/******************************************************************************/
#ifdef JEMALLOC_H_TYPES
+#define LARGE_MINCLASS (ZU(1) << LG_LARGE_MINCLASS)
+
/* Maximum number of regions in one run. */
#define LG_RUN_MAXREGS (LG_PAGE - LG_TINY_MIN)
#define RUN_MAXREGS (1U << LG_RUN_MAXREGS)
/*
* The minimum ratio of active:dirty pages per arena is computed as:
*
- * (nactive >> opt_lg_dirty_mult) >= ndirty
+ * (nactive >> lg_dirty_mult) >= ndirty
*
- * So, supposing that opt_lg_dirty_mult is 3, there can be no less than 8 times
- * as many active pages as dirty pages.
+ * So, supposing that lg_dirty_mult is 3, there can be no less than 8 times as
+ * many active pages as dirty pages.
*/
#define LG_DIRTY_MULT_DEFAULT 3
+typedef enum {
+ purge_mode_ratio = 0,
+ purge_mode_decay = 1,
+
+ purge_mode_limit = 2
+} purge_mode_t;
+#define PURGE_DEFAULT purge_mode_ratio
+/* Default decay time in seconds. */
+#define DECAY_TIME_DEFAULT 10
+/* Number of event ticks between time checks. */
+#define DECAY_NTICKS_PER_UPDATE 1000
+
+typedef struct arena_runs_dirty_link_s arena_runs_dirty_link_t;
typedef struct arena_run_s arena_run_t;
typedef struct arena_chunk_map_bits_s arena_chunk_map_bits_t;
typedef struct arena_chunk_map_misc_s arena_chunk_map_misc_t;
typedef struct arena_bin_info_s arena_bin_info_t;
typedef struct arena_bin_s arena_bin_t;
typedef struct arena_s arena_t;
+typedef struct arena_tdata_s arena_tdata_t;
#endif /* JEMALLOC_H_TYPES */
/******************************************************************************/
#ifdef JEMALLOC_H_STRUCTS
+#ifdef JEMALLOC_ARENA_STRUCTS_A
struct arena_run_s {
- /* Bin this run is associated with. */
- arena_bin_t *bin;
-
- /* Index of next region that has never been allocated, or nregs. */
- uint32_t nextind;
+ /* Index of bin this run is associated with. */
+ szind_t binind;
/* Number of free regions in run. */
unsigned nfree;
* Run address (or size) and various flags are stored together. The bit
* layout looks like (assuming 32-bit system):
*
- * ???????? ???????? ????nnnn nnnndula
+ * ???????? ???????? ???nnnnn nnndumla
*
* ? : Unallocated: Run address for first/last pages, unset for internal
* pages.
* Small: Run page offset.
- * Large: Run size for first page, unset for trailing pages.
+ * Large: Run page count for first page, unset for trailing pages.
* n : binind for small size class, BININD_INVALID for large size class.
* d : dirty?
* u : unzeroed?
+ * m : decommitted?
* l : large?
* a : allocated?
*
* x : don't care
* - : 0
* + : 1
- * [DULA] : bit set
- * [dula] : bit unset
+ * [DUMLA] : bit set
+ * [dumla] : bit unset
*
* Unallocated (clean):
- * ssssssss ssssssss ssss++++ ++++du-a
- * xxxxxxxx xxxxxxxx xxxxxxxx xxxx-Uxx
- * ssssssss ssssssss ssss++++ ++++dU-a
+ * ssssssss ssssssss sss+++++ +++dum-a
+ * xxxxxxxx xxxxxxxx xxxxxxxx xxx-Uxxx
+ * ssssssss ssssssss sss+++++ +++dUm-a
*
* Unallocated (dirty):
- * ssssssss ssssssss ssss++++ ++++D--a
+ * ssssssss ssssssss sss+++++ +++D-m-a
* xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx
- * ssssssss ssssssss ssss++++ ++++D--a
+ * ssssssss ssssssss sss+++++ +++D-m-a
*
* Small:
- * pppppppp pppppppp ppppnnnn nnnnd--A
- * pppppppp pppppppp ppppnnnn nnnn---A
- * pppppppp pppppppp ppppnnnn nnnnd--A
+ * pppppppp pppppppp pppnnnnn nnnd---A
+ * pppppppp pppppppp pppnnnnn nnn----A
+ * pppppppp pppppppp pppnnnnn nnnd---A
*
* Large:
- * ssssssss ssssssss ssss++++ ++++D-LA
+ * ssssssss ssssssss sss+++++ +++D--LA
* xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx
- * -------- -------- ----++++ ++++D-LA
+ * -------- -------- ---+++++ +++D--LA
*
- * Large (sampled, size <= PAGE):
- * ssssssss ssssssss ssssnnnn nnnnD-LA
+ * Large (sampled, size <= LARGE_MINCLASS):
+ * ssssssss ssssssss sssnnnnn nnnD--LA
+ * xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx
+ * -------- -------- ---+++++ +++D--LA
*
- * Large (not sampled, size == PAGE):
- * ssssssss ssssssss ssss++++ ++++D-LA
+ * Large (not sampled, size == LARGE_MINCLASS):
+ * ssssssss ssssssss sss+++++ +++D--LA
+ * xxxxxxxx xxxxxxxx xxxxxxxx xxxxxxxx
+ * -------- -------- ---+++++ +++D--LA
*/
size_t bits;
-#define CHUNK_MAP_BININD_SHIFT 4
+#define CHUNK_MAP_ALLOCATED ((size_t)0x01U)
+#define CHUNK_MAP_LARGE ((size_t)0x02U)
+#define CHUNK_MAP_STATE_MASK ((size_t)0x3U)
+
+#define CHUNK_MAP_DECOMMITTED ((size_t)0x04U)
+#define CHUNK_MAP_UNZEROED ((size_t)0x08U)
+#define CHUNK_MAP_DIRTY ((size_t)0x10U)
+#define CHUNK_MAP_FLAGS_MASK ((size_t)0x1cU)
+
+#define CHUNK_MAP_BININD_SHIFT 5
#define BININD_INVALID ((size_t)0xffU)
-/* CHUNK_MAP_BININD_MASK == (BININD_INVALID << CHUNK_MAP_BININD_SHIFT) */
-#define CHUNK_MAP_BININD_MASK ((size_t)0xff0U)
+#define CHUNK_MAP_BININD_MASK (BININD_INVALID << CHUNK_MAP_BININD_SHIFT)
#define CHUNK_MAP_BININD_INVALID CHUNK_MAP_BININD_MASK
-#define CHUNK_MAP_FLAGS_MASK ((size_t)0xcU)
-#define CHUNK_MAP_DIRTY ((size_t)0x8U)
-#define CHUNK_MAP_UNZEROED ((size_t)0x4U)
-#define CHUNK_MAP_LARGE ((size_t)0x2U)
-#define CHUNK_MAP_ALLOCATED ((size_t)0x1U)
-#define CHUNK_MAP_KEY CHUNK_MAP_ALLOCATED
+
+#define CHUNK_MAP_RUNIND_SHIFT (CHUNK_MAP_BININD_SHIFT + 8)
+#define CHUNK_MAP_SIZE_SHIFT (CHUNK_MAP_RUNIND_SHIFT - LG_PAGE)
+#define CHUNK_MAP_SIZE_MASK \
+ (~(CHUNK_MAP_BININD_MASK | CHUNK_MAP_FLAGS_MASK | CHUNK_MAP_STATE_MASK))
+};
+
+struct arena_runs_dirty_link_s {
+ qr(arena_runs_dirty_link_t) rd_link;
};
/*
*
* 1) arena_t's runs_avail tree.
* 2) arena_run_t conceptually uses this linkage for in-use non-full
- * runs, rather than directly embedding linkage.
+ * runs, rather than directly embedding linkage.
*/
rb_node(arena_chunk_map_misc_t) rb_link;
union {
/* Linkage for list of dirty runs. */
- ql_elm(arena_chunk_map_misc_t) dr_link;
+ arena_runs_dirty_link_t rd;
/* Profile counters, used for large object runs. */
- prof_tctx_t *prof_tctx;
+ union {
+ void *prof_tctx_pun;
+ prof_tctx_t *prof_tctx;
+ };
/* Small region run metadata. */
arena_run_t run;
};
};
-typedef rb_tree(arena_chunk_map_misc_t) arena_avail_tree_t;
typedef rb_tree(arena_chunk_map_misc_t) arena_run_tree_t;
-typedef ql_head(arena_chunk_map_misc_t) arena_chunk_miscelms_t;
+#endif /* JEMALLOC_ARENA_STRUCTS_A */
+#ifdef JEMALLOC_ARENA_STRUCTS_B
/* Arena chunk header. */
struct arena_chunk_s {
- /* Arena that owns the chunk. */
- arena_t *arena;
+ /*
+ * A pointer to the arena that owns the chunk is stored within the node.
+ * This field as a whole is used by chunks_rtree to support both
+ * ivsalloc() and core-based debugging.
+ */
+ extent_node_t node;
/*
* Map of pages within chunk that keeps track of free/large/small. The
*/
struct arena_bin_info_s {
/* Size of regions in a run for this bin's size class. */
- size_t reg_size;
+ size_t reg_size;
/* Redzone size. */
- size_t redzone_size;
+ size_t redzone_size;
/* Interval between regions (reg_size + (redzone_size << 1)). */
- size_t reg_interval;
+ size_t reg_interval;
/* Total size of a run for this bin's size class. */
- size_t run_size;
+ size_t run_size;
/* Total number of regions in a run for this bin's size class. */
- uint32_t nregs;
+ uint32_t nregs;
/*
* Metadata used to manipulate bitmaps for runs associated with this
* bin.
*/
- bitmap_info_t bitmap_info;
+ bitmap_info_t bitmap_info;
/* Offset of first region in a run for this bin's size class. */
- uint32_t reg0_offset;
+ uint32_t reg0_offset;
};
struct arena_bin_s {
* which may be acquired while holding one or more bin locks, but not
* vise versa.
*/
- malloc_mutex_t lock;
+ malloc_mutex_t lock;
/*
* Current run being used to service allocations of this bin's size
* class.
*/
- arena_run_t *runcur;
+ arena_run_t *runcur;
/*
* Tree of non-full runs. This tree is used when looking for an
* objects packed well, and it can also help reduce the number of
* almost-empty chunks.
*/
- arena_run_tree_t runs;
+ arena_run_tree_t runs;
/* Bin statistics. */
- malloc_bin_stats_t stats;
+ malloc_bin_stats_t stats;
};
struct arena_s {
/*
* Number of threads currently assigned to this arena. This field is
- * protected by arenas_lock.
+ * synchronized via atomic operations.
*/
unsigned nthreads;
/*
* There are three classes of arena operations from a locking
* perspective:
- * 1) Thread asssignment (modifies nthreads) is protected by
- * arenas_lock.
+ * 1) Thread assignment (modifies nthreads) is synchronized via atomics.
* 2) Bin-related operations are protected by bin locks.
* 3) Chunk- and run-related operations are protected by this mutex.
*/
arena_stats_t stats;
/*
* List of tcaches for extant threads associated with this arena.
- * Stats from these are merged incrementally, and at exit.
+ * Stats from these are merged incrementally, and at exit if
+ * opt_stats_print is enabled.
*/
ql_head(tcache_t) tcache_ql;
uint64_t prof_accumbytes;
+ /*
+ * PRNG state for cache index randomization of large allocation base
+ * pointers.
+ */
+ uint64_t offset_state;
+
dss_prec_t dss_prec;
/*
*/
arena_chunk_t *spare;
+ /* Minimum ratio (log base 2) of nactive:ndirty. */
+ ssize_t lg_dirty_mult;
+
+ /* True if a thread is currently executing arena_purge_to_limit(). */
+ bool purging;
+
/* Number of pages in active runs and huge regions. */
size_t nactive;
size_t ndirty;
/*
- * Size/address-ordered trees of this arena's available runs. The trees
- * are used for first-best-fit run allocation.
+ * Unused dirty memory this arena manages. Dirty memory is conceptually
+ * tracked as an arbitrarily interleaved LRU of dirty runs and cached
+ * chunks, but the list linkage is actually semi-duplicated in order to
+ * avoid extra arena_chunk_map_misc_t space overhead.
+ *
+ * LRU-----------------------------------------------------------MRU
+ *
+ * /-- arena ---\
+ * | |
+ * | |
+ * |------------| /- chunk -\
+ * ...->|chunks_cache|<--------------------------->| /----\ |<--...
+ * |------------| | |node| |
+ * | | | | | |
+ * | | /- run -\ /- run -\ | | | |
+ * | | | | | | | | | |
+ * | | | | | | | | | |
+ * |------------| |-------| |-------| | |----| |
+ * ...->|runs_dirty |<-->|rd |<-->|rd |<---->|rd |<----...
+ * |------------| |-------| |-------| | |----| |
+ * | | | | | | | | | |
+ * | | | | | | | \----/ |
+ * | | \-------/ \-------/ | |
+ * | | | |
+ * | | | |
+ * \------------/ \---------/
+ */
+ arena_runs_dirty_link_t runs_dirty;
+ extent_node_t chunks_cache;
+
+ /*
+ * Approximate time in seconds from the creation of a set of unused
+ * dirty pages until an equivalent set of unused dirty pages is purged
+ * and/or reused.
+ */
+ ssize_t decay_time;
+ /* decay_time / SMOOTHSTEP_NSTEPS. */
+ nstime_t decay_interval;
+ /*
+ * Time at which the current decay interval logically started. We do
+ * not actually advance to a new epoch until sometime after it starts
+ * because of scheduling and computation delays, and it is even possible
+ * to completely skip epochs. In all cases, during epoch advancement we
+ * merge all relevant activity into the most recently recorded epoch.
+ */
+ nstime_t decay_epoch;
+ /* decay_deadline randomness generator. */
+ uint64_t decay_jitter_state;
+ /*
+ * Deadline for current epoch. This is the sum of decay_interval and
+ * per epoch jitter which is a uniform random variable in
+ * [0..decay_interval). Epochs always advance by precise multiples of
+ * decay_interval, but we randomize the deadline to reduce the
+ * likelihood of arenas purging in lockstep.
+ */
+ nstime_t decay_deadline;
+ /*
+ * Number of dirty pages at beginning of current epoch. During epoch
+ * advancement we use the delta between decay_ndirty and ndirty to
+ * determine how many dirty pages, if any, were generated, and record
+ * the result in decay_backlog.
+ */
+ size_t decay_ndirty;
+ /*
+ * Memoized result of arena_decay_backlog_npages_limit() corresponding
+ * to the current contents of decay_backlog, i.e. the limit on how many
+ * pages are allowed to exist for the decay epochs.
+ */
+ size_t decay_backlog_npages_limit;
+ /*
+ * Trailing log of how many unused dirty pages were generated during
+ * each of the past SMOOTHSTEP_NSTEPS decay epochs, where the last
+ * element is the most recent epoch. Corresponding epoch times are
+ * relative to decay_epoch.
*/
- arena_avail_tree_t runs_avail;
+ size_t decay_backlog[SMOOTHSTEP_NSTEPS];
- /* List of dirty runs this arena manages. */
- arena_chunk_miscelms_t runs_dirty;
+ /* Extant huge allocations. */
+ ql_head(extent_node_t) huge;
+ /* Synchronizes all huge allocation/update/deallocation. */
+ malloc_mutex_t huge_mtx;
/*
- * user-configureable chunk allocation and deallocation functions.
+ * Trees of chunks that were previously allocated (trees differ only in
+ * node ordering). These are used when allocating chunks, in an attempt
+ * to re-use address space. Depending on function, different tree
+ * orderings are needed, which is why there are two trees with the same
+ * contents.
*/
- chunk_alloc_t *chunk_alloc;
- chunk_dalloc_t *chunk_dalloc;
+ extent_tree_t chunks_szad_cached;
+ extent_tree_t chunks_ad_cached;
+ extent_tree_t chunks_szad_retained;
+ extent_tree_t chunks_ad_retained;
+
+ malloc_mutex_t chunks_mtx;
+ /* Cache of nodes that were allocated via base_alloc(). */
+ ql_head(extent_node_t) node_cache;
+ malloc_mutex_t node_cache_mtx;
+
+ /* User-configurable chunk hook functions. */
+ chunk_hooks_t chunk_hooks;
/* bins is used to store trees of free regions. */
arena_bin_t bins[NBINS];
+
+ /*
+ * Quantized address-ordered trees of this arena's available runs. The
+ * trees are used for first-best-fit run allocation.
+ */
+ arena_run_tree_t runs_avail[1]; /* Dynamically sized. */
};
+/* Used in conjunction with tsd for fast arena-related context lookup. */
+struct arena_tdata_s {
+ ticker_t decay_ticker;
+};
+#endif /* JEMALLOC_ARENA_STRUCTS_B */
+
#endif /* JEMALLOC_H_STRUCTS */
/******************************************************************************/
#ifdef JEMALLOC_H_EXTERNS
-extern ssize_t opt_lg_dirty_mult;
-/*
- * small_size2bin_tab is a compact lookup table that rounds request sizes up to
- * size classes. In order to reduce cache footprint, the table is compressed,
- * and all accesses are via small_size2bin().
- */
-extern uint8_t const small_size2bin_tab[];
-/*
- * small_bin2size_tab duplicates information in arena_bin_info, but in a const
- * array, for which it is easier for the compiler to optimize repeated
- * dereferences.
- */
-extern uint32_t const small_bin2size_tab[NBINS];
+static const size_t large_pad =
+#ifdef JEMALLOC_CACHE_OBLIVIOUS
+ PAGE
+#else
+ 0
+#endif
+ ;
+
+extern purge_mode_t opt_purge;
+extern const char *purge_mode_names[];
+extern ssize_t opt_lg_dirty_mult;
+extern ssize_t opt_decay_time;
extern arena_bin_info_t arena_bin_info[NBINS];
-/* Number of large size classes. */
-#define nlclasses (chunk_npages - map_bias)
+extern size_t map_bias; /* Number of arena chunk header pages. */
+extern size_t map_misc_offset;
+extern size_t arena_maxrun; /* Max run size for arenas. */
+extern size_t large_maxclass; /* Max large size class. */
+extern size_t run_quantize_max; /* Max run_quantize_*() input. */
+extern unsigned nlclasses; /* Number of large size classes. */
+extern unsigned nhclasses; /* Number of huge size classes. */
-void *arena_chunk_alloc_huge(arena_t *arena, void *new_addr, size_t size,
- size_t alignment, bool *zero);
-void arena_chunk_dalloc_huge(arena_t *arena, void *chunk, size_t size);
-void arena_purge_all(arena_t *arena);
-void arena_tcache_fill_small(arena_t *arena, tcache_bin_t *tbin,
- size_t binind, uint64_t prof_accumbytes);
+#ifdef JEMALLOC_JET
+typedef size_t (run_quantize_t)(size_t);
+extern run_quantize_t *run_quantize_floor;
+extern run_quantize_t *run_quantize_ceil;
+#endif
+void arena_chunk_cache_maybe_insert(arena_t *arena, extent_node_t *node,
+ bool cache);
+void arena_chunk_cache_maybe_remove(arena_t *arena, extent_node_t *node,
+ bool cache);
+extent_node_t *arena_node_alloc(arena_t *arena);
+void arena_node_dalloc(arena_t *arena, extent_node_t *node);
+void *arena_chunk_alloc_huge(arena_t *arena, size_t usize, size_t alignment,
+ bool *zero);
+void arena_chunk_dalloc_huge(arena_t *arena, void *chunk, size_t usize);
+void arena_chunk_ralloc_huge_similar(arena_t *arena, void *chunk,
+ size_t oldsize, size_t usize);
+void arena_chunk_ralloc_huge_shrink(arena_t *arena, void *chunk,
+ size_t oldsize, size_t usize);
+bool arena_chunk_ralloc_huge_expand(arena_t *arena, void *chunk,
+ size_t oldsize, size_t usize, bool *zero);
+ssize_t arena_lg_dirty_mult_get(arena_t *arena);
+bool arena_lg_dirty_mult_set(arena_t *arena, ssize_t lg_dirty_mult);
+ssize_t arena_decay_time_get(arena_t *arena);
+bool arena_decay_time_set(arena_t *arena, ssize_t decay_time);
+void arena_maybe_purge(arena_t *arena);
+void arena_purge(arena_t *arena, bool all);
+void arena_tcache_fill_small(tsd_t *tsd, arena_t *arena, tcache_bin_t *tbin,
+ szind_t binind, uint64_t prof_accumbytes);
void arena_alloc_junk_small(void *ptr, arena_bin_info_t *bin_info,
bool zero);
#ifdef JEMALLOC_JET
void arena_dalloc_junk_small(void *ptr, arena_bin_info_t *bin_info);
#endif
void arena_quarantine_junk_small(void *ptr, size_t usize);
-void *arena_malloc_small(arena_t *arena, size_t size, bool zero);
-void *arena_malloc_large(arena_t *arena, size_t size, bool zero);
-void *arena_palloc(arena_t *arena, size_t size, size_t alignment, bool zero);
+void *arena_malloc_large(tsd_t *tsd, arena_t *arena, szind_t ind, bool zero);
+void *arena_malloc_hard(tsd_t *tsd, arena_t *arena, size_t size, szind_t ind,
+ bool zero, tcache_t *tcache);
+void *arena_palloc(tsd_t *tsd, arena_t *arena, size_t usize,
+ size_t alignment, bool zero, tcache_t *tcache);
void arena_prof_promoted(const void *ptr, size_t size);
-void arena_dalloc_bin_locked(arena_t *arena, arena_chunk_t *chunk, void *ptr,
- arena_chunk_map_bits_t *bitselm);
+void arena_dalloc_bin_junked_locked(arena_t *arena, arena_chunk_t *chunk,
+ void *ptr, arena_chunk_map_bits_t *bitselm);
void arena_dalloc_bin(arena_t *arena, arena_chunk_t *chunk, void *ptr,
size_t pageind, arena_chunk_map_bits_t *bitselm);
-void arena_dalloc_small(arena_t *arena, arena_chunk_t *chunk, void *ptr,
- size_t pageind);
+void arena_dalloc_small(tsd_t *tsd, arena_t *arena, arena_chunk_t *chunk,
+ void *ptr, size_t pageind);
#ifdef JEMALLOC_JET
typedef void (arena_dalloc_junk_large_t)(void *, size_t);
extern arena_dalloc_junk_large_t *arena_dalloc_junk_large;
+#else
+void arena_dalloc_junk_large(void *ptr, size_t usize);
#endif
-void arena_dalloc_large_locked(arena_t *arena, arena_chunk_t *chunk,
+void arena_dalloc_large_junked_locked(arena_t *arena, arena_chunk_t *chunk,
+ void *ptr);
+void arena_dalloc_large(tsd_t *tsd, arena_t *arena, arena_chunk_t *chunk,
void *ptr);
-void arena_dalloc_large(arena_t *arena, arena_chunk_t *chunk, void *ptr);
#ifdef JEMALLOC_JET
typedef void (arena_ralloc_junk_large_t)(void *, size_t, size_t);
extern arena_ralloc_junk_large_t *arena_ralloc_junk_large;
#endif
-bool arena_ralloc_no_move(void *ptr, size_t oldsize, size_t size,
+bool arena_ralloc_no_move(tsd_t *tsd, void *ptr, size_t oldsize, size_t size,
size_t extra, bool zero);
void *arena_ralloc(tsd_t *tsd, arena_t *arena, void *ptr, size_t oldsize,
- size_t size, size_t extra, size_t alignment, bool zero,
- bool try_tcache_alloc, bool try_tcache_dalloc);
+ size_t size, size_t alignment, bool zero, tcache_t *tcache);
dss_prec_t arena_dss_prec_get(arena_t *arena);
bool arena_dss_prec_set(arena_t *arena, dss_prec_t dss_prec);
-void arena_stats_merge(arena_t *arena, const char **dss, size_t *nactive,
+ssize_t arena_lg_dirty_mult_default_get(void);
+bool arena_lg_dirty_mult_default_set(ssize_t lg_dirty_mult);
+ssize_t arena_decay_time_default_get(void);
+bool arena_decay_time_default_set(ssize_t decay_time);
+void arena_basic_stats_merge(arena_t *arena, unsigned *nthreads,
+ const char **dss, ssize_t *lg_dirty_mult, ssize_t *decay_time,
+ size_t *nactive, size_t *ndirty);
+void arena_stats_merge(arena_t *arena, unsigned *nthreads, const char **dss,
+ ssize_t *lg_dirty_mult, ssize_t *decay_time, size_t *nactive,
size_t *ndirty, arena_stats_t *astats, malloc_bin_stats_t *bstats,
- malloc_large_stats_t *lstats);
-bool arena_new(arena_t *arena, unsigned ind);
-void arena_boot(void);
+ malloc_large_stats_t *lstats, malloc_huge_stats_t *hstats);
+unsigned arena_nthreads_get(arena_t *arena);
+void arena_nthreads_inc(arena_t *arena);
+void arena_nthreads_dec(arena_t *arena);
+arena_t *arena_new(unsigned ind);
+bool arena_boot(void);
void arena_prefork(arena_t *arena);
void arena_postfork_parent(arena_t *arena);
void arena_postfork_child(arena_t *arena);
#ifdef JEMALLOC_H_INLINES
#ifndef JEMALLOC_ENABLE_INLINE
-size_t small_size2bin_compute(size_t size);
-size_t small_size2bin_lookup(size_t size);
-size_t small_size2bin(size_t size);
-size_t small_bin2size_compute(size_t binind);
-size_t small_bin2size_lookup(size_t binind);
-size_t small_bin2size(size_t binind);
-size_t small_s2u_compute(size_t size);
-size_t small_s2u_lookup(size_t size);
-size_t small_s2u(size_t size);
arena_chunk_map_bits_t *arena_bitselm_get(arena_chunk_t *chunk,
size_t pageind);
arena_chunk_map_misc_t *arena_miscelm_get(arena_chunk_t *chunk,
size_t pageind);
-size_t arena_miscelm_to_pageind(arena_chunk_map_misc_t *miscelm);
+size_t arena_miscelm_to_pageind(const arena_chunk_map_misc_t *miscelm);
void *arena_miscelm_to_rpages(arena_chunk_map_misc_t *miscelm);
+arena_chunk_map_misc_t *arena_rd_to_miscelm(arena_runs_dirty_link_t *rd);
arena_chunk_map_misc_t *arena_run_to_miscelm(arena_run_t *run);
size_t *arena_mapbitsp_get(arena_chunk_t *chunk, size_t pageind);
size_t arena_mapbitsp_read(size_t *mapbitsp);
size_t arena_mapbits_get(arena_chunk_t *chunk, size_t pageind);
+size_t arena_mapbits_size_decode(size_t mapbits);
size_t arena_mapbits_unallocated_size_get(arena_chunk_t *chunk,
size_t pageind);
size_t arena_mapbits_large_size_get(arena_chunk_t *chunk, size_t pageind);
size_t arena_mapbits_small_runind_get(arena_chunk_t *chunk, size_t pageind);
-size_t arena_mapbits_binind_get(arena_chunk_t *chunk, size_t pageind);
+szind_t arena_mapbits_binind_get(arena_chunk_t *chunk, size_t pageind);
size_t arena_mapbits_dirty_get(arena_chunk_t *chunk, size_t pageind);
size_t arena_mapbits_unzeroed_get(arena_chunk_t *chunk, size_t pageind);
+size_t arena_mapbits_decommitted_get(arena_chunk_t *chunk, size_t pageind);
size_t arena_mapbits_large_get(arena_chunk_t *chunk, size_t pageind);
size_t arena_mapbits_allocated_get(arena_chunk_t *chunk, size_t pageind);
void arena_mapbitsp_write(size_t *mapbitsp, size_t mapbits);
+size_t arena_mapbits_size_encode(size_t size);
void arena_mapbits_unallocated_set(arena_chunk_t *chunk, size_t pageind,
size_t size, size_t flags);
void arena_mapbits_unallocated_size_set(arena_chunk_t *chunk, size_t pageind,
size_t size);
+void arena_mapbits_internal_set(arena_chunk_t *chunk, size_t pageind,
+ size_t flags);
void arena_mapbits_large_set(arena_chunk_t *chunk, size_t pageind,
size_t size, size_t flags);
void arena_mapbits_large_binind_set(arena_chunk_t *chunk, size_t pageind,
- size_t binind);
+ szind_t binind);
void arena_mapbits_small_set(arena_chunk_t *chunk, size_t pageind,
- size_t runind, size_t binind, size_t flags);
-void arena_mapbits_unzeroed_set(arena_chunk_t *chunk, size_t pageind,
- size_t unzeroed);
+ size_t runind, szind_t binind, size_t flags);
+void arena_metadata_allocated_add(arena_t *arena, size_t size);
+void arena_metadata_allocated_sub(arena_t *arena, size_t size);
+size_t arena_metadata_allocated_get(arena_t *arena);
bool arena_prof_accum_impl(arena_t *arena, uint64_t accumbytes);
bool arena_prof_accum_locked(arena_t *arena, uint64_t accumbytes);
bool arena_prof_accum(arena_t *arena, uint64_t accumbytes);
-size_t arena_ptr_small_binind_get(const void *ptr, size_t mapbits);
-size_t arena_bin_index(arena_t *arena, arena_bin_t *bin);
-unsigned arena_run_regind(arena_run_t *run, arena_bin_info_t *bin_info,
+szind_t arena_ptr_small_binind_get(const void *ptr, size_t mapbits);
+szind_t arena_bin_index(arena_t *arena, arena_bin_t *bin);
+size_t arena_run_regind(arena_run_t *run, arena_bin_info_t *bin_info,
const void *ptr);
prof_tctx_t *arena_prof_tctx_get(const void *ptr);
-void arena_prof_tctx_set(const void *ptr, prof_tctx_t *tctx);
-void *arena_malloc(tsd_t *tsd, arena_t *arena, size_t size, bool zero,
- bool try_tcache);
+void arena_prof_tctx_set(const void *ptr, size_t usize, prof_tctx_t *tctx);
+void arena_prof_tctx_reset(const void *ptr, size_t usize,
+ const void *old_ptr, prof_tctx_t *old_tctx);
+void arena_decay_ticks(tsd_t *tsd, arena_t *arena, unsigned nticks);
+void arena_decay_tick(tsd_t *tsd, arena_t *arena);
+void *arena_malloc(tsd_t *tsd, arena_t *arena, size_t size, szind_t ind,
+ bool zero, tcache_t *tcache, bool slow_path);
+arena_t *arena_aalloc(const void *ptr);
size_t arena_salloc(const void *ptr, bool demote);
-void arena_dalloc(tsd_t *tsd, arena_chunk_t *chunk, void *ptr,
- bool try_tcache);
-void arena_sdalloc(tsd_t *tsd, arena_chunk_t *chunk, void *ptr, size_t size,
- bool try_tcache);
+void arena_dalloc(tsd_t *tsd, void *ptr, tcache_t *tcache, bool slow_path);
+void arena_sdalloc(tsd_t *tsd, void *ptr, size_t size, tcache_t *tcache);
#endif
#if (defined(JEMALLOC_ENABLE_INLINE) || defined(JEMALLOC_ARENA_C_))
# ifdef JEMALLOC_ARENA_INLINE_A
-JEMALLOC_INLINE size_t
-small_size2bin_compute(size_t size)
-{
-#if (NTBINS != 0)
- if (size <= (ZU(1) << LG_TINY_MAXCLASS)) {
- size_t lg_tmin = LG_TINY_MAXCLASS - NTBINS + 1;
- size_t lg_ceil = lg_floor(pow2_ceil(size));
- return (lg_ceil < lg_tmin ? 0 : lg_ceil - lg_tmin);
- } else
-#endif
- {
- size_t x = lg_floor((size<<1)-1);
- size_t shift = (x < LG_SIZE_CLASS_GROUP + LG_QUANTUM) ? 0 :
- x - (LG_SIZE_CLASS_GROUP + LG_QUANTUM);
- size_t grp = shift << LG_SIZE_CLASS_GROUP;
-
- size_t lg_delta = (x < LG_SIZE_CLASS_GROUP + LG_QUANTUM + 1)
- ? LG_QUANTUM : x - LG_SIZE_CLASS_GROUP - 1;
-
- size_t delta_inverse_mask = ZI(-1) << lg_delta;
- size_t mod = ((((size-1) & delta_inverse_mask) >> lg_delta)) &
- ((ZU(1) << LG_SIZE_CLASS_GROUP) - 1);
-
- size_t bin = NTBINS + grp + mod;
- return (bin);
- }
-}
-
-JEMALLOC_ALWAYS_INLINE size_t
-small_size2bin_lookup(size_t size)
-{
-
- assert(size <= LOOKUP_MAXCLASS);
- {
- size_t ret = ((size_t)(small_size2bin_tab[(size-1) >>
- LG_TINY_MIN]));
- assert(ret == small_size2bin_compute(size));
- return (ret);
- }
-}
-
-JEMALLOC_ALWAYS_INLINE size_t
-small_size2bin(size_t size)
-{
-
- assert(size > 0);
- if (likely(size <= LOOKUP_MAXCLASS))
- return (small_size2bin_lookup(size));
- else
- return (small_size2bin_compute(size));
-}
-
-JEMALLOC_INLINE size_t
-small_bin2size_compute(size_t binind)
-{
-#if (NTBINS > 0)
- if (binind < NTBINS)
- return (ZU(1) << (LG_TINY_MAXCLASS - NTBINS + 1 + binind));
- else
-#endif
- {
- size_t reduced_binind = binind - NTBINS;
- size_t grp = reduced_binind >> LG_SIZE_CLASS_GROUP;
- size_t mod = reduced_binind & ((ZU(1) << LG_SIZE_CLASS_GROUP) -
- 1);
-
- size_t grp_size_mask = ~((!!grp)-1);
- size_t grp_size = ((ZU(1) << (LG_QUANTUM +
- (LG_SIZE_CLASS_GROUP-1))) << grp) & grp_size_mask;
-
- size_t shift = (grp == 0) ? 1 : grp;
- size_t lg_delta = shift + (LG_QUANTUM-1);
- size_t mod_size = (mod+1) << lg_delta;
-
- size_t usize = grp_size + mod_size;
- return (usize);
- }
-}
-
-JEMALLOC_ALWAYS_INLINE size_t
-small_bin2size_lookup(size_t binind)
-{
-
- assert(binind < NBINS);
- {
- size_t ret = (size_t)small_bin2size_tab[binind];
- assert(ret == small_bin2size_compute(binind));
- return (ret);
- }
-}
-
-JEMALLOC_ALWAYS_INLINE size_t
-small_bin2size(size_t binind)
-{
-
- return (small_bin2size_lookup(binind));
-}
-
-JEMALLOC_ALWAYS_INLINE size_t
-small_s2u_compute(size_t size)
-{
-#if (NTBINS > 0)
- if (size <= (ZU(1) << LG_TINY_MAXCLASS)) {
- size_t lg_tmin = LG_TINY_MAXCLASS - NTBINS + 1;
- size_t lg_ceil = lg_floor(pow2_ceil(size));
- return (lg_ceil < lg_tmin ? (ZU(1) << lg_tmin) :
- (ZU(1) << lg_ceil));
- } else
-#endif
- {
- size_t x = lg_floor((size<<1)-1);
- size_t lg_delta = (x < LG_SIZE_CLASS_GROUP + LG_QUANTUM + 1)
- ? LG_QUANTUM : x - LG_SIZE_CLASS_GROUP - 1;
- size_t delta = ZU(1) << lg_delta;
- size_t delta_mask = delta - 1;
- size_t usize = (size + delta_mask) & ~delta_mask;
- return (usize);
- }
-}
-
-JEMALLOC_ALWAYS_INLINE size_t
-small_s2u_lookup(size_t size)
-{
- size_t ret = small_bin2size(small_size2bin(size));
-
- assert(ret == small_s2u_compute(size));
- return (ret);
-}
-
-JEMALLOC_ALWAYS_INLINE size_t
-small_s2u(size_t size)
-{
-
- assert(size > 0);
- if (likely(size <= LOOKUP_MAXCLASS))
- return (small_s2u_lookup(size));
- else
- return (small_s2u_compute(size));
-}
-# endif /* JEMALLOC_ARENA_INLINE_A */
-
-# ifdef JEMALLOC_ARENA_INLINE_B
JEMALLOC_ALWAYS_INLINE arena_chunk_map_bits_t *
arena_bitselm_get(arena_chunk_t *chunk, size_t pageind)
{
}
JEMALLOC_ALWAYS_INLINE size_t
-arena_miscelm_to_pageind(arena_chunk_map_misc_t *miscelm)
+arena_miscelm_to_pageind(const arena_chunk_map_misc_t *miscelm)
{
arena_chunk_t *chunk = (arena_chunk_t *)CHUNK_ADDR2BASE(miscelm);
size_t pageind = ((uintptr_t)miscelm - ((uintptr_t)chunk +
return ((void *)((uintptr_t)chunk + (pageind << LG_PAGE)));
}
+JEMALLOC_ALWAYS_INLINE arena_chunk_map_misc_t *
+arena_rd_to_miscelm(arena_runs_dirty_link_t *rd)
+{
+ arena_chunk_map_misc_t *miscelm = (arena_chunk_map_misc_t
+ *)((uintptr_t)rd - offsetof(arena_chunk_map_misc_t, rd));
+
+ assert(arena_miscelm_to_pageind(miscelm) >= map_bias);
+ assert(arena_miscelm_to_pageind(miscelm) < chunk_npages);
+
+ return (miscelm);
+}
+
JEMALLOC_ALWAYS_INLINE arena_chunk_map_misc_t *
arena_run_to_miscelm(arena_run_t *run)
{
return (arena_mapbitsp_read(arena_mapbitsp_get(chunk, pageind)));
}
+JEMALLOC_ALWAYS_INLINE size_t
+arena_mapbits_size_decode(size_t mapbits)
+{
+ size_t size;
+
+#if CHUNK_MAP_SIZE_SHIFT > 0
+ size = (mapbits & CHUNK_MAP_SIZE_MASK) >> CHUNK_MAP_SIZE_SHIFT;
+#elif CHUNK_MAP_SIZE_SHIFT == 0
+ size = mapbits & CHUNK_MAP_SIZE_MASK;
+#else
+ size = (mapbits & CHUNK_MAP_SIZE_MASK) << -CHUNK_MAP_SIZE_SHIFT;
+#endif
+
+ return (size);
+}
+
JEMALLOC_ALWAYS_INLINE size_t
arena_mapbits_unallocated_size_get(arena_chunk_t *chunk, size_t pageind)
{
mapbits = arena_mapbits_get(chunk, pageind);
assert((mapbits & (CHUNK_MAP_LARGE|CHUNK_MAP_ALLOCATED)) == 0);
- return (mapbits & ~PAGE_MASK);
+ return (arena_mapbits_size_decode(mapbits));
}
JEMALLOC_ALWAYS_INLINE size_t
mapbits = arena_mapbits_get(chunk, pageind);
assert((mapbits & (CHUNK_MAP_LARGE|CHUNK_MAP_ALLOCATED)) ==
(CHUNK_MAP_LARGE|CHUNK_MAP_ALLOCATED));
- return (mapbits & ~PAGE_MASK);
+ return (arena_mapbits_size_decode(mapbits));
}
JEMALLOC_ALWAYS_INLINE size_t
mapbits = arena_mapbits_get(chunk, pageind);
assert((mapbits & (CHUNK_MAP_LARGE|CHUNK_MAP_ALLOCATED)) ==
CHUNK_MAP_ALLOCATED);
- return (mapbits >> LG_PAGE);
+ return (mapbits >> CHUNK_MAP_RUNIND_SHIFT);
}
-JEMALLOC_ALWAYS_INLINE size_t
+JEMALLOC_ALWAYS_INLINE szind_t
arena_mapbits_binind_get(arena_chunk_t *chunk, size_t pageind)
{
size_t mapbits;
- size_t binind;
+ szind_t binind;
mapbits = arena_mapbits_get(chunk, pageind);
binind = (mapbits & CHUNK_MAP_BININD_MASK) >> CHUNK_MAP_BININD_SHIFT;
size_t mapbits;
mapbits = arena_mapbits_get(chunk, pageind);
+ assert((mapbits & CHUNK_MAP_DECOMMITTED) == 0 || (mapbits &
+ (CHUNK_MAP_DIRTY|CHUNK_MAP_UNZEROED)) == 0);
return (mapbits & CHUNK_MAP_DIRTY);
}
size_t mapbits;
mapbits = arena_mapbits_get(chunk, pageind);
+ assert((mapbits & CHUNK_MAP_DECOMMITTED) == 0 || (mapbits &
+ (CHUNK_MAP_DIRTY|CHUNK_MAP_UNZEROED)) == 0);
return (mapbits & CHUNK_MAP_UNZEROED);
}
+JEMALLOC_ALWAYS_INLINE size_t
+arena_mapbits_decommitted_get(arena_chunk_t *chunk, size_t pageind)
+{
+ size_t mapbits;
+
+ mapbits = arena_mapbits_get(chunk, pageind);
+ assert((mapbits & CHUNK_MAP_DECOMMITTED) == 0 || (mapbits &
+ (CHUNK_MAP_DIRTY|CHUNK_MAP_UNZEROED)) == 0);
+ return (mapbits & CHUNK_MAP_DECOMMITTED);
+}
+
JEMALLOC_ALWAYS_INLINE size_t
arena_mapbits_large_get(arena_chunk_t *chunk, size_t pageind)
{
*mapbitsp = mapbits;
}
+JEMALLOC_ALWAYS_INLINE size_t
+arena_mapbits_size_encode(size_t size)
+{
+ size_t mapbits;
+
+#if CHUNK_MAP_SIZE_SHIFT > 0
+ mapbits = size << CHUNK_MAP_SIZE_SHIFT;
+#elif CHUNK_MAP_SIZE_SHIFT == 0
+ mapbits = size;
+#else
+ mapbits = size >> -CHUNK_MAP_SIZE_SHIFT;
+#endif
+
+ assert((mapbits & ~CHUNK_MAP_SIZE_MASK) == 0);
+ return (mapbits);
+}
+
JEMALLOC_ALWAYS_INLINE void
arena_mapbits_unallocated_set(arena_chunk_t *chunk, size_t pageind, size_t size,
size_t flags)
size_t *mapbitsp = arena_mapbitsp_get(chunk, pageind);
assert((size & PAGE_MASK) == 0);
- assert((flags & ~CHUNK_MAP_FLAGS_MASK) == 0);
- assert((flags & (CHUNK_MAP_DIRTY|CHUNK_MAP_UNZEROED)) == flags);
- arena_mapbitsp_write(mapbitsp, size | CHUNK_MAP_BININD_INVALID | flags);
+ assert((flags & CHUNK_MAP_FLAGS_MASK) == flags);
+ assert((flags & CHUNK_MAP_DECOMMITTED) == 0 || (flags &
+ (CHUNK_MAP_DIRTY|CHUNK_MAP_UNZEROED)) == 0);
+ arena_mapbitsp_write(mapbitsp, arena_mapbits_size_encode(size) |
+ CHUNK_MAP_BININD_INVALID | flags);
}
JEMALLOC_ALWAYS_INLINE void
assert((size & PAGE_MASK) == 0);
assert((mapbits & (CHUNK_MAP_LARGE|CHUNK_MAP_ALLOCATED)) == 0);
- arena_mapbitsp_write(mapbitsp, size | (mapbits & PAGE_MASK));
+ arena_mapbitsp_write(mapbitsp, arena_mapbits_size_encode(size) |
+ (mapbits & ~CHUNK_MAP_SIZE_MASK));
+}
+
+JEMALLOC_ALWAYS_INLINE void
+arena_mapbits_internal_set(arena_chunk_t *chunk, size_t pageind, size_t flags)
+{
+ size_t *mapbitsp = arena_mapbitsp_get(chunk, pageind);
+
+ assert((flags & CHUNK_MAP_UNZEROED) == flags);
+ arena_mapbitsp_write(mapbitsp, flags);
}
JEMALLOC_ALWAYS_INLINE void
size_t flags)
{
size_t *mapbitsp = arena_mapbitsp_get(chunk, pageind);
- size_t mapbits = arena_mapbitsp_read(mapbitsp);
- size_t unzeroed;
assert((size & PAGE_MASK) == 0);
- assert((flags & CHUNK_MAP_DIRTY) == flags);
- unzeroed = mapbits & CHUNK_MAP_UNZEROED; /* Preserve unzeroed. */
- arena_mapbitsp_write(mapbitsp, size | CHUNK_MAP_BININD_INVALID | flags
- | unzeroed | CHUNK_MAP_LARGE | CHUNK_MAP_ALLOCATED);
+ assert((flags & CHUNK_MAP_FLAGS_MASK) == flags);
+ assert((flags & CHUNK_MAP_DECOMMITTED) == 0 || (flags &
+ (CHUNK_MAP_DIRTY|CHUNK_MAP_UNZEROED)) == 0);
+ arena_mapbitsp_write(mapbitsp, arena_mapbits_size_encode(size) |
+ CHUNK_MAP_BININD_INVALID | flags | CHUNK_MAP_LARGE |
+ CHUNK_MAP_ALLOCATED);
}
JEMALLOC_ALWAYS_INLINE void
arena_mapbits_large_binind_set(arena_chunk_t *chunk, size_t pageind,
- size_t binind)
+ szind_t binind)
{
size_t *mapbitsp = arena_mapbitsp_get(chunk, pageind);
size_t mapbits = arena_mapbitsp_read(mapbitsp);
assert(binind <= BININD_INVALID);
- assert(arena_mapbits_large_size_get(chunk, pageind) == PAGE);
+ assert(arena_mapbits_large_size_get(chunk, pageind) == LARGE_MINCLASS +
+ large_pad);
arena_mapbitsp_write(mapbitsp, (mapbits & ~CHUNK_MAP_BININD_MASK) |
(binind << CHUNK_MAP_BININD_SHIFT));
}
JEMALLOC_ALWAYS_INLINE void
arena_mapbits_small_set(arena_chunk_t *chunk, size_t pageind, size_t runind,
- size_t binind, size_t flags)
+ szind_t binind, size_t flags)
{
size_t *mapbitsp = arena_mapbitsp_get(chunk, pageind);
- size_t mapbits = arena_mapbitsp_read(mapbitsp);
- size_t unzeroed;
assert(binind < BININD_INVALID);
assert(pageind - runind >= map_bias);
- assert((flags & CHUNK_MAP_DIRTY) == flags);
- unzeroed = mapbits & CHUNK_MAP_UNZEROED; /* Preserve unzeroed. */
- arena_mapbitsp_write(mapbitsp, (runind << LG_PAGE) | (binind <<
- CHUNK_MAP_BININD_SHIFT) | flags | unzeroed | CHUNK_MAP_ALLOCATED);
+ assert((flags & CHUNK_MAP_UNZEROED) == flags);
+ arena_mapbitsp_write(mapbitsp, (runind << CHUNK_MAP_RUNIND_SHIFT) |
+ (binind << CHUNK_MAP_BININD_SHIFT) | flags | CHUNK_MAP_ALLOCATED);
}
-JEMALLOC_ALWAYS_INLINE void
-arena_mapbits_unzeroed_set(arena_chunk_t *chunk, size_t pageind,
- size_t unzeroed)
+JEMALLOC_INLINE void
+arena_metadata_allocated_add(arena_t *arena, size_t size)
{
- size_t *mapbitsp = arena_mapbitsp_get(chunk, pageind);
- size_t mapbits = arena_mapbitsp_read(mapbitsp);
- arena_mapbitsp_write(mapbitsp, (mapbits & ~CHUNK_MAP_UNZEROED) |
- unzeroed);
+ atomic_add_z(&arena->stats.metadata_allocated, size);
+}
+
+JEMALLOC_INLINE void
+arena_metadata_allocated_sub(arena_t *arena, size_t size)
+{
+
+ atomic_sub_z(&arena->stats.metadata_allocated, size);
+}
+
+JEMALLOC_INLINE size_t
+arena_metadata_allocated_get(arena_t *arena)
+{
+
+ return (atomic_read_z(&arena->stats.metadata_allocated));
}
JEMALLOC_INLINE bool
}
}
-JEMALLOC_ALWAYS_INLINE size_t
+JEMALLOC_ALWAYS_INLINE szind_t
arena_ptr_small_binind_get(const void *ptr, size_t mapbits)
{
- size_t binind;
+ szind_t binind;
binind = (mapbits & CHUNK_MAP_BININD_MASK) >> CHUNK_MAP_BININD_SHIFT;
size_t rpages_ind;
arena_run_t *run;
arena_bin_t *bin;
- size_t actual_binind;
+ szind_t run_binind, actual_binind;
arena_bin_info_t *bin_info;
arena_chunk_map_misc_t *miscelm;
void *rpages;
assert(binind != BININD_INVALID);
assert(binind < NBINS);
chunk = (arena_chunk_t *)CHUNK_ADDR2BASE(ptr);
- arena = chunk->arena;
+ arena = extent_node_arena_get(&chunk->node);
pageind = ((uintptr_t)ptr - (uintptr_t)chunk) >> LG_PAGE;
actual_mapbits = arena_mapbits_get(chunk, pageind);
assert(mapbits == actual_mapbits);
pageind);
miscelm = arena_miscelm_get(chunk, rpages_ind);
run = &miscelm->run;
- bin = run->bin;
- actual_binind = bin - arena->bins;
- assert(binind == actual_binind);
+ run_binind = run->binind;
+ bin = &arena->bins[run_binind];
+ actual_binind = (szind_t)(bin - arena->bins);
+ assert(run_binind == actual_binind);
bin_info = &arena_bin_info[actual_binind];
rpages = arena_miscelm_to_rpages(miscelm);
assert(((uintptr_t)ptr - ((uintptr_t)rpages +
return (binind);
}
-# endif /* JEMALLOC_ARENA_INLINE_B */
+# endif /* JEMALLOC_ARENA_INLINE_A */
-# ifdef JEMALLOC_ARENA_INLINE_C
-JEMALLOC_INLINE size_t
+# ifdef JEMALLOC_ARENA_INLINE_B
+JEMALLOC_INLINE szind_t
arena_bin_index(arena_t *arena, arena_bin_t *bin)
{
- size_t binind = bin - arena->bins;
+ szind_t binind = (szind_t)(bin - arena->bins);
assert(binind < NBINS);
return (binind);
}
-JEMALLOC_INLINE unsigned
+JEMALLOC_INLINE size_t
arena_run_regind(arena_run_t *run, arena_bin_info_t *bin_info, const void *ptr)
{
- unsigned shift, diff, regind;
- size_t interval;
+ size_t diff, interval, shift, regind;
arena_chunk_map_misc_t *miscelm = arena_run_to_miscelm(run);
void *rpages = arena_miscelm_to_rpages(miscelm);
* Avoid doing division with a variable divisor if possible. Using
* actual division here can reduce allocator throughput by over 20%!
*/
- diff = (unsigned)((uintptr_t)ptr - (uintptr_t)rpages -
+ diff = (size_t)((uintptr_t)ptr - (uintptr_t)rpages -
bin_info->reg0_offset);
/* Rescale (factor powers of 2 out of the numerator and denominator). */
interval = bin_info->reg_interval;
- shift = jemalloc_ffs(interval) - 1;
+ shift = ffs_zu(interval) - 1;
diff >>= shift;
interval >>= shift;
* divide by 0, and 1 and 2 are both powers of two, which are
* handled above.
*/
-#define SIZE_INV_SHIFT ((sizeof(unsigned) << 3) - LG_RUN_MAXREGS)
-#define SIZE_INV(s) (((1U << SIZE_INV_SHIFT) / (s)) + 1)
- static const unsigned interval_invs[] = {
+#define SIZE_INV_SHIFT ((sizeof(size_t) << 3) - LG_RUN_MAXREGS)
+#define SIZE_INV(s) (((ZU(1) << SIZE_INV_SHIFT) / (s)) + 1)
+ static const size_t interval_invs[] = {
SIZE_INV(3),
SIZE_INV(4), SIZE_INV(5), SIZE_INV(6), SIZE_INV(7),
SIZE_INV(8), SIZE_INV(9), SIZE_INV(10), SIZE_INV(11),
SIZE_INV(28), SIZE_INV(29), SIZE_INV(30), SIZE_INV(31)
};
- if (likely(interval <= ((sizeof(interval_invs) /
- sizeof(unsigned)) + 2))) {
+ if (likely(interval <= ((sizeof(interval_invs) / sizeof(size_t))
+ + 2))) {
regind = (diff * interval_invs[interval - 3]) >>
SIZE_INV_SHIFT;
} else
{
prof_tctx_t *ret;
arena_chunk_t *chunk;
- size_t pageind, mapbits;
cassert(config_prof);
assert(ptr != NULL);
- assert(CHUNK_ADDR2BASE(ptr) != ptr);
chunk = (arena_chunk_t *)CHUNK_ADDR2BASE(ptr);
- pageind = ((uintptr_t)ptr - (uintptr_t)chunk) >> LG_PAGE;
- mapbits = arena_mapbits_get(chunk, pageind);
- assert((mapbits & CHUNK_MAP_ALLOCATED) != 0);
- if (likely((mapbits & CHUNK_MAP_LARGE) == 0))
- ret = (prof_tctx_t *)(uintptr_t)1U;
- else
- ret = arena_miscelm_get(chunk, pageind)->prof_tctx;
+ if (likely(chunk != ptr)) {
+ size_t pageind = ((uintptr_t)ptr - (uintptr_t)chunk) >> LG_PAGE;
+ size_t mapbits = arena_mapbits_get(chunk, pageind);
+ assert((mapbits & CHUNK_MAP_ALLOCATED) != 0);
+ if (likely((mapbits & CHUNK_MAP_LARGE) == 0))
+ ret = (prof_tctx_t *)(uintptr_t)1U;
+ else {
+ arena_chunk_map_misc_t *elm = arena_miscelm_get(chunk,
+ pageind);
+ ret = atomic_read_p(&elm->prof_tctx_pun);
+ }
+ } else
+ ret = huge_prof_tctx_get(ptr);
return (ret);
}
JEMALLOC_INLINE void
-arena_prof_tctx_set(const void *ptr, prof_tctx_t *tctx)
+arena_prof_tctx_set(const void *ptr, size_t usize, prof_tctx_t *tctx)
{
arena_chunk_t *chunk;
- size_t pageind;
cassert(config_prof);
assert(ptr != NULL);
- assert(CHUNK_ADDR2BASE(ptr) != ptr);
chunk = (arena_chunk_t *)CHUNK_ADDR2BASE(ptr);
- pageind = ((uintptr_t)ptr - (uintptr_t)chunk) >> LG_PAGE;
- assert(arena_mapbits_allocated_get(chunk, pageind) != 0);
+ if (likely(chunk != ptr)) {
+ size_t pageind = ((uintptr_t)ptr - (uintptr_t)chunk) >> LG_PAGE;
+
+ assert(arena_mapbits_allocated_get(chunk, pageind) != 0);
- if (unlikely(arena_mapbits_large_get(chunk, pageind) != 0))
- arena_miscelm_get(chunk, pageind)->prof_tctx = tctx;
+ if (unlikely(usize > SMALL_MAXCLASS || (uintptr_t)tctx >
+ (uintptr_t)1U)) {
+ arena_chunk_map_misc_t *elm;
+
+ assert(arena_mapbits_large_get(chunk, pageind) != 0);
+
+ elm = arena_miscelm_get(chunk, pageind);
+ atomic_write_p(&elm->prof_tctx_pun, tctx);
+ } else {
+ /*
+ * tctx must always be initialized for large runs.
+ * Assert that the surrounding conditional logic is
+ * equivalent to checking whether ptr refers to a large
+ * run.
+ */
+ assert(arena_mapbits_large_get(chunk, pageind) == 0);
+ }
+ } else
+ huge_prof_tctx_set(ptr, tctx);
+}
+
+JEMALLOC_INLINE void
+arena_prof_tctx_reset(const void *ptr, size_t usize, const void *old_ptr,
+ prof_tctx_t *old_tctx)
+{
+
+ cassert(config_prof);
+ assert(ptr != NULL);
+
+ if (unlikely(usize > SMALL_MAXCLASS || (ptr == old_ptr &&
+ (uintptr_t)old_tctx > (uintptr_t)1U))) {
+ arena_chunk_t *chunk = (arena_chunk_t *)CHUNK_ADDR2BASE(ptr);
+ if (likely(chunk != ptr)) {
+ size_t pageind;
+ arena_chunk_map_misc_t *elm;
+
+ pageind = ((uintptr_t)ptr - (uintptr_t)chunk) >>
+ LG_PAGE;
+ assert(arena_mapbits_allocated_get(chunk, pageind) !=
+ 0);
+ assert(arena_mapbits_large_get(chunk, pageind) != 0);
+
+ elm = arena_miscelm_get(chunk, pageind);
+ atomic_write_p(&elm->prof_tctx_pun,
+ (prof_tctx_t *)(uintptr_t)1U);
+ } else
+ huge_prof_tctx_reset(ptr);
+ }
+}
+
+JEMALLOC_ALWAYS_INLINE void
+arena_decay_ticks(tsd_t *tsd, arena_t *arena, unsigned nticks)
+{
+ ticker_t *decay_ticker;
+
+ if (unlikely(tsd == NULL))
+ return;
+ decay_ticker = decay_ticker_get(tsd, arena->ind);
+ if (unlikely(decay_ticker == NULL))
+ return;
+ if (unlikely(ticker_ticks(decay_ticker, nticks)))
+ arena_purge(arena, false);
+}
+
+JEMALLOC_ALWAYS_INLINE void
+arena_decay_tick(tsd_t *tsd, arena_t *arena)
+{
+
+ arena_decay_ticks(tsd, arena, 1);
}
JEMALLOC_ALWAYS_INLINE void *
-arena_malloc(tsd_t *tsd, arena_t *arena, size_t size, bool zero,
- bool try_tcache)
+arena_malloc(tsd_t *tsd, arena_t *arena, size_t size, szind_t ind, bool zero,
+ tcache_t *tcache, bool slow_path)
{
- tcache_t *tcache;
assert(size != 0);
- assert(size <= arena_maxclass);
- if (likely(size <= SMALL_MAXCLASS)) {
- if (likely(try_tcache) && likely((tcache = tcache_get(tsd,
- true)) != NULL))
- return (tcache_alloc_small(tcache, size, zero));
- else {
- return (arena_malloc_small(choose_arena(tsd, arena),
- size, zero));
+ if (likely(tcache != NULL)) {
+ if (likely(size <= SMALL_MAXCLASS)) {
+ return (tcache_alloc_small(tsd, arena, tcache, size,
+ ind, zero, slow_path));
}
- } else {
- /*
- * Initialize tcache after checking size in order to avoid
- * infinite recursion during tcache initialization.
- */
- if (try_tcache && size <= tcache_maxclass && likely((tcache =
- tcache_get(tsd, true)) != NULL))
- return (tcache_alloc_large(tcache, size, zero));
- else {
- return (arena_malloc_large(choose_arena(tsd, arena),
- size, zero));
+ if (likely(size <= tcache_maxclass)) {
+ return (tcache_alloc_large(tsd, arena, tcache, size,
+ ind, zero, slow_path));
}
+ /* (size > tcache_maxclass) case falls through. */
+ assert(size > tcache_maxclass);
}
+
+ return (arena_malloc_hard(tsd, arena, size, ind, zero, tcache));
+}
+
+JEMALLOC_ALWAYS_INLINE arena_t *
+arena_aalloc(const void *ptr)
+{
+ arena_chunk_t *chunk;
+
+ chunk = (arena_chunk_t *)CHUNK_ADDR2BASE(ptr);
+ if (likely(chunk != ptr))
+ return (extent_node_arena_get(&chunk->node));
+ else
+ return (huge_aalloc(ptr));
}
/* Return the size of the allocation pointed to by ptr. */
{
size_t ret;
arena_chunk_t *chunk;
- size_t pageind, binind;
+ size_t pageind;
+ szind_t binind;
assert(ptr != NULL);
- assert(CHUNK_ADDR2BASE(ptr) != ptr);
chunk = (arena_chunk_t *)CHUNK_ADDR2BASE(ptr);
- pageind = ((uintptr_t)ptr - (uintptr_t)chunk) >> LG_PAGE;
- assert(arena_mapbits_allocated_get(chunk, pageind) != 0);
- binind = arena_mapbits_binind_get(chunk, pageind);
- if (unlikely(binind == BININD_INVALID || (config_prof && !demote &&
- arena_mapbits_large_get(chunk, pageind) != 0))) {
- /*
- * Large allocation. In the common case (demote), and as this
- * is an inline function, most callers will only end up looking
- * at binind to determine that ptr is a small allocation.
- */
- assert(((uintptr_t)ptr & PAGE_MASK) == 0);
- ret = arena_mapbits_large_size_get(chunk, pageind);
- assert(ret != 0);
- assert(pageind + (ret>>LG_PAGE) <= chunk_npages);
- assert(ret == PAGE || arena_mapbits_large_size_get(chunk,
- pageind+(ret>>LG_PAGE)-1) == 0);
- assert(binind == arena_mapbits_binind_get(chunk,
- pageind+(ret>>LG_PAGE)-1));
- assert(arena_mapbits_dirty_get(chunk, pageind) ==
- arena_mapbits_dirty_get(chunk, pageind+(ret>>LG_PAGE)-1));
- } else {
- /* Small allocation (possibly promoted to a large object). */
- assert(arena_mapbits_large_get(chunk, pageind) != 0 ||
- arena_ptr_small_binind_get(ptr, arena_mapbits_get(chunk,
- pageind)) == binind);
- ret = small_bin2size(binind);
- }
+ if (likely(chunk != ptr)) {
+ pageind = ((uintptr_t)ptr - (uintptr_t)chunk) >> LG_PAGE;
+ assert(arena_mapbits_allocated_get(chunk, pageind) != 0);
+ binind = arena_mapbits_binind_get(chunk, pageind);
+ if (unlikely(binind == BININD_INVALID || (config_prof && !demote
+ && arena_mapbits_large_get(chunk, pageind) != 0))) {
+ /*
+ * Large allocation. In the common case (demote), and
+ * as this is an inline function, most callers will only
+ * end up looking at binind to determine that ptr is a
+ * small allocation.
+ */
+ assert(config_cache_oblivious || ((uintptr_t)ptr &
+ PAGE_MASK) == 0);
+ ret = arena_mapbits_large_size_get(chunk, pageind) -
+ large_pad;
+ assert(ret != 0);
+ assert(pageind + ((ret+large_pad)>>LG_PAGE) <=
+ chunk_npages);
+ assert(arena_mapbits_dirty_get(chunk, pageind) ==
+ arena_mapbits_dirty_get(chunk,
+ pageind+((ret+large_pad)>>LG_PAGE)-1));
+ } else {
+ /*
+ * Small allocation (possibly promoted to a large
+ * object).
+ */
+ assert(arena_mapbits_large_get(chunk, pageind) != 0 ||
+ arena_ptr_small_binind_get(ptr,
+ arena_mapbits_get(chunk, pageind)) == binind);
+ ret = index2size(binind);
+ }
+ } else
+ ret = huge_salloc(ptr);
return (ret);
}
JEMALLOC_ALWAYS_INLINE void
-arena_dalloc(tsd_t *tsd, arena_chunk_t *chunk, void *ptr, bool try_tcache)
+arena_dalloc(tsd_t *tsd, void *ptr, tcache_t *tcache, bool slow_path)
{
+ arena_chunk_t *chunk;
size_t pageind, mapbits;
- tcache_t *tcache;
assert(ptr != NULL);
- assert(CHUNK_ADDR2BASE(ptr) != ptr);
- pageind = ((uintptr_t)ptr - (uintptr_t)chunk) >> LG_PAGE;
- mapbits = arena_mapbits_get(chunk, pageind);
- assert(arena_mapbits_allocated_get(chunk, pageind) != 0);
- if (likely((mapbits & CHUNK_MAP_LARGE) == 0)) {
- /* Small allocation. */
- if (likely(try_tcache) && likely((tcache = tcache_get(tsd,
- false)) != NULL)) {
- size_t binind = arena_ptr_small_binind_get(ptr,
- mapbits);
- tcache_dalloc_small(tcache, ptr, binind);
- } else
- arena_dalloc_small(chunk->arena, chunk, ptr, pageind);
- } else {
- size_t size = arena_mapbits_large_size_get(chunk, pageind);
-
- assert(((uintptr_t)ptr & PAGE_MASK) == 0);
-
- if (try_tcache && size <= tcache_maxclass && likely((tcache =
- tcache_get(tsd, false)) != NULL)) {
- tcache_dalloc_large(tcache, ptr, size);
- } else
- arena_dalloc_large(chunk->arena, chunk, ptr);
- }
+ chunk = (arena_chunk_t *)CHUNK_ADDR2BASE(ptr);
+ if (likely(chunk != ptr)) {
+ pageind = ((uintptr_t)ptr - (uintptr_t)chunk) >> LG_PAGE;
+ mapbits = arena_mapbits_get(chunk, pageind);
+ assert(arena_mapbits_allocated_get(chunk, pageind) != 0);
+ if (likely((mapbits & CHUNK_MAP_LARGE) == 0)) {
+ /* Small allocation. */
+ if (likely(tcache != NULL)) {
+ szind_t binind = arena_ptr_small_binind_get(ptr,
+ mapbits);
+ tcache_dalloc_small(tsd, tcache, ptr, binind,
+ slow_path);
+ } else {
+ arena_dalloc_small(tsd, extent_node_arena_get(
+ &chunk->node), chunk, ptr, pageind);
+ }
+ } else {
+ size_t size = arena_mapbits_large_size_get(chunk,
+ pageind);
+
+ assert(config_cache_oblivious || ((uintptr_t)ptr &
+ PAGE_MASK) == 0);
+
+ if (likely(tcache != NULL) && size - large_pad <=
+ tcache_maxclass) {
+ tcache_dalloc_large(tsd, tcache, ptr, size -
+ large_pad, slow_path);
+ } else {
+ arena_dalloc_large(tsd, extent_node_arena_get(
+ &chunk->node), chunk, ptr);
+ }
+ }
+ } else
+ huge_dalloc(tsd, ptr, tcache);
}
JEMALLOC_ALWAYS_INLINE void
-arena_sdalloc(tsd_t *tsd, arena_chunk_t *chunk, void *ptr, size_t size,
- bool try_tcache)
+arena_sdalloc(tsd_t *tsd, void *ptr, size_t size, tcache_t *tcache)
{
- tcache_t *tcache;
+ arena_chunk_t *chunk;
- assert(ptr != NULL);
- assert(CHUNK_ADDR2BASE(ptr) != ptr);
-
- if (likely(size <= SMALL_MAXCLASS)) {
- /* Small allocation. */
- if (likely(try_tcache) && likely((tcache = tcache_get(tsd,
- false)) != NULL)) {
- size_t binind = small_size2bin(size);
- tcache_dalloc_small(tcache, ptr, binind);
- } else {
+ chunk = (arena_chunk_t *)CHUNK_ADDR2BASE(ptr);
+ if (likely(chunk != ptr)) {
+ if (config_prof && opt_prof) {
size_t pageind = ((uintptr_t)ptr - (uintptr_t)chunk) >>
LG_PAGE;
- arena_dalloc_small(chunk->arena, chunk, ptr, pageind);
+ assert(arena_mapbits_allocated_get(chunk, pageind) !=
+ 0);
+ if (arena_mapbits_large_get(chunk, pageind) != 0) {
+ /*
+ * Make sure to use promoted size, not request
+ * size.
+ */
+ size = arena_mapbits_large_size_get(chunk,
+ pageind) - large_pad;
+ }
}
- } else {
- assert(((uintptr_t)ptr & PAGE_MASK) == 0);
-
- if (try_tcache && size <= tcache_maxclass && (tcache =
- tcache_get(tsd, false)) != NULL) {
- tcache_dalloc_large(tcache, ptr, size);
- } else
- arena_dalloc_large(chunk->arena, chunk, ptr);
- }
+ assert(s2u(size) == s2u(arena_salloc(ptr, false)));
+
+ if (likely(size <= SMALL_MAXCLASS)) {
+ /* Small allocation. */
+ if (likely(tcache != NULL)) {
+ szind_t binind = size2index(size);
+ tcache_dalloc_small(tsd, tcache, ptr, binind,
+ true);
+ } else {
+ size_t pageind = ((uintptr_t)ptr -
+ (uintptr_t)chunk) >> LG_PAGE;
+ arena_dalloc_small(tsd, extent_node_arena_get(
+ &chunk->node), chunk, ptr, pageind);
+ }
+ } else {
+ assert(config_cache_oblivious || ((uintptr_t)ptr &
+ PAGE_MASK) == 0);
+
+ if (likely(tcache != NULL) && size <= tcache_maxclass) {
+ tcache_dalloc_large(tsd, tcache, ptr, size,
+ true);
+ } else {
+ arena_dalloc_large(tsd, extent_node_arena_get(
+ &chunk->node), chunk, ptr);
+ }
+ }
+ } else
+ huge_dalloc(tsd, ptr, tcache);
}
-# endif /* JEMALLOC_ARENA_INLINE_C */
+# endif /* JEMALLOC_ARENA_INLINE_B */
#endif
#endif /* JEMALLOC_H_INLINES */
projects / rustc.git / blobdiff