diff options
Diffstat (limited to 'src')
| -rw-r--r-- | src/arena.c | 970 | ||||
| -rw-r--r-- | src/bitmap.c | 2 | ||||
| -rw-r--r-- | src/chunk.c | 8 | ||||
| -rw-r--r-- | src/chunk_dss.c | 15 | ||||
| -rw-r--r-- | src/chunk_mmap.c | 4 | ||||
| -rw-r--r-- | src/ckh.c | 12 | ||||
| -rw-r--r-- | src/ctl.c | 247 | ||||
| -rw-r--r-- | src/huge.c | 46 | ||||
| -rw-r--r-- | src/jemalloc.c | 1157 | ||||
| -rw-r--r-- | src/mutex.c | 2 | ||||
| -rw-r--r-- | src/prof.c | 640 | ||||
| -rw-r--r-- | src/quarantine.c | 13 | ||||
| -rw-r--r-- | src/rtree.c | 78 | ||||
| -rw-r--r-- | src/stats.c | 8 | ||||
| -rw-r--r-- | src/tcache.c | 9 | ||||
| -rw-r--r-- | src/tsd.c | 36 | ||||
| -rw-r--r-- | src/util.c | 85 | ||||
| -rw-r--r-- | src/zone.c | 2 |
18 files changed, 2007 insertions, 1327 deletions
diff --git a/src/arena.c b/src/arena.c index d28b629a..4da6d50c 100644 --- a/src/arena.c +++ b/src/arena.c @@ -38,52 +38,18 @@ const uint8_t small_size2bin[] = { }; /******************************************************************************/ -/* Function prototypes for non-inline static functions. */ - -static void arena_avail_insert(arena_t *arena, arena_chunk_t *chunk, - size_t pageind, size_t npages, bool maybe_adjac_pred, - bool maybe_adjac_succ); -static void arena_avail_remove(arena_t *arena, arena_chunk_t *chunk, - size_t pageind, size_t npages, bool maybe_adjac_pred, - bool maybe_adjac_succ); -static void arena_run_split(arena_t *arena, arena_run_t *run, size_t size, - bool large, size_t binind, bool zero); -static arena_chunk_t *arena_chunk_alloc(arena_t *arena); -static void arena_chunk_dealloc(arena_t *arena, arena_chunk_t *chunk); -static arena_run_t *arena_run_alloc_helper(arena_t *arena, size_t size, - bool large, size_t binind, bool zero); -static arena_run_t *arena_run_alloc(arena_t *arena, size_t size, bool large, - size_t binind, bool zero); -static arena_chunk_t *chunks_dirty_iter_cb(arena_chunk_tree_t *tree, - arena_chunk_t *chunk, void *arg); +/* + * Function prototypes for static functions that are referenced prior to + * definition. + */ + static void arena_purge(arena_t *arena, bool all); static void arena_run_dalloc(arena_t *arena, arena_run_t *run, bool dirty, bool cleaned); -static void arena_run_trim_head(arena_t *arena, arena_chunk_t *chunk, - arena_run_t *run, size_t oldsize, size_t newsize); -static void arena_run_trim_tail(arena_t *arena, arena_chunk_t *chunk, - arena_run_t *run, size_t oldsize, size_t newsize, bool dirty); -static arena_run_t *arena_bin_runs_first(arena_bin_t *bin); -static void arena_bin_runs_insert(arena_bin_t *bin, arena_run_t *run); -static void arena_bin_runs_remove(arena_bin_t *bin, arena_run_t *run); -static arena_run_t *arena_bin_nonfull_run_tryget(arena_bin_t *bin); -static arena_run_t *arena_bin_nonfull_run_get(arena_t *arena, arena_bin_t *bin); -static void *arena_bin_malloc_hard(arena_t *arena, arena_bin_t *bin); -static void arena_dissociate_bin_run(arena_chunk_t *chunk, arena_run_t *run, - arena_bin_t *bin); static void arena_dalloc_bin_run(arena_t *arena, arena_chunk_t *chunk, arena_run_t *run, arena_bin_t *bin); static void arena_bin_lower_run(arena_t *arena, arena_chunk_t *chunk, arena_run_t *run, arena_bin_t *bin); -static void arena_ralloc_large_shrink(arena_t *arena, arena_chunk_t *chunk, - void *ptr, size_t oldsize, size_t size); -static bool arena_ralloc_large_grow(arena_t *arena, arena_chunk_t *chunk, - void *ptr, size_t oldsize, size_t size, size_t extra, bool zero); -static bool arena_ralloc_large(void *ptr, size_t oldsize, size_t size, - size_t extra, bool zero); -static size_t bin_info_run_size_calc(arena_bin_info_t *bin_info, - size_t min_run_size); -static void bin_info_init(void); /******************************************************************************/ @@ -388,50 +354,44 @@ arena_run_page_validate_zeroed(arena_chunk_t *chunk, size_t run_ind) } static void -arena_run_split(arena_t *arena, arena_run_t *run, size_t size, bool large, - size_t binind, bool zero) +arena_cactive_update(arena_t *arena, size_t add_pages, size_t sub_pages) { - arena_chunk_t *chunk; - size_t run_ind, total_pages, need_pages, rem_pages, i; - size_t flag_dirty; - assert((large && binind == BININD_INVALID) || (large == false && binind - != BININD_INVALID)); + if (config_stats) { + ssize_t cactive_diff = CHUNK_CEILING((arena->nactive + + add_pages) << LG_PAGE) - CHUNK_CEILING((arena->nactive - + sub_pages) << LG_PAGE); + if (cactive_diff != 0) + stats_cactive_add(cactive_diff); + } +} + +static void +arena_run_split_remove(arena_t *arena, arena_chunk_t *chunk, size_t run_ind, + size_t flag_dirty, size_t need_pages) +{ + size_t total_pages, rem_pages; - chunk = (arena_chunk_t *)CHUNK_ADDR2BASE(run); - run_ind = (unsigned)(((uintptr_t)run - (uintptr_t)chunk) >> LG_PAGE); - flag_dirty = arena_mapbits_dirty_get(chunk, run_ind); total_pages = arena_mapbits_unallocated_size_get(chunk, run_ind) >> LG_PAGE; assert(arena_mapbits_dirty_get(chunk, run_ind+total_pages-1) == flag_dirty); - need_pages = (size >> LG_PAGE); - assert(need_pages > 0); assert(need_pages <= total_pages); rem_pages = total_pages - need_pages; arena_avail_remove(arena, chunk, run_ind, total_pages, true, true); - if (config_stats) { - /* - * Update stats_cactive if nactive is crossing a chunk - * multiple. - */ - size_t cactive_diff = CHUNK_CEILING((arena->nactive + - need_pages) << LG_PAGE) - CHUNK_CEILING(arena->nactive << - LG_PAGE); - if (cactive_diff != 0) - stats_cactive_add(cactive_diff); - } + arena_cactive_update(arena, need_pages, 0); arena->nactive += need_pages; /* Keep track of trailing unused pages for later use. */ if (rem_pages > 0) { if (flag_dirty != 0) { - arena_mapbits_unallocated_set(chunk, run_ind+need_pages, - (rem_pages << LG_PAGE), CHUNK_MAP_DIRTY); + arena_mapbits_unallocated_set(chunk, + run_ind+need_pages, (rem_pages << LG_PAGE), + flag_dirty); arena_mapbits_unallocated_set(chunk, run_ind+total_pages-1, (rem_pages << LG_PAGE), - CHUNK_MAP_DIRTY); + flag_dirty); } else { arena_mapbits_unallocated_set(chunk, run_ind+need_pages, (rem_pages << LG_PAGE), @@ -445,166 +405,217 @@ arena_run_split(arena_t *arena, arena_run_t *run, size_t size, bool large, arena_avail_insert(arena, chunk, run_ind+need_pages, rem_pages, false, true); } +} - /* - * Update the page map separately for large vs. small runs, since it is - * possible to avoid iteration for large mallocs. - */ - if (large) { - if (zero) { - if (flag_dirty == 0) { - /* - * The run is clean, so some pages may be - * zeroed (i.e. never before touched). - */ - for (i = 0; i < need_pages; i++) { - if (arena_mapbits_unzeroed_get(chunk, - run_ind+i) != 0) { - arena_run_zero(chunk, run_ind+i, - 1); - } else if (config_debug) { - arena_run_page_validate_zeroed( - chunk, run_ind+i); - } else { - arena_run_page_mark_zeroed( - chunk, run_ind+i); - } +static void +arena_run_split_large_helper(arena_t *arena, arena_run_t *run, size_t size, + bool remove, bool zero) +{ + arena_chunk_t *chunk; + size_t flag_dirty, run_ind, need_pages, i; + + chunk = (arena_chunk_t *)CHUNK_ADDR2BASE(run); + run_ind = (unsigned)(((uintptr_t)run - (uintptr_t)chunk) >> LG_PAGE); + flag_dirty = arena_mapbits_dirty_get(chunk, run_ind); + need_pages = (size >> LG_PAGE); + assert(need_pages > 0); + + if (remove) { + arena_run_split_remove(arena, chunk, run_ind, flag_dirty, + need_pages); + } + + if (zero) { + if (flag_dirty == 0) { + /* + * The run is clean, so some pages may be zeroed (i.e. + * never before touched). + */ + for (i = 0; i < need_pages; i++) { + if (arena_mapbits_unzeroed_get(chunk, run_ind+i) + != 0) + arena_run_zero(chunk, run_ind+i, 1); + else if (config_debug) { + arena_run_page_validate_zeroed(chunk, + run_ind+i); + } else { + arena_run_page_mark_zeroed(chunk, + run_ind+i); } - } else { - /* - * The run is dirty, so all pages must be - * zeroed. - */ - arena_run_zero(chunk, run_ind, need_pages); } } else { - VALGRIND_MAKE_MEM_UNDEFINED((void *)((uintptr_t)chunk + - (run_ind << LG_PAGE)), (need_pages << LG_PAGE)); + /* The run is dirty, so all pages must be zeroed. */ + arena_run_zero(chunk, run_ind, need_pages); } - - /* - * Set the last element first, in case the run only contains one - * page (i.e. both statements set the same element). - */ - arena_mapbits_large_set(chunk, run_ind+need_pages-1, 0, - flag_dirty); - arena_mapbits_large_set(chunk, run_ind, size, flag_dirty); } else { - assert(zero == false); - /* - * Propagate the dirty and unzeroed flags to the allocated - * small run, so that arena_dalloc_bin_run() has the ability to - * conditionally trim clean pages. - */ - arena_mapbits_small_set(chunk, run_ind, 0, binind, flag_dirty); - /* - * The first page will always be dirtied during small run - * initialization, so a validation failure here would not - * actually cause an observable failure. - */ - if (config_debug && flag_dirty == 0 && - arena_mapbits_unzeroed_get(chunk, run_ind) == 0) - arena_run_page_validate_zeroed(chunk, run_ind); - for (i = 1; i < need_pages - 1; i++) { - arena_mapbits_small_set(chunk, run_ind+i, i, binind, 0); - if (config_debug && flag_dirty == 0 && - arena_mapbits_unzeroed_get(chunk, run_ind+i) == 0) { - arena_run_page_validate_zeroed(chunk, - run_ind+i); - } - } - arena_mapbits_small_set(chunk, run_ind+need_pages-1, - need_pages-1, binind, flag_dirty); - if (config_debug && flag_dirty == 0 && - arena_mapbits_unzeroed_get(chunk, run_ind+need_pages-1) == - 0) { - arena_run_page_validate_zeroed(chunk, - run_ind+need_pages-1); - } VALGRIND_MAKE_MEM_UNDEFINED((void *)((uintptr_t)chunk + (run_ind << LG_PAGE)), (need_pages << LG_PAGE)); } + + /* + * Set the last element first, in case the run only contains one page + * (i.e. both statements set the same element). + */ + arena_mapbits_large_set(chunk, run_ind+need_pages-1, 0, flag_dirty); + arena_mapbits_large_set(chunk, run_ind, size, flag_dirty); +} + +static void +arena_run_split_large(arena_t *arena, arena_run_t *run, size_t size, bool zero) +{ + + arena_run_split_large_helper(arena, run, size, true, zero); +} + +static void +arena_run_init_large(arena_t *arena, arena_run_t *run, size_t size, bool zero) +{ + + arena_run_split_large_helper(arena, run, size, false, zero); +} + +static void +arena_run_split_small(arena_t *arena, arena_run_t *run, size_t size, + size_t binind) +{ + arena_chunk_t *chunk; + size_t flag_dirty, run_ind, need_pages, i; + + assert(binind != BININD_INVALID); + + chunk = (arena_chunk_t *)CHUNK_ADDR2BASE(run); + run_ind = (unsigned)(((uintptr_t)run - (uintptr_t)chunk) >> LG_PAGE); + flag_dirty = arena_mapbits_dirty_get(chunk, run_ind); + need_pages = (size >> LG_PAGE); + assert(need_pages > 0); + + arena_run_split_remove(arena, chunk, run_ind, flag_dirty, need_pages); + + /* + * Propagate the dirty and unzeroed flags to the allocated small run, + * so that arena_dalloc_bin_run() has the ability to conditionally trim + * clean pages. + */ + arena_mapbits_small_set(chunk, run_ind, 0, binind, flag_dirty); + /* + * The first page will always be dirtied during small run + * initialization, so a validation failure here would not actually + * cause an observable failure. + */ + if (config_debug && flag_dirty == 0 && arena_mapbits_unzeroed_get(chunk, + run_ind) == 0) + arena_run_page_validate_zeroed(chunk, run_ind); + for (i = 1; i < need_pages - 1; i++) { + arena_mapbits_small_set(chunk, run_ind+i, i, binind, 0); + if (config_debug && flag_dirty == 0 && + arena_mapbits_unzeroed_get(chunk, run_ind+i) == 0) + arena_run_page_validate_zeroed(chunk, run_ind+i); + } + arena_mapbits_small_set(chunk, run_ind+need_pages-1, need_pages-1, + binind, flag_dirty); + if (config_debug && flag_dirty == 0 && arena_mapbits_unzeroed_get(chunk, + run_ind+need_pages-1) == 0) + arena_run_page_validate_zeroed(chunk, run_ind+need_pages-1); + VALGRIND_MAKE_MEM_UNDEFINED((void *)((uintptr_t)chunk + + (run_ind << LG_PAGE)), (need_pages << LG_PAGE)); } static arena_chunk_t * -arena_chunk_alloc(arena_t *arena) +arena_chunk_init_spare(arena_t *arena) { arena_chunk_t *chunk; - size_t i; - if (arena->spare != NULL) { - chunk = arena->spare; - arena->spare = NULL; + assert(arena->spare != NULL); - assert(arena_mapbits_allocated_get(chunk, map_bias) == 0); - assert(arena_mapbits_allocated_get(chunk, chunk_npages-1) == 0); - assert(arena_mapbits_unallocated_size_get(chunk, map_bias) == - arena_maxclass); - assert(arena_mapbits_unallocated_size_get(chunk, - chunk_npages-1) == arena_maxclass); - assert(arena_mapbits_dirty_get(chunk, map_bias) == - arena_mapbits_dirty_get(chunk, chunk_npages-1)); - } else { - bool zero; - size_t unzeroed; + chunk = arena->spare; + arena->spare = NULL; - zero = false; - malloc_mutex_unlock(&arena->lock); - chunk = (arena_chunk_t *)chunk_alloc(chunksize, chunksize, - false, &zero, arena->dss_prec); - malloc_mutex_lock(&arena->lock); - if (chunk == NULL) - return (NULL); - if (config_stats) - arena->stats.mapped += chunksize; + assert(arena_mapbits_allocated_get(chunk, map_bias) == 0); + assert(arena_mapbits_allocated_get(chunk, chunk_npages-1) == 0); + assert(arena_mapbits_unallocated_size_get(chunk, map_bias) == + arena_maxclass); + assert(arena_mapbits_unallocated_size_get(chunk, chunk_npages-1) == + arena_maxclass); + assert(arena_mapbits_dirty_get(chunk, map_bias) == + arena_mapbits_dirty_get(chunk, chunk_npages-1)); - chunk->arena = arena; + return (chunk); +} - /* - * Claim that no pages are in use, since the header is merely - * overhead. - */ - chunk->ndirty = 0; +static arena_chunk_t * +arena_chunk_init_hard(arena_t *arena) +{ + arena_chunk_t *chunk; + bool zero; + size_t unzeroed, i; - chunk->nruns_avail = 0; - chunk->nruns_adjac = 0; + assert(arena->spare == NULL); - /* - * Initialize the map to contain one maximal free untouched run. - * Mark the pages as zeroed iff chunk_alloc() returned a zeroed - * chunk. - */ - unzeroed = zero ? 0 : CHUNK_MAP_UNZEROED; - arena_mapbits_unallocated_set(chunk, map_bias, arena_maxclass, - unzeroed); - /* - * There is no need to initialize the internal page map entries - * unless the chunk is not zeroed. - */ - if (zero == false) { - VALGRIND_MAKE_MEM_UNDEFINED( - (void *)arena_mapp_get(chunk, map_bias+1), - (size_t)((uintptr_t) arena_mapp_get(chunk, - chunk_npages-1) - (uintptr_t)arena_mapp_get(chunk, - map_bias+1))); - for (i = map_bias+1; i < chunk_npages-1; i++) - arena_mapbits_unzeroed_set(chunk, i, unzeroed); - } else { - VALGRIND_MAKE_MEM_DEFINED( - (void *)arena_mapp_get(chunk, map_bias+1), - (size_t)((uintptr_t) arena_mapp_get(chunk, - chunk_npages-1) - (uintptr_t)arena_mapp_get(chunk, - map_bias+1))); - if (config_debug) { - for (i = map_bias+1; i < chunk_npages-1; i++) { - assert(arena_mapbits_unzeroed_get(chunk, - i) == unzeroed); - } + zero = false; + malloc_mutex_unlock(&arena->lock); + chunk = (arena_chunk_t *)chunk_alloc(chunksize, chunksize, false, + &zero, arena->dss_prec); + malloc_mutex_lock(&arena->lock); + if (chunk == NULL) + return (NULL); + if (config_stats) + arena->stats.mapped += chunksize; + + chunk->arena = arena; + + /* + * Claim that no pages are in use, since the header is merely overhead. + */ + chunk->ndirty = 0; + + chunk->nruns_avail = 0; + chunk->nruns_adjac = 0; + + /* + * Initialize the map to contain one maximal free untouched run. Mark + * the pages as zeroed iff chunk_alloc() returned a zeroed chunk. + */ + unzeroed = zero ? 0 : CHUNK_MAP_UNZEROED; + arena_mapbits_unallocated_set(chunk, map_bias, arena_maxclass, + unzeroed); + /* + * There is no need to initialize the internal page map entries unless + * the chunk is not zeroed. + */ + if (zero == false) { + VALGRIND_MAKE_MEM_UNDEFINED((void *)arena_mapp_get(chunk, + map_bias+1), (size_t)((uintptr_t) arena_mapp_get(chunk, + chunk_npages-1) - (uintptr_t)arena_mapp_get(chunk, + map_bias+1))); + for (i = map_bias+1; i < chunk_npages-1; i++) + arena_mapbits_unzeroed_set(chunk, i, unzeroed); + } else { + VALGRIND_MAKE_MEM_DEFINED((void *)arena_mapp_get(chunk, + map_bias+1), (size_t)((uintptr_t) arena_mapp_get(chunk, + chunk_npages-1) - (uintptr_t)arena_mapp_get(chunk, + map_bias+1))); + if (config_debug) { + for (i = map_bias+1; i < chunk_npages-1; i++) { + assert(arena_mapbits_unzeroed_get(chunk, i) == + unzeroed); } } - arena_mapbits_unallocated_set(chunk, chunk_npages-1, - arena_maxclass, unzeroed); } + arena_mapbits_unallocated_set(chunk, chunk_npages-1, arena_maxclass, + unzeroed); + + return (chunk); +} + +static arena_chunk_t * +arena_chunk_alloc(arena_t *arena) +{ + arena_chunk_t *chunk; + + if (arena->spare != NULL) + chunk = arena_chunk_init_spare(arena); + else + chunk = arena_chunk_init_hard(arena); /* Insert the run into the runs_avail tree. */ arena_avail_insert(arena, chunk, map_bias, chunk_npages-map_bias, @@ -646,8 +657,7 @@ arena_chunk_dealloc(arena_t *arena, arena_chunk_t *chunk) } static arena_run_t * -arena_run_alloc_helper(arena_t *arena, size_t size, bool large, size_t binind, - bool zero) +arena_run_alloc_large_helper(arena_t *arena, size_t size, bool zero) { arena_run_t *run; arena_chunk_map_t *mapelm, key; @@ -662,7 +672,7 @@ arena_run_alloc_helper(arena_t *arena, size_t size, bool large, size_t binind, run = (arena_run_t *)((uintptr_t)run_chunk + (pageind << LG_PAGE)); - arena_run_split(arena, run, size, large, binind, zero); + arena_run_split_large(arena, run, size, zero); return (run); } @@ -670,19 +680,16 @@ arena_run_alloc_helper(arena_t *arena, size_t size, bool large, size_t binind, } static arena_run_t * -arena_run_alloc(arena_t *arena, size_t size, bool large, size_t binind, - bool zero) +arena_run_alloc_large(arena_t *arena, size_t size, bool zero) { arena_chunk_t *chunk; arena_run_t *run; assert(size <= arena_maxclass); assert((size & PAGE_MASK) == 0); - assert((large && binind == BININD_INVALID) || (large == false && binind - != BININD_INVALID)); /* Search the arena's chunks for the lowest best fit. */ - run = arena_run_alloc_helper(arena, size, large, binind, zero); + run = arena_run_alloc_large_helper(arena, size, zero); if (run != NULL) return (run); @@ -692,7 +699,7 @@ arena_run_alloc(arena_t *arena, size_t size, bool large, size_t binind, chunk = arena_chunk_alloc(arena); if (chunk != NULL) { run = (arena_run_t *)((uintptr_t)chunk + (map_bias << LG_PAGE)); - arena_run_split(arena, run, size, large, binind, zero); + arena_run_split_large(arena, run, size, zero); return (run); } @@ -701,7 +708,63 @@ arena_run_alloc(arena_t *arena, size_t size, bool large, size_t binind, * sufficient memory available while this one dropped arena->lock in * arena_chunk_alloc(), so search one more time. */ - return (arena_run_alloc_helper(arena, size, large, binind, zero)); + return (arena_run_alloc_large_helper(arena, size, zero)); +} + +static arena_run_t * +arena_run_alloc_small_helper(arena_t *arena, size_t size, size_t binind) +{ + arena_run_t *run; + arena_chunk_map_t *mapelm, key; + + key.bits = size | CHUNK_MAP_KEY; + mapelm = arena_avail_tree_nsearch(&arena->runs_avail, &key); + if (mapelm != NULL) { + arena_chunk_t *run_chunk = CHUNK_ADDR2BASE(mapelm); + size_t pageind = (((uintptr_t)mapelm - + (uintptr_t)run_chunk->map) / sizeof(arena_chunk_map_t)) + + map_bias; + + run = (arena_run_t *)((uintptr_t)run_chunk + (pageind << + LG_PAGE)); + arena_run_split_small(arena, run, size, binind); + return (run); + } + + return (NULL); +} + +static arena_run_t * +arena_run_alloc_small(arena_t *arena, size_t size, size_t binind) +{ + arena_chunk_t *chunk; + arena_run_t *run; + + assert(size <= arena_maxclass); + assert((size & PAGE_MASK) == 0); + assert(binind != BININD_INVALID); + + /* Search the arena's chunks for the lowest best fit. */ + run = arena_run_alloc_small_helper(arena, size, binind); + if (run != NULL) + return (run); + + /* + * No usable runs. Create a new chunk from which to allocate the run. + */ + chunk = arena_chunk_alloc(arena); + if (chunk != NULL) { + run = (arena_run_t *)((uintptr_t)chunk + (map_bias << LG_PAGE)); + arena_run_split_small(arena, run, size, binind); + return (run); + } + + /* + * arena_chunk_alloc() failed, but another thread may have made + * sufficient memory available while this one dropped arena->lock in + * arena_chunk_alloc(), so search one more time. + */ + return (arena_run_alloc_small_helper(arena, size, binind)); } static inline void @@ -727,48 +790,42 @@ arena_maybe_purge(arena_t *arena) arena_purge(arena, false); } -static inline size_t -arena_chunk_purge(arena_t *arena, arena_chunk_t *chunk, bool all) +static arena_chunk_t * +chunks_dirty_iter_cb(arena_chunk_tree_t *tree, arena_chunk_t *chunk, void *arg) { - size_t npurged; - ql_head(arena_chunk_map_t) mapelms; - arena_chunk_map_t *mapelm; - size_t pageind, npages; - size_t nmadvise; + size_t *ndirty = (size_t *)arg; - ql_new(&mapelms); + assert(chunk->ndirty != 0); + *ndirty += chunk->ndirty; + return (NULL); +} + +static size_t +arena_compute_npurgatory(arena_t *arena, bool all) +{ + size_t npurgatory, npurgeable; /* - * If chunk is the spare, temporarily re-allocate it, 1) so that its - * run is reinserted into runs_avail, and 2) so that it cannot be - * completely discarded by another thread while arena->lock is dropped - * by this thread. Note that the arena_run_dalloc() call will - * implicitly deallocate the chunk, so no explicit action is required - * in this function to deallocate the chunk. - * - * Note that once a chunk contains dirty pages, it cannot again contain - * a single run unless 1) it is a dirty run, or 2) this function purges - * dirty pages and causes the transition to a single clean run. Thus - * (chunk == arena->spare) is possible, but it is not possible for - * this function to be called on the spare unless it contains a dirty - * run. + * Compute the minimum number of pages that this thread should try to + * purge. */ - if (chunk == arena->spare) { - assert(arena_mapbits_dirty_get(chunk, map_bias) != 0); - assert(arena_mapbits_dirty_get(chunk, chunk_npages-1) != 0); + npurgeable = arena->ndirty - arena->npurgatory; - arena_chunk_alloc(arena); - } + if (all == false) { + size_t threshold = (arena->nactive >> opt_lg_dirty_mult); - if (config_stats) - arena->stats.purged += chunk->ndirty; + npurgatory = npurgeable - threshold; + } else + npurgatory = npurgeable; - /* - * Operate on all dirty runs if there is no clean/dirty run - * fragmentation. - */ - if (chunk->nruns_adjac == 0) - all = true; + return (npurgatory); +} + +static void +arena_chunk_stash_dirty(arena_t *arena, arena_chunk_t *chunk, bool all, + arena_chunk_mapelms_t *mapelms) +{ + size_t pageind, npages; /* * Temporarily allocate free dirty runs within chunk. If all is false, @@ -776,7 +833,7 @@ arena_chunk_purge(arena_t *arena, arena_chunk_t *chunk, bool all) * all dirty runs. */ for (pageind = map_bias; pageind < chunk_npages; pageind += npages) { - mapelm = arena_mapp_get(chunk, pageind); + arena_chunk_map_t *mapelm = arena_mapp_get(chunk, pageind); if (arena_mapbits_allocated_get(chunk, pageind) == 0) { size_t run_size = arena_mapbits_unallocated_size_get(chunk, pageind); @@ -792,11 +849,11 @@ arena_chunk_purge(arena_t *arena, arena_chunk_t *chunk, bool all) arena_run_t *run = (arena_run_t *)((uintptr_t) chunk + (uintptr_t)(pageind << LG_PAGE)); - arena_run_split(arena, run, run_size, true, - BININD_INVALID, false); + arena_run_split_large(arena, run, run_size, + false); /* Append to list for later processing. */ ql_elm_new(mapelm, u.ql_link); - ql_tail_insert(&mapelms, mapelm, u.ql_link); + ql_tail_insert(mapelms, mapelm, u.ql_link); } } else { /* Skip run. */ @@ -820,12 +877,20 @@ arena_chunk_purge(arena_t *arena, arena_chunk_t *chunk, bool all) assert(pageind == chunk_npages); assert(chunk->ndirty == 0 || all == false); assert(chunk->nruns_adjac == 0); +} + +static size_t +arena_chunk_purge_stashed(arena_t *arena, arena_chunk_t *chunk, + arena_chunk_mapelms_t *mapelms) +{ + size_t npurged, pageind, npages, nmadvise; + arena_chunk_map_t *mapelm; malloc_mutex_unlock(&arena->lock); if (config_stats) nmadvise = 0; npurged = 0; - ql_foreach(mapelm, &mapelms, u.ql_link) { + ql_foreach(mapelm, mapelms, u.ql_link) { bool unzeroed; size_t flag_unzeroed, i; @@ -859,30 +924,75 @@ arena_chunk_purge(arena_t *arena, arena_chunk_t *chunk, bool all) if (config_stats) arena->stats.nmadvise += nmadvise; + return (npurged); +} + +static void +arena_chunk_unstash_purged(arena_t *arena, arena_chunk_t *chunk, + arena_chunk_mapelms_t *mapelms) +{ + arena_chunk_map_t *mapelm; + size_t pageind; + /* Deallocate runs. */ - for (mapelm = ql_first(&mapelms); mapelm != NULL; - mapelm = ql_first(&mapelms)) { + for (mapelm = ql_first(mapelms); mapelm != NULL; + mapelm = ql_first(mapelms)) { arena_run_t *run; pageind = (((uintptr_t)mapelm - (uintptr_t)chunk->map) / sizeof(arena_chunk_map_t)) + map_bias; run = (arena_run_t *)((uintptr_t)chunk + (uintptr_t)(pageind << LG_PAGE)); - ql_remove(&mapelms, mapelm, u.ql_link); + ql_remove(mapelms, mapelm, u.ql_link); arena_run_dalloc(arena, run, false, true); } - - return (npurged); } -static arena_chunk_t * -chunks_dirty_iter_cb(arena_chunk_tree_t *tree, arena_chunk_t *chunk, void *arg) +static inline size_t +arena_chunk_purge(arena_t *arena, arena_chunk_t *chunk, bool all) { - size_t *ndirty = (size_t *)arg; + size_t npurged; + arena_chunk_mapelms_t mapelms; - assert(chunk->ndirty != 0); - *ndirty += chunk->ndirty; - return (NULL); + ql_new(&mapelms); + + /* + * If chunk is the spare, temporarily re-allocate it, 1) so that its + * run is reinserted into runs_avail, and 2) so that it cannot be + * completely discarded by another thread while arena->lock is dropped + * by this thread. Note that the arena_run_dalloc() call will + * implicitly deallocate the chunk, so no explicit action is required + * in this function to deallocate the chunk. + * + * Note that once a chunk contains dirty pages, it cannot again contain + * a single run unless 1) it is a dirty run, or 2) this function purges + * dirty pages and causes the transition to a single clean run. Thus + * (chunk == arena->spare) is possible, but it is not possible for + * this function to be called on the spare unless it contains a dirty + * run. + */ + if (chunk == arena->spare) { + assert(arena_mapbits_dirty_get(chunk, map_bias) != 0); + assert(arena_mapbits_dirty_get(chunk, chunk_npages-1) != 0); + + arena_chunk_alloc(arena); + } + + if (config_stats) + arena->stats.purged += chunk->ndirty; + + /* + * Operate on all dirty runs if there is no clean/dirty run + * fragmentation. + */ + if (chunk->nruns_adjac == 0) + all = true; + + arena_chunk_stash_dirty(arena, chunk, all, &mapelms); + npurged = arena_chunk_purge_stashed(arena, chunk, &mapelms); + arena_chunk_unstash_purged(arena, chunk, &mapelms); + + return (npurged); } static void @@ -905,21 +1015,11 @@ arena_purge(arena_t *arena, bool all) arena->stats.npurge++; /* - * Compute the minimum number of pages that this thread should try to - * purge, and add the result to arena->npurgatory. This will keep - * multiple threads from racing to reduce ndirty below the threshold. + * Add the minimum number of pages this thread should try to purge to + * arena->npurgatory. This will keep multiple threads from racing to + * reduce ndirty below the threshold. */ - { - size_t npurgeable = arena->ndirty - arena->npurgatory; - - if (all == false) { - size_t threshold = (arena->nactive >> - opt_lg_dirty_mult); - - npurgatory = npurgeable - threshold; - } else - npurgatory = npurgeable; - } + npurgatory = arena_compute_npurgatory(arena, all); arena->npurgatory += npurgatory; while (npurgatory > 0) { @@ -986,61 +1086,12 @@ arena_purge_all(arena_t *arena) } static void -arena_run_dalloc(arena_t *arena, arena_run_t *run, bool dirty, bool cleaned) +arena_run_coalesce(arena_t *arena, arena_chunk_t *chunk, size_t *p_size, + size_t *p_run_ind, size_t *p_run_pages, size_t flag_dirty) { - arena_chunk_t *chunk; - size_t size, run_ind, run_pages, flag_dirty; - - chunk = (arena_chunk_t *)CHUNK_ADDR2BASE(run); - run_ind = (size_t)(((uintptr_t)run - (uintptr_t)chunk) >> LG_PAGE); - assert(run_ind >= map_bias); - assert(run_ind < chunk_npages); - if (arena_mapbits_large_get(chunk, run_ind) != 0) { - size = arena_mapbits_large_size_get(chunk, run_ind); - assert(size == PAGE || - arena_mapbits_large_size_get(chunk, - run_ind+(size>>LG_PAGE)-1) == 0); - } else { - size_t binind = arena_bin_index(arena, run->bin); - arena_bin_info_t *bin_info = &arena_bin_info[binind]; - size = bin_info->run_size; - } - run_pages = (size >> LG_PAGE); - if (config_stats) { - /* - * Update stats_cactive if nactive is crossing a chunk - * multiple. - */ - size_t cactive_diff = CHUNK_CEILING(arena->nactive << LG_PAGE) - - CHUNK_CEILING((arena->nactive - run_pages) << LG_PAGE); - if (cactive_diff != 0) - stats_cactive_sub(cactive_diff); - } - arena->nactive -= run_pages; - - /* - * The run is dirty if the caller claims to have dirtied it, as well as - * if it was already dirty before being allocated and the caller - * doesn't claim to have cleaned it. - */ - assert(arena_mapbits_dirty_get(chunk, run_ind) == - arena_mapbits_dirty_get(chunk, run_ind+run_pages-1)); - if (cleaned == false && arena_mapbits_dirty_get(chunk, run_ind) != 0) - dirty = true; - flag_dirty = dirty ? CHUNK_MAP_DIRTY : 0; - - /* Mark pages as unallocated in the chunk map. */ - if (dirty) { - arena_mapbits_unallocated_set(chunk, run_ind, size, - CHUNK_MAP_DIRTY); - arena_mapbits_unallocated_set(chunk, run_ind+run_pages-1, size, - CHUNK_MAP_DIRTY); - } else { - arena_mapbits_unallocated_set(chunk, run_ind, size, - arena_mapbits_unzeroed_get(chunk, run_ind)); - arena_mapbits_unallocated_set(chunk, run_ind+run_pages-1, size, - arena_mapbits_unzeroed_get(chunk, run_ind+run_pages-1)); - } + size_t size = *p_size; + size_t run_ind = *p_run_ind; + size_t run_pages = *p_run_pages; /* Try to coalesce forward. */ if (run_ind + run_pages < chunk_npages && @@ -1070,8 +1121,9 @@ arena_run_dalloc(arena_t *arena, arena_run_t *run, bool dirty, bool cleaned) } /* Try to coalesce backward. */ - if (run_ind > map_bias && arena_mapbits_allocated_get(chunk, run_ind-1) - == 0 && arena_mapbits_dirty_get(chunk, run_ind-1) == flag_dirty) { + if (run_ind > map_bias && arena_mapbits_allocated_get(chunk, + run_ind-1) == 0 && arena_mapbits_dirty_get(chunk, run_ind-1) == + flag_dirty) { size_t prun_size = arena_mapbits_unallocated_size_get(chunk, run_ind-1); size_t prun_pages = prun_size >> LG_PAGE; @@ -1096,6 +1148,62 @@ arena_run_dalloc(arena_t *arena, arena_run_t *run, bool dirty, bool cleaned) size); } + *p_size = size; + *p_run_ind = run_ind; + *p_run_pages = run_pages; +} + +static void +arena_run_dalloc(arena_t *arena, arena_run_t *run, bool dirty, bool cleaned) +{ + arena_chunk_t *chunk; + size_t size, run_ind, run_pages, flag_dirty; + + chunk = (arena_chunk_t *)CHUNK_ADDR2BASE(run); + run_ind = (size_t)(((uintptr_t)run - (uintptr_t)chunk) >> LG_PAGE); + assert(run_ind >= map_bias); + assert(run_ind < chunk_npages); + if (arena_mapbits_large_get(chunk, run_ind) != 0) { + size = arena_mapbits_large_size_get(chunk, run_ind); + assert(size == PAGE || + arena_mapbits_large_size_get(chunk, + run_ind+(size>>LG_PAGE)-1) == 0); + } else { + size_t binind = arena_bin_index(arena, run->bin); + arena_bin_info_t *bin_info = &arena_bin_info[binind]; + size = bin_info->run_size; + } + run_pages = (size >> LG_PAGE); + arena_cactive_update(arena, 0, run_pages); + arena->nactive -= run_pages; + + /* + * The run is dirty if the caller claims to have dirtied it, as well as + * if it was already dirty before being allocated and the caller + * doesn't claim to have cleaned it. + */ + assert(arena_mapbits_dirty_get(chunk, run_ind) == + arena_mapbits_dirty_get(chunk, run_ind+run_pages-1)); + if (cleaned == false && arena_mapbits_dirty_get(chunk, run_ind) != 0) + dirty = true; + flag_dirty = dirty ? CHUNK_MAP_DIRTY : 0; + + /* Mark pages as unallocated in the chunk map. */ + if (dirty) { + arena_mapbits_unallocated_set(chunk, run_ind, size, + CHUNK_MAP_DIRTY); + arena_mapbits_unallocated_set(chunk, run_ind+run_pages-1, size, + CHUNK_MAP_DIRTY); + } else { + arena_mapbits_unallocated_set(chunk, run_ind, size, + arena_mapbits_unzeroed_get(chunk, run_ind)); + arena_mapbits_unallocated_set(chunk, run_ind+run_pages-1, size, + arena_mapbits_unzeroed_get(chunk, run_ind+run_pages-1)); + } + + arena_run_coalesce(arena, chunk, &size, &run_ind, &run_pages, + flag_dirty); + /* Insert into runs_avail, now that coalescing is complete. */ assert(arena_mapbits_unallocated_size_get(chunk, run_ind) == arena_mapbits_unallocated_size_get(chunk, run_ind+run_pages-1)); @@ -1263,7 +1371,7 @@ arena_bin_nonfull_run_get(arena_t *arena, arena_bin_t *bin) malloc_mutex_unlock(&bin->lock); /******************************/ malloc_mutex_lock(&arena->lock); - run = arena_run_alloc(arena, bin_info->run_size, false, binind, false); + run = arena_run_alloc_small(arena, bin_info->run_size, binind); if (run != NULL) { bitmap_t *bitmap = (bitmap_t *)((uintptr_t)run + (uintptr_t)bin_info->bitmap_offset); @@ -1286,7 +1394,7 @@ arena_bin_nonfull_run_get(arena_t *arena, arena_bin_t *bin) } /* - * arena_run_alloc() failed, but another thread may have made + * arena_run_alloc_small() failed, but another thread may have made * sufficient memory available while this one dropped bin->lock above, * so search one more time. */ @@ -1321,12 +1429,12 @@ arena_bin_malloc_hard(arena_t *arena, arena_bin_t *bin) arena_chunk_t *chunk; /* - * arena_run_alloc() may have allocated run, or it may - * have pulled run from the bin's run tree. Therefore - * it is unsafe to make any assumptions about how run - * has previously been used, and arena_bin_lower_run() - * must be called, as if a region were just deallocated - * from the run. + * arena_run_alloc_small() may have allocated run, or + * it may have pulled run from the bin's run tree. + * Therefore it is unsafe to make any assumptions about + * how run has previously been used, and + * arena_bin_lower_run() must be called, as if a region + * were just deallocated from the run. */ chunk = (arena_chunk_t *)CHUNK_ADDR2BASE(run); if (run->nfree == bin_info->nregs) @@ -1404,8 +1512,28 @@ arena_alloc_junk_small(void *ptr, arena_bin_info_t *bin_info, bool zero) } } -void -arena_dalloc_junk_small(void *ptr, arena_bin_info_t *bin_info) +#ifdef JEMALLOC_JET +#undef arena_redzone_corruption +#define arena_redzone_corruption JEMALLOC_N(arena_redzone_corruption_impl) +#endif +static void +arena_redzone_corruption(void *ptr, size_t usize, bool after, + size_t offset, uint8_t byte) +{ + + malloc_printf("<jemalloc>: Corrupt redzone %zu byte%s %s %p " + "(size %zu), byte=%#x\n", offset, (offset == 1) ? "" : "s", + after ? "after" : "before", ptr, usize, byte); +} +#ifdef JEMALLOC_JET +#undef arena_redzone_corruption +#define arena_redzone_corruption JEMALLOC_N(arena_redzone_corruption) +arena_redzone_corruption_t *arena_redzone_corruption = + JEMALLOC_N(arena_redzone_corruption_impl); +#endif + +static void +arena_redzones_validate(void *ptr, arena_bin_info_t *bin_info, bool reset) { size_t size = bin_info->reg_size; size_t redzone_size = bin_info->redzone_size; @@ -1413,29 +1541,61 @@ arena_dalloc_junk_small(void *ptr, arena_bin_info_t *bin_info) bool error = false; for (i = 1; i <= redzone_size; i++) { - unsigned byte; - if ((byte = *(uint8_t *)((uintptr_t)ptr - i)) != 0xa5) { + uint8_t *byte = (uint8_t *)((uintptr_t)ptr - i); + if (*byte != 0xa5) { error = true; - malloc_printf("<jemalloc>: Corrupt redzone " - "%zu byte%s before %p (size %zu), byte=%#x\n", i, - (i == 1) ? "" : "s", ptr, size, byte); + arena_redzone_corruption(ptr, size, false, i, *byte); + if (reset) + *byte = 0xa5; } } for (i = 0; i < redzone_size; i++) { - unsigned byte; - if ((byte = *(uint8_t *)((uintptr_t)ptr + size + i)) != 0xa5) { + uint8_t *byte = (uint8_t *)((uintptr_t)ptr + size + i); + if (*byte != 0xa5) { error = true; - malloc_printf("<jemalloc>: Corrupt redzone " - "%zu byte%s after end of %p (size %zu), byte=%#x\n", - i, (i == 1) ? "" : "s", ptr, size, byte); + arena_redzone_corruption(ptr, size, true, i, *byte); + if (reset) + *byte = 0xa5; } } if (opt_abort && error) abort(); +} +#ifdef JEMALLOC_JET +#undef arena_dalloc_junk_small +#define arena_dalloc_junk_small JEMALLOC_N(arena_dalloc_junk_small_impl) +#endif +void +arena_dalloc_junk_small(void *ptr, arena_bin_info_t *bin_info) +{ + size_t redzone_size = bin_info->redzone_size; + + arena_redzones_validate(ptr, bin_info, false); memset((void *)((uintptr_t)ptr - redzone_size), 0x5a, bin_info->reg_interval); } +#ifdef JEMALLOC_JET +#undef arena_dalloc_junk_small +#define arena_dalloc_junk_small JEMALLOC_N(arena_dalloc_junk_small) +arena_dalloc_junk_small_t *arena_dalloc_junk_small = + JEMALLOC_N(arena_dalloc_junk_small_impl); +#endif + +void +arena_quarantine_junk_small(void *ptr, size_t usize) +{ + size_t binind; + arena_bin_info_t *bin_info; + cassert(config_fill); + assert(opt_junk); + assert(opt_quarantine); + assert(usize <= SMALL_MAXCLASS); + + binind = SMALL_SIZE2BIN(usize); + bin_info = &arena_bin_info[binind]; + arena_redzones_validate(ptr, bin_info, true); +} void * arena_malloc_small(arena_t *arena, size_t size, bool zero) @@ -1500,7 +1660,7 @@ arena_malloc_large(arena_t *arena, size_t size, bool zero) /* Large allocation. */ size = PAGE_CEILING(size); malloc_mutex_lock(&arena->lock); - ret = (void *)arena_run_alloc(arena, size, true, BININD_INVALID, zero); + ret = (void *)arena_run_alloc_large(arena, size, zero); if (ret == NULL) { malloc_mutex_unlock(&arena->lock); return (NULL); @@ -1546,7 +1706,7 @@ arena_palloc(arena_t *arena, size_t size, size_t alignment, bool zero) alloc_size = size + alignment - PAGE; malloc_mutex_lock(&arena->lock); - run = arena_run_alloc(arena, alloc_size, true, BININD_INVALID, zero); + run = arena_run_alloc_large(arena, alloc_size, false); if (run == NULL) { malloc_mutex_unlock(&arena->lock); return (NULL); @@ -1566,6 +1726,7 @@ arena_palloc(arena_t *arena, size_t size, size_t alignment, bool zero) arena_run_trim_tail(arena, chunk, ret, size + trailsize, size, false); } + arena_run_init_large(arena, (arena_run_t *)ret, size, zero); if (config_stats) { arena->stats.nmalloc_large++; @@ -1769,21 +1930,38 @@ arena_dalloc_small(arena_t *arena, arena_chunk_t *chunk, void *ptr, arena_dalloc_bin(arena, chunk, ptr, pageind, mapelm); } +#ifdef JEMALLOC_JET +#undef arena_dalloc_junk_large +#define arena_dalloc_junk_large JEMALLOC_N(arena_dalloc_junk_large_impl) +#endif +static void +arena_dalloc_junk_large(void *ptr, size_t usize) +{ + + if (config_fill && opt_junk) + memset(ptr, 0x5a, usize); +} +#ifdef JEMALLOC_JET +#undef arena_dalloc_junk_large +#define arena_dalloc_junk_large JEMALLOC_N(arena_dalloc_junk_large) +arena_dalloc_junk_large_t *arena_dalloc_junk_large = + JEMALLOC_N(arena_dalloc_junk_large_impl); +#endif + void arena_dalloc_large_locked(arena_t *arena, arena_chunk_t *chunk, void *ptr) { if (config_fill || config_stats) { size_t pageind = ((uintptr_t)ptr - (uintptr_t)chunk) >> LG_PAGE; - size_t size = arena_mapbits_large_size_get(chunk, pageind); + size_t usize = arena_mapbits_large_size_get(chunk, pageind); - if (config_fill && config_stats && opt_junk) - memset(ptr, 0x5a, size); + arena_dalloc_junk_large(ptr, usize); if (config_stats) { arena->stats.ndalloc_large++; - arena->stats.allocated_large -= size; - arena->stats.lstats[(size >> LG_PAGE) - 1].ndalloc++; - arena->stats.lstats[(size >> LG_PAGE) - 1].curruns--; + arena->stats.allocated_large -= usize; + arena->stats.lstats[(usize >> LG_PAGE) - 1].ndalloc++; + arena->stats.lstats[(usize >> LG_PAGE) - 1].curruns--; } } @@ -1854,9 +2032,8 @@ arena_ralloc_large_grow(arena_t *arena, arena_chunk_t *chunk, void *ptr, size_t flag_dirty; size_t splitsize = (oldsize + followsize <= size + extra) ? followsize : size + extra - oldsize; - arena_run_split(arena, (arena_run_t *)((uintptr_t)chunk + - ((pageind+npages) << LG_PAGE)), splitsize, true, - BININD_INVALID, zero); + arena_run_split_large(arena, (arena_run_t *)((uintptr_t)chunk + + ((pageind+npages) << LG_PAGE)), splitsize, zero); size = oldsize + splitsize; npages = size >> LG_PAGE; @@ -1895,6 +2072,26 @@ arena_ralloc_large_grow(arena_t *arena, arena_chunk_t *chunk, void *ptr, return (true); } +#ifdef JEMALLOC_JET +#undef arena_ralloc_junk_large +#define arena_ralloc_junk_large JEMALLOC_N(arena_ralloc_junk_large_impl) +#endif +static void +arena_ralloc_junk_large(void *ptr, size_t old_usize, size_t usize) +{ + + if (config_fill && opt_junk) { + memset((void *)((uintptr_t)ptr + usize), 0x5a, + old_usize - usize); + } +} +#ifdef JEMALLOC_JET +#undef arena_ralloc_junk_large +#define arena_ralloc_junk_large JEMALLOC_N(arena_ralloc_junk_large) +arena_ralloc_junk_large_t *arena_ralloc_junk_large = + JEMALLOC_N(arena_ralloc_junk_large_impl); +#endif + /* * Try to resize a large allocation, in order to avoid copying. This will * always fail if growing an object, and the following run is already in use. @@ -1908,10 +2105,6 @@ arena_ralloc_large(void *ptr, size_t oldsize, size_t size, size_t extra, psize = PAGE_CEILING(size + extra); if (psize == oldsize) { /* Same size class. */ - if (config_fill && opt_junk && size < oldsize) { - memset((void *)((uintptr_t)ptr + size), 0x5a, oldsize - - size); - } return (false); } else { arena_chunk_t *chunk; @@ -1922,10 +2115,7 @@ arena_ralloc_large(void *ptr, size_t oldsize, size_t size, size_t extra, if (psize < oldsize) { /* Fill before shrinking in order avoid a race. */ - if (config_fill && opt_junk) { - memset((void *)((uintptr_t)ptr + size), 0x5a, - oldsize - size); - } + arena_ralloc_junk_large(ptr, oldsize, psize); arena_ralloc_large_shrink(arena, chunk, ptr, oldsize, psize); return (false); @@ -1933,17 +2123,23 @@ arena_ralloc_large(void *ptr, size_t oldsize, size_t size, size_t extra, bool ret = arena_ralloc_large_grow(arena, chunk, ptr, oldsize, PAGE_CEILING(size), psize - PAGE_CEILING(size), zero); - if (config_fill && ret == false && zero == false && - opt_zero) { - memset((void *)((uintptr_t)ptr + oldsize), 0, - size - oldsize); + if (config_fill && ret == false && zero == false) { + if (opt_junk) { + memset((void *)((uintptr_t)ptr + + oldsize), 0xa5, isalloc(ptr, + config_prof) - oldsize); + } else if (opt_zero) { + memset((void *)((uintptr_t)ptr + + oldsize), 0, isalloc(ptr, + config_prof) - oldsize); + } } return (ret); } } } -void * +bool arena_ralloc_no_move(void *ptr, size_t oldsize, size_t size, size_t extra, bool zero) { @@ -1958,25 +2154,20 @@ arena_ralloc_no_move(void *ptr, size_t oldsize, size_t size, size_t extra, if ((size + extra <= SMALL_MAXCLASS && SMALL_SIZE2BIN(size + extra) == SMALL_SIZE2BIN(oldsize)) || (size <= oldsize && - size + extra >= oldsize)) { - if (config_fill && opt_junk && size < oldsize) { - memset((void *)((uintptr_t)ptr + size), - 0x5a, oldsize - size); - } - return (ptr); - } + size + extra >= oldsize)) + return (false); } else { assert(size <= arena_maxclass); if (size + extra > SMALL_MAXCLASS) { if (arena_ralloc_large(ptr, oldsize, size, extra, zero) == false) - return (ptr); + return (false); } } } /* Reallocation would require a move. */ - return (NULL); + return (true); } void * @@ -1988,9 +2179,8 @@ arena_ralloc(arena_t *arena, void *ptr, size_t oldsize, size_t size, size_t copysize; /* Try to avoid moving the allocation. */ - ret = arena_ralloc_no_move(ptr, oldsize, size, extra, zero); - if (ret != NULL) - return (ret); + if (arena_ralloc_no_move(ptr, oldsize, size, extra, zero) == false) + return (ptr); /* * size and oldsize are different enough that we need to move the @@ -2001,7 +2191,7 @@ arena_ralloc(arena_t *arena, void *ptr, size_t oldsize, size_t size, size_t usize = sa2u(size + extra, alignment); if (usize == 0) return (NULL); - ret = ipallocx(usize, alignment, zero, try_tcache_alloc, arena); + ret = ipalloct(usize, alignment, zero, try_tcache_alloc, arena); } else ret = arena_malloc(arena, size + extra, zero, try_tcache_alloc); @@ -2013,7 +2203,7 @@ arena_ralloc(arena_t *arena, void *ptr, size_t oldsize, size_t size, size_t usize = sa2u(size, alignment); if (usize == 0) return (NULL); - ret = ipallocx(usize, alignment, zero, try_tcache_alloc, + ret = ipalloct(usize, alignment, zero, try_tcache_alloc, arena); } else ret = arena_malloc(arena, size, zero, try_tcache_alloc); @@ -2031,7 +2221,7 @@ arena_ralloc(arena_t *arena, void *ptr, size_t oldsize, size_t size, copysize = (size < oldsize) ? size : oldsize; VALGRIND_MAKE_MEM_UNDEFINED(ret, copysize); memcpy(ret, ptr, copysize); - iqallocx(ptr, try_tcache_dalloc); + iqalloct(ptr, try_tcache_dalloc); return (ret); } diff --git a/src/bitmap.c b/src/bitmap.c index b47e2629..e2bd907d 100644 --- a/src/bitmap.c +++ b/src/bitmap.c @@ -1,4 +1,4 @@ -#define JEMALLOC_BITMAP_C_ +#define JEMALLOC_BITMAP_C_ #include "jemalloc/internal/jemalloc_internal.h" /******************************************************************************/ diff --git a/src/chunk.c b/src/chunk.c index b17f43f0..90ab116a 100644 --- a/src/chunk.c +++ b/src/chunk.c @@ -180,7 +180,7 @@ chunk_alloc(size_t size, size_t alignment, bool base, bool *zero, label_return: if (ret != NULL) { if (config_ivsalloc && base == false) { - if (rtree_set(chunks_rtree, (uintptr_t)ret, ret)) { + if (rtree_set(chunks_rtree, (uintptr_t)ret, 1)) { chunk_dealloc(ret, size, true); return (NULL); } @@ -321,7 +321,7 @@ chunk_dealloc(void *chunk, size_t size, bool unmap) assert((size & chunksize_mask) == 0); if (config_ivsalloc) - rtree_set(chunks_rtree, (uintptr_t)chunk, NULL); + rtree_set(chunks_rtree, (uintptr_t)chunk, 0); if (config_stats || config_prof) { malloc_mutex_lock(&chunks_mtx); assert(stats_chunks.curchunks >= (size / chunksize)); @@ -356,7 +356,7 @@ chunk_boot(void) extent_tree_ad_new(&chunks_ad_dss); if (config_ivsalloc) { chunks_rtree = rtree_new((ZU(1) << (LG_SIZEOF_PTR+3)) - - opt_lg_chunk); + opt_lg_chunk, base_alloc, NULL); if (chunks_rtree == NULL) return (true); } @@ -368,7 +368,7 @@ void chunk_prefork(void) { - malloc_mutex_lock(&chunks_mtx); + malloc_mutex_prefork(&chunks_mtx); if (config_ivsalloc) rtree_prefork(chunks_rtree); chunk_dss_prefork(); diff --git a/src/chunk_dss.c b/src/chunk_dss.c index 24781cc5..510bb8be 100644 --- a/src/chunk_dss.c +++ b/src/chunk_dss.c @@ -28,16 +28,17 @@ static void *dss_max; /******************************************************************************/ -#ifndef JEMALLOC_HAVE_SBRK static void * -sbrk(intptr_t increment) +chunk_dss_sbrk(intptr_t increment) { +#ifdef JEMALLOC_HAVE_SBRK + return (sbrk(increment)); +#else not_implemented(); - return (NULL); -} #endif +} dss_prec_t chunk_dss_prec_get(void) @@ -93,7 +94,7 @@ chunk_alloc_dss(size_t size, size_t alignment, bool *zero) */ do { /* Get the current end of the DSS. */ - dss_max = sbrk(0); + dss_max = chunk_dss_sbrk(0); /* * Calculate how much padding is necessary to * chunk-align the end of the DSS. @@ -117,7 +118,7 @@ chunk_alloc_dss(size_t size, size_t alignment, bool *zero) return (NULL); } incr = gap_size + cpad_size + size; - dss_prev = sbrk(incr); + dss_prev = chunk_dss_sbrk(incr); if (dss_prev == dss_max) { /* Success. */ dss_max = dss_next; @@ -163,7 +164,7 @@ chunk_dss_boot(void) if (malloc_mutex_init(&dss_mtx)) return (true); - dss_base = sbrk(0); + dss_base = chunk_dss_sbrk(0); dss_prev = dss_base; dss_max = dss_base; diff --git a/src/chunk_mmap.c b/src/chunk_mmap.c index 8a42e759..2056d793 100644 --- a/src/chunk_mmap.c +++ b/src/chunk_mmap.c @@ -43,7 +43,7 @@ pages_map(void *addr, size_t size) if (munmap(ret, size) == -1) { char buf[BUFERROR_BUF]; - buferror(buf, sizeof(buf)); + buferror(get_errno(), buf, sizeof(buf)); malloc_printf("<jemalloc: Error in munmap(): %s\n", buf); if (opt_abort) @@ -69,7 +69,7 @@ pages_unmap(void *addr, size_t size) { char buf[BUFERROR_BUF]; - buferror(buf, sizeof(buf)); + buferror(get_errno(), buf, sizeof(buf)); malloc_printf("<jemalloc>: Error in " #ifdef _WIN32 "VirtualFree" @@ -49,7 +49,7 @@ static void ckh_shrink(ckh_t *ckh); * Search bucket for key and return the cell number if found; SIZE_T_MAX * otherwise. */ -JEMALLOC_INLINE size_t +JEMALLOC_INLINE_C size_t ckh_bucket_search(ckh_t *ckh, size_t bucket, const void *key) { ckhc_t *cell; @@ -67,7 +67,7 @@ ckh_bucket_search(ckh_t *ckh, size_t bucket, const void *key) /* * Search table for key and return cell number if found; SIZE_T_MAX otherwise. */ -JEMALLOC_INLINE size_t +JEMALLOC_INLINE_C size_t ckh_isearch(ckh_t *ckh, const void *key) { size_t hashes[2], bucket, cell; @@ -88,7 +88,7 @@ ckh_isearch(ckh_t *ckh, const void *key) return (cell); } -JEMALLOC_INLINE bool +JEMALLOC_INLINE_C bool ckh_try_bucket_insert(ckh_t *ckh, size_t bucket, const void *key, const void *data) { @@ -120,7 +120,7 @@ ckh_try_bucket_insert(ckh_t *ckh, size_t bucket, const void *key, * eviction/relocation procedure until either success or detection of an * eviction/relocation bucket cycle. */ -JEMALLOC_INLINE bool +JEMALLOC_INLINE_C bool ckh_evict_reloc_insert(ckh_t *ckh, size_t argbucket, void const **argkey, void const **argdata) { @@ -190,7 +190,7 @@ ckh_evict_reloc_insert(ckh_t *ckh, size_t argbucket, void const **argkey, } } -JEMALLOC_INLINE bool +JEMALLOC_INLINE_C bool ckh_try_insert(ckh_t *ckh, void const**argkey, void const**argdata) { size_t hashes[2], bucket; @@ -219,7 +219,7 @@ ckh_try_insert(ckh_t *ckh, void const**argkey, void const**argdata) * Try to rebuild the hash table from scratch by inserting all items from the * old table into the new. */ -JEMALLOC_INLINE bool +JEMALLOC_INLINE_C bool ckh_rebuild(ckh_t *ckh, ckhc_t *aTab) { size_t count, i, nins; @@ -929,7 +929,7 @@ void ctl_prefork(void) { - malloc_mutex_lock(&ctl_mtx); + malloc_mutex_prefork(&ctl_mtx); } void @@ -1110,6 +1110,8 @@ label_return: \ return (ret); \ } +/******************************************************************************/ + CTL_RO_NL_GEN(version, JEMALLOC_VERSION, const char *) static int @@ -1131,49 +1133,52 @@ label_return: return (ret); } -static int -thread_tcache_enabled_ctl(const size_t *mib, size_t miblen, void *oldp, - size_t *oldlenp, void *newp, size_t newlen) -{ - int ret; - bool oldval; - - if (config_tcache == false) - return (ENOENT); - - oldval = tcache_enabled_get(); - if (newp != NULL) { - if (newlen != sizeof(bool)) { - ret = EINVAL; - goto label_return; - } - tcache_enabled_set(*(bool *)newp); - } - READ(oldval, bool); - - ret = 0; -label_return: - return (ret); -} - -static int -thread_tcache_flush_ctl(const size_t *mib, size_t miblen, void *oldp, - size_t *oldlenp, void *newp, size_t newlen) -{ - int ret; +/******************************************************************************/ - if (config_tcache == false) - return (ENOENT); +CTL_RO_BOOL_CONFIG_GEN(config_debug) +CTL_RO_BOOL_CONFIG_GEN(config_dss) +CTL_RO_BOOL_CONFIG_GEN(config_fill) +CTL_RO_BOOL_CONFIG_GEN(config_lazy_lock) +CTL_RO_BOOL_CONFIG_GEN(config_mremap) +CTL_RO_BOOL_CONFIG_GEN(config_munmap) +CTL_RO_BOOL_CONFIG_GEN(config_prof) +CTL_RO_BOOL_CONFIG_GEN(config_prof_libgcc) +CTL_RO_BOOL_CONFIG_GEN(config_prof_libunwind) +CTL_RO_BOOL_CONFIG_GEN(config_stats) +CTL_RO_BOOL_CONFIG_GEN(config_tcache) +CTL_RO_BOOL_CONFIG_GEN(config_tls) +CTL_RO_BOOL_CONFIG_GEN(config_utrace) +CTL_RO_BOOL_CONFIG_GEN(config_valgrind) +CTL_RO_BOOL_CONFIG_GEN(config_xmalloc) - READONLY(); - WRITEONLY(); +/******************************************************************************/ - tcache_flush(); +CTL_RO_NL_GEN(opt_abort, opt_abort, bool) +CTL_RO_NL_GEN(opt_dss, opt_dss, const char *) +CTL_RO_NL_GEN(opt_lg_chunk, opt_lg_chunk, size_t) +CTL_RO_NL_GEN(opt_narenas, opt_narenas, size_t) +CTL_RO_NL_GEN(opt_lg_dirty_mult, opt_lg_dirty_mult, ssize_t) +CTL_RO_NL_GEN(opt_stats_print, opt_stats_print, bool) +CTL_RO_NL_CGEN(config_fill, opt_junk, opt_junk, bool) +CTL_RO_NL_CGEN(config_fill, opt_quarantine, opt_quarantine, size_t) +CTL_RO_NL_CGEN(config_fill, opt_redzone, opt_redzone, bool) +CTL_RO_NL_CGEN(config_fill, opt_zero, opt_zero, bool) +CTL_RO_NL_CGEN(config_utrace, opt_utrace, opt_utrace, bool) +CTL_RO_NL_CGEN(config_valgrind, opt_valgrind, opt_valgrind, bool) +CTL_RO_NL_CGEN(config_xmalloc, opt_xmalloc, opt_xmalloc, bool) +CTL_RO_NL_CGEN(config_tcache, opt_tcache, opt_tcache, bool) +CTL_RO_NL_CGEN(config_tcache, opt_lg_tcache_max, opt_lg_tcache_max, ssize_t) +CTL_RO_NL_CGEN(config_prof, opt_prof, opt_prof, bool) +CTL_RO_NL_CGEN(config_prof, opt_prof_prefix, opt_prof_prefix, const char *) +CTL_RO_CGEN(config_prof, opt_prof_active, opt_prof_active, bool) /* Mutable. */ +CTL_RO_NL_CGEN(config_prof, opt_lg_prof_sample, opt_lg_prof_sample, size_t) +CTL_RO_NL_CGEN(config_prof, opt_prof_accum, opt_prof_accum, bool) +CTL_RO_NL_CGEN(config_prof, opt_lg_prof_interval, opt_lg_prof_interval, ssize_t) +CTL_RO_NL_CGEN(config_prof, opt_prof_gdump, opt_prof_gdump, bool) +CTL_RO_NL_CGEN(config_prof, opt_prof_final, opt_prof_final, bool) +CTL_RO_NL_CGEN(config_prof, opt_prof_leak, opt_prof_leak, bool) - ret = 0; -label_return: - return (ret); -} +/******************************************************************************/ static int thread_arena_ctl(const size_t *mib, size_t miblen, void *oldp, size_t *oldlenp, @@ -1235,50 +1240,49 @@ CTL_RO_NL_CGEN(config_stats, thread_deallocated, CTL_RO_NL_CGEN(config_stats, thread_deallocatedp, &thread_allocated_tsd_get()->deallocated, uint64_t *) -/******************************************************************************/ +static int +thread_tcache_enabled_ctl(const size_t *mib, size_t miblen, void *oldp, + size_t *oldlenp, void *newp, size_t newlen) +{ + int ret; + bool oldval; -CTL_RO_BOOL_CONFIG_GEN(config_debug) -CTL_RO_BOOL_CONFIG_GEN(config_dss) -CTL_RO_BOOL_CONFIG_GEN(config_fill) -CTL_RO_BOOL_CONFIG_GEN(config_lazy_lock) -CTL_RO_BOOL_CONFIG_GEN(config_mremap) -CTL_RO_BOOL_CONFIG_GEN(config_munmap) -CTL_RO_BOOL_CONFIG_GEN(config_prof) -CTL_RO_BOOL_CONFIG_GEN(config_prof_libgcc) -CTL_RO_BOOL_CONFIG_GEN(config_prof_libunwind) -CTL_RO_BOOL_CONFIG_GEN(config_stats) -CTL_RO_BOOL_CONFIG_GEN(config_tcache) -CTL_RO_BOOL_CONFIG_GEN(config_tls) -CTL_RO_BOOL_CONFIG_GEN(config_utrace) -CTL_RO_BOOL_CONFIG_GEN(config_valgrind) -CTL_RO_BOOL_CONFIG_GEN(config_xmalloc) + if (config_tcache == false) + return (ENOENT); -/******************************************************************************/ + oldval = tcache_enabled_get(); + if (newp != NULL) { + if (newlen != sizeof(bool)) { + ret = EINVAL; + goto label_return; + } + tcache_enabled_set(*(bool *)newp); + } + READ(oldval, bool); -CTL_RO_NL_GEN(opt_abort, opt_abort, bool) -CTL_RO_NL_GEN(opt_dss, opt_dss, const char *) -CTL_RO_NL_GEN(opt_lg_chunk, opt_lg_chunk, size_t) -CTL_RO_NL_GEN(opt_narenas, opt_narenas, size_t) -CTL_RO_NL_GEN(opt_lg_dirty_mult, opt_lg_dirty_mult, ssize_t) -CTL_RO_NL_GEN(opt_stats_print, opt_stats_print, bool) -CTL_RO_NL_CGEN(config_fill, opt_junk, opt_junk, bool) -CTL_RO_NL_CGEN(config_fill, opt_zero, opt_zero, bool) -CTL_RO_NL_CGEN(config_fill, opt_quarantine, opt_quarantine, size_t) -CTL_RO_NL_CGEN(config_fill, opt_redzone, opt_redzone, bool) -CTL_RO_NL_CGEN(config_utrace, opt_utrace, opt_utrace, bool) -CTL_RO_NL_CGEN(config_valgrind, opt_valgrind, opt_valgrind, bool) -CTL_RO_NL_CGEN(config_xmalloc, opt_xmalloc, opt_xmalloc, bool) -CTL_RO_NL_CGEN(config_tcache, opt_tcache, opt_tcache, bool) -CTL_RO_NL_CGEN(config_tcache, opt_lg_tcache_max, opt_lg_tcache_max, ssize_t) -CTL_RO_NL_CGEN(config_prof, opt_prof, opt_prof, bool) -CTL_RO_NL_CGEN(config_prof, opt_prof_prefix, opt_prof_prefix, const char *) -CTL_RO_CGEN(config_prof, opt_prof_active, opt_prof_active, bool) /* Mutable. */ -CTL_RO_NL_CGEN(config_prof, opt_lg_prof_sample, opt_lg_prof_sample, size_t) -CTL_RO_NL_CGEN(config_prof, opt_lg_prof_interval, opt_lg_prof_interval, ssize_t) -CTL_RO_NL_CGEN(config_prof, opt_prof_gdump, opt_prof_gdump, bool) -CTL_RO_NL_CGEN(config_prof, opt_prof_final, opt_prof_final, bool) -CTL_RO_NL_CGEN(config_prof, opt_prof_leak, opt_prof_leak, bool) -CTL_RO_NL_CGEN(config_prof, opt_prof_accum, opt_prof_accum, bool) + ret = 0; +label_return: + return (ret); +} + +static int +thread_tcache_flush_ctl(const size_t *mib, size_t miblen, void *oldp, + size_t *oldlenp, void *newp, size_t newlen) +{ + int ret; + + if (config_tcache == false) + return (ENOENT); + + READONLY(); + WRITEONLY(); + + tcache_flush(); + + ret = 0; +label_return: + return (ret); +} /******************************************************************************/ @@ -1390,31 +1394,8 @@ label_return: return (ret); } - /******************************************************************************/ -CTL_RO_NL_GEN(arenas_bin_i_size, arena_bin_info[mib[2]].reg_size, size_t) -CTL_RO_NL_GEN(arenas_bin_i_nregs, arena_bin_info[mib[2]].nregs, uint32_t) -CTL_RO_NL_GEN(arenas_bin_i_run_size, arena_bin_info[mib[2]].run_size, size_t) -static const ctl_named_node_t * -arenas_bin_i_index(const size_t *mib, size_t miblen, size_t i) -{ - - if (i > NBINS) - return (NULL); - return (super_arenas_bin_i_node); -} - -CTL_RO_NL_GEN(arenas_lrun_i_size, ((mib[2]+1) << LG_PAGE), size_t) -static const ctl_named_node_t * -arenas_lrun_i_index(const size_t *mib, size_t miblen, size_t i) -{ - - if (i > nlclasses) - return (NULL); - return (super_arenas_lrun_i_node); -} - static int arenas_narenas_ctl(const size_t *mib, size_t miblen, void *oldp, size_t *oldlenp, void *newp, size_t newlen) @@ -1468,7 +1449,28 @@ CTL_RO_NL_GEN(arenas_page, PAGE, size_t) CTL_RO_NL_CGEN(config_tcache, arenas_tcache_max, tcache_maxclass, size_t) CTL_RO_NL_GEN(arenas_nbins, NBINS, unsigned) CTL_RO_NL_CGEN(config_tcache, arenas_nhbins, nhbins, unsigned) +CTL_RO_NL_GEN(arenas_bin_i_size, arena_bin_info[mib[2]].reg_size, size_t) +CTL_RO_NL_GEN(arenas_bin_i_nregs, arena_bin_info[mib[2]].nregs, uint32_t) +CTL_RO_NL_GEN(arenas_bin_i_run_size, arena_bin_info[mib[2]].run_size, size_t) +static const ctl_named_node_t * +arenas_bin_i_index(const size_t *mib, size_t miblen, size_t i) +{ + + if (i > NBINS) + return (NULL); + return (super_arenas_bin_i_node); +} + CTL_RO_NL_GEN(arenas_nlruns, nlclasses, size_t) +CTL_RO_NL_GEN(arenas_lrun_i_size, ((mib[2]+1) << LG_PAGE), size_t) +static const ctl_named_node_t * +arenas_lrun_i_index(const size_t *mib, size_t miblen, size_t i) +{ + + if (i > nlclasses) + return (NULL); + return (super_arenas_lrun_i_node); +} static int arenas_purge_ctl(const size_t *mib, size_t miblen, void *oldp, size_t *oldlenp, @@ -1575,6 +1577,11 @@ CTL_RO_NL_CGEN(config_prof, prof_interval, prof_interval, uint64_t) /******************************************************************************/ +CTL_RO_CGEN(config_stats, stats_cactive, &stats_cactive, size_t *) +CTL_RO_CGEN(config_stats, stats_allocated, ctl_stats.allocated, size_t) +CTL_RO_CGEN(config_stats, stats_active, ctl_stats.active, size_t) +CTL_RO_CGEN(config_stats, stats_mapped, ctl_stats.mapped, size_t) + CTL_RO_CGEN(config_stats, stats_chunks_current, ctl_stats.chunks.current, size_t) CTL_RO_CGEN(config_stats, stats_chunks_total, ctl_stats.chunks.total, uint64_t) @@ -1582,6 +1589,20 @@ CTL_RO_CGEN(config_stats, stats_chunks_high, ctl_stats.chunks.high, size_t) CTL_RO_CGEN(config_stats, stats_huge_allocated, huge_allocated, size_t) CTL_RO_CGEN(config_stats, stats_huge_nmalloc, huge_nmalloc, uint64_t) CTL_RO_CGEN(config_stats, stats_huge_ndalloc, huge_ndalloc, uint64_t) + +CTL_RO_GEN(stats_arenas_i_dss, ctl_stats.arenas[mib[2]].dss, const char *) +CTL_RO_GEN(stats_arenas_i_nthreads, ctl_stats.arenas[mib[2]].nthreads, unsigned) +CTL_RO_GEN(stats_arenas_i_pactive, ctl_stats.arenas[mib[2]].pactive, size_t) +CTL_RO_GEN(stats_arenas_i_pdirty, ctl_stats.arenas[mib[2]].pdirty, size_t) +CTL_RO_CGEN(config_stats, stats_arenas_i_mapped, + ctl_stats.arenas[mib[2]].astats.mapped, size_t) +CTL_RO_CGEN(config_stats, stats_arenas_i_npurge, + ctl_stats.arenas[mib[2]].astats.npurge, uint64_t) +CTL_RO_CGEN(config_stats, stats_arenas_i_nmadvise, + ctl_stats.arenas[mib[2]].astats.nmadvise, uint64_t) +CTL_RO_CGEN(config_stats, stats_arenas_i_purged, + ctl_stats.arenas[mib[2]].astats.purged, uint64_t) + CTL_RO_CGEN(config_stats, stats_arenas_i_small_allocated, ctl_stats.arenas[mib[2]].allocated_small, size_t) CTL_RO_CGEN(config_stats, stats_arenas_i_small_nmalloc, @@ -1645,19 +1666,6 @@ stats_arenas_i_lruns_j_index(const size_t *mib, size_t miblen, size_t j) return (super_stats_arenas_i_lruns_j_node); } -CTL_RO_GEN(stats_arenas_i_nthreads, ctl_stats.arenas[mib[2]].nthreads, unsigned) -CTL_RO_GEN(stats_arenas_i_dss, ctl_stats.arenas[mib[2]].dss, const char *) -CTL_RO_GEN(stats_arenas_i_pactive, ctl_stats.arenas[mib[2]].pactive, size_t) -CTL_RO_GEN(stats_arenas_i_pdirty, ctl_stats.arenas[mib[2]].pdirty, size_t) -CTL_RO_CGEN(config_stats, stats_arenas_i_mapped, - ctl_stats.arenas[mib[2]].astats.mapped, size_t) -CTL_RO_CGEN(config_stats, stats_arenas_i_npurge, - ctl_stats.arenas[mib[2]].astats.npurge, uint64_t) -CTL_RO_CGEN(config_stats, stats_arenas_i_nmadvise, - ctl_stats.arenas[mib[2]].astats.nmadvise, uint64_t) -CTL_RO_CGEN(config_stats, stats_arenas_i_purged, - ctl_stats.arenas[mib[2]].astats.purged, uint64_t) - static const ctl_named_node_t * stats_arenas_i_index(const size_t *mib, size_t miblen, size_t i) { @@ -1674,8 +1682,3 @@ label_return: malloc_mutex_unlock(&ctl_mtx); return (ret); } - -CTL_RO_CGEN(config_stats, stats_cactive, &stats_cactive, size_t *) -CTL_RO_CGEN(config_stats, stats_allocated, ctl_stats.allocated, size_t) -CTL_RO_CGEN(config_stats, stats_active, ctl_stats.active, size_t) -CTL_RO_CGEN(config_stats, stats_mapped, ctl_stats.mapped, size_t) @@ -78,7 +78,7 @@ huge_palloc(size_t size, size_t alignment, bool zero) return (ret); } -void * +bool huge_ralloc_no_move(void *ptr, size_t oldsize, size_t size, size_t extra) { @@ -89,15 +89,11 @@ huge_ralloc_no_move(void *ptr, size_t oldsize, size_t size, size_t extra) && CHUNK_CEILING(oldsize) >= CHUNK_CEILING(size) && CHUNK_CEILING(oldsize) <= CHUNK_CEILING(size+extra)) { assert(CHUNK_CEILING(oldsize) == oldsize); - if (config_fill && opt_junk && size < oldsize) { - memset((void *)((uintptr_t)ptr + size), 0x5a, - oldsize - size); - } - return (ptr); + return (false); } /* Reallocation would require a move. */ - return (NULL); + return (true); } void * @@ -108,9 +104,8 @@ huge_ralloc(void *ptr, size_t oldsize, size_t size, size_t extra, size_t copysize; /* Try to avoid moving the allocation. */ - ret = huge_ralloc_no_move(ptr, oldsize, size, extra); - if (ret != NULL) - return (ret); + if (huge_ralloc_no_move(ptr, oldsize, size, extra) == false) + return (ptr); /* * size and oldsize are different enough that we need to use a @@ -169,7 +164,7 @@ huge_ralloc(void *ptr, size_t oldsize, size_t size, size_t extra, */ char buf[BUFERROR_BUF]; - buferror(buf, sizeof(buf)); + buferror(get_errno(), buf, sizeof(buf)); malloc_printf("<jemalloc>: Error in mremap(): %s\n", buf); if (opt_abort) @@ -181,11 +176,34 @@ huge_ralloc(void *ptr, size_t oldsize, size_t size, size_t extra, #endif { memcpy(ret, ptr, copysize); - iqallocx(ptr, try_tcache_dalloc); + iqalloct(ptr, try_tcache_dalloc); } return (ret); } +#ifdef JEMALLOC_JET +#undef huge_dalloc_junk +#define huge_dalloc_junk JEMALLOC_N(huge_dalloc_junk_impl) +#endif +static void +huge_dalloc_junk(void *ptr, size_t usize) +{ + + if (config_fill && config_dss && opt_junk) { + /* + * Only bother junk filling if the chunk isn't about to be + * unmapped. + */ + if (config_munmap == false || (config_dss && chunk_in_dss(ptr))) + memset(ptr, 0x5a, usize); + } +} +#ifdef JEMALLOC_JET +#undef huge_dalloc_junk +#define huge_dalloc_junk JEMALLOC_N(huge_dalloc_junk) +huge_dalloc_junk_t *huge_dalloc_junk = JEMALLOC_N(huge_dalloc_junk_impl); +#endif + void huge_dalloc(void *ptr, bool unmap) { @@ -208,8 +226,8 @@ huge_dalloc(void *ptr, bool unmap) malloc_mutex_unlock(&huge_mtx); - if (unmap && config_fill && config_dss && opt_junk) - memset(node->addr, 0x5a, node->size); + if (unmap) + huge_dalloc_junk(node->addr, node->size); chunk_dealloc(node->addr, node->size, unmap); diff --git a/src/jemalloc.c b/src/jemalloc.c index ae56db6b..563d99f8 100644 --- a/src/jemalloc.c +++ b/src/jemalloc.c @@ -100,18 +100,12 @@ typedef struct { #endif /******************************************************************************/ -/* Function prototypes for non-inline static functions. */ - -static void stats_print_atexit(void); -static unsigned malloc_ncpus(void); -static bool malloc_conf_next(char const **opts_p, char const **k_p, - size_t *klen_p, char const **v_p, size_t *vlen_p); -static void malloc_conf_error(const char *msg, const char *k, size_t klen, - const char *v, size_t vlen); -static void malloc_conf_init(void); +/* + * Function prototypes for static functions that are referenced prior to + * definition. + */ + static bool malloc_init_hard(void); -static int imemalign(void **memptr, size_t alignment, size_t size, - size_t min_alignment); /******************************************************************************/ /* @@ -252,7 +246,6 @@ stats_print_atexit(void) static unsigned malloc_ncpus(void) { - unsigned ret; long result; #ifdef _WIN32 @@ -262,14 +255,7 @@ malloc_ncpus(void) #else result = sysconf(_SC_NPROCESSORS_ONLN); #endif - if (result == -1) { - /* Error. */ - ret = 1; - } else { - ret = (unsigned)result; - } - - return (ret); + return ((result == -1) ? 1 : (unsigned)result); } void @@ -484,8 +470,7 @@ malloc_conf_init(void) } break; } default: - /* NOTREACHED */ - assert(false); + not_reached(); buf[0] = '\0'; opts = buf; } @@ -522,14 +507,15 @@ malloc_conf_init(void) "Invalid conf value", \ k, klen, v, vlen); \ } else if (clip) { \ - if (um < min) \ + if (min != 0 && um < min) \ o = min; \ else if (um > max) \ o = max; \ else \ o = um; \ } else { \ - if (um < min || um > max) { \ + if ((min != 0 && um < min) || \ + um > max) { \ malloc_conf_error( \ "Out-of-range " \ "conf value", \ @@ -695,17 +681,6 @@ malloc_init_hard(void) malloc_conf_init(); -#if (!defined(JEMALLOC_MUTEX_INIT_CB) && !defined(JEMALLOC_ZONE) \ - && !defined(_WIN32)) - /* Register fork handlers. */ - if (pthread_atfork(jemalloc_prefork, jemalloc_postfork_parent, - jemalloc_postfork_child) != 0) { - malloc_write("<jemalloc>: Error in pthread_atfork()\n"); - if (opt_abort) - abort(); - } -#endif - if (opt_stats_print) { /* Print statistics at exit. */ if (atexit(stats_print_atexit) != 0) { @@ -745,8 +720,10 @@ malloc_init_hard(void) return (true); } - if (malloc_mutex_init(&arenas_lock)) + if (malloc_mutex_init(&arenas_lock)) { + malloc_mutex_unlock(&init_lock); return (true); + } /* * Create enough scaffolding to allow recursive allocation in @@ -792,9 +769,25 @@ malloc_init_hard(void) return (true); } - /* Get number of CPUs. */ malloc_mutex_unlock(&init_lock); + /**********************************************************************/ + /* Recursive allocation may follow. */ + ncpus = malloc_ncpus(); + +#if (!defined(JEMALLOC_MUTEX_INIT_CB) && !defined(JEMALLOC_ZONE) \ + && !defined(_WIN32)) + /* LinuxThreads's pthread_atfork() allocates. */ + if (pthread_atfork(jemalloc_prefork, jemalloc_postfork_parent, + jemalloc_postfork_child) != 0) { + malloc_write("<jemalloc>: Error in pthread_atfork()\n"); + if (opt_abort) + abort(); + } +#endif + + /* Done recursively allocating. */ + /**********************************************************************/ malloc_mutex_lock(&init_lock); if (mutex_boot()) { @@ -841,6 +834,7 @@ malloc_init_hard(void) malloc_initialized = true; malloc_mutex_unlock(&init_lock); + return (false); } @@ -852,42 +846,88 @@ malloc_init_hard(void) * Begin malloc(3)-compatible functions. */ +static void * +imalloc_prof_sample(size_t usize, prof_thr_cnt_t *cnt) +{ + void *p; + + if (cnt == NULL) + return (NULL); + if (prof_promote && usize <= SMALL_MAXCLASS) { + p = imalloc(SMALL_MAXCLASS+1); + if (p == NULL) + return (NULL); + arena_prof_promoted(p, usize); + } else + p = imalloc(usize); + + return (p); +} + +JEMALLOC_ALWAYS_INLINE_C void * +imalloc_prof(size_t usize, prof_thr_cnt_t *cnt) +{ + void *p; + + if ((uintptr_t)cnt != (uintptr_t)1U) + p = imalloc_prof_sample(usize, cnt); + else + p = imalloc(usize); + if (p == NULL) + return (NULL); + prof_malloc(p, usize, cnt); + + return (p); +} + +/* + * MALLOC_BODY() is a macro rather than a function because its contents are in + * the fast path, but inlining would cause reliability issues when determining + * how many frames to discard from heap profiling backtraces. + */ +#define MALLOC_BODY(ret, size, usize) do { \ + if (malloc_init()) \ + ret = NULL; \ + else { \ + if (config_prof && opt_prof) { \ + prof_thr_cnt_t *cnt; \ + \ + usize = s2u(size); \ + /* \ + * Call PROF_ALLOC_PREP() here rather than in \ + * imalloc_prof() so that imalloc_prof() can be \ + * inlined without introducing uncertainty \ + * about the number of backtrace frames to \ + * ignore. imalloc_prof() is in the fast path \ + * when heap profiling is enabled, so inlining \ + * is critical to performance. (For \ + * consistency all callers of PROF_ALLOC_PREP() \ + * are structured similarly, even though e.g. \ + * realloc() isn't called enough for inlining \ + * to be critical.) \ + */ \ + PROF_ALLOC_PREP(1, usize, cnt); \ + ret = imalloc_prof(usize, cnt); \ + } else { \ + if (config_stats || (config_valgrind && \ + opt_valgrind)) \ + usize = s2u(size); \ + ret = imalloc(size); \ + } \ + } \ +} while (0) + void * je_malloc(size_t size) { void *ret; size_t usize JEMALLOC_CC_SILENCE_INIT(0); - prof_thr_cnt_t *cnt JEMALLOC_CC_SILENCE_INIT(NULL); - - if (malloc_init()) { - ret = NULL; - goto label_oom; - } if (size == 0) size = 1; - if (config_prof && opt_prof) { - usize = s2u(size); - PROF_ALLOC_PREP(1, usize, cnt); - if (cnt == NULL) { - ret = NULL; - goto label_oom; - } - if (prof_promote && (uintptr_t)cnt != (uintptr_t)1U && usize <= - SMALL_MAXCLASS) { - ret = imalloc(SMALL_MAXCLASS+1); - if (ret != NULL) - arena_prof_promoted(ret, usize); - } else - ret = imalloc(size); - } else { - if (config_stats || (config_valgrind && opt_valgrind)) - usize = s2u(size); - ret = imalloc(size); - } + MALLOC_BODY(ret, size, usize); -label_oom: if (ret == NULL) { if (config_xmalloc && opt_xmalloc) { malloc_write("<jemalloc>: Error in malloc(): " @@ -896,8 +936,6 @@ label_oom: } set_errno(ENOMEM); } - if (config_prof && opt_prof && ret != NULL) - prof_malloc(ret, usize, cnt); if (config_stats && ret != NULL) { assert(usize == isalloc(ret, config_prof)); thread_allocated_tsd_get()->allocated += usize; @@ -907,6 +945,42 @@ label_oom: return (ret); } +static void * +imemalign_prof_sample(size_t alignment, size_t usize, prof_thr_cnt_t *cnt) +{ + void *p; + + if (cnt == NULL) + return (NULL); + if (prof_promote && usize <= SMALL_MAXCLASS) { + assert(sa2u(SMALL_MAXCLASS+1, alignment) != 0); + p = ipalloc(sa2u(SMALL_MAXCLASS+1, alignment), alignment, + false); + if (p == NULL) + return (NULL); + arena_prof_promoted(p, usize); + } else + p = ipalloc(usize, alignment, false); + + return (p); +} + +JEMALLOC_ALWAYS_INLINE_C void * +imemalign_prof(size_t alignment, size_t usize, prof_thr_cnt_t *cnt) +{ + void *p; + + if ((uintptr_t)cnt != (uintptr_t)1U) + p = imemalign_prof_sample(alignment, usize, cnt); + else + p = ipalloc(usize, alignment, false); + if (p == NULL) + return (NULL); + prof_malloc(p, usize, cnt); + + return (p); +} + JEMALLOC_ATTR(nonnull(1)) #ifdef JEMALLOC_PROF /* @@ -916,19 +990,18 @@ JEMALLOC_ATTR(nonnull(1)) JEMALLOC_NOINLINE #endif static int -imemalign(void **memptr, size_t alignment, size_t size, - size_t min_alignment) +imemalign(void **memptr, size_t alignment, size_t size, size_t min_alignment) { int ret; size_t usize; void *result; - prof_thr_cnt_t *cnt JEMALLOC_CC_SILENCE_INIT(NULL); assert(min_alignment != 0); - if (malloc_init()) + if (malloc_init()) { result = NULL; - else { + goto label_oom; + } else { if (size == 0) size = 1; @@ -948,57 +1021,38 @@ imemalign(void **memptr, size_t alignment, size_t size, usize = sa2u(size, alignment); if (usize == 0) { result = NULL; - ret = ENOMEM; - goto label_return; + goto label_oom; } if (config_prof && opt_prof) { + prof_thr_cnt_t *cnt; + PROF_ALLOC_PREP(2, usize, cnt); - if (cnt == NULL) { - result = NULL; - ret = EINVAL; - } else { - if (prof_promote && (uintptr_t)cnt != - (uintptr_t)1U && usize <= SMALL_MAXCLASS) { - assert(sa2u(SMALL_MAXCLASS+1, - alignment) != 0); - result = ipalloc(sa2u(SMALL_MAXCLASS+1, - alignment), alignment, false); - if (result != NULL) { - arena_prof_promoted(result, - usize); - } - } else { - result = ipalloc(usize, alignment, - false); - } - } + result = imemalign_prof(alignment, usize, cnt); } else result = ipalloc(usize, alignment, false); - } - - if (result == NULL) { - if (config_xmalloc && opt_xmalloc) { - malloc_write("<jemalloc>: Error allocating aligned " - "memory: out of memory\n"); - abort(); - } - ret = ENOMEM; - goto label_return; + if (result == NULL) + goto label_oom; } *memptr = result; ret = 0; - label_return: if (config_stats && result != NULL) { assert(usize == isalloc(result, config_prof)); thread_allocated_tsd_get()->allocated += usize; } - if (config_prof && opt_prof && result != NULL) - prof_malloc(result, usize, cnt); UTRACE(0, size, result); return (ret); +label_oom: + assert(result == NULL); + if (config_xmalloc && opt_xmalloc) { + malloc_write("<jemalloc>: Error allocating aligned memory: " + "out of memory\n"); + abort(); + } + ret = ENOMEM; + goto label_return; } int @@ -1025,13 +1079,46 @@ je_aligned_alloc(size_t alignment, size_t size) return (ret); } +static void * +icalloc_prof_sample(size_t usize, prof_thr_cnt_t *cnt) +{ + void *p; + + if (cnt == NULL) + return (NULL); + if (prof_promote && usize <= SMALL_MAXCLASS) { + p = icalloc(SMALL_MAXCLASS+1); + if (p == NULL) + return (NULL); + arena_prof_promoted(p, usize); + } else + p = icalloc(usize); + + return (p); +} + +JEMALLOC_ALWAYS_INLINE_C void * +icalloc_prof(size_t usize, prof_thr_cnt_t *cnt) +{ + void *p; + + if ((uintptr_t)cnt != (uintptr_t)1U) + p = icalloc_prof_sample(usize, cnt); + else + p = icalloc(usize); + if (p == NULL) + return (NULL); + prof_malloc(p, usize, cnt); + + return (p); +} + void * je_calloc(size_t num, size_t size) { void *ret; size_t num_size; size_t usize JEMALLOC_CC_SILENCE_INIT(0); - prof_thr_cnt_t *cnt JEMALLOC_CC_SILENCE_INIT(NULL); if (malloc_init()) { num_size = 0; @@ -1060,19 +1147,11 @@ je_calloc(size_t num, size_t size) } if (config_prof && opt_prof) { + prof_thr_cnt_t *cnt; + usize = s2u(num_size); PROF_ALLOC_PREP(1, usize, cnt); - if (cnt == NULL) { - ret = NULL; - goto label_return; - } - if (prof_promote && (uintptr_t)cnt != (uintptr_t)1U && usize - <= SMALL_MAXCLASS) { - ret = icalloc(SMALL_MAXCLASS+1); - if (ret != NULL) - arena_prof_promoted(ret, usize); - } else - ret = icalloc(num_size); + ret = icalloc_prof(usize, cnt); } else { if (config_stats || (config_valgrind && opt_valgrind)) usize = s2u(num_size); @@ -1088,9 +1167,6 @@ label_return: } set_errno(ENOMEM); } - - if (config_prof && opt_prof && ret != NULL) - prof_malloc(ret, usize, cnt); if (config_stats && ret != NULL) { assert(usize == isalloc(ret, config_prof)); thread_allocated_tsd_get()->allocated += usize; @@ -1100,152 +1176,126 @@ label_return: return (ret); } +static void * +irealloc_prof_sample(void *oldptr, size_t usize, prof_thr_cnt_t *cnt) +{ + void *p; + + if (cnt == NULL) + return (NULL); + if (prof_promote && usize <= SMALL_MAXCLASS) { + p = iralloc(oldptr, SMALL_MAXCLASS+1, 0, 0, false); + if (p == NULL) + return (NULL); + arena_prof_promoted(p, usize); + } else + p = iralloc(oldptr, usize, 0, 0, false); + + return (p); +} + +JEMALLOC_ALWAYS_INLINE_C void * +irealloc_prof(void *oldptr, size_t old_usize, size_t usize, prof_thr_cnt_t *cnt) +{ + void *p; + prof_ctx_t *old_ctx; + + old_ctx = prof_ctx_get(oldptr); + if ((uintptr_t)cnt != (uintptr_t)1U) + p = irealloc_prof_sample(oldptr, usize, cnt); + else + p = iralloc(oldptr, usize, 0, 0, false); + if (p == NULL) + return (NULL); + prof_realloc(p, usize, cnt, old_usize, old_ctx); + + return (p); +} + +JEMALLOC_INLINE_C void +ifree(void *ptr) +{ + size_t usize; + UNUSED size_t rzsize JEMALLOC_CC_SILENCE_INIT(0); + + assert(ptr != NULL); + assert(malloc_initialized || IS_INITIALIZER); + + if (config_prof && opt_prof) { + usize = isalloc(ptr, config_prof); + prof_free(ptr, usize); + } else if (config_stats || config_valgrind) + usize = isalloc(ptr, config_prof); + if (config_stats) + thread_allocated_tsd_get()->deallocated += usize; + if (config_valgrind && opt_valgrind) + rzsize = p2rz(ptr); + iqalloc(ptr); + JEMALLOC_VALGRIND_FREE(ptr, rzsize); +} + void * je_realloc(void *ptr, size_t size) { void *ret; size_t usize JEMALLOC_CC_SILENCE_INIT(0); - size_t old_size = 0; - size_t old_rzsize JEMALLOC_CC_SILENCE_INIT(0); - prof_thr_cnt_t *cnt JEMALLOC_CC_SILENCE_INIT(NULL); - prof_ctx_t *old_ctx JEMALLOC_CC_SILENCE_INIT(NULL); + size_t old_usize = 0; + UNUSED size_t old_rzsize JEMALLOC_CC_SILENCE_INIT(0); if (size == 0) { if (ptr != NULL) { - /* realloc(ptr, 0) is equivalent to free(p). */ - assert(malloc_initialized || IS_INITIALIZER); - if (config_prof) { - old_size = isalloc(ptr, true); - if (config_valgrind && opt_valgrind) - old_rzsize = p2rz(ptr); - } else if (config_stats) { - old_size = isalloc(ptr, false); - if (config_valgrind && opt_valgrind) - old_rzsize = u2rz(old_size); - } else if (config_valgrind && opt_valgrind) { - old_size = isalloc(ptr, false); - old_rzsize = u2rz(old_size); - } - if (config_prof && opt_prof) { - old_ctx = prof_ctx_get(ptr); - cnt = NULL; - } - iqalloc(ptr); - ret = NULL; - goto label_return; - } else - size = 1; + /* realloc(ptr, 0) is equivalent to free(ptr). */ + UTRACE(ptr, 0, 0); + ifree(ptr); + return (NULL); + } + size = 1; } if (ptr != NULL) { assert(malloc_initialized || IS_INITIALIZER); malloc_thread_init(); - if (config_prof) { - old_size = isalloc(ptr, true); - if (config_valgrind && opt_valgrind) - old_rzsize = p2rz(ptr); - } else if (config_stats) { - old_size = isalloc(ptr, false); - if (config_valgrind && opt_valgrind) - old_rzsize = u2rz(old_size); - } else if (config_valgrind && opt_valgrind) { - old_size = isalloc(ptr, false); - old_rzsize = u2rz(old_size); - } + if ((config_prof && opt_prof) || config_stats || + (config_valgrind && opt_valgrind)) + old_usize = isalloc(ptr, config_prof); + if (config_valgrind && opt_valgrind) + old_rzsize = config_prof ? p2rz(ptr) : u2rz(old_usize); + if (config_prof && opt_prof) { + prof_thr_cnt_t *cnt; + usize = s2u(size); - old_ctx = prof_ctx_get(ptr); PROF_ALLOC_PREP(1, usize, cnt); - if (cnt == NULL) { - old_ctx = NULL; - ret = NULL; - goto label_oom; - } - if (prof_promote && (uintptr_t)cnt != (uintptr_t)1U && - usize <= SMALL_MAXCLASS) { - ret = iralloc(ptr, SMALL_MAXCLASS+1, 0, 0, - false, false); - if (ret != NULL) - arena_prof_promoted(ret, usize); - else - old_ctx = NULL; - } else { - ret = iralloc(ptr, size, 0, 0, false, false); - if (ret == NULL) - old_ctx = NULL; - } + ret = irealloc_prof(ptr, old_usize, usize, cnt); } else { if (config_stats || (config_valgrind && opt_valgrind)) usize = s2u(size); - ret = iralloc(ptr, size, 0, 0, false, false); - } - -label_oom: - if (ret == NULL) { - if (config_xmalloc && opt_xmalloc) { - malloc_write("<jemalloc>: Error in realloc(): " - "out of memory\n"); - abort(); - } - set_errno(ENOMEM); + ret = iralloc(ptr, size, 0, 0, false); } } else { /* realloc(NULL, size) is equivalent to malloc(size). */ - if (config_prof && opt_prof) - old_ctx = NULL; - if (malloc_init()) { - if (config_prof && opt_prof) - cnt = NULL; - ret = NULL; - } else { - if (config_prof && opt_prof) { - usize = s2u(size); - PROF_ALLOC_PREP(1, usize, cnt); - if (cnt == NULL) - ret = NULL; - else { - if (prof_promote && (uintptr_t)cnt != - (uintptr_t)1U && usize <= - SMALL_MAXCLASS) { - ret = imalloc(SMALL_MAXCLASS+1); - if (ret != NULL) { - arena_prof_promoted(ret, - usize); - } - } else - ret = imalloc(size); - } - } else { - if (config_stats || (config_valgrind && - opt_valgrind)) - usize = s2u(size); - ret = imalloc(size); - } - } + MALLOC_BODY(ret, size, usize); + } - if (ret == NULL) { - if (config_xmalloc && opt_xmalloc) { - malloc_write("<jemalloc>: Error in realloc(): " - "out of memory\n"); - abort(); - } - set_errno(ENOMEM); + if (ret == NULL) { + if (config_xmalloc && opt_xmalloc) { + malloc_write("<jemalloc>: Error in realloc(): " + "out of memory\n"); + abort(); } + set_errno(ENOMEM); } - -label_return: - if (config_prof && opt_prof) - prof_realloc(ret, usize, cnt, old_size, old_ctx); if (config_stats && ret != NULL) { thread_allocated_t *ta; assert(usize == isalloc(ret, config_prof)); ta = thread_allocated_tsd_get(); ta->allocated += usize; - ta->deallocated += old_size; + ta->deallocated += old_usize; } UTRACE(ptr, size, ret); - JEMALLOC_VALGRIND_REALLOC(ret, usize, ptr, old_size, old_rzsize, false); + JEMALLOC_VALGRIND_REALLOC(ret, usize, ptr, old_usize, old_rzsize, + false); return (ret); } @@ -1254,24 +1304,8 @@ je_free(void *ptr) { UTRACE(ptr, 0, 0); - if (ptr != NULL) { - size_t usize; - size_t rzsize JEMALLOC_CC_SILENCE_INIT(0); - - assert(malloc_initialized || IS_INITIALIZER); - - if (config_prof && opt_prof) { - usize = isalloc(ptr, config_prof); - prof_free(ptr, usize); - } else if (config_stats || config_valgrind) - usize = isalloc(ptr, config_prof); - if (config_stats) - thread_allocated_tsd_get()->deallocated += usize; - if (config_valgrind && opt_valgrind) - rzsize = p2rz(ptr); - iqalloc(ptr); - JEMALLOC_VALGRIND_FREE(ptr, rzsize); - } + if (ptr != NULL) + ifree(ptr); } /* @@ -1337,208 +1371,344 @@ JEMALLOC_EXPORT void *(* __memalign_hook)(size_t alignment, size_t size) = * Begin non-standard functions. */ -size_t -je_malloc_usable_size(JEMALLOC_USABLE_SIZE_CONST void *ptr) +JEMALLOC_ALWAYS_INLINE_C void * +imallocx(size_t usize, size_t alignment, bool zero, bool try_tcache, + arena_t *arena) { - size_t ret; - assert(malloc_initialized || IS_INITIALIZER); - malloc_thread_init(); + assert(usize == ((alignment == 0) ? s2u(usize) : sa2u(usize, + alignment))); - if (config_ivsalloc) - ret = ivsalloc(ptr, config_prof); + if (alignment != 0) + return (ipalloct(usize, alignment, zero, try_tcache, arena)); + else if (zero) + return (icalloct(usize, try_tcache, arena)); else - ret = (ptr != NULL) ? isalloc(ptr, config_prof) : 0; - - return (ret); + return (imalloct(usize, try_tcache, arena)); } -void -je_malloc_stats_print(void (*write_cb)(void *, const char *), void *cbopaque, - const char *opts) +static void * +imallocx_prof_sample(size_t usize, size_t alignment, bool zero, bool try_tcache, + arena_t *arena, prof_thr_cnt_t *cnt) { + void *p; - stats_print(write_cb, cbopaque, opts); + if (cnt == NULL) + return (NULL); + if (prof_promote && usize <= SMALL_MAXCLASS) { + size_t usize_promoted = (alignment == 0) ? + s2u(SMALL_MAXCLASS+1) : sa2u(SMALL_MAXCLASS+1, alignment); + assert(usize_promoted != 0); + p = imallocx(usize_promoted, alignment, zero, try_tcache, + arena); + if (p == NULL) + return (NULL); + arena_prof_promoted(p, usize); + } else + p = imallocx(usize, alignment, zero, try_tcache, arena); + + return (p); } -int -je_mallctl(const char *name, void *oldp, size_t *oldlenp, void *newp, - size_t newlen) +JEMALLOC_ALWAYS_INLINE_C void * +imallocx_prof(size_t usize, size_t alignment, bool zero, bool try_tcache, + arena_t *arena, prof_thr_cnt_t *cnt) { + void *p; - if (malloc_init()) - return (EAGAIN); + if ((uintptr_t)cnt != (uintptr_t)1U) { + p = imallocx_prof_sample(usize, alignment, zero, try_tcache, + arena, cnt); + } else + p = imallocx(usize, alignment, zero, try_tcache, arena); + if (p == NULL) + return (NULL); + prof_malloc(p, usize, cnt); - return (ctl_byname(name, oldp, oldlenp, newp, newlen)); + return (p); } -int -je_mallctlnametomib(const char *name, size_t *mibp, size_t *miblenp) +void * +je_mallocx(size_t size, int flags) { + void *p; + size_t usize; + size_t alignment = (ZU(1) << (flags & MALLOCX_LG_ALIGN_MASK) + & (SIZE_T_MAX-1)); + bool zero = flags & MALLOCX_ZERO; + unsigned arena_ind = ((unsigned)(flags >> 8)) - 1; + arena_t *arena; + bool try_tcache; + + assert(size != 0); if (malloc_init()) - return (EAGAIN); + goto label_oom; - return (ctl_nametomib(name, mibp, miblenp)); + if (arena_ind != UINT_MAX) { + arena = arenas[arena_ind]; + try_tcache = false; + } else { + arena = NULL; + try_tcache = true; + } + + usize = (alignment == 0) ? s2u(size) : sa2u(size, alignment); + assert(usize != 0); + + if (config_prof && opt_prof) { + prof_thr_cnt_t *cnt; + + PROF_ALLOC_PREP(1, usize, cnt); + p = imallocx_prof(usize, alignment, zero, try_tcache, arena, + cnt); + } else + p = imallocx(usize, alignment, zero, try_tcache, arena); + if (p == NULL) + goto label_oom; + + if (config_stats) { + assert(usize == isalloc(p, config_prof)); + thread_allocated_tsd_get()->allocated += usize; + } + UTRACE(0, size, p); + JEMALLOC_VALGRIND_MALLOC(true, p, usize, zero); + return (p); +label_oom: + if (config_xmalloc && opt_xmalloc) { + malloc_write("<jemalloc>: Error in mallocx(): out of memory\n"); + abort(); + } + UTRACE(0, size, 0); + return (NULL); } -int -je_mallctlbymib(const size_t *mib, size_t miblen, void *oldp, size_t *oldlenp, - void *newp, size_t newlen) +static void * +irallocx_prof_sample(void *oldptr, size_t size, size_t alignment, size_t usize, + bool zero, bool try_tcache_alloc, bool try_tcache_dalloc, arena_t *arena, + prof_thr_cnt_t *cnt) { + void *p; - if (malloc_init()) - return (EAGAIN); + if (cnt == NULL) + return (NULL); + if (prof_promote && usize <= SMALL_MAXCLASS) { + p = iralloct(oldptr, SMALL_MAXCLASS+1, (SMALL_MAXCLASS+1 >= + size) ? 0 : size - (SMALL_MAXCLASS+1), alignment, zero, + try_tcache_alloc, try_tcache_dalloc, arena); + if (p == NULL) + return (NULL); + arena_prof_promoted(p, usize); + } else { + p = iralloct(oldptr, size, 0, alignment, zero, + try_tcache_alloc, try_tcache_dalloc, arena); + } - return (ctl_bymib(mib, miblen, oldp, oldlenp, newp, newlen)); + return (p); } -/* - * End non-standard functions. - */ -/******************************************************************************/ -/* - * Begin experimental functions. - */ -#ifdef JEMALLOC_EXPERIMENTAL - JEMALLOC_ALWAYS_INLINE_C void * -iallocm(size_t usize, size_t alignment, bool zero, bool try_tcache, - arena_t *arena) +irallocx_prof(void *oldptr, size_t old_usize, size_t size, size_t alignment, + size_t *usize, bool zero, bool try_tcache_alloc, bool try_tcache_dalloc, + arena_t *arena, prof_thr_cnt_t *cnt) { + void *p; + prof_ctx_t *old_ctx; - assert(usize == ((alignment == 0) ? s2u(usize) : sa2u(usize, - alignment))); + old_ctx = prof_ctx_get(oldptr); + if ((uintptr_t)cnt != (uintptr_t)1U) + p = irallocx_prof_sample(oldptr, size, alignment, *usize, zero, + try_tcache_alloc, try_tcache_dalloc, arena, cnt); + else { + p = iralloct(oldptr, size, 0, alignment, zero, + try_tcache_alloc, try_tcache_dalloc, arena); + } + if (p == NULL) + return (NULL); - if (alignment != 0) - return (ipallocx(usize, alignment, zero, try_tcache, arena)); - else if (zero) - return (icallocx(usize, try_tcache, arena)); - else - return (imallocx(usize, try_tcache, arena)); + if (p == oldptr && alignment != 0) { + /* + * The allocation did not move, so it is possible that the size + * class is smaller than would guarantee the requested + * alignment, and that the alignment constraint was + * serendipitously satisfied. Additionally, old_usize may not + * be the same as the current usize because of in-place large + * reallocation. Therefore, query the actual value of usize. + */ + *usize = isalloc(p, config_prof); + } + prof_realloc(p, *usize, cnt, old_usize, old_ctx); + + return (p); } -int -je_allocm(void **ptr, size_t *rsize, size_t size, int flags) +void * +je_rallocx(void *ptr, size_t size, int flags) { void *p; - size_t usize; - size_t alignment = (ZU(1) << (flags & ALLOCM_LG_ALIGN_MASK) + size_t usize, old_usize; + UNUSED size_t old_rzsize JEMALLOC_CC_SILENCE_INIT(0); + size_t alignment = (ZU(1) << (flags & MALLOCX_LG_ALIGN_MASK) & (SIZE_T_MAX-1)); - bool zero = flags & ALLOCM_ZERO; + bool zero = flags & MALLOCX_ZERO; unsigned arena_ind = ((unsigned)(flags >> 8)) - 1; + bool try_tcache_alloc, try_tcache_dalloc; arena_t *arena; - bool try_tcache; assert(ptr != NULL); assert(size != 0); - - if (malloc_init()) - goto label_oom; + assert(malloc_initialized || IS_INITIALIZER); + malloc_thread_init(); if (arena_ind != UINT_MAX) { + arena_chunk_t *chunk; + try_tcache_alloc = false; + chunk = (arena_chunk_t *)CHUNK_ADDR2BASE(ptr); + try_tcache_dalloc = (chunk == ptr || chunk->arena != + arenas[arena_ind]); arena = arenas[arena_ind]; - try_tcache = false; } else { + try_tcache_alloc = true; + try_tcache_dalloc = true; arena = NULL; - try_tcache = true; } - usize = (alignment == 0) ? s2u(size) : sa2u(size, alignment); - if (usize == 0) - goto label_oom; + if ((config_prof && opt_prof) || config_stats || + (config_valgrind && opt_valgrind)) + old_usize = isalloc(ptr, config_prof); + if (config_valgrind && opt_valgrind) + old_rzsize = u2rz(old_usize); if (config_prof && opt_prof) { prof_thr_cnt_t *cnt; + usize = (alignment == 0) ? s2u(size) : sa2u(size, alignment); + assert(usize != 0); PROF_ALLOC_PREP(1, usize, cnt); - if (cnt == NULL) + p = irallocx_prof(ptr, old_usize, size, alignment, &usize, zero, + try_tcache_alloc, try_tcache_dalloc, arena, cnt); + if (p == NULL) goto label_oom; - if (prof_promote && (uintptr_t)cnt != (uintptr_t)1U && usize <= - SMALL_MAXCLASS) { - size_t usize_promoted = (alignment == 0) ? - s2u(SMALL_MAXCLASS+1) : sa2u(SMALL_MAXCLASS+1, - alignment); - assert(usize_promoted != 0); - p = iallocm(usize_promoted, alignment, zero, - try_tcache, arena); - if (p == NULL) - goto label_oom; - arena_prof_promoted(p, usize); - } else { - p = iallocm(usize, alignment, zero, try_tcache, arena); - if (p == NULL) - goto label_oom; - } - prof_malloc(p, usize, cnt); } else { - p = iallocm(usize, alignment, zero, try_tcache, arena); + p = iralloct(ptr, size, 0, alignment, zero, try_tcache_alloc, + try_tcache_dalloc, arena); if (p == NULL) goto label_oom; + if (config_stats || (config_valgrind && opt_valgrind)) + usize = isalloc(p, config_prof); } - if (rsize != NULL) - *rsize = usize; - *ptr = p; if (config_stats) { - assert(usize == isalloc(p, config_prof)); - thread_allocated_tsd_get()->allocated += usize; + thread_allocated_t *ta; + ta = thread_allocated_tsd_get(); + ta->allocated += usize; + ta->deallocated += old_usize; } - UTRACE(0, size, p); - JEMALLOC_VALGRIND_MALLOC(true, p, usize, zero); - return (ALLOCM_SUCCESS); + UTRACE(ptr, size, p); + JEMALLOC_VALGRIND_REALLOC(p, usize, ptr, old_usize, old_rzsize, zero); + return (p); label_oom: if (config_xmalloc && opt_xmalloc) { - malloc_write("<jemalloc>: Error in allocm(): " - "out of memory\n"); + malloc_write("<jemalloc>: Error in rallocx(): out of memory\n"); abort(); } - *ptr = NULL; - UTRACE(0, size, 0); - return (ALLOCM_ERR_OOM); + UTRACE(ptr, size, 0); + return (NULL); } -int -je_rallocm(void **ptr, size_t *rsize, size_t size, size_t extra, int flags) +JEMALLOC_ALWAYS_INLINE_C size_t +ixallocx_helper(void *ptr, size_t old_usize, size_t size, size_t extra, + size_t alignment, bool zero, arena_t *arena) +{ + size_t usize; + + if (ixalloc(ptr, size, extra, alignment, zero)) + return (old_usize); + usize = isalloc(ptr, config_prof); + + return (usize); +} + +static size_t +ixallocx_prof_sample(void *ptr, size_t old_usize, size_t size, size_t extra, + size_t alignment, size_t max_usize, bool zero, arena_t *arena, + prof_thr_cnt_t *cnt) +{ + size_t usize; + + if (cnt == NULL) + return (old_usize); + /* Use minimum usize to determine whether promotion may happen. */ + if (prof_promote && ((alignment == 0) ? s2u(size) : sa2u(size, + alignment)) <= SMALL_MAXCLASS) { + if (ixalloc(ptr, SMALL_MAXCLASS+1, (SMALL_MAXCLASS+1 >= + size+extra) ? 0 : size+extra - (SMALL_MAXCLASS+1), + alignment, zero)) + return (old_usize); + usize = isalloc(ptr, config_prof); + if (max_usize < PAGE) + arena_prof_promoted(ptr, usize); + } else { + usize = ixallocx_helper(ptr, old_usize, size, extra, alignment, + zero, arena); + } + + return (usize); +} + +JEMALLOC_ALWAYS_INLINE_C size_t +ixallocx_prof(void *ptr, size_t old_usize, size_t size, size_t extra, + size_t alignment, size_t max_usize, bool zero, arena_t *arena, + prof_thr_cnt_t *cnt) { - void *p, *q; size_t usize; - size_t old_size; - size_t old_rzsize JEMALLOC_CC_SILENCE_INIT(0); - size_t alignment = (ZU(1) << (flags & ALLOCM_LG_ALIGN_MASK) + prof_ctx_t *old_ctx; + + old_ctx = prof_ctx_get(ptr); + if ((uintptr_t)cnt != (uintptr_t)1U) { + usize = ixallocx_prof_sample(ptr, old_usize, size, extra, + alignment, zero, max_usize, arena, cnt); + } else { + usize = ixallocx_helper(ptr, old_usize, size, extra, alignment, + zero, arena); + } + if (usize == old_usize) + return (usize); + prof_realloc(ptr, usize, cnt, old_usize, old_ctx); + + return (usize); +} + +size_t +je_xallocx(void *ptr, size_t size, size_t extra, int flags) +{ + size_t usize, old_usize; + UNUSED size_t old_rzsize JEMALLOC_CC_SILENCE_INIT(0); + size_t alignment = (ZU(1) << (flags & MALLOCX_LG_ALIGN_MASK) & (SIZE_T_MAX-1)); - bool zero = flags & ALLOCM_ZERO; - bool no_move = flags & ALLOCM_NO_MOVE; + bool zero = flags & MALLOCX_ZERO; unsigned arena_ind = ((unsigned)(flags >> 8)) - 1; - bool try_tcache_alloc, try_tcache_dalloc; arena_t *arena; assert(ptr != NULL); - assert(*ptr != NULL); assert(size != 0); assert(SIZE_T_MAX - size >= extra); assert(malloc_initialized || IS_INITIALIZER); malloc_thread_init(); - if (arena_ind != UINT_MAX) { - arena_chunk_t *chunk; - try_tcache_alloc = true; - chunk = (arena_chunk_t *)CHUNK_ADDR2BASE(*ptr); - try_tcache_dalloc = (chunk == *ptr || chunk->arena != - arenas[arena_ind]); + if (arena_ind != UINT_MAX) arena = arenas[arena_ind]; - } else { - try_tcache_alloc = true; - try_tcache_dalloc = true; + else arena = NULL; - } - p = *ptr; + old_usize = isalloc(ptr, config_prof); + if (config_valgrind && opt_valgrind) + old_rzsize = u2rz(old_usize); + if (config_prof && opt_prof) { prof_thr_cnt_t *cnt; - /* - * usize isn't knowable before iralloc() returns when extra is + * usize isn't knowable before ixalloc() returns when extra is * non-zero. Therefore, compute its maximum possible value and * use that in PROF_ALLOC_PREP() to decide whether to capture a * backtrace. prof_realloc() will use the actual usize to @@ -1546,112 +1716,51 @@ je_rallocm(void **ptr, size_t *rsize, size_t size, size_t extra, int flags) */ size_t max_usize = (alignment == 0) ? s2u(size+extra) : sa2u(size+extra, alignment); - prof_ctx_t *old_ctx = prof_ctx_get(p); - old_size = isalloc(p, true); - if (config_valgrind && opt_valgrind) - old_rzsize = p2rz(p); PROF_ALLOC_PREP(1, max_usize, cnt); - if (cnt == NULL) - goto label_oom; - /* - * Use minimum usize to determine whether promotion may happen. - */ - if (prof_promote && (uintptr_t)cnt != (uintptr_t)1U - && ((alignment == 0) ? s2u(size) : sa2u(size, alignment)) - <= SMALL_MAXCLASS) { - q = irallocx(p, SMALL_MAXCLASS+1, (SMALL_MAXCLASS+1 >= - size+extra) ? 0 : size+extra - (SMALL_MAXCLASS+1), - alignment, zero, no_move, try_tcache_alloc, - try_tcache_dalloc, arena); - if (q == NULL) - goto label_err; - if (max_usize < PAGE) { - usize = max_usize; - arena_prof_promoted(q, usize); - } else - usize = isalloc(q, config_prof); - } else { - q = irallocx(p, size, extra, alignment, zero, no_move, - try_tcache_alloc, try_tcache_dalloc, arena); - if (q == NULL) - goto label_err; - usize = isalloc(q, config_prof); - } - prof_realloc(q, usize, cnt, old_size, old_ctx); - if (rsize != NULL) - *rsize = usize; + usize = ixallocx_prof(ptr, old_usize, size, extra, alignment, + max_usize, zero, arena, cnt); } else { - if (config_stats) { - old_size = isalloc(p, false); - if (config_valgrind && opt_valgrind) - old_rzsize = u2rz(old_size); - } else if (config_valgrind && opt_valgrind) { - old_size = isalloc(p, false); - old_rzsize = u2rz(old_size); - } - q = irallocx(p, size, extra, alignment, zero, no_move, - try_tcache_alloc, try_tcache_dalloc, arena); - if (q == NULL) - goto label_err; - if (config_stats) - usize = isalloc(q, config_prof); - if (rsize != NULL) { - if (config_stats == false) - usize = isalloc(q, config_prof); - *rsize = usize; - } + usize = ixallocx_helper(ptr, old_usize, size, extra, alignment, + zero, arena); } + if (usize == old_usize) + goto label_not_resized; - *ptr = q; if (config_stats) { thread_allocated_t *ta; ta = thread_allocated_tsd_get(); ta->allocated += usize; - ta->deallocated += old_size; - } - UTRACE(p, size, q); - JEMALLOC_VALGRIND_REALLOC(q, usize, p, old_size, old_rzsize, zero); - return (ALLOCM_SUCCESS); -label_err: - if (no_move) { - UTRACE(p, size, q); - return (ALLOCM_ERR_NOT_MOVED); - } -label_oom: - if (config_xmalloc && opt_xmalloc) { - malloc_write("<jemalloc>: Error in rallocm(): " - "out of memory\n"); - abort(); + ta->deallocated += old_usize; } - UTRACE(p, size, 0); - return (ALLOCM_ERR_OOM); + JEMALLOC_VALGRIND_REALLOC(ptr, usize, ptr, old_usize, old_rzsize, zero); +label_not_resized: + UTRACE(ptr, size, ptr); + return (usize); } -int -je_sallocm(const void *ptr, size_t *rsize, int flags) +size_t +je_sallocx(const void *ptr, int flags) { - size_t sz; + size_t usize; assert(malloc_initialized || IS_INITIALIZER); malloc_thread_init(); if (config_ivsalloc) - sz = ivsalloc(ptr, config_prof); + usize = ivsalloc(ptr, config_prof); else { assert(ptr != NULL); - sz = isalloc(ptr, config_prof); + usize = isalloc(ptr, config_prof); } - assert(rsize != NULL); - *rsize = sz; - return (ALLOCM_SUCCESS); + return (usize); } -int -je_dallocm(void *ptr, int flags) +void +je_dallocx(void *ptr, int flags) { size_t usize; - size_t rzsize JEMALLOC_CC_SILENCE_INIT(0); + UNUSED size_t rzsize JEMALLOC_CC_SILENCE_INIT(0); unsigned arena_ind = ((unsigned)(flags >> 8)) - 1; bool try_tcache; @@ -1677,28 +1786,162 @@ je_dallocm(void *ptr, int flags) thread_allocated_tsd_get()->deallocated += usize; if (config_valgrind && opt_valgrind) rzsize = p2rz(ptr); - iqallocx(ptr, try_tcache); + iqalloct(ptr, try_tcache); JEMALLOC_VALGRIND_FREE(ptr, rzsize); - - return (ALLOCM_SUCCESS); } -int -je_nallocm(size_t *rsize, size_t size, int flags) +size_t +je_nallocx(size_t size, int flags) { size_t usize; - size_t alignment = (ZU(1) << (flags & ALLOCM_LG_ALIGN_MASK) + size_t alignment = (ZU(1) << (flags & MALLOCX_LG_ALIGN_MASK) & (SIZE_T_MAX-1)); assert(size != 0); if (malloc_init()) - return (ALLOCM_ERR_OOM); + return (0); usize = (alignment == 0) ? s2u(size) : sa2u(size, alignment); - if (usize == 0) + assert(usize != 0); + return (usize); +} + +int +je_mallctl(const char *name, void *oldp, size_t *oldlenp, void *newp, + size_t newlen) +{ + + if (malloc_init()) + return (EAGAIN); + + return (ctl_byname(name, oldp, oldlenp, newp, newlen)); +} + +int +je_mallctlnametomib(const char *name, size_t *mibp, size_t *miblenp) +{ + + if (malloc_init()) + return (EAGAIN); + + return (ctl_nametomib(name, mibp, miblenp)); +} + +int +je_mallctlbymib(const size_t *mib, size_t miblen, void *oldp, size_t *oldlenp, + void *newp, size_t newlen) +{ + + if (malloc_init()) + return (EAGAIN); + + return (ctl_bymib(mib, miblen, oldp, oldlenp, newp, newlen)); +} + +void +je_malloc_stats_print(void (*write_cb)(void *, const char *), void *cbopaque, + const char *opts) +{ + + stats_print(write_cb, cbopaque, opts); +} + +size_t +je_malloc_usable_size(JEMALLOC_USABLE_SIZE_CONST void *ptr) +{ + size_t ret; + + assert(malloc_initialized || IS_INITIALIZER); + malloc_thread_init(); + + if (config_ivsalloc) + ret = ivsalloc(ptr, config_prof); + else + ret = (ptr != NULL) ? isalloc(ptr, config_prof) : 0; + + return (ret); +} + +/* + * End non-standard functions. + */ +/******************************************************************************/ +/* + * Begin experimental functions. + */ +#ifdef JEMALLOC_EXPERIMENTAL + +int +je_allocm(void **ptr, size_t *rsize, size_t size, int flags) +{ + void *p; + + assert(ptr != NULL); + + p = je_mallocx(size, flags); + if (p == NULL) return (ALLOCM_ERR_OOM); + if (rsize != NULL) + *rsize = isalloc(p, config_prof); + *ptr = p; + return (ALLOCM_SUCCESS); +} + +int +je_rallocm(void **ptr, size_t *rsize, size_t size, size_t extra, int flags) +{ + int ret; + bool no_move = flags & ALLOCM_NO_MOVE; + + assert(ptr != NULL); + assert(*ptr != NULL); + assert(size != 0); + assert(SIZE_T_MAX - size >= extra); + + if (no_move) { + size_t usize = je_xallocx(*ptr, size, extra, flags); + ret = (usize >= size) ? ALLOCM_SUCCESS : ALLOCM_ERR_NOT_MOVED; + if (rsize != NULL) + *rsize = usize; + } else { + void *p = je_rallocx(*ptr, size+extra, flags); + if (p != NULL) { + *ptr = p; + ret = ALLOCM_SUCCESS; + } else + ret = ALLOCM_ERR_OOM; + if (rsize != NULL) + *rsize = isalloc(*ptr, config_prof); + } + return (ret); +} + +int +je_sallocm(const void *ptr, size_t *rsize, int flags) +{ + assert(rsize != NULL); + *rsize = je_sallocx(ptr, flags); + return (ALLOCM_SUCCESS); +} + +int +je_dallocm(void *ptr, int flags) +{ + + je_dallocx(ptr, flags); + return (ALLOCM_SUCCESS); +} + +int +je_nallocm(size_t *rsize, size_t size, int flags) +{ + size_t usize; + + usize = je_nallocx(size, flags); + if (usize == 0) + return (ALLOCM_ERR_OOM); if (rsize != NULL) *rsize = usize; return (ALLOCM_SUCCESS); diff --git a/src/mutex.c b/src/mutex.c index 55e18c23..788eca38 100644 --- a/src/mutex.c +++ b/src/mutex.c @@ -6,7 +6,7 @@ #endif #ifndef _CRT_SPINCOUNT -#define _CRT_SPINCOUNT 4000 +#define _CRT_SPINCOUNT 4000 #endif /******************************************************************************/ @@ -24,7 +24,12 @@ bool opt_prof_gdump = false; bool opt_prof_final = true; bool opt_prof_leak = false; bool opt_prof_accum = false; -char opt_prof_prefix[PATH_MAX + 1]; +char opt_prof_prefix[ + /* Minimize memory bloat for non-prof builds. */ +#ifdef JEMALLOC_PROF + PATH_MAX + +#endif + 1]; uint64_t prof_interval = 0; bool prof_promote; @@ -54,10 +59,17 @@ static uint64_t prof_dump_useq; /* * This buffer is rather large for stack allocation, so use a single buffer for - * all profile dumps. The buffer is implicitly protected by bt2ctx_mtx, since - * it must be locked anyway during dumping. + * all profile dumps. */ -static char prof_dump_buf[PROF_DUMP_BUFSIZE]; +static malloc_mutex_t prof_dump_mtx; +static char prof_dump_buf[ + /* Minimize memory bloat for non-prof builds. */ +#ifdef JEMALLOC_PROF + PROF_DUMP_BUFSIZE +#else + 1 +#endif +]; static unsigned prof_dump_buf_end; static int prof_dump_fd; @@ -65,36 +77,6 @@ static int prof_dump_fd; static bool prof_booted = false; /******************************************************************************/ -/* Function prototypes for non-inline static functions. */ - -static prof_bt_t *bt_dup(prof_bt_t *bt); -static void bt_destroy(prof_bt_t *bt); -#ifdef JEMALLOC_PROF_LIBGCC -static _Unwind_Reason_Code prof_unwind_init_callback( - struct _Unwind_Context *context, void *arg); -static _Unwind_Reason_Code prof_unwind_callback( - struct _Unwind_Context *context, void *arg); -#endif -static bool prof_flush(bool propagate_err); -static bool prof_write(bool propagate_err, const char *s); -static bool prof_printf(bool propagate_err, const char *format, ...) - JEMALLOC_ATTR(format(printf, 2, 3)); -static void prof_ctx_sum(prof_ctx_t *ctx, prof_cnt_t *cnt_all, - size_t *leak_nctx); -static void prof_ctx_destroy(prof_ctx_t *ctx); -static void prof_ctx_merge(prof_ctx_t *ctx, prof_thr_cnt_t *cnt); -static bool prof_dump_ctx(bool propagate_err, prof_ctx_t *ctx, - prof_bt_t *bt); -static bool prof_dump_maps(bool propagate_err); -static bool prof_dump(bool propagate_err, const char *filename, - bool leakcheck); -static void prof_dump_filename(char *filename, char v, int64_t vseq); -static void prof_fdump(void); -static void prof_bt_hash(const void *key, size_t r_hash[2]); -static bool prof_bt_keycomp(const void *k1, const void *k2); -static malloc_mutex_t *prof_ctx_mutex_choose(void); - -/******************************************************************************/ void bt_init(prof_bt_t *bt, void **vec) @@ -423,10 +405,169 @@ prof_backtrace(prof_bt_t *bt, unsigned nignore) { cassert(config_prof); - assert(false); + not_reached(); } #endif +static malloc_mutex_t * +prof_ctx_mutex_choose(void) +{ + unsigned nctxs = atomic_add_u(&cum_ctxs, 1); + + return (&ctx_locks[(nctxs - 1) % PROF_NCTX_LOCKS]); +} + +static void +prof_ctx_init(prof_ctx_t *ctx, prof_bt_t *bt) +{ + + ctx->bt = bt; + ctx->lock = prof_ctx_mutex_choose(); + /* + * Set nlimbo to 1, in order to avoid a race condition with + * prof_ctx_merge()/prof_ctx_destroy(). + */ + ctx->nlimbo = 1; + ql_elm_new(ctx, dump_link); + memset(&ctx->cnt_merged, 0, sizeof(prof_cnt_t)); + ql_new(&ctx->cnts_ql); +} + +static void +prof_ctx_destroy(prof_ctx_t *ctx) +{ + prof_tdata_t *prof_tdata; + + cassert(config_prof); + + /* + * Check that ctx is still unused by any thread cache before destroying + * it. prof_lookup() increments ctx->nlimbo in order to avoid a race + * condition with this function, as does prof_ctx_merge() in order to + * avoid a race between the main body of prof_ctx_merge() and entry + * into this function. + */ + prof_tdata = prof_tdata_get(false); + assert((uintptr_t)prof_tdata > (uintptr_t)PROF_TDATA_STATE_MAX); + prof_enter(prof_tdata); + malloc_mutex_lock(ctx->lock); + if (ql_first(&ctx->cnts_ql) == NULL && ctx->cnt_merged.curobjs == 0 && + ctx->nlimbo == 1) { + assert(ctx->cnt_merged.curbytes == 0); + assert(ctx->cnt_merged.accumobjs == 0); + assert(ctx->cnt_merged.accumbytes == 0); + /* Remove ctx from bt2ctx. */ + if (ckh_remove(&bt2ctx, ctx->bt, NULL, NULL)) + not_reached(); + prof_leave(prof_tdata); + /* Destroy ctx. */ + malloc_mutex_unlock(ctx->lock); + bt_destroy(ctx->bt); + idalloc(ctx); + } else { + /* + * Compensate for increment in prof_ctx_merge() or + * prof_lookup(). + */ + ctx->nlimbo--; + malloc_mutex_unlock(ctx->lock); + prof_leave(prof_tdata); + } +} + +static void +prof_ctx_merge(prof_ctx_t *ctx, prof_thr_cnt_t *cnt) +{ + bool destroy; + + cassert(config_prof); + + /* Merge cnt stats and detach from ctx. */ + malloc_mutex_lock(ctx->lock); + ctx->cnt_merged.curobjs += cnt->cnts.curobjs; + ctx->cnt_merged.curbytes += cnt->cnts.curbytes; + ctx->cnt_merged.accumobjs += cnt->cnts.accumobjs; + ctx->cnt_merged.accumbytes += cnt->cnts.accumbytes; + ql_remove(&ctx->cnts_ql, cnt, cnts_link); + if (opt_prof_accum == false && ql_first(&ctx->cnts_ql) == NULL && + ctx->cnt_merged.curobjs == 0 && ctx->nlimbo == 0) { + /* + * Increment ctx->nlimbo in order to keep another thread from + * winning the race to destroy ctx while this one has ctx->lock + * dropped. Without this, it would be possible for another + * thread to: + * + * 1) Sample an allocation associated with ctx. + * 2) Deallocate the sampled object. + * 3) Successfully prof_ctx_destroy(ctx). + * + * The result would be that ctx no longer exists by the time + * this thread accesses it in prof_ctx_destroy(). + */ + ctx->nlimbo++; + destroy = true; + } else + destroy = false; + malloc_mutex_unlock(ctx->lock); + if (destroy) + prof_ctx_destroy(ctx); +} + +static bool +prof_lookup_global(prof_bt_t *bt, prof_tdata_t *prof_tdata, void **p_btkey, + prof_ctx_t **p_ctx, bool *p_new_ctx) +{ + union { + prof_ctx_t *p; + void *v; + } ctx; + union { + prof_bt_t *p; + void *v; + } btkey; + bool new_ctx; + + prof_enter(prof_tdata); + if (ckh_search(&bt2ctx, bt, &btkey.v, &ctx.v)) { + /* bt has never been seen before. Insert it. */ + ctx.v = imalloc(sizeof(prof_ctx_t)); + if (ctx.v == NULL) { + prof_leave(prof_tdata); + return (true); + } + btkey.p = bt_dup(bt); + if (btkey.v == NULL) { + prof_leave(prof_tdata); + idalloc(ctx.v); + return (true); + } + prof_ctx_init(ctx.p, btkey.p); + if (ckh_insert(&bt2ctx, btkey.v, ctx.v)) { + /* OOM. */ + prof_leave(prof_tdata); + idalloc(btkey.v); + idalloc(ctx.v); + return (true); + } + new_ctx = true; + } else { + /* + * Increment nlimbo, in order to avoid a race condition with + * prof_ctx_merge()/prof_ctx_destroy(). + */ + malloc_mutex_lock(ctx.p->lock); + ctx.p->nlimbo++; + malloc_mutex_unlock(ctx.p->lock); + new_ctx = false; + } + prof_leave(prof_tdata); + + *p_btkey = btkey.v; + *p_ctx = ctx.p; + *p_new_ctx = new_ctx; + return (false); +} + prof_thr_cnt_t * prof_lookup(prof_bt_t *bt) { @@ -443,62 +584,16 @@ prof_lookup(prof_bt_t *bt) return (NULL); if (ckh_search(&prof_tdata->bt2cnt, bt, NULL, &ret.v)) { - union { - prof_bt_t *p; - void *v; - } btkey; - union { - prof_ctx_t *p; - void *v; - } ctx; + void *btkey; + prof_ctx_t *ctx; bool new_ctx; /* * This thread's cache lacks bt. Look for it in the global * cache. */ - prof_enter(prof_tdata); - if (ckh_search(&bt2ctx, bt, &btkey.v, &ctx.v)) { - /* bt has never been seen before. Insert it. */ - ctx.v = imalloc(sizeof(prof_ctx_t)); - if (ctx.v == NULL) { - prof_leave(prof_tdata); - return (NULL); - } - btkey.p = bt_dup(bt); - if (btkey.v == NULL) { - prof_leave(prof_tdata); - idalloc(ctx.v); - return (NULL); - } - ctx.p->bt = btkey.p; - ctx.p->lock = prof_ctx_mutex_choose(); - /* - * Set nlimbo to 1, in order to avoid a race condition - * with prof_ctx_merge()/prof_ctx_destroy(). - */ - ctx.p->nlimbo = 1; - memset(&ctx.p->cnt_merged, 0, sizeof(prof_cnt_t)); - ql_new(&ctx.p->cnts_ql); - if (ckh_insert(&bt2ctx, btkey.v, ctx.v)) { - /* OOM. */ - prof_leave(prof_tdata); - idalloc(btkey.v); - idalloc(ctx.v); - return (NULL); - } - new_ctx = true; - } else { - /* - * Increment nlimbo, in order to avoid a race condition - * with prof_ctx_merge()/prof_ctx_destroy(). - */ - malloc_mutex_lock(ctx.p->lock); - ctx.p->nlimbo++; - malloc_mutex_unlock(ctx.p->lock); - new_ctx = false; - } - prof_leave(prof_tdata); + if (prof_lookup_global(bt, prof_tdata, &btkey, &ctx, &new_ctx)) + return (NULL); /* Link a prof_thd_cnt_t into ctx for this thread. */ if (ckh_count(&prof_tdata->bt2cnt) == PROF_TCMAX) { @@ -511,7 +606,7 @@ prof_lookup(prof_bt_t *bt) assert(ret.v != NULL); if (ckh_remove(&prof_tdata->bt2cnt, ret.p->ctx->bt, NULL, NULL)) - assert(false); + not_reached(); ql_remove(&prof_tdata->lru_ql, ret.p, lru_link); prof_ctx_merge(ret.p->ctx, ret.p); /* ret can now be re-used. */ @@ -521,27 +616,27 @@ prof_lookup(prof_bt_t *bt) ret.v = imalloc(sizeof(prof_thr_cnt_t)); if (ret.p == NULL) { if (new_ctx) - prof_ctx_destroy(ctx.p); + prof_ctx_destroy(ctx); return (NULL); } ql_elm_new(ret.p, cnts_link); ql_elm_new(ret.p, lru_link); } /* Finish initializing ret. */ - ret.p->ctx = ctx.p; + ret.p->ctx = ctx; ret.p->epoch = 0; memset(&ret.p->cnts, 0, sizeof(prof_cnt_t)); - if (ckh_insert(&prof_tdata->bt2cnt, btkey.v, ret.v)) { + if (ckh_insert(&prof_tdata->bt2cnt, btkey, ret.v)) { if (new_ctx) - prof_ctx_destroy(ctx.p); + prof_ctx_destroy(ctx); idalloc(ret.v); return (NULL); } ql_head_insert(&prof_tdata->lru_ql, ret.p, lru_link); - malloc_mutex_lock(ctx.p->lock); - ql_tail_insert(&ctx.p->cnts_ql, ret.p, cnts_link); - ctx.p->nlimbo--; - malloc_mutex_unlock(ctx.p->lock); + malloc_mutex_lock(ctx->lock); + ql_tail_insert(&ctx->cnts_ql, ret.p, cnts_link); + ctx->nlimbo--; + malloc_mutex_unlock(ctx->lock); } else { /* Move ret to the front of the LRU. */ ql_remove(&prof_tdata->lru_ql, ret.p, lru_link); @@ -551,8 +646,52 @@ prof_lookup(prof_bt_t *bt) return (ret.p); } +#ifdef JEMALLOC_JET +size_t +prof_bt_count(void) +{ + size_t bt_count; + prof_tdata_t *prof_tdata; + + prof_tdata = prof_tdata_get(false); + if ((uintptr_t)prof_tdata <= (uintptr_t)PROF_TDATA_STATE_MAX) + return (0); + + prof_enter(prof_tdata); + bt_count = ckh_count(&bt2ctx); + prof_leave(prof_tdata); + + return (bt_count); +} +#endif + +#ifdef JEMALLOC_JET +#undef prof_dump_open +#define prof_dump_open JEMALLOC_N(prof_dump_open_impl) +#endif +static int +prof_dump_open(bool propagate_err, const char *filename) +{ + int fd; + + fd = creat(filename, 0644); + if (fd == -1 && propagate_err == false) { + malloc_printf("<jemalloc>: creat(\"%s\"), 0644) failed\n", + filename); + if (opt_abort) + abort(); + } + + return (fd); +} +#ifdef JEMALLOC_JET +#undef prof_dump_open +#define prof_dump_open JEMALLOC_N(prof_dump_open) +prof_dump_open_t *prof_dump_open = JEMALLOC_N(prof_dump_open_impl); +#endif + static bool -prof_flush(bool propagate_err) +prof_dump_flush(bool propagate_err) { bool ret = false; ssize_t err; @@ -575,7 +714,20 @@ prof_flush(bool propagate_err) } static bool -prof_write(bool propagate_err, const char *s) +prof_dump_close(bool propagate_err) +{ + bool ret; + + assert(prof_dump_fd != -1); + ret = prof_dump_flush(propagate_err); + close(prof_dump_fd); + prof_dump_fd = -1; + + return (ret); +} + +static bool +prof_dump_write(bool propagate_err, const char *s) { unsigned i, slen, n; @@ -586,7 +738,7 @@ prof_write(bool propagate_err, const char *s) while (i < slen) { /* Flush the buffer if it is full. */ if (prof_dump_buf_end == PROF_DUMP_BUFSIZE) - if (prof_flush(propagate_err) && propagate_err) + if (prof_dump_flush(propagate_err) && propagate_err) return (true); if (prof_dump_buf_end + slen <= PROF_DUMP_BUFSIZE) { @@ -606,7 +758,7 @@ prof_write(bool propagate_err, const char *s) JEMALLOC_ATTR(format(printf, 2, 3)) static bool -prof_printf(bool propagate_err, const char *format, ...) +prof_dump_printf(bool propagate_err, const char *format, ...) { bool ret; va_list ap; @@ -615,13 +767,14 @@ prof_printf(bool propagate_err, const char *format, ...) va_start(ap, format); malloc_vsnprintf(buf, sizeof(buf), format, ap); va_end(ap); - ret = prof_write(propagate_err, buf); + ret = prof_dump_write(propagate_err, buf); return (ret); } static void -prof_ctx_sum(prof_ctx_t *ctx, prof_cnt_t *cnt_all, size_t *leak_nctx) +prof_dump_ctx_prep(prof_ctx_t *ctx, prof_cnt_t *cnt_all, size_t *leak_nctx, + prof_ctx_list_t *ctx_ql) { prof_thr_cnt_t *thr_cnt; prof_cnt_t tcnt; @@ -630,6 +783,14 @@ prof_ctx_sum(prof_ctx_t *ctx, prof_cnt_t *cnt_all, size_t *leak_nctx) malloc_mutex_lock(ctx->lock); + /* + * Increment nlimbo so that ctx won't go away before dump. + * Additionally, link ctx into the dump list so that it is included in + * prof_dump()'s second pass. + */ + ctx->nlimbo++; + ql_tail_insert(ctx_ql, ctx, dump_link); + memcpy(&ctx->cnt_summed, &ctx->cnt_merged, sizeof(prof_cnt_t)); ql_foreach(thr_cnt, &ctx->cnts_ql, cnts_link) { volatile unsigned *epoch = &thr_cnt->epoch; @@ -670,89 +831,52 @@ prof_ctx_sum(prof_ctx_t *ctx, prof_cnt_t *cnt_all, size_t *leak_nctx) malloc_mutex_unlock(ctx->lock); } -static void -prof_ctx_destroy(prof_ctx_t *ctx) +static bool +prof_dump_header(bool propagate_err, const prof_cnt_t *cnt_all) { - prof_tdata_t *prof_tdata; - cassert(config_prof); - - /* - * Check that ctx is still unused by any thread cache before destroying - * it. prof_lookup() increments ctx->nlimbo in order to avoid a race - * condition with this function, as does prof_ctx_merge() in order to - * avoid a race between the main body of prof_ctx_merge() and entry - * into this function. - */ - prof_tdata = prof_tdata_get(false); - assert((uintptr_t)prof_tdata > (uintptr_t)PROF_TDATA_STATE_MAX); - prof_enter(prof_tdata); - malloc_mutex_lock(ctx->lock); - if (ql_first(&ctx->cnts_ql) == NULL && ctx->cnt_merged.curobjs == 0 && - ctx->nlimbo == 1) { - assert(ctx->cnt_merged.curbytes == 0); - assert(ctx->cnt_merged.accumobjs == 0); - assert(ctx->cnt_merged.accumbytes == 0); - /* Remove ctx from bt2ctx. */ - if (ckh_remove(&bt2ctx, ctx->bt, NULL, NULL)) - assert(false); - prof_leave(prof_tdata); - /* Destroy ctx. */ - malloc_mutex_unlock(ctx->lock); - bt_destroy(ctx->bt); - idalloc(ctx); + if (opt_lg_prof_sample == 0) { + if (prof_dump_printf(propagate_err, + "heap profile: %"PRId64": %"PRId64 + " [%"PRIu64": %"PRIu64"] @ heapprofile\n", + cnt_all->curobjs, cnt_all->curbytes, + cnt_all->accumobjs, cnt_all->accumbytes)) + return (true); } else { - /* - * Compensate for increment in prof_ctx_merge() or - * prof_lookup(). - */ - ctx->nlimbo--; - malloc_mutex_unlock(ctx->lock); - prof_leave(prof_tdata); + if (prof_dump_printf(propagate_err, + "heap profile: %"PRId64": %"PRId64 + " [%"PRIu64": %"PRIu64"] @ heap_v2/%"PRIu64"\n", + cnt_all->curobjs, cnt_all->curbytes, + cnt_all->accumobjs, cnt_all->accumbytes, + ((uint64_t)1U << opt_lg_prof_sample))) + return (true); } + + return (false); } static void -prof_ctx_merge(prof_ctx_t *ctx, prof_thr_cnt_t *cnt) +prof_dump_ctx_cleanup_locked(prof_ctx_t *ctx, prof_ctx_list_t *ctx_ql) { - bool destroy; - cassert(config_prof); + ctx->nlimbo--; + ql_remove(ctx_ql, ctx, dump_link); +} + +static void +prof_dump_ctx_cleanup(prof_ctx_t *ctx, prof_ctx_list_t *ctx_ql) +{ - /* Merge cnt stats and detach from ctx. */ malloc_mutex_lock(ctx->lock); - ctx->cnt_merged.curobjs += cnt->cnts.curobjs; - ctx->cnt_merged.curbytes += cnt->cnts.curbytes; - ctx->cnt_merged.accumobjs += cnt->cnts.accumobjs; - ctx->cnt_merged.accumbytes += cnt->cnts.accumbytes; - ql_remove(&ctx->cnts_ql, cnt, cnts_link); - if (opt_prof_accum == false && ql_first(&ctx->cnts_ql) == NULL && - ctx->cnt_merged.curobjs == 0 && ctx->nlimbo == 0) { - /* - * Increment ctx->nlimbo in order to keep another thread from - * winning the race to destroy ctx while this one has ctx->lock - * dropped. Without this, it would be possible for another - * thread to: - * - * 1) Sample an allocation associated with ctx. - * 2) Deallocate the sampled object. - * 3) Successfully prof_ctx_destroy(ctx). - * - * The result would be that ctx no longer exists by the time - * this thread accesses it in prof_ctx_destroy(). - */ - ctx->nlimbo++; - destroy = true; - } else - destroy = false; + prof_dump_ctx_cleanup_locked(ctx, ctx_ql); malloc_mutex_unlock(ctx->lock); - if (destroy) - prof_ctx_destroy(ctx); } static bool -prof_dump_ctx(bool propagate_err, prof_ctx_t *ctx, prof_bt_t *bt) +prof_dump_ctx(bool propagate_err, prof_ctx_t *ctx, const prof_bt_t *bt, + prof_ctx_list_t *ctx_ql) { + bool ret; unsigned i; cassert(config_prof); @@ -764,36 +888,49 @@ prof_dump_ctx(bool propagate_err, prof_ctx_t *ctx, prof_bt_t *bt) * filled in. Avoid dumping any ctx that is an artifact of either * implementation detail. */ + malloc_mutex_lock(ctx->lock); if ((opt_prof_accum == false && ctx->cnt_summed.curobjs == 0) || (opt_prof_accum && ctx->cnt_summed.accumobjs == 0)) { assert(ctx->cnt_summed.curobjs == 0); assert(ctx->cnt_summed.curbytes == 0); assert(ctx->cnt_summed.accumobjs == 0); assert(ctx->cnt_summed.accumbytes == 0); - return (false); + ret = false; + goto label_return; } - if (prof_printf(propagate_err, "%"PRId64": %"PRId64 + if (prof_dump_printf(propagate_err, "%"PRId64": %"PRId64 " [%"PRIu64": %"PRIu64"] @", ctx->cnt_summed.curobjs, ctx->cnt_summed.curbytes, - ctx->cnt_summed.accumobjs, ctx->cnt_summed.accumbytes)) - return (true); + ctx->cnt_summed.accumobjs, ctx->cnt_summed.accumbytes)) { + ret = true; + goto label_return; + } for (i = 0; i < bt->len; i++) { - if (prof_printf(propagate_err, " %#"PRIxPTR, - (uintptr_t)bt->vec[i])) - return (true); + if (prof_dump_printf(propagate_err, " %#"PRIxPTR, + (uintptr_t)bt->vec[i])) { + ret = true; + goto label_return; + } } - if (prof_write(propagate_err, "\n")) - return (true); + if (prof_dump_write(propagate_err, "\n")) { + ret = true; + goto label_return; + } - return (false); + ret = false; +label_return: + prof_dump_ctx_cleanup_locked(ctx, ctx_ql); + malloc_mutex_unlock(ctx->lock); + return (ret); } static bool prof_dump_maps(bool propagate_err) { + bool ret; int mfd; char filename[PATH_MAX + 1]; @@ -805,25 +942,52 @@ prof_dump_maps(bool propagate_err) if (mfd != -1) { ssize_t nread; - if (prof_write(propagate_err, "\nMAPPED_LIBRARIES:\n") && - propagate_err) - return (true); + if (prof_dump_write(propagate_err, "\nMAPPED_LIBRARIES:\n") && + propagate_err) { + ret = true; + goto label_return; + } nread = 0; do { prof_dump_buf_end += nread; if (prof_dump_buf_end == PROF_DUMP_BUFSIZE) { /* Make space in prof_dump_buf before read(). */ - if (prof_flush(propagate_err) && propagate_err) - return (true); + if (prof_dump_flush(propagate_err) && + propagate_err) { + ret = true; + goto label_return; + } } nread = read(mfd, &prof_dump_buf[prof_dump_buf_end], PROF_DUMP_BUFSIZE - prof_dump_buf_end); } while (nread > 0); + } else { + ret = true; + goto label_return; + } + + ret = false; +label_return: + if (mfd != -1) close(mfd); - } else - return (true); + return (ret); +} - return (false); +static void +prof_leakcheck(const prof_cnt_t *cnt_all, size_t leak_nctx, + const char *filename) +{ + + if (cnt_all->curbytes != 0) { + malloc_printf("<jemalloc>: Leak summary: %"PRId64" byte%s, %" + PRId64" object%s, %zu context%s\n", + cnt_all->curbytes, (cnt_all->curbytes != 1) ? "s" : "", + cnt_all->curobjs, (cnt_all->curobjs != 1) ? "s" : "", + leak_nctx, (leak_nctx != 1) ? "s" : ""); + malloc_printf( + "<jemalloc>: Run pprof on \"%s\" for leak detail\n", + filename); + } } static bool @@ -833,98 +997,74 @@ prof_dump(bool propagate_err, const char *filename, bool leakcheck) prof_cnt_t cnt_all; size_t tabind; union { - prof_bt_t *p; - void *v; - } bt; - union { prof_ctx_t *p; void *v; } ctx; size_t leak_nctx; + prof_ctx_list_t ctx_ql; cassert(config_prof); prof_tdata = prof_tdata_get(false); if ((uintptr_t)prof_tdata <= (uintptr_t)PROF_TDATA_STATE_MAX) return (true); - prof_enter(prof_tdata); - prof_dump_fd = creat(filename, 0644); - if (prof_dump_fd == -1) { - if (propagate_err == false) { - malloc_printf( - "<jemalloc>: creat(\"%s\"), 0644) failed\n", - filename); - if (opt_abort) - abort(); - } - goto label_error; - } + + malloc_mutex_lock(&prof_dump_mtx); /* Merge per thread profile stats, and sum them in cnt_all. */ memset(&cnt_all, 0, sizeof(prof_cnt_t)); leak_nctx = 0; + ql_new(&ctx_ql); + prof_enter(prof_tdata); for (tabind = 0; ckh_iter(&bt2ctx, &tabind, NULL, &ctx.v) == false;) - prof_ctx_sum(ctx.p, &cnt_all, &leak_nctx); + prof_dump_ctx_prep(ctx.p, &cnt_all, &leak_nctx, &ctx_ql); + prof_leave(prof_tdata); + + /* Create dump file. */ + if ((prof_dump_fd = prof_dump_open(propagate_err, filename)) == -1) + goto label_open_close_error; /* Dump profile header. */ - if (opt_lg_prof_sample == 0) { - if (prof_printf(propagate_err, - "heap profile: %"PRId64": %"PRId64 - " [%"PRIu64": %"PRIu64"] @ heapprofile\n", - cnt_all.curobjs, cnt_all.curbytes, - cnt_all.accumobjs, cnt_all.accumbytes)) - goto label_error; - } else { - if (prof_printf(propagate_err, - "heap profile: %"PRId64": %"PRId64 - " [%"PRIu64": %"PRIu64"] @ heap_v2/%"PRIu64"\n", - cnt_all.curobjs, cnt_all.curbytes, - cnt_all.accumobjs, cnt_all.accumbytes, - ((uint64_t)1U << opt_lg_prof_sample))) - goto label_error; - } + if (prof_dump_header(propagate_err, &cnt_all)) + goto label_write_error; - /* Dump per ctx profile stats. */ - for (tabind = 0; ckh_iter(&bt2ctx, &tabind, &bt.v, &ctx.v) - == false;) { - if (prof_dump_ctx(propagate_err, ctx.p, bt.p)) - goto label_error; + /* Dump per ctx profile stats. */ + while ((ctx.p = ql_first(&ctx_ql)) != NULL) { + if (prof_dump_ctx(propagate_err, ctx.p, ctx.p->bt, &ctx_ql)) + goto label_write_error; } /* Dump /proc/<pid>/maps if possible. */ if (prof_dump_maps(propagate_err)) - goto label_error; + goto label_write_error; - if (prof_flush(propagate_err)) - goto label_error; - close(prof_dump_fd); - prof_leave(prof_tdata); + if (prof_dump_close(propagate_err)) + goto label_open_close_error; - if (leakcheck && cnt_all.curbytes != 0) { - malloc_printf("<jemalloc>: Leak summary: %"PRId64" byte%s, %" - PRId64" object%s, %zu context%s\n", - cnt_all.curbytes, (cnt_all.curbytes != 1) ? "s" : "", - cnt_all.curobjs, (cnt_all.curobjs != 1) ? "s" : "", - leak_nctx, (leak_nctx != 1) ? "s" : ""); - malloc_printf( - "<jemalloc>: Run pprof on \"%s\" for leak detail\n", - filename); - } + malloc_mutex_unlock(&prof_dump_mtx); + + if (leakcheck) + prof_leakcheck(&cnt_all, leak_nctx, filename); return (false); -label_error: - prof_leave(prof_tdata); +label_write_error: + prof_dump_close(propagate_err); +label_open_close_error: + while ((ctx.p = ql_first(&ctx_ql)) != NULL) + prof_dump_ctx_cleanup(ctx.p, &ctx_ql); + malloc_mutex_unlock(&prof_dump_mtx); return (true); } #define DUMP_FILENAME_BUFSIZE (PATH_MAX + 1) +#define VSEQ_INVALID UINT64_C(0xffffffffffffffff) static void prof_dump_filename(char *filename, char v, int64_t vseq) { cassert(config_prof); - if (vseq != UINT64_C(0xffffffffffffffff)) { + if (vseq != VSEQ_INVALID) { /* "<prefix>.<pid>.<seq>.v<vseq>.heap" */ malloc_snprintf(filename, DUMP_FILENAME_BUFSIZE, "%s.%d.%"PRIu64".%c%"PRId64".heap", @@ -950,7 +1090,7 @@ prof_fdump(void) if (opt_prof_final && opt_prof_prefix[0] != '\0') { malloc_mutex_lock(&prof_dump_seq_mtx); - prof_dump_filename(filename, 'f', UINT64_C(0xffffffffffffffff)); + prof_dump_filename(filename, 'f', VSEQ_INVALID); malloc_mutex_unlock(&prof_dump_seq_mtx); prof_dump(false, filename, opt_prof_leak); } @@ -1056,14 +1196,6 @@ prof_bt_keycomp(const void *k1, const void *k2) return (memcmp(bt1->vec, bt2->vec, bt1->len * sizeof(void *)) == 0); } -static malloc_mutex_t * -prof_ctx_mutex_choose(void) -{ - unsigned nctxs = atomic_add_u(&cum_ctxs, 1); - - return (&ctx_locks[(nctxs - 1) % PROF_NCTX_LOCKS]); -} - prof_tdata_t * prof_tdata_init(void) { @@ -1208,6 +1340,8 @@ prof_boot2(void) if (malloc_mutex_init(&prof_dump_seq_mtx)) return (true); + if (malloc_mutex_init(&prof_dump_mtx)) + return (true); if (atexit(prof_fdump) != 0) { malloc_write("<jemalloc>: Error in atexit()\n"); @@ -1245,10 +1379,10 @@ prof_prefork(void) if (opt_prof) { unsigned i; - malloc_mutex_lock(&bt2ctx_mtx); - malloc_mutex_lock(&prof_dump_seq_mtx); + malloc_mutex_prefork(&bt2ctx_mtx); + malloc_mutex_prefork(&prof_dump_seq_mtx); for (i = 0; i < PROF_NCTX_LOCKS; i++) - malloc_mutex_lock(&ctx_locks[i]); + malloc_mutex_prefork(&ctx_locks[i]); } } diff --git a/src/quarantine.c b/src/quarantine.c index f96a948d..54315116 100644 --- a/src/quarantine.c +++ b/src/quarantine.c @@ -141,8 +141,17 @@ quarantine(void *ptr) obj->usize = usize; quarantine->curbytes += usize; quarantine->curobjs++; - if (opt_junk) - memset(ptr, 0x5a, usize); + if (config_fill && opt_junk) { + /* + * Only do redzone validation if Valgrind isn't in + * operation. + */ + if ((config_valgrind == false || opt_valgrind == false) + && usize <= SMALL_MAXCLASS) + arena_quarantine_junk_small(ptr, usize); + else + memset(ptr, 0x5a, usize); + } } else { assert(quarantine->curbytes == 0); idalloc(ptr); diff --git a/src/rtree.c b/src/rtree.c index 90c6935a..205957ac 100644 --- a/src/rtree.c +++ b/src/rtree.c @@ -2,42 +2,55 @@ #include "jemalloc/internal/jemalloc_internal.h" rtree_t * -rtree_new(unsigned bits) +rtree_new(unsigned bits, rtree_alloc_t *alloc, rtree_dalloc_t *dalloc) { rtree_t *ret; - unsigned bits_per_level, height, i; + unsigned bits_per_level, bits_in_leaf, height, i; + + assert(bits > 0 && bits <= (sizeof(uintptr_t) << 3)); bits_per_level = ffs(pow2_ceil((RTREE_NODESIZE / sizeof(void *)))) - 1; - height = bits / bits_per_level; - if (height * bits_per_level != bits) - height++; - assert(height * bits_per_level >= bits); + bits_in_leaf = ffs(pow2_ceil((RTREE_NODESIZE / sizeof(uint8_t)))) - 1; + if (bits > bits_in_leaf) { + height = 1 + (bits - bits_in_leaf) / bits_per_level; + if ((height-1) * bits_per_level + bits_in_leaf != bits) + height++; + } else { + height = 1; + } + assert((height-1) * bits_per_level + bits_in_leaf >= bits); - ret = (rtree_t*)base_alloc(offsetof(rtree_t, level2bits) + + ret = (rtree_t*)alloc(offsetof(rtree_t, level2bits) + (sizeof(unsigned) * height)); if (ret == NULL) return (NULL); memset(ret, 0, offsetof(rtree_t, level2bits) + (sizeof(unsigned) * height)); + ret->alloc = alloc; + ret->dalloc = dalloc; if (malloc_mutex_init(&ret->mutex)) { - /* Leak the rtree. */ + if (dalloc != NULL) + dalloc(ret); return (NULL); } ret->height = height; - if (bits_per_level * height > bits) - ret->level2bits[0] = bits % bits_per_level; - else - ret->level2bits[0] = bits_per_level; - for (i = 1; i < height; i++) - ret->level2bits[i] = bits_per_level; - - ret->root = (void**)base_alloc(sizeof(void *) << ret->level2bits[0]); + if (height > 1) { + if ((height-1) * bits_per_level + bits_in_leaf > bits) { + ret->level2bits[0] = (bits - bits_in_leaf) % + bits_per_level; + } else + ret->level2bits[0] = bits_per_level; + for (i = 1; i < height-1; i++) + ret->level2bits[i] = bits_per_level; + ret->level2bits[height-1] = bits_in_leaf; + } else + ret->level2bits[0] = bits; + + ret->root = (void**)alloc(sizeof(void *) << ret->level2bits[0]); if (ret->root == NULL) { - /* - * We leak the rtree here, since there's no generic base - * deallocation. - */ + if (dalloc != NULL) + dalloc(ret); return (NULL); } memset(ret->root, 0, sizeof(void *) << ret->level2bits[0]); @@ -45,6 +58,31 @@ rtree_new(unsigned bits) return (ret); } +static void +rtree_delete_subtree(rtree_t *rtree, void **node, unsigned level) +{ + + if (level < rtree->height - 1) { + size_t nchildren, i; + + nchildren = ZU(1) << rtree->level2bits[level]; + for (i = 0; i < nchildren; i++) { + void **child = (void **)node[i]; + if (child != NULL) + rtree_delete_subtree(rtree, child, level + 1); + } + } + rtree->dalloc(node); +} + +void +rtree_delete(rtree_t *rtree) +{ + + rtree_delete_subtree(rtree, rtree->root, 0); + rtree->dalloc(rtree); +} + void rtree_prefork(rtree_t *rtree) { diff --git a/src/stats.c b/src/stats.c index 43f87af6..bef2ab33 100644 --- a/src/stats.c +++ b/src/stats.c @@ -345,25 +345,25 @@ stats_print(void (*write_cb)(void *, const char *), void *cbopaque, malloc_cprintf(write_cb, cbopaque, "Assertions %s\n", bv ? "enabled" : "disabled"); -#define OPT_WRITE_BOOL(n) \ +#define OPT_WRITE_BOOL(n) \ if ((err = je_mallctl("opt."#n, &bv, &bsz, NULL, 0)) \ == 0) { \ malloc_cprintf(write_cb, cbopaque, \ " opt."#n": %s\n", bv ? "true" : "false"); \ } -#define OPT_WRITE_SIZE_T(n) \ +#define OPT_WRITE_SIZE_T(n) \ if ((err = je_mallctl("opt."#n, &sv, &ssz, NULL, 0)) \ == 0) { \ malloc_cprintf(write_cb, cbopaque, \ " opt."#n": %zu\n", sv); \ } -#define OPT_WRITE_SSIZE_T(n) \ +#define OPT_WRITE_SSIZE_T(n) \ if ((err = je_mallctl("opt."#n, &ssv, &sssz, NULL, 0)) \ == 0) { \ malloc_cprintf(write_cb, cbopaque, \ " opt."#n": %zd\n", ssv); \ } -#define OPT_WRITE_CHAR_P(n) \ +#define OPT_WRITE_CHAR_P(n) \ if ((err = je_mallctl("opt."#n, &cpv, &cpsz, NULL, 0)) \ == 0) { \ malloc_cprintf(write_cb, cbopaque, \ diff --git a/src/tcache.c b/src/tcache.c index 98ed19ed..6de92960 100644 --- a/src/tcache.c +++ b/src/tcache.c @@ -260,8 +260,8 @@ tcache_arena_dissociate(tcache_t *tcache) /* Unlink from list of extant tcaches. */ malloc_mutex_lock(&tcache->arena->lock); ql_remove(&tcache->arena->tcache_ql, tcache, link); - malloc_mutex_unlock(&tcache->arena->lock); tcache_stats_merge(tcache, tcache->arena); + malloc_mutex_unlock(&tcache->arena->lock); } } @@ -292,7 +292,7 @@ tcache_create(arena_t *arena) else if (size <= tcache_maxclass) tcache = (tcache_t *)arena_malloc_large(arena, size, true); else - tcache = (tcache_t *)icallocx(size, false, arena); + tcache = (tcache_t *)icalloct(size, false, arena); if (tcache == NULL) return (NULL); @@ -366,7 +366,7 @@ tcache_destroy(tcache_t *tcache) arena_dalloc_large(arena, chunk, tcache); } else - idallocx(tcache, false); + idalloct(tcache, false); } void @@ -399,11 +399,14 @@ tcache_thread_cleanup(void *arg) } } +/* Caller must own arena->lock. */ void tcache_stats_merge(tcache_t *tcache, arena_t *arena) { unsigned i; + cassert(config_stats); + /* Merge and reset tcache stats. */ for (i = 0; i < NBINS; i++) { arena_bin_t *bin = &arena->bins[i]; @@ -21,7 +21,7 @@ void malloc_tsd_dalloc(void *wrapper) { - idalloc(wrapper); + idalloct(wrapper, false); } void @@ -105,3 +105,37 @@ JEMALLOC_SECTION(".CRT$XLY") JEMALLOC_ATTR(used) static const BOOL (WINAPI *tls_callback)(HINSTANCE hinstDLL, DWORD fdwReason, LPVOID lpvReserved) = _tls_callback; #endif + +#if (!defined(JEMALLOC_MALLOC_THREAD_CLEANUP) && !defined(JEMALLOC_TLS) && \ + !defined(_WIN32)) +void * +tsd_init_check_recursion(tsd_init_head_t *head, tsd_init_block_t *block) +{ + pthread_t self = pthread_self(); + tsd_init_block_t *iter; + + /* Check whether this thread has already inserted into the list. */ + malloc_mutex_lock(&head->lock); + ql_foreach(iter, &head->blocks, link) { + if (iter->thread == self) { + malloc_mutex_unlock(&head->lock); + return (iter->data); + } + } + /* Insert block into list. */ + ql_elm_new(block, link); + block->thread = self; + ql_tail_insert(&head->blocks, block, link); + malloc_mutex_unlock(&head->lock); + return (NULL); +} + +void +tsd_init_finish(tsd_init_head_t *head, tsd_init_block_t *block) +{ + + malloc_mutex_lock(&head->lock); + ql_remove(&head->blocks, block, link); + malloc_mutex_unlock(&head->lock); +} +#endif @@ -77,7 +77,7 @@ malloc_write(const char *s) * provide a wrapper. */ int -buferror(char *buf, size_t buflen) +buferror(int err, char *buf, size_t buflen) { #ifdef _WIN32 @@ -85,34 +85,36 @@ buferror(char *buf, size_t buflen) (LPSTR)buf, buflen, NULL); return (0); #elif defined(_GNU_SOURCE) - char *b = strerror_r(errno, buf, buflen); + char *b = strerror_r(err, buf, buflen); if (b != buf) { strncpy(buf, b, buflen); buf[buflen-1] = '\0'; } return (0); #else - return (strerror_r(errno, buf, buflen)); + return (strerror_r(err, buf, buflen)); #endif } uintmax_t -malloc_strtoumax(const char *nptr, char **endptr, int base) +malloc_strtoumax(const char *restrict nptr, char **restrict endptr, int base) { uintmax_t ret, digit; int b; bool neg; const char *p, *ns; + p = nptr; if (base < 0 || base == 1 || base > 36) { + ns = p; set_errno(EINVAL); - return (UINTMAX_MAX); + ret = UINTMAX_MAX; + goto label_return; } b = base; /* Swallow leading whitespace and get sign, if any. */ neg = false; - p = nptr; while (true) { switch (*p) { case '\t': case '\n': case '\v': case '\f': case '\r': case ' ': @@ -146,7 +148,7 @@ malloc_strtoumax(const char *nptr, char **endptr, int base) if (b == 8) p++; break; - case 'x': + case 'X': case 'x': switch (p[2]) { case '0': case '1': case '2': case '3': case '4': case '5': case '6': case '7': case '8': case '9': @@ -164,7 +166,9 @@ malloc_strtoumax(const char *nptr, char **endptr, int base) } break; default: - break; + p++; + ret = 0; + goto label_return; } } if (b == 0) @@ -181,13 +185,22 @@ malloc_strtoumax(const char *nptr, char **endptr, int base) if (ret < pret) { /* Overflow. */ set_errno(ERANGE); - return (UINTMAX_MAX); + ret = UINTMAX_MAX; + goto label_return; } p++; } if (neg) ret = -ret; + if (p == ns) { + /* No conversion performed. */ + set_errno(EINVAL); + ret = UINTMAX_MAX; + goto label_return; + } + +label_return: if (endptr != NULL) { if (p == ns) { /* No characters were converted. */ @@ -195,7 +208,6 @@ malloc_strtoumax(const char *nptr, char **endptr, int base) } else *endptr = (char *)p; } - return (ret); } @@ -331,7 +343,7 @@ malloc_vsnprintf(char *str, size_t size, const char *format, va_list ap) APPEND_C(' '); \ } \ } while (0) -#define GET_ARG_NUMERIC(val, len) do { \ +#define GET_ARG_NUMERIC(val, len) do { \ switch (len) { \ case '?': \ val = va_arg(ap, int); \ @@ -354,6 +366,9 @@ malloc_vsnprintf(char *str, size_t size, const char *format, va_list ap) case 'j': \ val = va_arg(ap, intmax_t); \ break; \ + case 'j' | 0x80: \ + val = va_arg(ap, uintmax_t); \ + break; \ case 't': \ val = va_arg(ap, ptrdiff_t); \ break; \ @@ -385,11 +400,6 @@ malloc_vsnprintf(char *str, size_t size, const char *format, va_list ap) unsigned char len = '?'; f++; - if (*f == '%') { - /* %% */ - APPEND_C(*f); - break; - } /* Flags. */ while (true) { switch (*f) { @@ -419,6 +429,10 @@ malloc_vsnprintf(char *str, size_t size, const char *format, va_list ap) case '*': width = va_arg(ap, int); f++; + if (width < 0) { + left_justify = true; + width = -width; + } break; case '0': case '1': case '2': case '3': case '4': case '5': case '6': case '7': case '8': case '9': { @@ -428,19 +442,16 @@ malloc_vsnprintf(char *str, size_t size, const char *format, va_list ap) assert(uwidth != UINTMAX_MAX || get_errno() != ERANGE); width = (int)uwidth; - if (*f == '.') { - f++; - goto label_precision; - } else - goto label_length; break; - } case '.': - f++; - goto label_precision; - default: goto label_length; + } default: + break; } + /* Width/precision separator. */ + if (*f == '.') + f++; + else + goto label_length; /* Precision. */ - label_precision: switch (*f) { case '*': prec = va_arg(ap, int); @@ -469,16 +480,8 @@ malloc_vsnprintf(char *str, size_t size, const char *format, va_list ap) } else len = 'l'; break; - case 'j': - len = 'j'; - f++; - break; - case 't': - len = 't'; - f++; - break; - case 'z': - len = 'z'; + case 'q': case 'j': case 't': case 'z': + len = *f; f++; break; default: break; @@ -487,6 +490,11 @@ malloc_vsnprintf(char *str, size_t size, const char *format, va_list ap) switch (*f) { char *s; size_t slen; + case '%': + /* %% */ + APPEND_C(*f); + f++; + break; case 'd': case 'i': { intmax_t val JEMALLOC_CC_SILENCE_INIT(0); char buf[D2S_BUFSIZE]; @@ -540,7 +548,7 @@ malloc_vsnprintf(char *str, size_t size, const char *format, va_list ap) assert(len == '?' || len == 'l'); assert_not_implemented(len != 'l'); s = va_arg(ap, char *); - slen = (prec == -1) ? strlen(s) : prec; + slen = (prec < 0) ? strlen(s) : prec; APPEND_PADDED_S(s, slen, width, left_justify); f++; break; @@ -553,8 +561,7 @@ malloc_vsnprintf(char *str, size_t size, const char *format, va_list ap) APPEND_PADDED_S(s, slen, width, left_justify); f++; break; - } - default: not_implemented(); + } default: not_reached(); } break; } default: { @@ -137,7 +137,7 @@ zone_destroy(malloc_zone_t *zone) { /* This function should never be called. */ - assert(false); + not_reached(); return (NULL); } |