1 //===-- sanitizer_allocator_primary32.h -------------------------*- C++ -*-===//
2 //
3 // Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
4 // See https://llvm.org/LICENSE.txt for license information.
5 // SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
6 //
7 //===----------------------------------------------------------------------===//
8 //
9 // Part of the Sanitizer Allocator.
10 //
11 //===----------------------------------------------------------------------===//
12 #ifndef SANITIZER_ALLOCATOR_H
13 #error This file must be included inside sanitizer_allocator.h
14 #endif
15 
16 template<class SizeClassAllocator> struct SizeClassAllocator32LocalCache;
17 
18 // SizeClassAllocator32 -- allocator for 32-bit address space.
19 // This allocator can theoretically be used on 64-bit arch, but there it is less
20 // efficient than SizeClassAllocator64.
21 //
22 // [kSpaceBeg, kSpaceBeg + kSpaceSize) is the range of addresses which can
23 // be returned by MmapOrDie().
24 //
25 // Region:
26 //   a result of a single call to MmapAlignedOrDieOnFatalError(kRegionSize,
27 //                                                             kRegionSize).
28 // Since the regions are aligned by kRegionSize, there are exactly
29 // kNumPossibleRegions possible regions in the address space and so we keep
30 // a ByteMap possible_regions to store the size classes of each Region.
31 // 0 size class means the region is not used by the allocator.
32 //
33 // One Region is used to allocate chunks of a single size class.
34 // A Region looks like this:
35 // UserChunk1 .. UserChunkN <gap> MetaChunkN .. MetaChunk1
36 //
37 // In order to avoid false sharing the objects of this class should be
38 // chache-line aligned.
39 
40 struct SizeClassAllocator32FlagMasks {  //  Bit masks.
41   enum {
42     kRandomShuffleChunks = 1,
43     kUseSeparateSizeClassForBatch = 2,
44   };
45 };
46 
47 template <class Params>
48 class SizeClassAllocator32 {
49  private:
50   static const u64 kTwoLevelByteMapSize1 =
51       (Params::kSpaceSize >> Params::kRegionSizeLog) >> 12;
52   static const u64 kMinFirstMapSizeTwoLevelByteMap = 4;
53 
54  public:
55   using AddressSpaceView = typename Params::AddressSpaceView;
56   static const uptr kSpaceBeg = Params::kSpaceBeg;
57   static const u64 kSpaceSize = Params::kSpaceSize;
58   static const uptr kMetadataSize = Params::kMetadataSize;
59   typedef typename Params::SizeClassMap SizeClassMap;
60   static const uptr kRegionSizeLog = Params::kRegionSizeLog;
61   typedef typename Params::MapUnmapCallback MapUnmapCallback;
62   using ByteMap = typename conditional<
63       (kTwoLevelByteMapSize1 < kMinFirstMapSizeTwoLevelByteMap),
64       FlatByteMap<(Params::kSpaceSize >> Params::kRegionSizeLog),
65                   AddressSpaceView>,
66       TwoLevelByteMap<kTwoLevelByteMapSize1, 1 << 12, AddressSpaceView>>::type;
67 
68   COMPILER_CHECK(!SANITIZER_SIGN_EXTENDED_ADDRESSES ||
69                  (kSpaceSize & (kSpaceSize - 1)) == 0);
70 
71   static const bool kRandomShuffleChunks = Params::kFlags &
72       SizeClassAllocator32FlagMasks::kRandomShuffleChunks;
73   static const bool kUseSeparateSizeClassForBatch = Params::kFlags &
74       SizeClassAllocator32FlagMasks::kUseSeparateSizeClassForBatch;
75 
76   struct TransferBatch {
77     static const uptr kMaxNumCached = SizeClassMap::kMaxNumCachedHint - 2;
78     void SetFromArray(void *batch[], uptr count) {
79       DCHECK_LE(count, kMaxNumCached);
80       count_ = count;
81       for (uptr i = 0; i < count; i++)
82         batch_[i] = batch[i];
83     }
84     uptr Count() const { return count_; }
85     void Clear() { count_ = 0; }
86     void Add(void *ptr) {
87       batch_[count_++] = ptr;
88       DCHECK_LE(count_, kMaxNumCached);
89     }
90     void CopyToArray(void *to_batch[]) const {
91       for (uptr i = 0, n = Count(); i < n; i++)
92         to_batch[i] = batch_[i];
93     }
94 
95     // How much memory do we need for a batch containing n elements.
96     static uptr AllocationSizeRequiredForNElements(uptr n) {
97       return sizeof(uptr) * 2 + sizeof(void *) * n;
98     }
99     static uptr MaxCached(uptr size) {
100       return Min(kMaxNumCached, SizeClassMap::MaxCachedHint(size));
101     }
102 
103     TransferBatch *next;
104 
105    private:
106     uptr count_;
107     void *batch_[kMaxNumCached];
108   };
109 
110   static const uptr kBatchSize = sizeof(TransferBatch);
111   COMPILER_CHECK((kBatchSize & (kBatchSize - 1)) == 0);
112   COMPILER_CHECK(kBatchSize == SizeClassMap::kMaxNumCachedHint * sizeof(uptr));
113 
114   static uptr ClassIdToSize(uptr class_id) {
115     return (class_id == SizeClassMap::kBatchClassID) ?
116         kBatchSize : SizeClassMap::Size(class_id);
117   }
118 
119   typedef SizeClassAllocator32<Params> ThisT;
120   typedef SizeClassAllocator32LocalCache<ThisT> AllocatorCache;
121 
122   void Init(s32 release_to_os_interval_ms) {
123     possible_regions.Init();
124     internal_memset(size_class_info_array, 0, sizeof(size_class_info_array));
125   }
126 
127   s32 ReleaseToOSIntervalMs() const {
128     return kReleaseToOSIntervalNever;
129   }
130 
131   void SetReleaseToOSIntervalMs(s32 release_to_os_interval_ms) {
132     // This is empty here. Currently only implemented in 64-bit allocator.
133   }
134 
135   void ForceReleaseToOS() {
136     // Currently implemented in 64-bit allocator only.
137   }
138 
139   void *MapWithCallback(uptr size) {
140     void *res = MmapOrDie(size, PrimaryAllocatorName);
141     MapUnmapCallback().OnMap((uptr)res, size);
142     return res;
143   }
144 
145   void UnmapWithCallback(uptr beg, uptr size) {
146     MapUnmapCallback().OnUnmap(beg, size);
147     UnmapOrDie(reinterpret_cast<void *>(beg), size);
148   }
149 
150   static bool CanAllocate(uptr size, uptr alignment) {
151     return size <= SizeClassMap::kMaxSize &&
152       alignment <= SizeClassMap::kMaxSize;
153   }
154 
155   void *GetMetaData(const void *p) {
156     CHECK(PointerIsMine(p));
157     uptr mem = reinterpret_cast<uptr>(p);
158     uptr beg = ComputeRegionBeg(mem);
159     uptr size = ClassIdToSize(GetSizeClass(p));
160     u32 offset = mem - beg;
161     uptr n = offset / (u32)size;  // 32-bit division
162     uptr meta = (beg + kRegionSize) - (n + 1) * kMetadataSize;
163     return reinterpret_cast<void*>(meta);
164   }
165 
166   NOINLINE TransferBatch *AllocateBatch(AllocatorStats *stat, AllocatorCache *c,
167                                         uptr class_id) {
168     DCHECK_LT(class_id, kNumClasses);
169     SizeClassInfo *sci = GetSizeClassInfo(class_id);
170     SpinMutexLock l(&sci->mutex);
171     if (sci->free_list.empty()) {
172       if (UNLIKELY(!PopulateFreeList(stat, c, sci, class_id)))
173         return nullptr;
174       DCHECK(!sci->free_list.empty());
175     }
176     TransferBatch *b = sci->free_list.front();
177     sci->free_list.pop_front();
178     return b;
179   }
180 
181   NOINLINE void DeallocateBatch(AllocatorStats *stat, uptr class_id,
182                                 TransferBatch *b) {
183     DCHECK_LT(class_id, kNumClasses);
184     CHECK_GT(b->Count(), 0);
185     SizeClassInfo *sci = GetSizeClassInfo(class_id);
186     SpinMutexLock l(&sci->mutex);
187     sci->free_list.push_front(b);
188   }
189 
190   bool PointerIsMine(const void *p) {
191     uptr mem = reinterpret_cast<uptr>(p);
192     if (SANITIZER_SIGN_EXTENDED_ADDRESSES)
193       mem &= (kSpaceSize - 1);
194     if (mem < kSpaceBeg || mem >= kSpaceBeg + kSpaceSize)
195       return false;
196     return GetSizeClass(p) != 0;
197   }
198 
199   uptr GetSizeClass(const void *p) {
200     return possible_regions[ComputeRegionId(reinterpret_cast<uptr>(p))];
201   }
202 
203   void *GetBlockBegin(const void *p) {
204     CHECK(PointerIsMine(p));
205     uptr mem = reinterpret_cast<uptr>(p);
206     uptr beg = ComputeRegionBeg(mem);
207     uptr size = ClassIdToSize(GetSizeClass(p));
208     u32 offset = mem - beg;
209     u32 n = offset / (u32)size;  // 32-bit division
210     uptr res = beg + (n * (u32)size);
211     return reinterpret_cast<void*>(res);
212   }
213 
214   uptr GetActuallyAllocatedSize(void *p) {
215     CHECK(PointerIsMine(p));
216     return ClassIdToSize(GetSizeClass(p));
217   }
218 
219   static uptr ClassID(uptr size) { return SizeClassMap::ClassID(size); }
220 
221   uptr TotalMemoryUsed() {
222     // No need to lock here.
223     uptr res = 0;
224     for (uptr i = 0; i < kNumPossibleRegions; i++)
225       if (possible_regions[i])
226         res += kRegionSize;
227     return res;
228   }
229 
230   void TestOnlyUnmap() {
231     for (uptr i = 0; i < kNumPossibleRegions; i++)
232       if (possible_regions[i])
233         UnmapWithCallback((i * kRegionSize), kRegionSize);
234   }
235 
236   // ForceLock() and ForceUnlock() are needed to implement Darwin malloc zone
237   // introspection API.
238   void ForceLock() {
239     for (uptr i = 0; i < kNumClasses; i++) {
240       GetSizeClassInfo(i)->mutex.Lock();
241     }
242   }
243 
244   void ForceUnlock() {
245     for (int i = kNumClasses - 1; i >= 0; i--) {
246       GetSizeClassInfo(i)->mutex.Unlock();
247     }
248   }
249 
250   // Iterate over all existing chunks.
251   // The allocator must be locked when calling this function.
252   void ForEachChunk(ForEachChunkCallback callback, void *arg) {
253     for (uptr region = 0; region < kNumPossibleRegions; region++)
254       if (possible_regions[region]) {
255         uptr chunk_size = ClassIdToSize(possible_regions[region]);
256         uptr max_chunks_in_region = kRegionSize / (chunk_size + kMetadataSize);
257         uptr region_beg = region * kRegionSize;
258         for (uptr chunk = region_beg;
259              chunk < region_beg + max_chunks_in_region * chunk_size;
260              chunk += chunk_size) {
261           // Too slow: CHECK_EQ((void *)chunk, GetBlockBegin((void *)chunk));
262           callback(chunk, arg);
263         }
264       }
265   }
266 
267   void PrintStats() {}
268 
269   static uptr AdditionalSize() { return 0; }
270 
271   typedef SizeClassMap SizeClassMapT;
272   static const uptr kNumClasses = SizeClassMap::kNumClasses;
273 
274  private:
275   static const uptr kRegionSize = 1 << kRegionSizeLog;
276   static const uptr kNumPossibleRegions = kSpaceSize / kRegionSize;
277 
278   struct ALIGNED(SANITIZER_CACHE_LINE_SIZE) SizeClassInfo {
279     StaticSpinMutex mutex;
280     IntrusiveList<TransferBatch> free_list;
281     u32 rand_state;
282   };
283   COMPILER_CHECK(sizeof(SizeClassInfo) % kCacheLineSize == 0);
284 
285   uptr ComputeRegionId(uptr mem) const {
286     if (SANITIZER_SIGN_EXTENDED_ADDRESSES)
287       mem &= (kSpaceSize - 1);
288     const uptr res = mem >> kRegionSizeLog;
289     CHECK_LT(res, kNumPossibleRegions);
290     return res;
291   }
292 
293   uptr ComputeRegionBeg(uptr mem) {
294     return mem & ~(kRegionSize - 1);
295   }
296 
297   uptr AllocateRegion(AllocatorStats *stat, uptr class_id) {
298     DCHECK_LT(class_id, kNumClasses);
299     const uptr res = reinterpret_cast<uptr>(MmapAlignedOrDieOnFatalError(
300         kRegionSize, kRegionSize, PrimaryAllocatorName));
301     if (UNLIKELY(!res))
302       return 0;
303     MapUnmapCallback().OnMap(res, kRegionSize);
304     stat->Add(AllocatorStatMapped, kRegionSize);
305     CHECK(IsAligned(res, kRegionSize));
306     possible_regions.set(ComputeRegionId(res), static_cast<u8>(class_id));
307     return res;
308   }
309 
310   SizeClassInfo *GetSizeClassInfo(uptr class_id) {
311     DCHECK_LT(class_id, kNumClasses);
312     return &size_class_info_array[class_id];
313   }
314 
315   bool PopulateBatches(AllocatorCache *c, SizeClassInfo *sci, uptr class_id,
316                        TransferBatch **current_batch, uptr max_count,
317                        uptr *pointers_array, uptr count) {
318     // If using a separate class for batches, we do not need to shuffle it.
319     if (kRandomShuffleChunks && (!kUseSeparateSizeClassForBatch ||
320         class_id != SizeClassMap::kBatchClassID))
321       RandomShuffle(pointers_array, count, &sci->rand_state);
322     TransferBatch *b = *current_batch;
323     for (uptr i = 0; i < count; i++) {
324       if (!b) {
325         b = c->CreateBatch(class_id, this, (TransferBatch*)pointers_array[i]);
326         if (UNLIKELY(!b))
327           return false;
328         b->Clear();
329       }
330       b->Add((void*)pointers_array[i]);
331       if (b->Count() == max_count) {
332         sci->free_list.push_back(b);
333         b = nullptr;
334       }
335     }
336     *current_batch = b;
337     return true;
338   }
339 
340   bool PopulateFreeList(AllocatorStats *stat, AllocatorCache *c,
341                         SizeClassInfo *sci, uptr class_id) {
342     const uptr region = AllocateRegion(stat, class_id);
343     if (UNLIKELY(!region))
344       return false;
345     if (kRandomShuffleChunks)
346       if (UNLIKELY(sci->rand_state == 0))
347         // The random state is initialized from ASLR (PIE) and time.
348         sci->rand_state = reinterpret_cast<uptr>(sci) ^ NanoTime();
349     const uptr size = ClassIdToSize(class_id);
350     const uptr n_chunks = kRegionSize / (size + kMetadataSize);
351     const uptr max_count = TransferBatch::MaxCached(size);
352     DCHECK_GT(max_count, 0);
353     TransferBatch *b = nullptr;
354     constexpr uptr kShuffleArraySize = 48;
355     uptr shuffle_array[kShuffleArraySize];
356     uptr count = 0;
357     for (uptr i = region; i < region + n_chunks * size; i += size) {
358       shuffle_array[count++] = i;
359       if (count == kShuffleArraySize) {
360         if (UNLIKELY(!PopulateBatches(c, sci, class_id, &b, max_count,
361                                       shuffle_array, count)))
362           return false;
363         count = 0;
364       }
365     }
366     if (count) {
367       if (UNLIKELY(!PopulateBatches(c, sci, class_id, &b, max_count,
368                                     shuffle_array, count)))
369         return false;
370     }
371     if (b) {
372       CHECK_GT(b->Count(), 0);
373       sci->free_list.push_back(b);
374     }
375     return true;
376   }
377 
378   ByteMap possible_regions;
379   SizeClassInfo size_class_info_array[kNumClasses];
380 };
381