project-base/server-skynet-source-3rd-jemalloc
2
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
7d63fed0fd
server-skynet-source-3rd-je.../include/jemalloc/internal/extent.h

376 lines
8.7 KiB
C
Raw Normal View History

Refactor jemalloc.c into multiple source files. Fix a stats bug in large object curruns accounting. Replace tcache_bin_fill() with arena_tcache_fill(), and fix a bug in an OOM error path. Fix API name mangling to coexist with __attribute__((malloc)).
2010-01-17 01:53:50 +08:00
/******************************************************************************/
#ifdef JEMALLOC_H_TYPES
Rename extent_node_t to extent_t.
2016-03-24 12:09:28 +08:00
typedef struct extent_s extent_t;
Refactor jemalloc.c into multiple source files. Fix a stats bug in large object curruns accounting. Replace tcache_bin_fill() with arena_tcache_fill(), and fix a bug in an OOM error path. Fix API name mangling to coexist with __attribute__((malloc)).
2010-01-17 01:53:50 +08:00
#endif /* JEMALLOC_H_TYPES */
/******************************************************************************/
#ifdef JEMALLOC_H_STRUCTS
Rename extent_node_t to extent_t.
2016-03-24 12:09:28 +08:00
/* Extent (span of pages). Use accessor functions for e_* fields. */
struct extent_s {
Move centralized chunk management into arenas. Migrate all centralized data structures related to huge allocations and recyclable chunks into arena_t, so that each arena can manage huge allocations and recyclable virtual memory completely independently of other arenas. Add chunk node caching to arenas, in order to avoid contention on the base allocator. Use chunks_rtree to look up huge allocations rather than a red-black tree. Maintain a per arena unsorted list of huge allocations (which will be needed to enumerate huge allocations during arena reset). Remove the --enable-ivsalloc option, make ivsalloc() always available, and use it for size queries if --enable-debug is enabled. The only practical implications to this removal are that 1) ivsalloc() is now always available during live debugging (and the underlying radix tree is available during core-based debugging), and 2) size query validation can no longer be enabled independent of --enable-debug. Remove the stats.chunks.{current,total,high} mallctls, and replace their underlying statistics with simpler atomically updated counters used exclusively for gdump triggering. These statistics are no longer very useful because each arena manages chunks independently, and per arena statistics provide similar information. Simplify chunk synchronization code, now that base chunk allocation cannot cause recursive lock acquisition.
2015-02-12 04:24:27 +08:00
/* Arena from which this extent came, if any. */
Rename extent_node_t to extent_t.
2016-03-24 12:09:28 +08:00
arena_t *e_arena;
Implement allocation profiling and leack checking. Add the --enable-prof and --enable-prof-libunwind configure options. Add the B/b, F/f, I/i, L/l, and U/u JEMALLOC_OPTIONS. Interval-based profile dump triggering is not yet implemented. Add supporting generic code: * Add memory barriers. * Add prn (LCG PRNG). * Add hash (Murmur hash function). * Add ckh (cuckoo hash tables).
2010-02-11 02:37:56 +08:00
Rename extent_node_t to extent_t.
2016-03-24 12:09:28 +08:00
/* Pointer to the extent that this structure is responsible for. */
void *e_addr;
Integrate whole chunks into unused dirty page purging machinery. Extend per arena unused dirty page purging to manage unused dirty chunks in aaddtion to unused dirty runs. Rather than immediately unmapping deallocated chunks (or purging them in the --disable-munmap case), store them in a separate set of trees, chunks_[sz]ad_dirty. Preferrentially allocate dirty chunks. When excessive unused dirty pages accumulate, purge runs and chunks in ingegrated LRU order (and unmap chunks in the --enable-munmap case). Refactor extent_node_t to provide accessor functions.
2015-02-16 10:04:46 +08:00
Use huge size class infrastructure for large size classes.
2016-05-28 15:17:28 +08:00
/* Extent size. */
Rename extent_node_t to extent_t.
2016-03-24 12:09:28 +08:00
size_t e_size;
Refactor jemalloc.c into multiple source files. Fix a stats bug in large object curruns accounting. Replace tcache_bin_fill() with arena_tcache_fill(), and fix a bug in an OOM error path. Fix API name mangling to coexist with __attribute__((malloc)).
2010-01-17 01:53:50 +08:00
Use huge size class infrastructure for large size classes.
2016-05-28 15:17:28 +08:00
/*
* Usable size, typically smaller than extent size due to large_pad or
* promotion of sampled small regions.
*/
size_t e_usize;
Add extent_active_[gs]et(). Always initialize extents' runs_dirty and chunks_cache linkage.
2016-03-28 18:17:10 +08:00
/* True if extent is active (in use). */
bool e_active;
Implement chunk hook support for page run commit/decommit. Cascade from decommit to purge when purging unused dirty pages, so that it is possible to decommit cleaned memory rather than just purging. For non-Windows debug builds, decommit runs rather than purging them, since this causes access of deallocated runs to segfault. This resolves #251.
2015-08-05 01:49:46 +08:00
/*
* The zeroed flag is used by chunk recycling code to track whether
* memory is zero-filled.
*/
Rename extent_node_t to extent_t.
2016-03-24 12:09:28 +08:00
bool e_zeroed;
Implement chunk hook support for page run commit/decommit. Cascade from decommit to purge when purging unused dirty pages, so that it is possible to decommit cleaned memory rather than just purging. For non-Windows debug builds, decommit runs rather than purging them, since this causes access of deallocated runs to segfault. This resolves #251.
2015-08-05 01:49:46 +08:00
Generalize chunk management hooks. Add the "arena.<i>.chunk_hooks" mallctl, which replaces and expands on the "arena.<i>.chunk.{alloc,dalloc,purge}" mallctls. The chunk hooks allow control over chunk allocation/deallocation, decommit/commit, purging, and splitting/merging, such that the application can rely on jemalloc's internal chunk caching and retaining functionality, yet implement a variety of chunk management mechanisms and policies. Merge the chunks_[sz]ad_{mmap,dss} red-black trees into chunks_[sz]ad_retained. This slightly reduces how hard jemalloc tries to honor the dss precedence setting; prior to this change the precedence setting was also consulted when recycling chunks. Fix chunk purging. Don't purge chunks in arena_purge_stashed(); instead deallocate them in arena_unstash_purged(), so that the dirty memory linkage remains valid until after the last time it is used. This resolves #176 and #201.
2015-07-28 23:28:19 +08:00
/*
* True if physical memory is committed to the extent, whether
* explicitly or implicitly as on a system that overcommits and
Arena chunk decommit cleanups and fixes. Decommit arena chunk header during chunk deallocation if the rest of the chunk is decommitted.
2015-08-10 07:47:27 +08:00
* satisfies physical memory needs on demand via soft page faults.
Generalize chunk management hooks. Add the "arena.<i>.chunk_hooks" mallctl, which replaces and expands on the "arena.<i>.chunk.{alloc,dalloc,purge}" mallctls. The chunk hooks allow control over chunk allocation/deallocation, decommit/commit, purging, and splitting/merging, such that the application can rely on jemalloc's internal chunk caching and retaining functionality, yet implement a variety of chunk management mechanisms and policies. Merge the chunks_[sz]ad_{mmap,dss} red-black trees into chunks_[sz]ad_retained. This slightly reduces how hard jemalloc tries to honor the dss precedence setting; prior to this change the precedence setting was also consulted when recycling chunks. Fix chunk purging. Don't purge chunks in arena_purge_stashed(); instead deallocate them in arena_unstash_purged(), so that the dirty memory linkage remains valid until after the last time it is used. This resolves #176 and #201.
2015-07-28 23:28:19 +08:00
*/
Rename extent_node_t to extent_t.
2016-03-24 12:09:28 +08:00
bool e_committed;
Generalize chunk management hooks. Add the "arena.<i>.chunk_hooks" mallctl, which replaces and expands on the "arena.<i>.chunk.{alloc,dalloc,purge}" mallctls. The chunk hooks allow control over chunk allocation/deallocation, decommit/commit, purging, and splitting/merging, such that the application can rely on jemalloc's internal chunk caching and retaining functionality, yet implement a variety of chunk management mechanisms and policies. Merge the chunks_[sz]ad_{mmap,dss} red-black trees into chunks_[sz]ad_retained. This slightly reduces how hard jemalloc tries to honor the dss precedence setting; prior to this change the precedence setting was also consulted when recycling chunks. Fix chunk purging. Don't purge chunks in arena_purge_stashed(); instead deallocate them in arena_unstash_purged(), so that the dirty memory linkage remains valid until after the last time it is used. This resolves #176 and #201.
2015-07-28 23:28:19 +08:00
Reduce extent_node_t size to fit in one cache line.
2015-01-31 13:21:16 +08:00
/*
Replace extent_achunk_[gs]et() with extent_slab_[gs]et().
2016-04-07 22:24:14 +08:00
* The slab flag indicates whether the extent is used for a slab of
* small regions. This helps differentiate small size classes, and it
* indicates whether interior pointers can be looked up via iealloc().
Reduce extent_node_t size to fit in one cache line.
2015-01-31 13:21:16 +08:00
*/
Replace extent_achunk_[gs]et() with extent_slab_[gs]et().
2016-04-07 22:24:14 +08:00
bool e_slab;
Integrate whole chunks into unused dirty page purging machinery. Extend per arena unused dirty page purging to manage unused dirty chunks in aaddtion to unused dirty runs. Rather than immediately unmapping deallocated chunks (or purging them in the --disable-munmap case), store them in a separate set of trees, chunks_[sz]ad_dirty. Preferrentially allocate dirty chunks. When excessive unused dirty pages accumulate, purge runs and chunks in ingegrated LRU order (and unmap chunks in the --enable-munmap case). Refactor extent_node_t to provide accessor functions.
2015-02-16 10:04:46 +08:00
Make extent_prof_tctx_[gs]et() atomic.
2016-05-25 12:13:36 +08:00
union {
Move slabs out of chunks.
2016-05-30 09:34:50 +08:00
/* Small region slab metadata. */
arena_slab_data_t e_slab_data;
Rename huge to large.
2016-06-01 05:50:21 +08:00
/* Profile counters, used for large objects. */
Move slabs out of chunks.
2016-05-30 09:34:50 +08:00
union {
void *e_prof_tctx_pun;
prof_tctx_t *e_prof_tctx;
};
Make extent_prof_tctx_[gs]et() atomic.
2016-05-25 12:13:36 +08:00
};
Simplify extent_node_t and add extent_node_init().
2015-02-18 07:13:52 +08:00
Move slabs out of chunks.
2016-05-30 09:34:50 +08:00
/*
* Linkage for arena's extents_dirty and arena_bin_t's slabs_full rings.
*/
qr(extent_t) qr_link;
Move centralized chunk management into arenas. Migrate all centralized data structures related to huge allocations and recyclable chunks into arena_t, so that each arena can manage huge allocations and recyclable virtual memory completely independently of other arenas. Add chunk node caching to arenas, in order to avoid contention on the base allocator. Use chunks_rtree to look up huge allocations rather than a red-black tree. Maintain a per arena unsorted list of huge allocations (which will be needed to enumerate huge allocations during arena reset). Remove the --enable-ivsalloc option, make ivsalloc() always available, and use it for size queries if --enable-debug is enabled. The only practical implications to this removal are that 1) ivsalloc() is now always available during live debugging (and the underlying radix tree is available during core-based debugging), and 2) size query validation can no longer be enabled independent of --enable-debug. Remove the stats.chunks.{current,total,high} mallctls, and replace their underlying statistics with simpler atomically updated counters used exclusively for gdump triggering. These statistics are no longer very useful because each arena manages chunks independently, and per arena statistics provide similar information. Simplify chunk synchronization code, now that base chunk allocation cannot cause recursive lock acquisition.
2015-02-12 04:24:27 +08:00
union {
Replace extent_tree_szad_* with extent_heap_*.
2016-05-18 05:58:56 +08:00
/* Linkage for per size class address-ordered heaps. */
phn(extent_t) ph_link;
Move centralized chunk management into arenas. Migrate all centralized data structures related to huge allocations and recyclable chunks into arena_t, so that each arena can manage huge allocations and recyclable virtual memory completely independently of other arenas. Add chunk node caching to arenas, in order to avoid contention on the base allocator. Use chunks_rtree to look up huge allocations rather than a red-black tree. Maintain a per arena unsorted list of huge allocations (which will be needed to enumerate huge allocations during arena reset). Remove the --enable-ivsalloc option, make ivsalloc() always available, and use it for size queries if --enable-debug is enabled. The only practical implications to this removal are that 1) ivsalloc() is now always available during live debugging (and the underlying radix tree is available during core-based debugging), and 2) size query validation can no longer be enabled independent of --enable-debug. Remove the stats.chunks.{current,total,high} mallctls, and replace their underlying statistics with simpler atomically updated counters used exclusively for gdump triggering. These statistics are no longer very useful because each arena manages chunks independently, and per arena statistics provide similar information. Simplify chunk synchronization code, now that base chunk allocation cannot cause recursive lock acquisition.
2015-02-12 04:24:27 +08:00
Rename huge to large.
2016-06-01 05:50:21 +08:00
/* Linkage for arena's large and extent_cache lists. */
Rename extent_node_t to extent_t.
2016-03-24 12:09:28 +08:00
ql_elm(extent_t) ql_link;
Move centralized chunk management into arenas. Migrate all centralized data structures related to huge allocations and recyclable chunks into arena_t, so that each arena can manage huge allocations and recyclable virtual memory completely independently of other arenas. Add chunk node caching to arenas, in order to avoid contention on the base allocator. Use chunks_rtree to look up huge allocations rather than a red-black tree. Maintain a per arena unsorted list of huge allocations (which will be needed to enumerate huge allocations during arena reset). Remove the --enable-ivsalloc option, make ivsalloc() always available, and use it for size queries if --enable-debug is enabled. The only practical implications to this removal are that 1) ivsalloc() is now always available during live debugging (and the underlying radix tree is available during core-based debugging), and 2) size query validation can no longer be enabled independent of --enable-debug. Remove the stats.chunks.{current,total,high} mallctls, and replace their underlying statistics with simpler atomically updated counters used exclusively for gdump triggering. These statistics are no longer very useful because each arena manages chunks independently, and per arena statistics provide similar information. Simplify chunk synchronization code, now that base chunk allocation cannot cause recursive lock acquisition.
2015-02-12 04:24:27 +08:00
};
Refactor jemalloc.c into multiple source files. Fix a stats bug in large object curruns accounting. Replace tcache_bin_fill() with arena_tcache_fill(), and fix a bug in an OOM error path. Fix API name mangling to coexist with __attribute__((malloc)).
2010-01-17 01:53:50 +08:00
};
Replace extent_tree_szad_* with extent_heap_*.
2016-05-18 05:58:56 +08:00
typedef ph(extent_t) extent_heap_t;
Refactor jemalloc.c into multiple source files. Fix a stats bug in large object curruns accounting. Replace tcache_bin_fill() with arena_tcache_fill(), and fix a bug in an OOM error path. Fix API name mangling to coexist with __attribute__((malloc)).
2010-01-17 01:53:50 +08:00
#endif /* JEMALLOC_H_STRUCTS */
/******************************************************************************/
#ifdef JEMALLOC_H_EXTERNS
Refactor chunk_dalloc_{cache,wrapper}() to take extent arguments. Rename arena_extent_[d]alloc() to extent_[d]alloc(). Move all chunk [de]registration responsibility into chunk.c.
2016-05-24 05:56:35 +08:00
extent_t *extent_alloc(tsdn_t *tsdn, arena_t *arena);
void extent_dalloc(tsdn_t *tsdn, arena_t *arena, extent_t *extent);
Replace extent_tree_szad_* with extent_heap_*.
2016-05-18 05:58:56 +08:00
#ifdef JEMALLOC_JET
typedef size_t (extent_size_quantize_t)(size_t);
extern extent_size_quantize_t *extent_size_quantize_floor;
extern extent_size_quantize_t *extent_size_quantize_ceil;
#else
size_t extent_size_quantize_floor(size_t size);
size_t extent_size_quantize_ceil(size_t size);
#endif
ph_proto(, extent_heap_, extent_heap_t, extent_t)
Refactor jemalloc.c into multiple source files. Fix a stats bug in large object curruns accounting. Replace tcache_bin_fill() with arena_tcache_fill(), and fix a bug in an OOM error path. Fix API name mangling to coexist with __attribute__((malloc)).
2010-01-17 01:53:50 +08:00
#endif /* JEMALLOC_H_EXTERNS */
/******************************************************************************/
#ifdef JEMALLOC_H_INLINES
Integrate whole chunks into unused dirty page purging machinery. Extend per arena unused dirty page purging to manage unused dirty chunks in aaddtion to unused dirty runs. Rather than immediately unmapping deallocated chunks (or purging them in the --disable-munmap case), store them in a separate set of trees, chunks_[sz]ad_dirty. Preferrentially allocate dirty chunks. When excessive unused dirty pages accumulate, purge runs and chunks in ingegrated LRU order (and unmap chunks in the --enable-munmap case). Refactor extent_node_t to provide accessor functions.
2015-02-16 10:04:46 +08:00
#ifndef JEMALLOC_ENABLE_INLINE
Rename extent_node_t to extent_t.
2016-03-24 12:09:28 +08:00
arena_t *extent_arena_get(const extent_t *extent);
Implement cache-oblivious support for huge size classes.
2016-05-28 09:57:15 +08:00
void *extent_base_get(const extent_t *extent);
Rename extent_node_t to extent_t.
2016-03-24 12:09:28 +08:00
void *extent_addr_get(const extent_t *extent);
size_t extent_size_get(const extent_t *extent);
Implement cache-oblivious support for huge size classes.
2016-05-28 09:57:15 +08:00
size_t extent_usize_get(const extent_t *extent);
Allow chunks to not be naturally aligned. Precisely size extents for huge size classes that aren't multiples of chunksize.
2016-05-27 13:12:38 +08:00
void *extent_before_get(const extent_t *extent);
void *extent_last_get(const extent_t *extent);
Add extent_past_get().
2016-04-07 22:34:26 +08:00
void *extent_past_get(const extent_t *extent);
Add extent_active_[gs]et(). Always initialize extents' runs_dirty and chunks_cache linkage.
2016-03-28 18:17:10 +08:00
bool extent_active_get(const extent_t *extent);
Use rtree rather than [sz]ad trees for chunk split/coalesce operations.
2016-05-17 04:25:18 +08:00
bool extent_retained_get(const extent_t *extent);
Rename extent_node_t to extent_t.
2016-03-24 12:09:28 +08:00
bool extent_zeroed_get(const extent_t *extent);
bool extent_committed_get(const extent_t *extent);
Replace extent_achunk_[gs]et() with extent_slab_[gs]et().
2016-04-07 22:24:14 +08:00
bool extent_slab_get(const extent_t *extent);
Move slabs out of chunks.
2016-05-30 09:34:50 +08:00
arena_slab_data_t *extent_slab_data_get(extent_t *extent);
const arena_slab_data_t *extent_slab_data_get_const(const extent_t *extent);
Rename extent_node_t to extent_t.
2016-03-24 12:09:28 +08:00
prof_tctx_t *extent_prof_tctx_get(const extent_t *extent);
void extent_arena_set(extent_t *extent, arena_t *arena);
void extent_addr_set(extent_t *extent, void *addr);
Implement cache-oblivious support for huge size classes.
2016-05-28 09:57:15 +08:00
void extent_addr_randomize(tsdn_t *tsdn, extent_t *extent, size_t alignment);
Rename extent_node_t to extent_t.
2016-03-24 12:09:28 +08:00
void extent_size_set(extent_t *extent, size_t size);
Use huge size class infrastructure for large size classes.
2016-05-28 15:17:28 +08:00
void extent_usize_set(extent_t *extent, size_t usize);
Add extent_active_[gs]et(). Always initialize extents' runs_dirty and chunks_cache linkage.
2016-03-28 18:17:10 +08:00
void extent_active_set(extent_t *extent, bool active);
Rename extent_node_t to extent_t.
2016-03-24 12:09:28 +08:00
void extent_zeroed_set(extent_t *extent, bool zeroed);
void extent_committed_set(extent_t *extent, bool committed);
Replace extent_achunk_[gs]et() with extent_slab_[gs]et().
2016-04-07 22:24:14 +08:00
void extent_slab_set(extent_t *extent, bool slab);
Rename extent_node_t to extent_t.
2016-03-24 12:09:28 +08:00
void extent_prof_tctx_set(extent_t *extent, prof_tctx_t *tctx);
void extent_init(extent_t *extent, arena_t *arena, void *addr,
Move slabs out of chunks.
2016-05-30 09:34:50 +08:00
size_t size, size_t usize, bool active, bool zeroed, bool committed,
bool slab);
void extent_ring_insert(extent_t *sentinel, extent_t *extent);
void extent_ring_remove(extent_t *extent);
Integrate whole chunks into unused dirty page purging machinery. Extend per arena unused dirty page purging to manage unused dirty chunks in aaddtion to unused dirty runs. Rather than immediately unmapping deallocated chunks (or purging them in the --disable-munmap case), store them in a separate set of trees, chunks_[sz]ad_dirty. Preferrentially allocate dirty chunks. When excessive unused dirty pages accumulate, purge runs and chunks in ingegrated LRU order (and unmap chunks in the --enable-munmap case). Refactor extent_node_t to provide accessor functions.
2015-02-16 10:04:46 +08:00
#endif
#if (defined(JEMALLOC_ENABLE_INLINE) || defined(JEMALLOC_EXTENT_C_))
JEMALLOC_INLINE arena_t *
Rename extent_node_t to extent_t.
2016-03-24 12:09:28 +08:00
extent_arena_get(const extent_t *extent)
Integrate whole chunks into unused dirty page purging machinery. Extend per arena unused dirty page purging to manage unused dirty chunks in aaddtion to unused dirty runs. Rather than immediately unmapping deallocated chunks (or purging them in the --disable-munmap case), store them in a separate set of trees, chunks_[sz]ad_dirty. Preferrentially allocate dirty chunks. When excessive unused dirty pages accumulate, purge runs and chunks in ingegrated LRU order (and unmap chunks in the --enable-munmap case). Refactor extent_node_t to provide accessor functions.
2015-02-16 10:04:46 +08:00
{
Rename extent_node_t to extent_t.
2016-03-24 12:09:28 +08:00
return (extent->e_arena);
Integrate whole chunks into unused dirty page purging machinery. Extend per arena unused dirty page purging to manage unused dirty chunks in aaddtion to unused dirty runs. Rather than immediately unmapping deallocated chunks (or purging them in the --disable-munmap case), store them in a separate set of trees, chunks_[sz]ad_dirty. Preferrentially allocate dirty chunks. When excessive unused dirty pages accumulate, purge runs and chunks in ingegrated LRU order (and unmap chunks in the --enable-munmap case). Refactor extent_node_t to provide accessor functions.
2015-02-16 10:04:46 +08:00
}
Implement cache-oblivious support for huge size classes.
2016-05-28 09:57:15 +08:00
JEMALLOC_INLINE void *
extent_base_get(const extent_t *extent)
{
assert(extent->e_addr == PAGE_ADDR2BASE(extent->e_addr) ||
!extent->e_slab);
return (PAGE_ADDR2BASE(extent->e_addr));
}
Integrate whole chunks into unused dirty page purging machinery. Extend per arena unused dirty page purging to manage unused dirty chunks in aaddtion to unused dirty runs. Rather than immediately unmapping deallocated chunks (or purging them in the --disable-munmap case), store them in a separate set of trees, chunks_[sz]ad_dirty. Preferrentially allocate dirty chunks. When excessive unused dirty pages accumulate, purge runs and chunks in ingegrated LRU order (and unmap chunks in the --enable-munmap case). Refactor extent_node_t to provide accessor functions.
2015-02-16 10:04:46 +08:00
JEMALLOC_INLINE void *
Rename extent_node_t to extent_t.
2016-03-24 12:09:28 +08:00
extent_addr_get(const extent_t *extent)
Integrate whole chunks into unused dirty page purging machinery. Extend per arena unused dirty page purging to manage unused dirty chunks in aaddtion to unused dirty runs. Rather than immediately unmapping deallocated chunks (or purging them in the --disable-munmap case), store them in a separate set of trees, chunks_[sz]ad_dirty. Preferrentially allocate dirty chunks. When excessive unused dirty pages accumulate, purge runs and chunks in ingegrated LRU order (and unmap chunks in the --enable-munmap case). Refactor extent_node_t to provide accessor functions.
2015-02-16 10:04:46 +08:00
{
Implement cache-oblivious support for huge size classes.
2016-05-28 09:57:15 +08:00
assert(extent->e_addr == PAGE_ADDR2BASE(extent->e_addr) ||
!extent->e_slab);
Rename extent_node_t to extent_t.
2016-03-24 12:09:28 +08:00
return (extent->e_addr);
Integrate whole chunks into unused dirty page purging machinery. Extend per arena unused dirty page purging to manage unused dirty chunks in aaddtion to unused dirty runs. Rather than immediately unmapping deallocated chunks (or purging them in the --disable-munmap case), store them in a separate set of trees, chunks_[sz]ad_dirty. Preferrentially allocate dirty chunks. When excessive unused dirty pages accumulate, purge runs and chunks in ingegrated LRU order (and unmap chunks in the --enable-munmap case). Refactor extent_node_t to provide accessor functions.
2015-02-16 10:04:46 +08:00
}
JEMALLOC_INLINE size_t
Rename extent_node_t to extent_t.
2016-03-24 12:09:28 +08:00
extent_size_get(const extent_t *extent)
Integrate whole chunks into unused dirty page purging machinery. Extend per arena unused dirty page purging to manage unused dirty chunks in aaddtion to unused dirty runs. Rather than immediately unmapping deallocated chunks (or purging them in the --disable-munmap case), store them in a separate set of trees, chunks_[sz]ad_dirty. Preferrentially allocate dirty chunks. When excessive unused dirty pages accumulate, purge runs and chunks in ingegrated LRU order (and unmap chunks in the --enable-munmap case). Refactor extent_node_t to provide accessor functions.
2015-02-16 10:04:46 +08:00
{
Rename extent_node_t to extent_t.
2016-03-24 12:09:28 +08:00
return (extent->e_size);
Integrate whole chunks into unused dirty page purging machinery. Extend per arena unused dirty page purging to manage unused dirty chunks in aaddtion to unused dirty runs. Rather than immediately unmapping deallocated chunks (or purging them in the --disable-munmap case), store them in a separate set of trees, chunks_[sz]ad_dirty. Preferrentially allocate dirty chunks. When excessive unused dirty pages accumulate, purge runs and chunks in ingegrated LRU order (and unmap chunks in the --enable-munmap case). Refactor extent_node_t to provide accessor functions.
2015-02-16 10:04:46 +08:00
}
Implement cache-oblivious support for huge size classes.
2016-05-28 09:57:15 +08:00
JEMALLOC_INLINE size_t
extent_usize_get(const extent_t *extent)
{
assert(!extent->e_slab);
Use huge size class infrastructure for large size classes.
2016-05-28 15:17:28 +08:00
return (extent->e_usize);
Implement cache-oblivious support for huge size classes.
2016-05-28 09:57:15 +08:00
}
Allow chunks to not be naturally aligned. Precisely size extents for huge size classes that aren't multiples of chunksize.
2016-05-27 13:12:38 +08:00
JEMALLOC_INLINE void *
extent_before_get(const extent_t *extent)
{
return ((void *)(uintptr_t)extent->e_addr - PAGE);
}
JEMALLOC_INLINE void *
extent_last_get(const extent_t *extent)
{
Use huge size class infrastructure for large size classes.
2016-05-28 15:17:28 +08:00
return ((void *)(uintptr_t)extent->e_addr + extent_size_get(extent) -
PAGE);
Allow chunks to not be naturally aligned. Precisely size extents for huge size classes that aren't multiples of chunksize.
2016-05-27 13:12:38 +08:00
}
Add extent_past_get().
2016-04-07 22:34:26 +08:00
JEMALLOC_INLINE void *
extent_past_get(const extent_t *extent)
{
Use huge size class infrastructure for large size classes.
2016-05-28 15:17:28 +08:00
return ((void *)(uintptr_t)extent->e_addr + extent_size_get(extent));
Add extent_past_get().
2016-04-07 22:34:26 +08:00
}
Add extent_active_[gs]et(). Always initialize extents' runs_dirty and chunks_cache linkage.
2016-03-28 18:17:10 +08:00
JEMALLOC_INLINE bool
extent_active_get(const extent_t *extent)
{
return (extent->e_active);
}
Use rtree rather than [sz]ad trees for chunk split/coalesce operations.
2016-05-17 04:25:18 +08:00
JEMALLOC_INLINE bool
extent_retained_get(const extent_t *extent)
{
Move slabs out of chunks.
2016-05-30 09:34:50 +08:00
return (qr_next(extent, qr_link) == extent);
Use rtree rather than [sz]ad trees for chunk split/coalesce operations.
2016-05-17 04:25:18 +08:00
}
Generalize chunk management hooks. Add the "arena.<i>.chunk_hooks" mallctl, which replaces and expands on the "arena.<i>.chunk.{alloc,dalloc,purge}" mallctls. The chunk hooks allow control over chunk allocation/deallocation, decommit/commit, purging, and splitting/merging, such that the application can rely on jemalloc's internal chunk caching and retaining functionality, yet implement a variety of chunk management mechanisms and policies. Merge the chunks_[sz]ad_{mmap,dss} red-black trees into chunks_[sz]ad_retained. This slightly reduces how hard jemalloc tries to honor the dss precedence setting; prior to this change the precedence setting was also consulted when recycling chunks. Fix chunk purging. Don't purge chunks in arena_purge_stashed(); instead deallocate them in arena_unstash_purged(), so that the dirty memory linkage remains valid until after the last time it is used. This resolves #176 and #201.
2015-07-28 23:28:19 +08:00
JEMALLOC_INLINE bool
Rename extent_node_t to extent_t.
2016-03-24 12:09:28 +08:00
extent_zeroed_get(const extent_t *extent)
Generalize chunk management hooks. Add the "arena.<i>.chunk_hooks" mallctl, which replaces and expands on the "arena.<i>.chunk.{alloc,dalloc,purge}" mallctls. The chunk hooks allow control over chunk allocation/deallocation, decommit/commit, purging, and splitting/merging, such that the application can rely on jemalloc's internal chunk caching and retaining functionality, yet implement a variety of chunk management mechanisms and policies. Merge the chunks_[sz]ad_{mmap,dss} red-black trees into chunks_[sz]ad_retained. This slightly reduces how hard jemalloc tries to honor the dss precedence setting; prior to this change the precedence setting was also consulted when recycling chunks. Fix chunk purging. Don't purge chunks in arena_purge_stashed(); instead deallocate them in arena_unstash_purged(), so that the dirty memory linkage remains valid until after the last time it is used. This resolves #176 and #201.
2015-07-28 23:28:19 +08:00
{
Rename extent_node_t to extent_t.
2016-03-24 12:09:28 +08:00
return (extent->e_zeroed);
Generalize chunk management hooks. Add the "arena.<i>.chunk_hooks" mallctl, which replaces and expands on the "arena.<i>.chunk.{alloc,dalloc,purge}" mallctls. The chunk hooks allow control over chunk allocation/deallocation, decommit/commit, purging, and splitting/merging, such that the application can rely on jemalloc's internal chunk caching and retaining functionality, yet implement a variety of chunk management mechanisms and policies. Merge the chunks_[sz]ad_{mmap,dss} red-black trees into chunks_[sz]ad_retained. This slightly reduces how hard jemalloc tries to honor the dss precedence setting; prior to this change the precedence setting was also consulted when recycling chunks. Fix chunk purging. Don't purge chunks in arena_purge_stashed(); instead deallocate them in arena_unstash_purged(), so that the dirty memory linkage remains valid until after the last time it is used. This resolves #176 and #201.
2015-07-28 23:28:19 +08:00
}
Integrate whole chunks into unused dirty page purging machinery. Extend per arena unused dirty page purging to manage unused dirty chunks in aaddtion to unused dirty runs. Rather than immediately unmapping deallocated chunks (or purging them in the --disable-munmap case), store them in a separate set of trees, chunks_[sz]ad_dirty. Preferrentially allocate dirty chunks. When excessive unused dirty pages accumulate, purge runs and chunks in ingegrated LRU order (and unmap chunks in the --enable-munmap case). Refactor extent_node_t to provide accessor functions.
2015-02-16 10:04:46 +08:00
JEMALLOC_INLINE bool
Rename extent_node_t to extent_t.
2016-03-24 12:09:28 +08:00
extent_committed_get(const extent_t *extent)
Integrate whole chunks into unused dirty page purging machinery. Extend per arena unused dirty page purging to manage unused dirty chunks in aaddtion to unused dirty runs. Rather than immediately unmapping deallocated chunks (or purging them in the --disable-munmap case), store them in a separate set of trees, chunks_[sz]ad_dirty. Preferrentially allocate dirty chunks. When excessive unused dirty pages accumulate, purge runs and chunks in ingegrated LRU order (and unmap chunks in the --enable-munmap case). Refactor extent_node_t to provide accessor functions.
2015-02-16 10:04:46 +08:00
{
Rename extent_node_t to extent_t.
2016-03-24 12:09:28 +08:00
return (extent->e_committed);
Integrate whole chunks into unused dirty page purging machinery. Extend per arena unused dirty page purging to manage unused dirty chunks in aaddtion to unused dirty runs. Rather than immediately unmapping deallocated chunks (or purging them in the --disable-munmap case), store them in a separate set of trees, chunks_[sz]ad_dirty. Preferrentially allocate dirty chunks. When excessive unused dirty pages accumulate, purge runs and chunks in ingegrated LRU order (and unmap chunks in the --enable-munmap case). Refactor extent_node_t to provide accessor functions.
2015-02-16 10:04:46 +08:00
}
JEMALLOC_INLINE bool
Replace extent_achunk_[gs]et() with extent_slab_[gs]et().
2016-04-07 22:24:14 +08:00
extent_slab_get(const extent_t *extent)
Integrate whole chunks into unused dirty page purging machinery. Extend per arena unused dirty page purging to manage unused dirty chunks in aaddtion to unused dirty runs. Rather than immediately unmapping deallocated chunks (or purging them in the --disable-munmap case), store them in a separate set of trees, chunks_[sz]ad_dirty. Preferrentially allocate dirty chunks. When excessive unused dirty pages accumulate, purge runs and chunks in ingegrated LRU order (and unmap chunks in the --enable-munmap case). Refactor extent_node_t to provide accessor functions.
2015-02-16 10:04:46 +08:00
{
Replace extent_achunk_[gs]et() with extent_slab_[gs]et().
2016-04-07 22:24:14 +08:00
return (extent->e_slab);
Integrate whole chunks into unused dirty page purging machinery. Extend per arena unused dirty page purging to manage unused dirty chunks in aaddtion to unused dirty runs. Rather than immediately unmapping deallocated chunks (or purging them in the --disable-munmap case), store them in a separate set of trees, chunks_[sz]ad_dirty. Preferrentially allocate dirty chunks. When excessive unused dirty pages accumulate, purge runs and chunks in ingegrated LRU order (and unmap chunks in the --enable-munmap case). Refactor extent_node_t to provide accessor functions.
2015-02-16 10:04:46 +08:00
}
Move slabs out of chunks.
2016-05-30 09:34:50 +08:00
JEMALLOC_INLINE arena_slab_data_t *
extent_slab_data_get(extent_t *extent)
{
assert(extent->e_slab);
return (&extent->e_slab_data);
}
JEMALLOC_INLINE const arena_slab_data_t *
extent_slab_data_get_const(const extent_t *extent)
{
assert(extent->e_slab);
return (&extent->e_slab_data);
}
Integrate whole chunks into unused dirty page purging machinery. Extend per arena unused dirty page purging to manage unused dirty chunks in aaddtion to unused dirty runs. Rather than immediately unmapping deallocated chunks (or purging them in the --disable-munmap case), store them in a separate set of trees, chunks_[sz]ad_dirty. Preferrentially allocate dirty chunks. When excessive unused dirty pages accumulate, purge runs and chunks in ingegrated LRU order (and unmap chunks in the --enable-munmap case). Refactor extent_node_t to provide accessor functions.
2015-02-16 10:04:46 +08:00
JEMALLOC_INLINE prof_tctx_t *
Rename extent_node_t to extent_t.
2016-03-24 12:09:28 +08:00
extent_prof_tctx_get(const extent_t *extent)
Integrate whole chunks into unused dirty page purging machinery. Extend per arena unused dirty page purging to manage unused dirty chunks in aaddtion to unused dirty runs. Rather than immediately unmapping deallocated chunks (or purging them in the --disable-munmap case), store them in a separate set of trees, chunks_[sz]ad_dirty. Preferrentially allocate dirty chunks. When excessive unused dirty pages accumulate, purge runs and chunks in ingegrated LRU order (and unmap chunks in the --enable-munmap case). Refactor extent_node_t to provide accessor functions.
2015-02-16 10:04:46 +08:00
{
Make extent_prof_tctx_[gs]et() atomic.
2016-05-25 12:13:36 +08:00
return ((prof_tctx_t *)atomic_read_p(
&((extent_t *)extent)->e_prof_tctx_pun));
Integrate whole chunks into unused dirty page purging machinery. Extend per arena unused dirty page purging to manage unused dirty chunks in aaddtion to unused dirty runs. Rather than immediately unmapping deallocated chunks (or purging them in the --disable-munmap case), store them in a separate set of trees, chunks_[sz]ad_dirty. Preferrentially allocate dirty chunks. When excessive unused dirty pages accumulate, purge runs and chunks in ingegrated LRU order (and unmap chunks in the --enable-munmap case). Refactor extent_node_t to provide accessor functions.
2015-02-16 10:04:46 +08:00
}
JEMALLOC_INLINE void
Rename extent_node_t to extent_t.
2016-03-24 12:09:28 +08:00
extent_arena_set(extent_t *extent, arena_t *arena)
Integrate whole chunks into unused dirty page purging machinery. Extend per arena unused dirty page purging to manage unused dirty chunks in aaddtion to unused dirty runs. Rather than immediately unmapping deallocated chunks (or purging them in the --disable-munmap case), store them in a separate set of trees, chunks_[sz]ad_dirty. Preferrentially allocate dirty chunks. When excessive unused dirty pages accumulate, purge runs and chunks in ingegrated LRU order (and unmap chunks in the --enable-munmap case). Refactor extent_node_t to provide accessor functions.
2015-02-16 10:04:46 +08:00
{
Rename extent_node_t to extent_t.
2016-03-24 12:09:28 +08:00
extent->e_arena = arena;
Integrate whole chunks into unused dirty page purging machinery. Extend per arena unused dirty page purging to manage unused dirty chunks in aaddtion to unused dirty runs. Rather than immediately unmapping deallocated chunks (or purging them in the --disable-munmap case), store them in a separate set of trees, chunks_[sz]ad_dirty. Preferrentially allocate dirty chunks. When excessive unused dirty pages accumulate, purge runs and chunks in ingegrated LRU order (and unmap chunks in the --enable-munmap case). Refactor extent_node_t to provide accessor functions.
2015-02-16 10:04:46 +08:00
}
JEMALLOC_INLINE void
Rename extent_node_t to extent_t.
2016-03-24 12:09:28 +08:00
extent_addr_set(extent_t *extent, void *addr)
Integrate whole chunks into unused dirty page purging machinery. Extend per arena unused dirty page purging to manage unused dirty chunks in aaddtion to unused dirty runs. Rather than immediately unmapping deallocated chunks (or purging them in the --disable-munmap case), store them in a separate set of trees, chunks_[sz]ad_dirty. Preferrentially allocate dirty chunks. When excessive unused dirty pages accumulate, purge runs and chunks in ingegrated LRU order (and unmap chunks in the --enable-munmap case). Refactor extent_node_t to provide accessor functions.
2015-02-16 10:04:46 +08:00
{
Rename extent_node_t to extent_t.
2016-03-24 12:09:28 +08:00
extent->e_addr = addr;
Integrate whole chunks into unused dirty page purging machinery. Extend per arena unused dirty page purging to manage unused dirty chunks in aaddtion to unused dirty runs. Rather than immediately unmapping deallocated chunks (or purging them in the --disable-munmap case), store them in a separate set of trees, chunks_[sz]ad_dirty. Preferrentially allocate dirty chunks. When excessive unused dirty pages accumulate, purge runs and chunks in ingegrated LRU order (and unmap chunks in the --enable-munmap case). Refactor extent_node_t to provide accessor functions.
2015-02-16 10:04:46 +08:00
}
Implement cache-oblivious support for huge size classes.
2016-05-28 09:57:15 +08:00
JEMALLOC_INLINE void
extent_addr_randomize(tsdn_t *tsdn, extent_t *extent, size_t alignment)
{
assert(extent_base_get(extent) == extent_addr_get(extent));
if (alignment < PAGE) {
unsigned lg_range = LG_PAGE -
lg_floor(CACHELINE_CEILING(alignment));
uint64_t r =
prng_lg_range(&extent_arena_get(extent)->offset_state,
lg_range, true);
Use huge size class infrastructure for large size classes.
2016-05-28 15:17:28 +08:00
uintptr_t random_offset = ((uintptr_t)r) << (LG_PAGE -
lg_range);
Implement cache-oblivious support for huge size classes.
2016-05-28 09:57:15 +08:00
extent->e_addr = (void *)((uintptr_t)extent->e_addr +
random_offset);
Use huge size class infrastructure for large size classes.
2016-05-28 15:17:28 +08:00
assert(ALIGNMENT_ADDR2BASE(extent->e_addr, alignment) ==
extent->e_addr);
Implement cache-oblivious support for huge size classes.
2016-05-28 09:57:15 +08:00
}
}
Integrate whole chunks into unused dirty page purging machinery. Extend per arena unused dirty page purging to manage unused dirty chunks in aaddtion to unused dirty runs. Rather than immediately unmapping deallocated chunks (or purging them in the --disable-munmap case), store them in a separate set of trees, chunks_[sz]ad_dirty. Preferrentially allocate dirty chunks. When excessive unused dirty pages accumulate, purge runs and chunks in ingegrated LRU order (and unmap chunks in the --enable-munmap case). Refactor extent_node_t to provide accessor functions.
2015-02-16 10:04:46 +08:00
JEMALLOC_INLINE void
Rename extent_node_t to extent_t.
2016-03-24 12:09:28 +08:00
extent_size_set(extent_t *extent, size_t size)
Integrate whole chunks into unused dirty page purging machinery. Extend per arena unused dirty page purging to manage unused dirty chunks in aaddtion to unused dirty runs. Rather than immediately unmapping deallocated chunks (or purging them in the --disable-munmap case), store them in a separate set of trees, chunks_[sz]ad_dirty. Preferrentially allocate dirty chunks. When excessive unused dirty pages accumulate, purge runs and chunks in ingegrated LRU order (and unmap chunks in the --enable-munmap case). Refactor extent_node_t to provide accessor functions.
2015-02-16 10:04:46 +08:00
{
Rename extent_node_t to extent_t.
2016-03-24 12:09:28 +08:00
extent->e_size = size;
Integrate whole chunks into unused dirty page purging machinery. Extend per arena unused dirty page purging to manage unused dirty chunks in aaddtion to unused dirty runs. Rather than immediately unmapping deallocated chunks (or purging them in the --disable-munmap case), store them in a separate set of trees, chunks_[sz]ad_dirty. Preferrentially allocate dirty chunks. When excessive unused dirty pages accumulate, purge runs and chunks in ingegrated LRU order (and unmap chunks in the --enable-munmap case). Refactor extent_node_t to provide accessor functions.
2015-02-16 10:04:46 +08:00
}
Use huge size class infrastructure for large size classes.
2016-05-28 15:17:28 +08:00
JEMALLOC_INLINE void
extent_usize_set(extent_t *extent, size_t usize)
{
extent->e_usize = usize;
}
Add extent_active_[gs]et(). Always initialize extents' runs_dirty and chunks_cache linkage.
2016-03-28 18:17:10 +08:00
JEMALLOC_INLINE void
extent_active_set(extent_t *extent, bool active)
{
extent->e_active = active;
}
Generalize chunk management hooks. Add the "arena.<i>.chunk_hooks" mallctl, which replaces and expands on the "arena.<i>.chunk.{alloc,dalloc,purge}" mallctls. The chunk hooks allow control over chunk allocation/deallocation, decommit/commit, purging, and splitting/merging, such that the application can rely on jemalloc's internal chunk caching and retaining functionality, yet implement a variety of chunk management mechanisms and policies. Merge the chunks_[sz]ad_{mmap,dss} red-black trees into chunks_[sz]ad_retained. This slightly reduces how hard jemalloc tries to honor the dss precedence setting; prior to this change the precedence setting was also consulted when recycling chunks. Fix chunk purging. Don't purge chunks in arena_purge_stashed(); instead deallocate them in arena_unstash_purged(), so that the dirty memory linkage remains valid until after the last time it is used. This resolves #176 and #201.
2015-07-28 23:28:19 +08:00
JEMALLOC_INLINE void
Rename extent_node_t to extent_t.
2016-03-24 12:09:28 +08:00
extent_zeroed_set(extent_t *extent, bool zeroed)
Generalize chunk management hooks. Add the "arena.<i>.chunk_hooks" mallctl, which replaces and expands on the "arena.<i>.chunk.{alloc,dalloc,purge}" mallctls. The chunk hooks allow control over chunk allocation/deallocation, decommit/commit, purging, and splitting/merging, such that the application can rely on jemalloc's internal chunk caching and retaining functionality, yet implement a variety of chunk management mechanisms and policies. Merge the chunks_[sz]ad_{mmap,dss} red-black trees into chunks_[sz]ad_retained. This slightly reduces how hard jemalloc tries to honor the dss precedence setting; prior to this change the precedence setting was also consulted when recycling chunks. Fix chunk purging. Don't purge chunks in arena_purge_stashed(); instead deallocate them in arena_unstash_purged(), so that the dirty memory linkage remains valid until after the last time it is used. This resolves #176 and #201.
2015-07-28 23:28:19 +08:00
{
Rename extent_node_t to extent_t.
2016-03-24 12:09:28 +08:00
extent->e_zeroed = zeroed;
Generalize chunk management hooks. Add the "arena.<i>.chunk_hooks" mallctl, which replaces and expands on the "arena.<i>.chunk.{alloc,dalloc,purge}" mallctls. The chunk hooks allow control over chunk allocation/deallocation, decommit/commit, purging, and splitting/merging, such that the application can rely on jemalloc's internal chunk caching and retaining functionality, yet implement a variety of chunk management mechanisms and policies. Merge the chunks_[sz]ad_{mmap,dss} red-black trees into chunks_[sz]ad_retained. This slightly reduces how hard jemalloc tries to honor the dss precedence setting; prior to this change the precedence setting was also consulted when recycling chunks. Fix chunk purging. Don't purge chunks in arena_purge_stashed(); instead deallocate them in arena_unstash_purged(), so that the dirty memory linkage remains valid until after the last time it is used. This resolves #176 and #201.
2015-07-28 23:28:19 +08:00
}
Integrate whole chunks into unused dirty page purging machinery. Extend per arena unused dirty page purging to manage unused dirty chunks in aaddtion to unused dirty runs. Rather than immediately unmapping deallocated chunks (or purging them in the --disable-munmap case), store them in a separate set of trees, chunks_[sz]ad_dirty. Preferrentially allocate dirty chunks. When excessive unused dirty pages accumulate, purge runs and chunks in ingegrated LRU order (and unmap chunks in the --enable-munmap case). Refactor extent_node_t to provide accessor functions.
2015-02-16 10:04:46 +08:00
JEMALLOC_INLINE void
Rename extent_node_t to extent_t.
2016-03-24 12:09:28 +08:00
extent_committed_set(extent_t *extent, bool committed)
Integrate whole chunks into unused dirty page purging machinery. Extend per arena unused dirty page purging to manage unused dirty chunks in aaddtion to unused dirty runs. Rather than immediately unmapping deallocated chunks (or purging them in the --disable-munmap case), store them in a separate set of trees, chunks_[sz]ad_dirty. Preferrentially allocate dirty chunks. When excessive unused dirty pages accumulate, purge runs and chunks in ingegrated LRU order (and unmap chunks in the --enable-munmap case). Refactor extent_node_t to provide accessor functions.
2015-02-16 10:04:46 +08:00
{
Rename extent_node_t to extent_t.
2016-03-24 12:09:28 +08:00
extent->e_committed = committed;
Integrate whole chunks into unused dirty page purging machinery. Extend per arena unused dirty page purging to manage unused dirty chunks in aaddtion to unused dirty runs. Rather than immediately unmapping deallocated chunks (or purging them in the --disable-munmap case), store them in a separate set of trees, chunks_[sz]ad_dirty. Preferrentially allocate dirty chunks. When excessive unused dirty pages accumulate, purge runs and chunks in ingegrated LRU order (and unmap chunks in the --enable-munmap case). Refactor extent_node_t to provide accessor functions.
2015-02-16 10:04:46 +08:00
}
JEMALLOC_INLINE void
Replace extent_achunk_[gs]et() with extent_slab_[gs]et().
2016-04-07 22:24:14 +08:00
extent_slab_set(extent_t *extent, bool slab)
Integrate whole chunks into unused dirty page purging machinery. Extend per arena unused dirty page purging to manage unused dirty chunks in aaddtion to unused dirty runs. Rather than immediately unmapping deallocated chunks (or purging them in the --disable-munmap case), store them in a separate set of trees, chunks_[sz]ad_dirty. Preferrentially allocate dirty chunks. When excessive unused dirty pages accumulate, purge runs and chunks in ingegrated LRU order (and unmap chunks in the --enable-munmap case). Refactor extent_node_t to provide accessor functions.
2015-02-16 10:04:46 +08:00
{
Replace extent_achunk_[gs]et() with extent_slab_[gs]et().
2016-04-07 22:24:14 +08:00
extent->e_slab = slab;
Integrate whole chunks into unused dirty page purging machinery. Extend per arena unused dirty page purging to manage unused dirty chunks in aaddtion to unused dirty runs. Rather than immediately unmapping deallocated chunks (or purging them in the --disable-munmap case), store them in a separate set of trees, chunks_[sz]ad_dirty. Preferrentially allocate dirty chunks. When excessive unused dirty pages accumulate, purge runs and chunks in ingegrated LRU order (and unmap chunks in the --enable-munmap case). Refactor extent_node_t to provide accessor functions.
2015-02-16 10:04:46 +08:00
}
JEMALLOC_INLINE void
Rename extent_node_t to extent_t.
2016-03-24 12:09:28 +08:00
extent_prof_tctx_set(extent_t *extent, prof_tctx_t *tctx)
Integrate whole chunks into unused dirty page purging machinery. Extend per arena unused dirty page purging to manage unused dirty chunks in aaddtion to unused dirty runs. Rather than immediately unmapping deallocated chunks (or purging them in the --disable-munmap case), store them in a separate set of trees, chunks_[sz]ad_dirty. Preferrentially allocate dirty chunks. When excessive unused dirty pages accumulate, purge runs and chunks in ingegrated LRU order (and unmap chunks in the --enable-munmap case). Refactor extent_node_t to provide accessor functions.
2015-02-16 10:04:46 +08:00
{
Make extent_prof_tctx_[gs]et() atomic.
2016-05-25 12:13:36 +08:00
atomic_write_p(&extent->e_prof_tctx_pun, tctx);
Integrate whole chunks into unused dirty page purging machinery. Extend per arena unused dirty page purging to manage unused dirty chunks in aaddtion to unused dirty runs. Rather than immediately unmapping deallocated chunks (or purging them in the --disable-munmap case), store them in a separate set of trees, chunks_[sz]ad_dirty. Preferrentially allocate dirty chunks. When excessive unused dirty pages accumulate, purge runs and chunks in ingegrated LRU order (and unmap chunks in the --enable-munmap case). Refactor extent_node_t to provide accessor functions.
2015-02-16 10:04:46 +08:00
}
Simplify extent_node_t and add extent_node_init().
2015-02-18 07:13:52 +08:00
JEMALLOC_INLINE void
Rename extent_node_t to extent_t.
2016-03-24 12:09:28 +08:00
extent_init(extent_t *extent, arena_t *arena, void *addr, size_t size,
Move slabs out of chunks.
2016-05-30 09:34:50 +08:00
size_t usize, bool active, bool zeroed, bool committed, bool slab)
Simplify extent_node_t and add extent_node_init().
2015-02-18 07:13:52 +08:00
{
Implement cache-oblivious support for huge size classes.
2016-05-28 09:57:15 +08:00
assert(addr == PAGE_ADDR2BASE(addr) || !slab);
Rename extent_node_t to extent_t.
2016-03-24 12:09:28 +08:00
extent_arena_set(extent, arena);
extent_addr_set(extent, addr);
extent_size_set(extent, size);
Use huge size class infrastructure for large size classes.
2016-05-28 15:17:28 +08:00
extent_usize_set(extent, usize);
Add extent_active_[gs]et(). Always initialize extents' runs_dirty and chunks_cache linkage.
2016-03-28 18:17:10 +08:00
extent_active_set(extent, active);
Rename extent_node_t to extent_t.
2016-03-24 12:09:28 +08:00
extent_zeroed_set(extent, zeroed);
extent_committed_set(extent, committed);
Replace extent_achunk_[gs]et() with extent_slab_[gs]et().
2016-04-07 22:24:14 +08:00
extent_slab_set(extent, slab);
Simplify extent_node_t and add extent_node_init().
2015-02-18 07:13:52 +08:00
if (config_prof)
Rename extent_node_t to extent_t.
2016-03-24 12:09:28 +08:00
extent_prof_tctx_set(extent, NULL);
Move slabs out of chunks.
2016-05-30 09:34:50 +08:00
qr_new(extent, qr_link);
arena_chunk_dirty_node_init() --> extent_node_dirty_linkage_init()
2015-02-18 14:23:10 +08:00
}
Rename "dirty chunks" to "cached chunks". Rename "dirty chunks" to "cached chunks", in order to avoid overloading the term "dirty". Fix the regression caused by 339c2b23b2d61993ac768afcc72af135662c6771 (Fix chunk_unmap() to propagate dirty state.), and actually address what that change attempted, which is to only purge chunks once, and propagate whether zeroed pages resulted into chunk_record().
2015-02-18 17:15:50 +08:00
JEMALLOC_INLINE void
Move slabs out of chunks.
2016-05-30 09:34:50 +08:00
extent_ring_insert(extent_t *sentinel, extent_t *extent)
Rename "dirty chunks" to "cached chunks". Rename "dirty chunks" to "cached chunks", in order to avoid overloading the term "dirty". Fix the regression caused by 339c2b23b2d61993ac768afcc72af135662c6771 (Fix chunk_unmap() to propagate dirty state.), and actually address what that change attempted, which is to only purge chunks once, and propagate whether zeroed pages resulted into chunk_record().
2015-02-18 17:15:50 +08:00
{
Move slabs out of chunks.
2016-05-30 09:34:50 +08:00
qr_meld(sentinel, extent, qr_link);
Rename "dirty chunks" to "cached chunks". Rename "dirty chunks" to "cached chunks", in order to avoid overloading the term "dirty". Fix the regression caused by 339c2b23b2d61993ac768afcc72af135662c6771 (Fix chunk_unmap() to propagate dirty state.), and actually address what that change attempted, which is to only purge chunks once, and propagate whether zeroed pages resulted into chunk_record().
2015-02-18 17:15:50 +08:00
}
JEMALLOC_INLINE void
Move slabs out of chunks.
2016-05-30 09:34:50 +08:00
extent_ring_remove(extent_t *extent)
Rename "dirty chunks" to "cached chunks". Rename "dirty chunks" to "cached chunks", in order to avoid overloading the term "dirty". Fix the regression caused by 339c2b23b2d61993ac768afcc72af135662c6771 (Fix chunk_unmap() to propagate dirty state.), and actually address what that change attempted, which is to only purge chunks once, and propagate whether zeroed pages resulted into chunk_record().
2015-02-18 17:15:50 +08:00
{
Move slabs out of chunks.
2016-05-30 09:34:50 +08:00
qr_remove(extent, qr_link);
Rename "dirty chunks" to "cached chunks". Rename "dirty chunks" to "cached chunks", in order to avoid overloading the term "dirty". Fix the regression caused by 339c2b23b2d61993ac768afcc72af135662c6771 (Fix chunk_unmap() to propagate dirty state.), and actually address what that change attempted, which is to only purge chunks once, and propagate whether zeroed pages resulted into chunk_record().
2015-02-18 17:15:50 +08:00
}
Integrate whole chunks into unused dirty page purging machinery. Extend per arena unused dirty page purging to manage unused dirty chunks in aaddtion to unused dirty runs. Rather than immediately unmapping deallocated chunks (or purging them in the --disable-munmap case), store them in a separate set of trees, chunks_[sz]ad_dirty. Preferrentially allocate dirty chunks. When excessive unused dirty pages accumulate, purge runs and chunks in ingegrated LRU order (and unmap chunks in the --enable-munmap case). Refactor extent_node_t to provide accessor functions.
2015-02-16 10:04:46 +08:00
#endif
Refactor jemalloc.c into multiple source files. Fix a stats bug in large object curruns accounting. Replace tcache_bin_fill() with arena_tcache_fill(), and fix a bug in an OOM error path. Fix API name mangling to coexist with __attribute__((malloc)).
2010-01-17 01:53:50 +08:00
#endif /* JEMALLOC_H_INLINES */
/******************************************************************************/
Reference in New Issue Copy Permalink