Delete chrome.mediaGalleriesPrivate because the functionality unique to it has since...
[chromium-blink-merge.git] / third_party / tcmalloc / chromium / src / memory_region_map.cc
blobd7338f2a64cd715d2b043f646c8bb7a4b6f96b99
1 /* Copyright (c) 2006, Google Inc.
2 * All rights reserved.
3 *
4 * Redistribution and use in source and binary forms, with or without
5 * modification, are permitted provided that the following conditions are
6 * met:
7 *
8 * * Redistributions of source code must retain the above copyright
9 * notice, this list of conditions and the following disclaimer.
10 * * Redistributions in binary form must reproduce the above
11 * copyright notice, this list of conditions and the following disclaimer
12 * in the documentation and/or other materials provided with the
13 * distribution.
14 * * Neither the name of Google Inc. nor the names of its
15 * contributors may be used to endorse or promote products derived from
16 * this software without specific prior written permission.
18 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
19 * "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
20 * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
21 * A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
22 * OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
23 * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
24 * LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
25 * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
26 * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
27 * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
28 * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
30 * ---
31 * Author: Maxim Lifantsev
35 // Background and key design points of MemoryRegionMap.
37 // MemoryRegionMap is a low-level module with quite atypical requirements that
38 // result in some degree of non-triviality of the implementation and design.
40 // MemoryRegionMap collects info about *all* memory regions created with
41 // mmap, munmap, mremap, sbrk.
42 // They key word above is 'all': all that are happening in a process
43 // during its lifetime frequently starting even before global object
44 // constructor execution.
46 // This is needed by the primary client of MemoryRegionMap:
47 // HeapLeakChecker uses the regions and the associated stack traces
48 // to figure out what part of the memory is the heap:
49 // if MemoryRegionMap were to miss some (early) regions, leak checking would
50 // stop working correctly.
52 // To accomplish the goal of functioning before/during global object
53 // constructor execution MemoryRegionMap is done as a singleton service
54 // that relies on own on-demand initialized static constructor-less data,
55 // and only relies on other low-level modules that can also function properly
56 // even before global object constructors run.
58 // Accomplishing the goal of collecting data about all mmap, munmap, mremap,
59 // sbrk occurrences is a more involved: conceptually to do this one needs to
60 // record some bits of data in particular about any mmap or sbrk call,
61 // but to do that one needs to allocate memory for that data at some point,
62 // but all memory allocations in the end themselves come from an mmap
63 // or sbrk call (that's how the address space of the process grows).
65 // Also note that we need to do all the above recording from
66 // within an mmap/sbrk hook which is sometimes/frequently is made by a memory
67 // allocator, including the allocator MemoryRegionMap itself must rely on.
68 // In the case of heap-checker usage this includes even the very first
69 // mmap/sbrk call happening in the program: heap-checker gets activated due to
70 // a link-time installed mmap/sbrk hook and it initializes MemoryRegionMap
71 // and asks it to record info about this very first call right from that
72 // very first hook invocation.
74 // MemoryRegionMap is doing its memory allocations via LowLevelAlloc:
75 // unlike more complex standard memory allocator, LowLevelAlloc cooperates with
76 // MemoryRegionMap by not holding any of its own locks while it calls mmap
77 // to get memory, thus we are able to call LowLevelAlloc from
78 // our mmap/sbrk hooks without causing a deadlock in it.
79 // For the same reason of deadlock prevention the locking in MemoryRegionMap
80 // itself is write-recursive which is an exception to Google's mutex usage.
82 // We still need to break the infinite cycle of mmap calling our hook,
83 // which asks LowLevelAlloc for memory to record this mmap,
84 // which (sometimes) causes mmap, which calls our hook, and so on.
85 // We do this as follows: on a recursive call of MemoryRegionMap's
86 // mmap/sbrk/mremap hook we record the data about the allocation in a
87 // static fixed-sized stack (saved_regions and saved_buckets), when the
88 // recursion unwinds but before returning from the outer hook call we unwind
89 // this stack and move the data from saved_regions and saved_buckets to its
90 // permanent place in the RegionSet and "bucket_table" respectively,
91 // which can cause more allocations and mmap-s and recursion and unwinding,
92 // but the whole process ends eventually due to the fact that for the small
93 // allocations we are doing LowLevelAlloc reuses one mmap call and parcels out
94 // the memory it created to satisfy several of our allocation requests.
97 // ========================================================================= //
99 #include <config.h>
101 #ifdef HAVE_UNISTD_H
102 #include <unistd.h>
103 #endif
104 #ifdef HAVE_INTTYPES_H
105 #include <inttypes.h>
106 #endif
107 #ifdef HAVE_MMAP
108 #include <sys/mman.h>
109 #elif !defined(MAP_FAILED)
110 #define MAP_FAILED -1 // the only thing we need from mman.h
111 #endif
112 #ifdef HAVE_PTHREAD
113 #include <pthread.h> // for pthread_t, pthread_self()
114 #endif
115 #include <stddef.h>
117 #include <algorithm>
118 #include <set>
120 #include "memory_region_map.h"
122 #include "base/logging.h"
123 #include "base/low_level_alloc.h"
124 #include "malloc_hook-inl.h"
126 #include <gperftools/stacktrace.h>
127 #include <gperftools/malloc_hook.h>
129 // MREMAP_FIXED is a linux extension. How it's used in this file,
130 // setting it to 0 is equivalent to saying, "This feature isn't
131 // supported", which is right.
132 #ifndef MREMAP_FIXED
133 # define MREMAP_FIXED 0
134 #endif
136 using std::max;
138 // ========================================================================= //
140 int MemoryRegionMap::client_count_ = 0;
141 int MemoryRegionMap::max_stack_depth_ = 0;
142 MemoryRegionMap::RegionSet* MemoryRegionMap::regions_ = NULL;
143 LowLevelAlloc::Arena* MemoryRegionMap::arena_ = NULL;
144 SpinLock MemoryRegionMap::lock_(SpinLock::LINKER_INITIALIZED);
145 SpinLock MemoryRegionMap::owner_lock_( // ACQUIRED_AFTER(lock_)
146 SpinLock::LINKER_INITIALIZED);
147 int MemoryRegionMap::recursion_count_ = 0; // GUARDED_BY(owner_lock_)
148 pthread_t MemoryRegionMap::lock_owner_tid_; // GUARDED_BY(owner_lock_)
149 int64 MemoryRegionMap::map_size_ = 0;
150 int64 MemoryRegionMap::unmap_size_ = 0;
151 HeapProfileBucket** MemoryRegionMap::bucket_table_ = NULL; // GUARDED_BY(lock_)
152 int MemoryRegionMap::num_buckets_ = 0; // GUARDED_BY(lock_)
153 int MemoryRegionMap::saved_buckets_count_ = 0; // GUARDED_BY(lock_)
154 HeapProfileBucket MemoryRegionMap::saved_buckets_[20]; // GUARDED_BY(lock_)
156 // GUARDED_BY(lock_)
157 const void* MemoryRegionMap::saved_buckets_keys_[20][kMaxStackDepth];
159 // ========================================================================= //
161 // Simple hook into execution of global object constructors,
162 // so that we do not call pthread_self() when it does not yet work.
163 static bool libpthread_initialized = false;
164 static bool initializer = (libpthread_initialized = true, true);
166 static inline bool current_thread_is(pthread_t should_be) {
167 // Before main() runs, there's only one thread, so we're always that thread
168 if (!libpthread_initialized) return true;
169 // this starts working only sometime well into global constructor execution:
170 return pthread_equal(pthread_self(), should_be);
173 // ========================================================================= //
175 // Constructor-less place-holder to store a RegionSet in.
176 union MemoryRegionMap::RegionSetRep {
177 char rep[sizeof(RegionSet)];
178 void* align_it; // do not need a better alignment for 'rep' than this
179 RegionSet* region_set() { return reinterpret_cast<RegionSet*>(rep); }
182 // The bytes where MemoryRegionMap::regions_ will point to.
183 // We use RegionSetRep with noop c-tor so that global construction
184 // does not interfere.
185 static MemoryRegionMap::RegionSetRep regions_rep;
187 // ========================================================================= //
189 // Has InsertRegionLocked been called recursively
190 // (or rather should we *not* use regions_ to record a hooked mmap).
191 static bool recursive_insert = false;
193 void MemoryRegionMap::Init(int max_stack_depth, bool use_buckets) {
194 RAW_VLOG(10, "MemoryRegionMap Init");
195 RAW_CHECK(max_stack_depth >= 0, "");
196 // Make sure we don't overflow the memory in region stacks:
197 RAW_CHECK(max_stack_depth <= kMaxStackDepth,
198 "need to increase kMaxStackDepth?");
199 Lock();
200 client_count_ += 1;
201 max_stack_depth_ = max(max_stack_depth_, max_stack_depth);
202 if (client_count_ > 1) {
203 // not first client: already did initialization-proper
204 Unlock();
205 RAW_VLOG(10, "MemoryRegionMap Init increment done");
206 return;
208 // Set our hooks and make sure they were installed:
209 RAW_CHECK(MallocHook::AddMmapHook(&MmapHook), "");
210 RAW_CHECK(MallocHook::AddMremapHook(&MremapHook), "");
211 RAW_CHECK(MallocHook::AddSbrkHook(&SbrkHook), "");
212 RAW_CHECK(MallocHook::AddMunmapHook(&MunmapHook), "");
213 // We need to set recursive_insert since the NewArena call itself
214 // will already do some allocations with mmap which our hooks will catch
215 // recursive_insert allows us to buffer info about these mmap calls.
216 // Note that Init() can be (and is) sometimes called
217 // already from within an mmap/sbrk hook.
218 recursive_insert = true;
219 arena_ = LowLevelAlloc::NewArena(0, LowLevelAlloc::DefaultArena());
220 recursive_insert = false;
221 HandleSavedRegionsLocked(&InsertRegionLocked); // flush the buffered ones
222 // Can't instead use HandleSavedRegionsLocked(&DoInsertRegionLocked) before
223 // recursive_insert = false; as InsertRegionLocked will also construct
224 // regions_ on demand for us.
225 if (use_buckets) {
226 const int table_bytes = kHashTableSize * sizeof(*bucket_table_);
227 recursive_insert = true;
228 bucket_table_ = static_cast<HeapProfileBucket**>(
229 MyAllocator::Allocate(table_bytes));
230 recursive_insert = false;
231 memset(bucket_table_, 0, table_bytes);
232 num_buckets_ = 0;
234 if (regions_ == NULL) // init regions_
235 InitRegionSetLocked();
236 Unlock();
237 RAW_VLOG(10, "MemoryRegionMap Init done");
240 bool MemoryRegionMap::Shutdown() {
241 RAW_VLOG(10, "MemoryRegionMap Shutdown");
242 Lock();
243 RAW_CHECK(client_count_ > 0, "");
244 client_count_ -= 1;
245 if (client_count_ != 0) { // not last client; need not really shutdown
246 Unlock();
247 RAW_VLOG(10, "MemoryRegionMap Shutdown decrement done");
248 return true;
250 if (bucket_table_ != NULL) {
251 for (int i = 0; i < kHashTableSize; i++) {
252 for (HeapProfileBucket* curr = bucket_table_[i]; curr != 0; /**/) {
253 HeapProfileBucket* bucket = curr;
254 curr = curr->next;
255 MyAllocator::Free(bucket->stack, 0);
256 MyAllocator::Free(bucket, 0);
259 MyAllocator::Free(bucket_table_, 0);
260 num_buckets_ = 0;
261 bucket_table_ = NULL;
263 RAW_CHECK(MallocHook::RemoveMmapHook(&MmapHook), "");
264 RAW_CHECK(MallocHook::RemoveMremapHook(&MremapHook), "");
265 RAW_CHECK(MallocHook::RemoveSbrkHook(&SbrkHook), "");
266 RAW_CHECK(MallocHook::RemoveMunmapHook(&MunmapHook), "");
267 if (regions_) regions_->~RegionSet();
268 regions_ = NULL;
269 bool deleted_arena = LowLevelAlloc::DeleteArena(arena_);
270 if (deleted_arena) {
271 arena_ = 0;
272 } else {
273 RAW_LOG(WARNING, "Can't delete LowLevelAlloc arena: it's being used");
275 Unlock();
276 RAW_VLOG(10, "MemoryRegionMap Shutdown done");
277 return deleted_arena;
280 bool MemoryRegionMap::IsRecordingLocked() {
281 RAW_CHECK(LockIsHeld(), "should be held (by this thread)");
282 return client_count_ > 0;
285 // Invariants (once libpthread_initialized is true):
286 // * While lock_ is not held, recursion_count_ is 0 (and
287 // lock_owner_tid_ is the previous owner, but we don't rely on
288 // that).
289 // * recursion_count_ and lock_owner_tid_ are only written while
290 // both lock_ and owner_lock_ are held. They may be read under
291 // just owner_lock_.
292 // * At entry and exit of Lock() and Unlock(), the current thread
293 // owns lock_ iff pthread_equal(lock_owner_tid_, pthread_self())
294 // && recursion_count_ > 0.
295 void MemoryRegionMap::Lock() {
297 SpinLockHolder l(&owner_lock_);
298 if (recursion_count_ > 0 && current_thread_is(lock_owner_tid_)) {
299 RAW_CHECK(lock_.IsHeld(), "Invariants violated");
300 recursion_count_++;
301 RAW_CHECK(recursion_count_ <= 5,
302 "recursive lock nesting unexpectedly deep");
303 return;
306 lock_.Lock();
308 SpinLockHolder l(&owner_lock_);
309 RAW_CHECK(recursion_count_ == 0,
310 "Last Unlock didn't reset recursion_count_");
311 if (libpthread_initialized)
312 lock_owner_tid_ = pthread_self();
313 recursion_count_ = 1;
317 void MemoryRegionMap::Unlock() {
318 SpinLockHolder l(&owner_lock_);
319 RAW_CHECK(recursion_count_ > 0, "unlock when not held");
320 RAW_CHECK(lock_.IsHeld(),
321 "unlock when not held, and recursion_count_ is wrong");
322 RAW_CHECK(current_thread_is(lock_owner_tid_), "unlock by non-holder");
323 recursion_count_--;
324 if (recursion_count_ == 0) {
325 lock_.Unlock();
329 bool MemoryRegionMap::LockIsHeld() {
330 SpinLockHolder l(&owner_lock_);
331 return lock_.IsHeld() && current_thread_is(lock_owner_tid_);
334 const MemoryRegionMap::Region*
335 MemoryRegionMap::DoFindRegionLocked(uintptr_t addr) {
336 RAW_CHECK(LockIsHeld(), "should be held (by this thread)");
337 if (regions_ != NULL) {
338 Region sample;
339 sample.SetRegionSetKey(addr);
340 RegionSet::iterator region = regions_->lower_bound(sample);
341 if (region != regions_->end()) {
342 RAW_CHECK(addr <= region->end_addr, "");
343 if (region->start_addr <= addr && addr < region->end_addr) {
344 return &(*region);
348 return NULL;
351 bool MemoryRegionMap::FindRegion(uintptr_t addr, Region* result) {
352 Lock();
353 const Region* region = DoFindRegionLocked(addr);
354 if (region != NULL) *result = *region; // create it as an independent copy
355 Unlock();
356 return region != NULL;
359 bool MemoryRegionMap::FindAndMarkStackRegion(uintptr_t stack_top,
360 Region* result) {
361 Lock();
362 const Region* region = DoFindRegionLocked(stack_top);
363 if (region != NULL) {
364 RAW_VLOG(10, "Stack at %p is inside region %p..%p",
365 reinterpret_cast<void*>(stack_top),
366 reinterpret_cast<void*>(region->start_addr),
367 reinterpret_cast<void*>(region->end_addr));
368 const_cast<Region*>(region)->set_is_stack(); // now we know
369 // cast is safe (set_is_stack does not change the set ordering key)
370 *result = *region; // create *result as an independent copy
372 Unlock();
373 return region != NULL;
376 HeapProfileBucket* MemoryRegionMap::GetBucket(int depth,
377 const void* const key[]) {
378 RAW_CHECK(LockIsHeld(), "should be held (by this thread)");
379 // Make hash-value
380 uintptr_t hash = 0;
381 for (int i = 0; i < depth; i++) {
382 hash += reinterpret_cast<uintptr_t>(key[i]);
383 hash += hash << 10;
384 hash ^= hash >> 6;
386 hash += hash << 3;
387 hash ^= hash >> 11;
389 // Lookup stack trace in table
390 unsigned int hash_index = (static_cast<unsigned int>(hash)) % kHashTableSize;
391 for (HeapProfileBucket* bucket = bucket_table_[hash_index];
392 bucket != 0;
393 bucket = bucket->next) {
394 if ((bucket->hash == hash) && (bucket->depth == depth) &&
395 std::equal(key, key + depth, bucket->stack)) {
396 return bucket;
400 // Create new bucket
401 const size_t key_size = sizeof(key[0]) * depth;
402 HeapProfileBucket* bucket;
403 if (recursive_insert) { // recursion: save in saved_buckets_
404 const void** key_copy = saved_buckets_keys_[saved_buckets_count_];
405 std::copy(key, key + depth, key_copy);
406 bucket = &saved_buckets_[saved_buckets_count_];
407 memset(bucket, 0, sizeof(*bucket));
408 ++saved_buckets_count_;
409 bucket->stack = key_copy;
410 bucket->next = NULL;
411 } else {
412 recursive_insert = true;
413 const void** key_copy = static_cast<const void**>(
414 MyAllocator::Allocate(key_size));
415 recursive_insert = false;
416 std::copy(key, key + depth, key_copy);
417 recursive_insert = true;
418 bucket = static_cast<HeapProfileBucket*>(
419 MyAllocator::Allocate(sizeof(HeapProfileBucket)));
420 recursive_insert = false;
421 memset(bucket, 0, sizeof(*bucket));
422 bucket->stack = key_copy;
423 bucket->next = bucket_table_[hash_index];
425 bucket->hash = hash;
426 bucket->depth = depth;
427 bucket_table_[hash_index] = bucket;
428 ++num_buckets_;
429 return bucket;
432 MemoryRegionMap::RegionIterator MemoryRegionMap::BeginRegionLocked() {
433 RAW_CHECK(LockIsHeld(), "should be held (by this thread)");
434 RAW_CHECK(regions_ != NULL, "");
435 return regions_->begin();
438 MemoryRegionMap::RegionIterator MemoryRegionMap::EndRegionLocked() {
439 RAW_CHECK(LockIsHeld(), "should be held (by this thread)");
440 RAW_CHECK(regions_ != NULL, "");
441 return regions_->end();
444 inline void MemoryRegionMap::DoInsertRegionLocked(const Region& region) {
445 RAW_VLOG(12, "Inserting region %p..%p from %p",
446 reinterpret_cast<void*>(region.start_addr),
447 reinterpret_cast<void*>(region.end_addr),
448 reinterpret_cast<void*>(region.caller()));
449 RegionSet::const_iterator i = regions_->lower_bound(region);
450 if (i != regions_->end() && i->start_addr <= region.start_addr) {
451 RAW_DCHECK(region.end_addr <= i->end_addr, ""); // lower_bound ensures this
452 return; // 'region' is a subset of an already recorded region; do nothing
453 // We can be stricter and allow this only when *i has been created via
454 // an mmap with MAP_NORESERVE flag set.
456 if (DEBUG_MODE) {
457 RAW_CHECK(i == regions_->end() || !region.Overlaps(*i),
458 "Wow, overlapping memory regions");
459 Region sample;
460 sample.SetRegionSetKey(region.start_addr);
461 i = regions_->lower_bound(sample);
462 RAW_CHECK(i == regions_->end() || !region.Overlaps(*i),
463 "Wow, overlapping memory regions");
465 region.AssertIsConsistent(); // just making sure
466 // This inserts and allocates permanent storage for region
467 // and its call stack data: it's safe to do it now:
468 regions_->insert(region);
469 RAW_VLOG(12, "Inserted region %p..%p :",
470 reinterpret_cast<void*>(region.start_addr),
471 reinterpret_cast<void*>(region.end_addr));
472 if (VLOG_IS_ON(12)) LogAllLocked();
475 // These variables are local to MemoryRegionMap::InsertRegionLocked()
476 // and MemoryRegionMap::HandleSavedRegionsLocked()
477 // and are file-level to ensure that they are initialized at load time.
479 // Number of unprocessed region inserts.
480 static int saved_regions_count = 0;
482 // Unprocessed inserts (must be big enough to hold all allocations that can
483 // be caused by a InsertRegionLocked call).
484 // Region has no constructor, so that c-tor execution does not interfere
485 // with the any-time use of the static memory behind saved_regions.
486 static MemoryRegionMap::Region saved_regions[20];
488 inline void MemoryRegionMap::HandleSavedRegionsLocked(
489 void (*insert_func)(const Region& region)) {
490 while (saved_regions_count > 0) {
491 // Making a local-var copy of the region argument to insert_func
492 // including its stack (w/o doing any memory allocations) is important:
493 // in many cases the memory in saved_regions
494 // will get written-to during the (*insert_func)(r) call below.
495 Region r = saved_regions[--saved_regions_count];
496 (*insert_func)(r);
500 void MemoryRegionMap::RestoreSavedBucketsLocked() {
501 RAW_CHECK(LockIsHeld(), "should be held (by this thread)");
502 while (saved_buckets_count_ > 0) {
503 HeapProfileBucket bucket = saved_buckets_[--saved_buckets_count_];
504 unsigned int hash_index =
505 static_cast<unsigned int>(bucket.hash) % kHashTableSize;
506 bool is_found = false;
507 for (HeapProfileBucket* curr = bucket_table_[hash_index];
508 curr != 0;
509 curr = curr->next) {
510 if ((curr->hash == bucket.hash) && (curr->depth == bucket.depth) &&
511 std::equal(bucket.stack, bucket.stack + bucket.depth, curr->stack)) {
512 curr->allocs += bucket.allocs;
513 curr->alloc_size += bucket.alloc_size;
514 curr->frees += bucket.frees;
515 curr->free_size += bucket.free_size;
516 is_found = true;
517 break;
520 if (is_found) continue;
522 const size_t key_size = sizeof(bucket.stack[0]) * bucket.depth;
523 const void** key_copy = static_cast<const void**>(
524 MyAllocator::Allocate(key_size));
525 std::copy(bucket.stack, bucket.stack + bucket.depth, key_copy);
526 HeapProfileBucket* new_bucket = static_cast<HeapProfileBucket*>(
527 MyAllocator::Allocate(sizeof(HeapProfileBucket)));
528 memset(new_bucket, 0, sizeof(*new_bucket));
529 new_bucket->hash = bucket.hash;
530 new_bucket->depth = bucket.depth;
531 new_bucket->stack = key_copy;
532 new_bucket->next = bucket_table_[hash_index];
533 bucket_table_[hash_index] = new_bucket;
534 ++num_buckets_;
538 inline void MemoryRegionMap::InitRegionSetLocked() {
539 RAW_VLOG(12, "Initializing region set");
540 regions_ = regions_rep.region_set();
541 recursive_insert = true;
542 new(regions_) RegionSet();
543 HandleSavedRegionsLocked(&DoInsertRegionLocked);
544 recursive_insert = false;
547 inline void MemoryRegionMap::InsertRegionLocked(const Region& region) {
548 RAW_CHECK(LockIsHeld(), "should be held (by this thread)");
549 // We can be called recursively, because RegionSet constructor
550 // and DoInsertRegionLocked() (called below) can call the allocator.
551 // recursive_insert tells us if that's the case. When this happens,
552 // region insertion information is recorded in saved_regions[],
553 // and taken into account when the recursion unwinds.
554 // Do the insert:
555 if (recursive_insert) { // recursion: save in saved_regions
556 RAW_VLOG(12, "Saving recursive insert of region %p..%p from %p",
557 reinterpret_cast<void*>(region.start_addr),
558 reinterpret_cast<void*>(region.end_addr),
559 reinterpret_cast<void*>(region.caller()));
560 RAW_CHECK(saved_regions_count < arraysize(saved_regions), "");
561 // Copy 'region' to saved_regions[saved_regions_count]
562 // together with the contents of its call_stack,
563 // then increment saved_regions_count.
564 saved_regions[saved_regions_count++] = region;
565 } else { // not a recusrive call
566 if (regions_ == NULL) // init regions_
567 InitRegionSetLocked();
568 recursive_insert = true;
569 // Do the actual insertion work to put new regions into regions_:
570 DoInsertRegionLocked(region);
571 HandleSavedRegionsLocked(&DoInsertRegionLocked);
572 recursive_insert = false;
576 // We strip out different number of stack frames in debug mode
577 // because less inlining happens in that case
578 #ifdef NDEBUG
579 static const int kStripFrames = 1;
580 #else
581 static const int kStripFrames = 3;
582 #endif
584 void MemoryRegionMap::RecordRegionAddition(const void* start, size_t size) {
585 // Record start/end info about this memory acquisition call in a new region:
586 Region region;
587 region.Create(start, size);
588 // First get the call stack info into the local varible 'region':
589 const int depth =
590 max_stack_depth_ > 0
591 ? MallocHook::GetCallerStackTrace(const_cast<void**>(region.call_stack),
592 max_stack_depth_, kStripFrames + 1)
593 : 0;
594 region.set_call_stack_depth(depth); // record stack info fully
595 RAW_VLOG(10, "New global region %p..%p from %p",
596 reinterpret_cast<void*>(region.start_addr),
597 reinterpret_cast<void*>(region.end_addr),
598 reinterpret_cast<void*>(region.caller()));
599 // Note: none of the above allocates memory.
600 Lock(); // recursively lock
601 map_size_ += size;
602 InsertRegionLocked(region);
603 // This will (eventually) allocate storage for and copy over the stack data
604 // from region.call_stack_data_ that is pointed by region.call_stack().
605 if (bucket_table_ != NULL) {
606 HeapProfileBucket* b = GetBucket(depth, region.call_stack);
607 ++b->allocs;
608 b->alloc_size += size;
609 if (!recursive_insert) {
610 recursive_insert = true;
611 RestoreSavedBucketsLocked();
612 recursive_insert = false;
615 Unlock();
618 void MemoryRegionMap::RecordRegionRemoval(const void* start, size_t size) {
619 Lock();
620 if (recursive_insert) {
621 // First remove the removed region from saved_regions, if it's
622 // there, to prevent overrunning saved_regions in recursive
623 // map/unmap call sequences, and also from later inserting regions
624 // which have already been unmapped.
625 uintptr_t start_addr = reinterpret_cast<uintptr_t>(start);
626 uintptr_t end_addr = start_addr + size;
627 int put_pos = 0;
628 int old_count = saved_regions_count;
629 for (int i = 0; i < old_count; ++i, ++put_pos) {
630 Region& r = saved_regions[i];
631 if (r.start_addr == start_addr && r.end_addr == end_addr) {
632 // An exact match, so it's safe to remove.
633 RecordRegionRemovalInBucket(r.call_stack_depth, r.call_stack, size);
634 --saved_regions_count;
635 --put_pos;
636 RAW_VLOG(10, ("Insta-Removing saved region %p..%p; "
637 "now have %d saved regions"),
638 reinterpret_cast<void*>(start_addr),
639 reinterpret_cast<void*>(end_addr),
640 saved_regions_count);
641 } else {
642 if (put_pos < i) {
643 saved_regions[put_pos] = saved_regions[i];
648 if (regions_ == NULL) { // We must have just unset the hooks,
649 // but this thread was already inside the hook.
650 Unlock();
651 return;
653 if (!recursive_insert) {
654 HandleSavedRegionsLocked(&InsertRegionLocked);
656 // first handle adding saved regions if any
657 uintptr_t start_addr = reinterpret_cast<uintptr_t>(start);
658 uintptr_t end_addr = start_addr + size;
659 // subtract start_addr, end_addr from all the regions
660 RAW_VLOG(10, "Removing global region %p..%p; have %" PRIuS " regions",
661 reinterpret_cast<void*>(start_addr),
662 reinterpret_cast<void*>(end_addr),
663 regions_->size());
664 Region sample;
665 sample.SetRegionSetKey(start_addr);
666 // Only iterate over the regions that might overlap start_addr..end_addr:
667 for (RegionSet::iterator region = regions_->lower_bound(sample);
668 region != regions_->end() && region->start_addr < end_addr;
669 /*noop*/) {
670 RAW_VLOG(13, "Looking at region %p..%p",
671 reinterpret_cast<void*>(region->start_addr),
672 reinterpret_cast<void*>(region->end_addr));
673 if (start_addr <= region->start_addr &&
674 region->end_addr <= end_addr) { // full deletion
675 RAW_VLOG(12, "Deleting region %p..%p",
676 reinterpret_cast<void*>(region->start_addr),
677 reinterpret_cast<void*>(region->end_addr));
678 RecordRegionRemovalInBucket(region->call_stack_depth, region->call_stack,
679 region->end_addr - region->start_addr);
680 RegionSet::iterator d = region;
681 ++region;
682 regions_->erase(d);
683 continue;
684 } else if (region->start_addr < start_addr &&
685 end_addr < region->end_addr) { // cutting-out split
686 RAW_VLOG(12, "Splitting region %p..%p in two",
687 reinterpret_cast<void*>(region->start_addr),
688 reinterpret_cast<void*>(region->end_addr));
689 RecordRegionRemovalInBucket(region->call_stack_depth, region->call_stack,
690 end_addr - start_addr);
691 // Make another region for the start portion:
692 // The new region has to be the start portion because we can't
693 // just modify region->end_addr as it's the sorting key.
694 Region r = *region;
695 r.set_end_addr(start_addr);
696 InsertRegionLocked(r);
697 // cut *region from start:
698 const_cast<Region&>(*region).set_start_addr(end_addr);
699 } else if (end_addr > region->start_addr &&
700 start_addr <= region->start_addr) { // cut from start
701 RAW_VLOG(12, "Start-chopping region %p..%p",
702 reinterpret_cast<void*>(region->start_addr),
703 reinterpret_cast<void*>(region->end_addr));
704 RecordRegionRemovalInBucket(region->call_stack_depth, region->call_stack,
705 end_addr - region->start_addr);
706 const_cast<Region&>(*region).set_start_addr(end_addr);
707 } else if (start_addr > region->start_addr &&
708 start_addr < region->end_addr) { // cut from end
709 RAW_VLOG(12, "End-chopping region %p..%p",
710 reinterpret_cast<void*>(region->start_addr),
711 reinterpret_cast<void*>(region->end_addr));
712 RecordRegionRemovalInBucket(region->call_stack_depth, region->call_stack,
713 region->end_addr - start_addr);
714 // Can't just modify region->end_addr (it's the sorting key):
715 Region r = *region;
716 r.set_end_addr(start_addr);
717 RegionSet::iterator d = region;
718 ++region;
719 // It's safe to erase before inserting since r is independent of *d:
720 // r contains an own copy of the call stack:
721 regions_->erase(d);
722 InsertRegionLocked(r);
723 continue;
725 ++region;
727 RAW_VLOG(12, "Removed region %p..%p; have %" PRIuS " regions",
728 reinterpret_cast<void*>(start_addr),
729 reinterpret_cast<void*>(end_addr),
730 regions_->size());
731 if (VLOG_IS_ON(12)) LogAllLocked();
732 unmap_size_ += size;
733 Unlock();
736 void MemoryRegionMap::RecordRegionRemovalInBucket(int depth,
737 const void* const stack[],
738 size_t size) {
739 RAW_CHECK(LockIsHeld(), "should be held (by this thread)");
740 if (bucket_table_ == NULL) return;
741 HeapProfileBucket* b = GetBucket(depth, stack);
742 ++b->frees;
743 b->free_size += size;
746 void MemoryRegionMap::MmapHook(const void* result,
747 const void* start, size_t size,
748 int prot, int flags,
749 int fd, off_t offset) {
750 // TODO(maxim): replace all 0x%"PRIxS" by %p when RAW_VLOG uses a safe
751 // snprintf reimplementation that does not malloc to pretty-print NULL
752 RAW_VLOG(10, "MMap = 0x%" PRIxPTR " of %" PRIuS " at %" PRIu64 " "
753 "prot %d flags %d fd %d offs %" PRId64,
754 reinterpret_cast<uintptr_t>(result), size,
755 reinterpret_cast<uint64>(start), prot, flags, fd,
756 static_cast<int64>(offset));
757 if (result != reinterpret_cast<void*>(MAP_FAILED) && size != 0) {
758 RecordRegionAddition(result, size);
762 void MemoryRegionMap::MunmapHook(const void* ptr, size_t size) {
763 RAW_VLOG(10, "MUnmap of %p %" PRIuS, ptr, size);
764 if (size != 0) {
765 RecordRegionRemoval(ptr, size);
769 void MemoryRegionMap::MremapHook(const void* result,
770 const void* old_addr, size_t old_size,
771 size_t new_size, int flags,
772 const void* new_addr) {
773 RAW_VLOG(10, "MRemap = 0x%" PRIxPTR " of 0x%" PRIxPTR " %" PRIuS " "
774 "to %" PRIuS " flags %d new_addr=0x%" PRIxPTR,
775 (uintptr_t)result, (uintptr_t)old_addr,
776 old_size, new_size, flags,
777 flags & MREMAP_FIXED ? (uintptr_t)new_addr : 0);
778 if (result != reinterpret_cast<void*>(-1)) {
779 RecordRegionRemoval(old_addr, old_size);
780 RecordRegionAddition(result, new_size);
784 extern "C" void* __sbrk(ptrdiff_t increment); // defined in libc
786 void MemoryRegionMap::SbrkHook(const void* result, ptrdiff_t increment) {
787 RAW_VLOG(10, "Sbrk = 0x%" PRIxPTR " of %" PRIdS,
788 (uintptr_t)result, increment);
789 if (result != reinterpret_cast<void*>(-1)) {
790 if (increment > 0) {
791 void* new_end = sbrk(0);
792 RecordRegionAddition(result, reinterpret_cast<uintptr_t>(new_end) -
793 reinterpret_cast<uintptr_t>(result));
794 } else if (increment < 0) {
795 void* new_end = sbrk(0);
796 RecordRegionRemoval(new_end, reinterpret_cast<uintptr_t>(result) -
797 reinterpret_cast<uintptr_t>(new_end));
802 void MemoryRegionMap::LogAllLocked() {
803 RAW_CHECK(LockIsHeld(), "should be held (by this thread)");
804 RAW_LOG(INFO, "List of regions:");
805 uintptr_t previous = 0;
806 for (RegionSet::const_iterator r = regions_->begin();
807 r != regions_->end(); ++r) {
808 RAW_LOG(INFO, "Memory region 0x%" PRIxPTR "..0x%" PRIxPTR " "
809 "from 0x%" PRIxPTR " stack=%d",
810 r->start_addr, r->end_addr, r->caller(), r->is_stack);
811 RAW_CHECK(previous < r->end_addr, "wow, we messed up the set order");
812 // this must be caused by uncontrolled recursive operations on regions_
813 previous = r->end_addr;
815 RAW_LOG(INFO, "End of regions list");