image.hpp
@@ -35,7 +35,6 @@ struct vm_parameters {
 	cell young_size, aging_size, tenured_size;
 	cell code_size;
 	bool fep;
-	bool verbose_gc;
 	bool console;
 	bool signals;
 	cell max_pic_size;
diff --git a/vm/mark_bits.hpp b/vm/mark_bits.hpp
index 8b6b0c75eb..1b8fa913a6 100644
--- a/vm/mark_bits.hpp
+++ b/vm/mark_bits.hpp
@@ -154,6 +154,20 @@ template<typename Block> struct mark_bits {
 #endif
 		return new_block;
 	}
+
+	/* Find the next allocated block without calling size() on unmarked
+	objects. */
+	cell unmarked_space_starting_at(Block *original)
+	{
+		char *start = (char *)original;
+		char *scan = start;
+		char *end = (char *)(this->start + this->size);
+
+		while(scan != end && !marked_p((Block *)scan))
+			scan += block_granularity;
+
+		return scan - start;
+	}
 };
 
 }
diff --git a/vm/master.hpp b/vm/master.hpp
index 0873480add..9b525bd22e 100755
--- a/vm/master.hpp
+++ b/vm/master.hpp
@@ -59,6 +59,7 @@ namespace factor
 #include "aging_space.hpp"
 #include "tenured_space.hpp"
 #include "data_heap.hpp"
+#include "code_heap.hpp"
 #include "gc.hpp"
 #include "debug.hpp"
 #include "strings.hpp"
@@ -68,7 +69,6 @@ namespace factor
 #include "io.hpp"
 #include "image.hpp"
 #include "alien.hpp"
-#include "code_heap.hpp"
 #include "callbacks.hpp"
 #include "vm.hpp"
 #include "tagged.hpp"
diff --git a/vm/tenured_space.hpp b/vm/tenured_space.hpp
index 1b3baeaf52..f77f17a56a 100644
--- a/vm/tenured_space.hpp
+++ b/vm/tenured_space.hpp
@@ -20,30 +20,16 @@ struct tenured_space : free_list_allocator<object> {
 			return NULL;
 	}
 
-	object *first_allocated_block_after(object *block)
-	{
-		while(block != this->last_block() && block->free_p())
-		{
-			free_heap_block *free_block = (free_heap_block *)block;
-			block = (object *)((cell)free_block + free_block->size());
-		}
-
-		if(block == this->last_block())
-			return NULL;
-		else
-			return block;
-	}
-
 	cell first_object()
 	{
-		return (cell)first_allocated_block_after(this->first_block());
+		return (cell)next_allocated_block_after(this->first_block());
 	}
 
 	cell next_object_after(cell scan)
 	{
 		cell size = ((object *)scan)->size();
 		object *next = (object *)(scan + size);
-		return (cell)first_allocated_block_after(next);
+		return (cell)next_allocated_block_after(next);
 	}
 
 	void clear_mark_bits()
diff --git a/vm/vm.hpp b/vm/vm.hpp
index 128982fb9c..0844b47e51 100755
--- a/vm/vm.hpp
+++ b/vm/vm.hpp
@@ -46,9 +46,6 @@ struct factor_vm
 	/* GC is off during heap walking */
 	bool gc_off;
 
-	/* GC logging */
-	bool verbose_gc;
-
 	/* Data heap */
 	data_heap *data;
 
@@ -223,6 +220,7 @@ struct factor_vm
 	void set_data_heap(data_heap *data_);
 	void init_data_heap(cell young_size, cell aging_size, cell tenured_size);
 	void primitive_size();
+	data_heap_room data_room();
 	void primitive_data_room();
 	void begin_scan();
 	void end_scan();
@@ -502,6 +500,7 @@ struct factor_vm
 	void update_code_heap_words_and_literals();
 	void relocate_code_heap();
 	void primitive_modify_code_heap();
+	code_heap_room code_room();
 	void primitive_code_room();
 	void primitive_strip_stack_traces();