Reland of "[heap] Switch to 500k pages"

Decrease regular heap object size to 400k. In a follow up, we can now get rid of
the new space border page while keeping the 1M minimum new space size.

BUG=chromium:636331

This reverts commit 555c9619.

Review-Url: https://codereview.chromium.org/2232653003
Cr-Commit-Position: refs/heads/master@{#38608}

src/base/build_config.h

@@ -200,7 +200,7 @@
 // Bump up for Power Linux due to larger (64K) page size.
 const int kPageSizeBits = 22;
 #else
-const int kPageSizeBits = 20;
+const int kPageSizeBits = 19;
 #endif
 
 #endif  // V8_BASE_BUILD_CONFIG_H_

src/heap/heap.cc

@@ -77,7 +77,7 @@ Heap::Heap()
       // semispace_size_ should be a power of 2 and old_generation_size_ should
       // be a multiple of Page::kPageSize.
       max_semi_space_size_(8 * (kPointerSize / 4) * MB),
-      initial_semispace_size_(Page::kPageSize),
+      initial_semispace_size_(MB),
       max_old_generation_size_(700ul * (kPointerSize / 4) * MB),
       initial_old_generation_size_(max_old_generation_size_ /
                                    kInitalOldGenerationLimitFactor),
@@ -5429,16 +5429,19 @@ void Heap::PrintAlloctionsHash() {
 
 
 void Heap::NotifyDeserializationComplete() {
-  deserialization_complete_ = true;
-#ifdef DEBUG
-  // All pages right after bootstrapping must be marked as never-evacuate.
+  DCHECK_EQ(0, gc_count());
   PagedSpaces spaces(this);
   for (PagedSpace* s = spaces.next(); s != NULL; s = spaces.next()) {
+    if (isolate()->snapshot_available()) s->ShrinkImmortalImmovablePages();
+#ifdef DEBUG
+    // All pages right after bootstrapping must be marked as never-evacuate.
     for (Page* p : *s) {
       CHECK(p->NeverEvacuate());
     }
-  }
 #endif  // DEBUG
+  }
+
+  deserialization_complete_ = true;
 }
 
 void Heap::SetEmbedderHeapTracer(EmbedderHeapTracer* tracer) {

src/heap/mark-compact.cc

@@ -600,7 +600,7 @@ void MarkCompactCollector::ComputeEvacuationHeuristics(
   // For memory reducing and optimize for memory mode we directly define both
   // constants.
   const int kTargetFragmentationPercentForReduceMemory = 20;
-  const int kMaxEvacuatedBytesForReduceMemory = 12 * Page::kPageSize;
+  const int kMaxEvacuatedBytesForReduceMemory = 12 * MB;
   const int kTargetFragmentationPercentForOptimizeMemory = 20;
   const int kMaxEvacuatedBytesForOptimizeMemory = 6 * MB;
 
@@ -608,10 +608,10 @@ void MarkCompactCollector::ComputeEvacuationHeuristics(
   // defaults to start and switch to a trace-based (using compaction speed)
   // approach as soon as we have enough samples.
   const int kTargetFragmentationPercent = 70;
-  const int kMaxEvacuatedBytes = 4 * Page::kPageSize;
+  const int kMaxEvacuatedBytes = 4 * MB;
   // Time to take for a single area (=payload of page). Used as soon as there
   // exist enough compaction speed samples.
-  const int kTargetMsPerArea = 1;
+  const float kTargetMsPerArea = 0.5;
 
   if (heap()->ShouldReduceMemory()) {
     *target_fragmentation_percent = kTargetFragmentationPercentForReduceMemory;
@@ -3223,7 +3223,7 @@ int MarkCompactCollector::NumberOfParallelCompactionTasks(int pages,
   // The number of parallel compaction tasks is limited by:
   // - #evacuation pages
   // - (#cores - 1)
-  const double kTargetCompactionTimeInMs = 1;
+  const double kTargetCompactionTimeInMs = .5;
   const int kNumSweepingTasks = 3;
 
   double compaction_speed =

src/heap/spaces.cc

@@ -596,6 +596,21 @@ void MemoryChunk::Unlink() {
   set_next_chunk(NULL);
 }
 
+void MemoryAllocator::ShrinkChunk(MemoryChunk* chunk, size_t bytes_to_shrink) {
+  DCHECK_GE(bytes_to_shrink, static_cast<size_t>(base::OS::CommitPageSize()));
+  DCHECK_EQ(0, bytes_to_shrink % base::OS::CommitPageSize());
+  Address free_start = chunk->area_end_ - bytes_to_shrink;
+  // Don't adjust the size of the page. The area is just uncomitted but not
+  // released.
+  chunk->area_end_ -= bytes_to_shrink;
+  UncommitBlock(free_start, bytes_to_shrink);
+  if (chunk->IsFlagSet(MemoryChunk::IS_EXECUTABLE)) {
+    if (chunk->reservation_.IsReserved())
+      chunk->reservation_.Guard(chunk->area_end_);
+    else
+      base::OS::Guard(chunk->area_end_, base::OS::CommitPageSize());
+  }
+}
 
 MemoryChunk* MemoryAllocator::AllocateChunk(intptr_t reserve_area_size,
                                             intptr_t commit_area_size,
@@ -743,6 +758,47 @@ void Page::ResetFreeListStatistics() {
   available_in_free_list_ = 0;
 }
 
+size_t Page::ShrinkToHighWaterMark() {
+  // Shrink pages to high water mark. The water mark points either to a filler
+  // or the area_end.
+  HeapObject* filler = HeapObject::FromAddress(HighWaterMark());
+  if (filler->address() == area_end()) return 0;
+  CHECK(filler->IsFiller());
+  if (!filler->IsFreeSpace()) return 0;
+
+#ifdef DEBUG
+  // Check the the filler is indeed the last filler on the page.
+  HeapObjectIterator it(this);
+  HeapObject* filler2 = nullptr;
+  for (HeapObject* obj = it.Next(); obj != nullptr; obj = it.Next()) {
+    filler2 = HeapObject::FromAddress(obj->address() + obj->Size());
+  }
+  if (filler2 == nullptr || filler2->address() == area_end()) return 0;
+  DCHECK(filler2->IsFiller());
+  DCHECK_EQ(filler->address(), filler2->address());
+#endif  // DEBUG
+
+  size_t unused = RoundDown(
+      static_cast<size_t>(area_end() - filler->address() - FreeSpace::kSize),
+      base::OS::CommitPageSize());
+  if (unused > 0) {
+    if (FLAG_trace_gc_verbose) {
+      PrintIsolate(heap()->isolate(), "Shrinking page %p: end %p -> %p\n",
+                   reinterpret_cast<void*>(this),
+                   reinterpret_cast<void*>(area_end()),
+                   reinterpret_cast<void*>(area_end() - unused));
+    }
+    heap()->CreateFillerObjectAt(
+        filler->address(),
+        static_cast<int>(area_end() - filler->address() - unused),
+        ClearRecordedSlots::kNo);
+    heap()->memory_allocator()->ShrinkChunk(this, unused);
+    CHECK(filler->IsFiller());
+    CHECK_EQ(filler->address() + filler->Size(), area_end());
+  }
+  return unused;
+}
+
 void MemoryAllocator::PartialFreeMemory(MemoryChunk* chunk,
                                         Address start_free) {
   // We do not allow partial shrink for code.
@@ -1214,17 +1270,25 @@ Object* PagedSpace::FindObject(Address addr) {
   return Smi::FromInt(0);
 }
 
-bool PagedSpace::Expand() {
-  int size = AreaSize();
-  if (snapshotable() && !HasPages()) {
-    size = Snapshot::SizeOfFirstPage(heap()->isolate(), identity());
+void PagedSpace::ShrinkImmortalImmovablePages() {
+  DCHECK(!heap()->deserialization_complete());
+  MemoryChunk::UpdateHighWaterMark(allocation_info_.top());
+  EmptyAllocationInfo();
+  ResetFreeList();
+
+  for (Page* page : *this) {
+    DCHECK(page->IsFlagSet(Page::NEVER_EVACUATE));
+    size_t unused = page->ShrinkToHighWaterMark();
+    accounting_stats_.DecreaseCapacity(static_cast<intptr_t>(unused));
+    AccountUncommitted(unused);
   }
+}
 
+bool PagedSpace::Expand() {
+  const int size = AreaSize();
   if (!heap()->CanExpandOldGeneration(size)) return false;
-
   Page* p = heap()->memory_allocator()->AllocatePage(size, this, executable());
   if (p == nullptr) return false;
-
   AccountCommitted(static_cast<intptr_t>(p->size()));
 
   // Pages created during bootstrapping may contain immortal immovable objects.
@@ -1315,7 +1379,6 @@ void PagedSpace::IncreaseCapacity(int size) {
 
 void PagedSpace::ReleasePage(Page* page) {
   DCHECK_EQ(page->LiveBytes(), 0);
-  DCHECK_EQ(AreaSize(), page->area_size());
   DCHECK_EQ(page->owner(), this);
 
   free_list_.EvictFreeListItems(page);
@@ -1334,10 +1397,8 @@ void PagedSpace::ReleasePage(Page* page) {
   }
 
   AccountUncommitted(static_cast<intptr_t>(page->size()));
+  accounting_stats_.ShrinkSpace(page->area_size());
   heap()->memory_allocator()->Free<MemoryAllocator::kPreFreeAndQueue>(page);
-
-  DCHECK(Capacity() > 0);
-  accounting_stats_.ShrinkSpace(AreaSize());
 }
 
 #ifdef DEBUG

src/heap/spaces.h

@@ -235,7 +235,10 @@ class MemoryChunk {
     IN_TO_SPACE,    // All pages in new space has one of these two set.
     NEW_SPACE_BELOW_AGE_MARK,
     EVACUATION_CANDIDATE,
-    NEVER_EVACUATE,  // May contain immortal immutables.
+
+    // |NEVER_EVACUATE|: A page tagged with this flag will never be selected
+    // for evacuation. Typically used for immortal immovable pages.
+    NEVER_EVACUATE,
 
     // Large objects can have a progress bar in their page header. These object
     // are scanned in increments and will be kept black while being scanned.
@@ -722,7 +725,7 @@ class Page : public MemoryChunk {
   // account.
   // TODO(hpayer): This limit should be way smaller but we currently have
   // short living objects >256K.
-  static const int kMaxRegularHeapObjectSize = 600 * KB;
+  static const int kMaxRegularHeapObjectSize = 400 * KB;
 
   static inline Page* ConvertNewToOld(Page* old_page, PagedSpace* new_owner);
 
@@ -823,6 +826,8 @@ class Page : public MemoryChunk {
     available_in_free_list_.Increment(available);
   }
 
+  size_t ShrinkToHighWaterMark();
+
 #ifdef DEBUG
   void Print();
 #endif  // DEBUG
@@ -1303,6 +1308,8 @@ class MemoryAllocator {
                              intptr_t commit_area_size,
                              Executability executable, Space* space);
 
+  void ShrinkChunk(MemoryChunk* chunk, size_t bytes_to_shrink);
+
   Address ReserveAlignedMemory(size_t requested, size_t alignment,
                                base::VirtualMemory* controller);
   Address AllocateAlignedMemory(size_t reserve_size, size_t commit_size,
@@ -1618,6 +1625,7 @@ class AllocationStats BASE_EMBEDDED {
     capacity_ -= size_in_bytes;
     size_ -= size_in_bytes;
     CHECK_GE(size_, 0);
+    CHECK_GE(capacity_, 0);
   }
 
   // Allocate from available bytes (available -> size).
@@ -2183,6 +2191,10 @@ class PagedSpace : public Space {
   iterator begin() { return iterator(anchor_.next_page()); }
   iterator end() { return iterator(&anchor_); }
 
+  // Shrink immortal immovable pages of the space to be exactly the size needed
+  // using the high water mark.
+  void ShrinkImmortalImmovablePages();
+
  protected:
   // PagedSpaces that should be included in snapshots have different, i.e.,
   // smaller, initial pages.

src/isolate.cc

@@ -2438,13 +2438,19 @@ bool Isolate::Init(Deserializer* des) {
   runtime_profiler_ = new RuntimeProfiler(this);
 
   // If we are deserializing, read the state into the now-empty heap.
-  if (!create_heap_objects) {
-    des->Deserialize(this);
-  }
-  load_stub_cache_->Initialize();
-  store_stub_cache_->Initialize();
-  if (FLAG_ignition || serializer_enabled()) {
-    interpreter_->Initialize();
+  {
+    AlwaysAllocateScope always_allocate(this);
+
+    if (!create_heap_objects) {
+      des->Deserialize(this);
+    }
+    load_stub_cache_->Initialize();
+    store_stub_cache_->Initialize();
+    if (FLAG_ignition || serializer_enabled()) {
+      interpreter_->Initialize();
+    }
+
+    heap_.NotifyDeserializationComplete();
   }
 
   // Finish initialization of ThreadLocal after deserialization is done.
@@ -2475,8 +2481,6 @@ bool Isolate::Init(Deserializer* des) {
 
   time_millis_at_init_ = heap_.MonotonicallyIncreasingTimeInMs();
 
-  heap_.NotifyDeserializationComplete();
-
   if (!create_heap_objects) {
     // Now that the heap is consistent, it's OK to generate the code for the
     // deopt entry table that might have been referred to by optimized code in

src/objects.h

@@ -4712,6 +4712,7 @@ class FreeSpace: public HeapObject {
   // Size is smi tagged when it is stored.
   static const int kSizeOffset = HeapObject::kHeaderSize;
   static const int kNextOffset = POINTER_SIZE_ALIGN(kSizeOffset + kPointerSize);
+  static const int kSize = kNextOffset + kPointerSize;
 
  private:
   DISALLOW_IMPLICIT_CONSTRUCTORS(FreeSpace);
@@ -10467,12 +10468,12 @@ class JSArray: public JSObject {
   static const int kLengthOffset = JSObject::kHeaderSize;
   static const int kSize = kLengthOffset + kPointerSize;
 
-  // 600 * KB is the Page::kMaxRegularHeapObjectSize defined in spaces.h which
+  // 400 * KB is the Page::kMaxRegularHeapObjectSize defined in spaces.h which
   // we do not want to include in objects.h
   // Note that Page::kMaxRegularHeapObjectSize has to be in sync with
   // kInitialMaxFastElementArray which is checked in a DCHECK in heap.cc.
   static const int kInitialMaxFastElementArray =
-      (600 * KB - FixedArray::kHeaderSize - kSize - AllocationMemento::kSize) /
+      (400 * KB - FixedArray::kHeaderSize - kSize - AllocationMemento::kSize) /
       kPointerSize;
 
  private:

src/snapshot/snapshot-common.cc

@@ -31,19 +31,6 @@ bool Snapshot::HasContextSnapshot(Isolate* isolate, size_t index) {
   return index < num_contexts;
 }
 
-
-uint32_t Snapshot::SizeOfFirstPage(Isolate* isolate, AllocationSpace space) {
-  DCHECK(space >= FIRST_PAGED_SPACE && space <= LAST_PAGED_SPACE);
-  if (!isolate->snapshot_available()) {
-    return static_cast<uint32_t>(MemoryAllocator::PageAreaSize(space));
-  }
-  uint32_t size;
-  int offset = kFirstPageSizesOffset + (space - FIRST_PAGED_SPACE) * kInt32Size;
-  memcpy(&size, isolate->snapshot_blob()->data + offset, kInt32Size);
-  return size;
-}
-
-
 bool Snapshot::Initialize(Isolate* isolate) {
   if (!isolate->snapshot_available()) return false;
   base::ElapsedTimer timer;
@@ -89,25 +76,8 @@ MaybeHandle<Context> Snapshot::NewContextFromSnapshot(
   return Handle<Context>::cast(result);
 }
 
-void UpdateMaxRequirementPerPage(
-    uint32_t* requirements,
-    Vector<const SerializedData::Reservation> reservations) {
-  int space = 0;
-  uint32_t current_requirement = 0;
-  for (const auto& reservation : reservations) {
-    current_requirement += reservation.chunk_size();
-    if (reservation.is_last()) {
-      requirements[space] = std::max(requirements[space], current_requirement);
-      current_requirement = 0;
-      space++;
-    }
-  }
-  DCHECK_EQ(i::Serializer::kNumberOfSpaces, space);
-}
-
-void CalculateFirstPageSizes(const SnapshotData* startup_snapshot,
-                             const List<SnapshotData*>* context_snapshots,
-                             uint32_t* sizes_out) {
+void ProfileDeserialization(const SnapshotData* startup_snapshot,
+                            const List<SnapshotData*>* context_snapshots) {
   if (FLAG_profile_deserialization) {
     int startup_total = 0;
     PrintF("Deserialization will reserve:\n");
@@ -123,36 +93,6 @@ void CalculateFirstPageSizes(const SnapshotData* startup_snapshot,
       PrintF("%10d bytes per context #%d\n", context_total, i);
     }
   }
-
-  uint32_t startup_requirements[i::Serializer::kNumberOfSpaces];
-  uint32_t context_requirements[i::Serializer::kNumberOfSpaces];
-  for (int space = 0; space < i::Serializer::kNumberOfSpaces; space++) {
-    startup_requirements[space] = 0;
-    context_requirements[space] = 0;
-  }
-
-  UpdateMaxRequirementPerPage(startup_requirements,
-                              startup_snapshot->Reservations());
-  for (const auto& context_snapshot : *context_snapshots) {
-    UpdateMaxRequirementPerPage(context_requirements,
-                                context_snapshot->Reservations());
-  }
-
-  for (int space = 0; space < i::Serializer::kNumberOfSpaces; space++) {
-    // If the space requirement for a page is less than a page size, we consider
-    // limiting the size of the first page in order to save memory on startup.
-    uint32_t required = startup_requirements[space] +
-                        2 * context_requirements[space] +
-                        Page::kObjectStartOffset;
-    // Add a small allowance to the code space for small scripts.
-    if (space == CODE_SPACE) required += 32 * KB;
-
-    if (space >= FIRST_PAGED_SPACE && space <= LAST_PAGED_SPACE) {
-      uint32_t max_size =
-          MemoryAllocator::PageAreaSize(static_cast<AllocationSpace>(space));
-      sizes_out[space - FIRST_PAGED_SPACE] = std::min(required, max_size);
-    }
-  }
 }
 
 v8::StartupData Snapshot::CreateSnapshotBlob(
@@ -166,13 +106,9 @@ v8::StartupData Snapshot::CreateSnapshotBlob(
     total_length += context_snapshot->RawData().length();
   }
 
-  uint32_t first_page_sizes[kNumPagedSpaces];
-  CalculateFirstPageSizes(startup_snapshot, context_snapshots,
-                          first_page_sizes);
+  ProfileDeserialization(startup_snapshot, context_snapshots);
 
   char* data = new char[total_length];
-  memcpy(data + kFirstPageSizesOffset, first_page_sizes,
-         kNumPagedSpaces * kInt32Size);
   memcpy(data + kNumberOfContextsOffset, &num_contexts, kInt32Size);
   int payload_offset = StartupSnapshotOffset(num_contexts);
   int payload_length = startup_snapshot->RawData().length();

src/snapshot/snapshot.h

@@ -67,9 +67,6 @@ class Snapshot : public AllStatic {
 
   static bool EmbedsScript(Isolate* isolate);
 
-  static uint32_t SizeOfFirstPage(Isolate* isolate, AllocationSpace space);
-
-
   // To be implemented by the snapshot source.
   static const v8::StartupData* DefaultSnapshotBlob();
 
@@ -88,21 +85,16 @@ class Snapshot : public AllStatic {
                                                int index);
 
   // Snapshot blob layout:
-  // [0 - 5] pre-calculated first page sizes for paged spaces
-  // [6] number of contexts N
-  // [7] offset to context 0
-  // [8] offset to context 1
+  // [0] number of contexts N
+  // [1] offset to context 0
+  // [2] offset to context 1
   // ...
   // ... offset to context N - 1
   // ... startup snapshot data
   // ... context 0 snapshot data
   // ... context 1 snapshot data
 
-  static const int kNumPagedSpaces = LAST_PAGED_SPACE - FIRST_PAGED_SPACE + 1;
-
-  static const int kFirstPageSizesOffset = 0;
-  static const int kNumberOfContextsOffset =
-      kFirstPageSizesOffset + kNumPagedSpaces * kInt32Size;
+  static const int kNumberOfContextsOffset = 0;
   static const int kFirstContextOffsetOffset =
       kNumberOfContextsOffset + kInt32Size;
 

test/cctest/heap/heap-utils.cc

@@ -28,6 +28,35 @@ int FixedArrayLenFromSize(int size) {
   return (size - FixedArray::kHeaderSize) / kPointerSize;
 }
 
+std::vector<Handle<FixedArray>> FillOldSpacePageWithFixedArrays(Heap* heap,
+                                                                int remainder) {
+  std::vector<Handle<FixedArray>> handles;
+  Isolate* isolate = heap->isolate();
+  const int kArraySize = 128;
+  const int kArrayLen = heap::FixedArrayLenFromSize(kArraySize);
+  CHECK_EQ(Page::kAllocatableMemory % kArraySize, 0);
+  Handle<FixedArray> array;
+  for (size_t allocated = 0;
+       allocated != (Page::kAllocatableMemory - remainder);
+       allocated += array->Size()) {
+    if (allocated == (Page::kAllocatableMemory - kArraySize)) {
+      array = isolate->factory()->NewFixedArray(
+          heap::FixedArrayLenFromSize(kArraySize - remainder), TENURED);
+      CHECK_EQ(kArraySize - remainder, array->Size());
+    } else {
+      array = isolate->factory()->NewFixedArray(kArrayLen, TENURED);
+      CHECK_EQ(kArraySize, array->Size());
+    }
+    if (handles.empty()) {
+      // Check that allocations started on a new page.
+      CHECK_EQ(array->address(),
+               Page::FromAddress(array->address())->area_start());
+    }
+    handles.push_back(array);
+  }
+  return handles;
+}
+
 std::vector<Handle<FixedArray>> CreatePadding(Heap* heap, int padding_size,
                                               PretenureFlag tenure,
                                               int object_size) {

test/cctest/heap/heap-utils.h

@@ -15,6 +15,12 @@ void SealCurrentObjects(Heap* heap);
 
 int FixedArrayLenFromSize(int size);
 
+// Fill a page with fixed arrays leaving remainder behind. The function does
+// not create additional fillers and assumes that the space has just been
+// sealed.
+std::vector<Handle<FixedArray>> FillOldSpacePageWithFixedArrays(Heap* heap,
+                                                                int remainder);
+
 std::vector<Handle<FixedArray>> CreatePadding(
     Heap* heap, int padding_size, PretenureFlag tenure,
     int object_size = Page::kMaxRegularHeapObjectSize);

test/cctest/heap/test-heap.cc

@@ -2218,6 +2218,18 @@ static Address AlignOldSpace(AllocationAlignment alignment, int offset) {
 // Test the case where allocation must be done from the free list, so filler
 // may precede or follow the object.
 TEST(TestAlignedOverAllocation) {
+  Heap* heap = CcTest::heap();
+  // Test checks for fillers before and behind objects and requires a fresh
+  // page and empty free list.
+  heap::AbandonCurrentlyFreeMemory(heap->old_space());
+  // Allocate a dummy object to properly set up the linear allocation info.
+  AllocationResult dummy =
+      heap->old_space()->AllocateRawUnaligned(kPointerSize);
+  CHECK(!dummy.IsRetry());
+  heap->CreateFillerObjectAt(
+      HeapObject::cast(dummy.ToObjectChecked())->address(), kPointerSize,
+      ClearRecordedSlots::kNo);
+
   // Double misalignment is 4 on 32-bit platforms, 0 on 64-bit ones.
   const intptr_t double_misalignment = kDoubleSize - kPointerSize;
   Address start;
@@ -2359,6 +2371,9 @@ static void FillUpNewSpace(NewSpace* new_space) {
 
 TEST(GrowAndShrinkNewSpace) {
   CcTest::InitializeVM();
+  // Avoid shrinking new space in GC epilogue. This can happen if allocation
+  // throughput samples have been taken while executing the benchmark.
+  i::FLAG_predictable = true;
   Heap* heap = CcTest::heap();
   NewSpace* new_space = heap->new_space();
 
@@ -3601,8 +3616,12 @@ TEST(ReleaseOverReservedPages) {
   i::FLAG_page_promotion = false;
   CcTest::InitializeVM();
   Isolate* isolate = CcTest::i_isolate();
+  // If there's snapshot available, we don't know whether 20 small arrays will
+  // fit on the initial pages.
+  if (!isolate->snapshot_available()) return;
   Factory* factory = isolate->factory();
   Heap* heap = isolate->heap();
+
   v8::HandleScope scope(CcTest::isolate());
   static const int number_of_test_pages = 20;
 
@@ -3632,14 +3651,8 @@ TEST(ReleaseOverReservedPages) {
                           "triggered by test 2");
   CHECK_GE(overall_page_count, old_space->CountTotalPages() * 2);
 
-  // Triggering a last-resort GC should cause all pages to be released to the
-  // OS so that other processes can seize the memory.  If we get a failure here
-  // where there are 2 pages left instead of 1, then we should increase the
-  // size of the first page a little in SizeOfFirstPage in spaces.cc.  The
-  // first page should be small in order to reduce memory used when the VM
-  // boots, but if the 20 small arrays don't fit on the first page then that's
-  // an indication that it is too small.
   heap->CollectAllAvailableGarbage("triggered really hard");
+  // Triggering a last-resort GC should release all additional pages.
   CHECK_EQ(initial_page_count, old_space->CountTotalPages());
 }
 

test/cctest/heap/test-spaces.cc

@@ -32,6 +32,7 @@
 #include "src/v8.h"
 #include "test/cctest/cctest.h"
 #include "test/cctest/heap/heap-tester.h"
+#include "test/cctest/heap/heap-utils.h"
 
 namespace v8 {
 namespace internal {
@@ -478,8 +479,7 @@ TEST(LargeObjectSpace) {
   CHECK(lo->AllocateRaw(lo_size, NOT_EXECUTABLE).IsRetry());
 }
 
-
-TEST(SizeOfFirstPageIsLargeEnough) {
+TEST(SizeOfInitialHeap) {
   if (i::FLAG_always_opt) return;
   // Bootstrapping without a snapshot causes more allocations.
   CcTest::InitializeVM();
@@ -495,22 +495,31 @@ TEST(SizeOfFirstPageIsLargeEnough) {
     return;
   }
 
-  // If this test fails due to enabling experimental natives that are not part
-  // of the snapshot, we may need to adjust CalculateFirstPageSizes.
+  // The limit for each space for an empty isolate containing just the
+  // snapshot.
+  const size_t kMaxInitialSizePerSpace = 1536 * KB;  // 1.5MB
 
-  // Freshly initialized VM gets by with one page per space.
+  // Freshly initialized VM gets by with the snapshot size (which is below
+  // kMaxInitialSizePerSpace per space).
+  Heap* heap = isolate->heap();
+  int page_count[LAST_PAGED_SPACE + 1] = {0, 0, 0, 0};
   for (int i = FIRST_PAGED_SPACE; i <= LAST_PAGED_SPACE; i++) {
     // Debug code can be very large, so skip CODE_SPACE if we are generating it.
     if (i == CODE_SPACE && i::FLAG_debug_code) continue;
-    CHECK_EQ(1, isolate->heap()->paged_space(i)->CountTotalPages());
+
+    page_count[i] = heap->paged_space(i)->CountTotalPages();
+    // Check that the initial heap is also below the limit.
+    CHECK_LT(static_cast<size_t>(heap->paged_space(i)->CommittedMemory()),
+             kMaxInitialSizePerSpace);
   }
 
-  // Executing the empty script gets by with one page per space.
+  // Executing the empty script gets by with the same number of pages, i.e.,
+  // requires no extra space.
   CompileRun("/*empty*/");
   for (int i = FIRST_PAGED_SPACE; i <= LAST_PAGED_SPACE; i++) {
     // Debug code can be very large, so skip CODE_SPACE if we are generating it.
     if (i == CODE_SPACE && i::FLAG_debug_code) continue;
-    CHECK_EQ(1, isolate->heap()->paged_space(i)->CountTotalPages());
+    CHECK_EQ(page_count[i], isolate->heap()->paged_space(i)->CountTotalPages());
   }
 
   // No large objects required to perform the above steps.
@@ -681,5 +690,105 @@ UNINITIALIZED_TEST(InlineAllocationObserverCadence) {
   isolate->Dispose();
 }
 
+TEST(ShrinkPageToHighWaterMarkFreeSpaceEnd) {
+  CcTest::InitializeVM();
+  Isolate* isolate = CcTest::i_isolate();
+  HandleScope scope(isolate);
+
+  heap::SealCurrentObjects(CcTest::heap());
+
+  // Prepare page that only contains a single object and a trailing FreeSpace
+  // filler.
+  Handle<FixedArray> array = isolate->factory()->NewFixedArray(128, TENURED);
+  Page* page = Page::FromAddress(array->address());
+
+  // Reset space so high water mark is consistent.
+  CcTest::heap()->old_space()->ResetFreeList();
+  CcTest::heap()->old_space()->EmptyAllocationInfo();
+
+  HeapObject* filler =
+      HeapObject::FromAddress(array->address() + array->Size());
+  CHECK(filler->IsFreeSpace());
+  size_t shrinked = page->ShrinkToHighWaterMark();
+  size_t should_have_shrinked =
+      RoundDown(static_cast<size_t>(Page::kAllocatableMemory - array->Size()),
+                base::OS::CommitPageSize());
+  CHECK_EQ(should_have_shrinked, shrinked);
+}
+
+TEST(ShrinkPageToHighWaterMarkNoFiller) {
+  CcTest::InitializeVM();
+  Isolate* isolate = CcTest::i_isolate();
+  HandleScope scope(isolate);
+
+  heap::SealCurrentObjects(CcTest::heap());
+
+  const int kFillerSize = 0;
+  std::vector<Handle<FixedArray>> arrays =
+      heap::FillOldSpacePageWithFixedArrays(CcTest::heap(), kFillerSize);
+  Handle<FixedArray> array = arrays.back();
+  Page* page = Page::FromAddress(array->address());
+  CHECK_EQ(page->area_end(), array->address() + array->Size() + kFillerSize);
+
+  // Reset space so high water mark and fillers are consistent.
+  CcTest::heap()->old_space()->ResetFreeList();
+  CcTest::heap()->old_space()->EmptyAllocationInfo();
+
+  const size_t shrinked = page->ShrinkToHighWaterMark();
+  CHECK_EQ(0, shrinked);
+}
+
+TEST(ShrinkPageToHighWaterMarkOneWordFiller) {
+  CcTest::InitializeVM();
+  Isolate* isolate = CcTest::i_isolate();
+  HandleScope scope(isolate);
+
+  heap::SealCurrentObjects(CcTest::heap());
+
+  const int kFillerSize = kPointerSize;
+  std::vector<Handle<FixedArray>> arrays =
+      heap::FillOldSpacePageWithFixedArrays(CcTest::heap(), kFillerSize);
+  Handle<FixedArray> array = arrays.back();
+  Page* page = Page::FromAddress(array->address());
+  CHECK_EQ(page->area_end(), array->address() + array->Size() + kFillerSize);
+
+  // Reset space so high water mark and fillers are consistent.
+  CcTest::heap()->old_space()->ResetFreeList();
+  CcTest::heap()->old_space()->EmptyAllocationInfo();
+
+  HeapObject* filler =
+      HeapObject::FromAddress(array->address() + array->Size());
+  CHECK_EQ(filler->map(), CcTest::heap()->one_pointer_filler_map());
+
+  const size_t shrinked = page->ShrinkToHighWaterMark();
+  CHECK_EQ(0, shrinked);
+}
+
+TEST(ShrinkPageToHighWaterMarkTwoWordFiller) {
+  CcTest::InitializeVM();
+  Isolate* isolate = CcTest::i_isolate();
+  HandleScope scope(isolate);
+
+  heap::SealCurrentObjects(CcTest::heap());
+
+  const int kFillerSize = 2 * kPointerSize;
+  std::vector<Handle<FixedArray>> arrays =
+      heap::FillOldSpacePageWithFixedArrays(CcTest::heap(), kFillerSize);
+  Handle<FixedArray> array = arrays.back();
+  Page* page = Page::FromAddress(array->address());
+  CHECK_EQ(page->area_end(), array->address() + array->Size() + kFillerSize);
+
+  // Reset space so high water mark and fillers are consistent.
+  CcTest::heap()->old_space()->ResetFreeList();
+  CcTest::heap()->old_space()->EmptyAllocationInfo();
+
+  HeapObject* filler =
+      HeapObject::FromAddress(array->address() + array->Size());
+  CHECK_EQ(filler->map(), CcTest::heap()->two_pointer_filler_map());
+
+  const size_t shrinked = page->ShrinkToHighWaterMark();
+  CHECK_EQ(0, shrinked);
+}
+
 }  // namespace internal
 }  // namespace v8

test/cctest/test-serialize.cc

@@ -1177,13 +1177,13 @@ TEST(CodeSerializerThreeBigStrings) {
 
   Vector<const uint8_t> source_b =
       ConstructSource(STATIC_CHAR_VECTOR("var b = \""), STATIC_CHAR_VECTOR("b"),
-                      STATIC_CHAR_VECTOR("\";"), 600000);
+                      STATIC_CHAR_VECTOR("\";"), 400000);
   Handle<String> source_b_str =
       f->NewStringFromOneByte(source_b).ToHandleChecked();
 
   Vector<const uint8_t> source_c =
       ConstructSource(STATIC_CHAR_VECTOR("var c = \""), STATIC_CHAR_VECTOR("c"),
-                      STATIC_CHAR_VECTOR("\";"), 500000);
+                      STATIC_CHAR_VECTOR("\";"), 400000);
   Handle<String> source_c_str =
       f->NewStringFromOneByte(source_c).ToHandleChecked();
 
@@ -1216,10 +1216,10 @@ TEST(CodeSerializerThreeBigStrings) {
   v8::Maybe<int32_t> result =
       CompileRun("(a + b).length")
           ->Int32Value(v8::Isolate::GetCurrent()->GetCurrentContext());
-  CHECK_EQ(600000 + 700000, result.FromJust());
+  CHECK_EQ(400000 + 700000, result.FromJust());
   result = CompileRun("(b + c).length")
                ->Int32Value(v8::Isolate::GetCurrent()->GetCurrentContext());
-  CHECK_EQ(500000 + 600000, result.FromJust());
+  CHECK_EQ(400000 + 400000, result.FromJust());
   Heap* heap = isolate->heap();
   v8::Local<v8::String> result_str =
       CompileRun("a")