lihao/jemalloc
1
0
Fork 0
mirror of https://github.com/jemalloc/jemalloc.git synced 2026-07-24 13:53:11 +00:00
Code Issues Packages Projects Releases Wiki Activity

Introducing a new usize calculation policy

Browse Source 
Converting size to usize is what jemalloc has been done by ceiling
size to the closest size class. However, this causes lots of memory
wastes with HPA enabled.  This commit changes how usize is calculated so
that the gap between two contiguous usize is no larger than a page.
Specifically, this commit includes the following changes:

1. Adding a build-time config option (--enable-limit-usize-gap) and a
runtime one (limit_usize_gap) to guard the changes.
When build-time
config is enabled, some minor CPU overhead is expected because usize
will be stored and accessed apart from index.  When runtime option is
also enabled (it can only be enabled with the build-time config
enabled). a new usize calculation approach wil be employed.  This new
calculation will ceil size to the closest multiple of PAGE for all sizes
larger than USIZE_GROW_SLOW_THRESHOLD instead of using the size classes.
Note when the build-time config is enabled, the runtime option is
default on.

2. Prepare tcache for size to grow by PAGE over GROUP*PAGE.
To prepare for the upcoming changes where size class grows by PAGE when
larger than NGROUP * PAGE, disable the tcache when it is larger than 2 *
NGROUP * PAGE. The threshold for tcache is set higher to prevent perf
regression as much as possible while usizes between NGROUP * PAGE and 2 *
NGROUP * PAGE happen to grow by PAGE.

3. Prepare pac and hpa psset for size to grow by PAGE over GROUP*PAGE
For PAC, to avoid having too many bins, arena bins still have the same
layout.  This means some extra search is needed for a page-level request that
is not aligned with the orginal size class: it should also search the heap
before the current index since the previous heap might also be able to
have some allocations satisfying it.  The same changes apply to HPA's
psset.
This search relies on the enumeration of the heap because not all allocs in
the previous heap are guaranteed to satisfy the request.  To balance the
memory and CPU overhead, we currently enumerate at most a fixed number
of nodes before concluding none can satisfy the request during an
enumeration.

4. Add bytes counter to arena large stats.
To prepare for the upcoming usize changes, stats collected by
multiplying alive allocations and the bin size is no longer accurate.
Thus, add separate counters to record the bytes malloced and dalloced.

5. Change structs use when freeing to avoid using index2size for large sizes.
  - Change the definition of emap_alloc_ctx_t
  - Change the read of both from edata_t.
  - Change the assignment and usage of emap_alloc_ctx_t.
  - Change other callsites of index2size.
Note for the changes in the data structure, i.e., emap_alloc_ctx_t,
will be used when the build-time config (--enable-limit-usize-gap) is
enabled but they will store the same value as index2size(szind) if the
runtime option (opt_limit_usize_gap) is not enabled.

6. Adapt hpa to the usize changes.
Change the settings in sec to limit is usage for sizes larger than
USIZE_GROW_SLOW_THRESHOLD and modify corresponding tests.

7. Modify usize calculation and corresponding tests.
Change the sz_s2u_compute. Note sz_index2size is not always safe now
while sz_size2index still works as expected.
This commit is contained in:
guangli-dai
2024-03-26 14:35:29 -07:00
committed by Qi Wang
parent ac279d7e71
commit c067a55c79
33 changed files with 713 additions and 74 deletions
Download Patch File Download Diff File Expand all files Collapse all files

58
src/jemalloc.c
View File

@@ -123,6 +123,13 @@ zero_realloc_action_t opt_zero_realloc_action =
atomic_zu_t zero_realloc_count = ATOMIC_INIT(0);
bool opt_limit_usize_gap =
#ifdef LIMIT_USIZE_GAP
true;
#else
false;
#endif
const char *const zero_realloc_mode_names[] = {
"alloc",
"free",
@@ -1578,8 +1585,8 @@ malloc_conf_init_helper(sc_data_t *sc_data, unsigned bin_shard_sizes[SC_NBINS],
"hpa_sec_nshards", 0, 0, CONF_CHECK_MIN,
CONF_DONT_CHECK_MAX, true);
CONF_HANDLE_SIZE_T(opt_hpa_sec_opts.max_alloc,
"hpa_sec_max_alloc", PAGE, 0, CONF_CHECK_MIN,
CONF_DONT_CHECK_MAX, true);
"hpa_sec_max_alloc", PAGE, USIZE_GROW_SLOW_THRESHOLD,
CONF_CHECK_MIN, CONF_CHECK_MAX, true);
CONF_HANDLE_SIZE_T(opt_hpa_sec_opts.max_bytes,
"hpa_sec_max_bytes", PAGE, 0, CONF_CHECK_MIN,
CONF_DONT_CHECK_MAX, true);
@@ -1763,6 +1770,11 @@ malloc_conf_init_helper(sc_data_t *sc_data, unsigned bin_shard_sizes[SC_NBINS],
"san_guard_large", 0, SIZE_T_MAX,
CONF_DONT_CHECK_MIN, CONF_DONT_CHECK_MAX, false)
if (config_limit_usize_gap) {
CONF_HANDLE_BOOL(opt_limit_usize_gap,
"limit_usize_gap");
}
CONF_ERROR("Invalid conf pair", k, klen, v, vlen);
#undef CONF_ERROR
#undef CONF_CONTINUE
@@ -2182,6 +2194,17 @@ static bool
malloc_init_hard(void) {
tsd_t *tsd;
if (config_limit_usize_gap) {
assert(TCACHE_MAXCLASS_LIMIT <= USIZE_GROW_SLOW_THRESHOLD);
assert(SC_LOOKUP_MAXCLASS <= USIZE_GROW_SLOW_THRESHOLD);
/*
* This asserts an extreme case where TINY_MAXCLASS is larger
* than LARGE_MINCLASS. It could only happen if some constants
* are configured miserably wrong.
*/
assert(SC_LG_TINY_MAXCLASS <=
(size_t)1ULL << (LG_PAGE + SC_LG_NGROUP));
}
#if defined(_WIN32) && _WIN32_WINNT < 0x0600
_init_init_lock();
#endif
@@ -2376,7 +2399,8 @@ aligned_usize_get(size_t size, size_t alignment, size_t *usize, szind_t *ind,
if (unlikely(*ind >= SC_NSIZES)) {
return true;
}
*usize = sz_index2size(*ind);
*usize = sz_limit_usize_gap_enabled()? sz_s2u(size):
sz_index2size(*ind);
assert(*usize > 0 && *usize <= SC_LARGE_MAXCLASS);
return false;
}
@@ -2924,7 +2948,7 @@ ifree(tsd_t *tsd, void *ptr, tcache_t *tcache, bool slow_path) {
&alloc_ctx);
assert(alloc_ctx.szind != SC_NSIZES);
size_t usize = sz_index2size(alloc_ctx.szind);
size_t usize = emap_alloc_ctx_usize_get(&alloc_ctx);
if (config_prof && opt_prof) {
prof_free(tsd, ptr, usize, &alloc_ctx);
}
@@ -2956,35 +2980,41 @@ isfree(tsd_t *tsd, void *ptr, size_t usize, tcache_t *tcache, bool slow_path) {
assert(malloc_initialized() || IS_INITIALIZER);
emap_alloc_ctx_t alloc_ctx;
szind_t szind = sz_size2index(usize);
if (!config_prof) {
alloc_ctx.szind = sz_size2index(usize);
alloc_ctx.slab = (alloc_ctx.szind < SC_NBINS);
emap_alloc_ctx_init(&alloc_ctx, szind, (szind < SC_NBINS),
usize);
} else {
if (likely(!prof_sample_aligned(ptr))) {
/*
* When the ptr is not page aligned, it was not sampled.
* usize can be trusted to determine szind and slab.
*/
alloc_ctx.szind = sz_size2index(usize);
alloc_ctx.slab = (alloc_ctx.szind < SC_NBINS);
emap_alloc_ctx_init(&alloc_ctx, szind,
(szind < SC_NBINS), usize);
} else if (opt_prof) {
/*
* Small sampled allocs promoted can still get correct
* usize here. Check comments in edata_usize_get.
*/
emap_alloc_ctx_lookup(tsd_tsdn(tsd), &arena_emap_global,
ptr, &alloc_ctx);
if (config_opt_safety_checks) {
/* Small alloc may have !slab (sampled). */
size_t true_size =
emap_alloc_ctx_usize_get(&alloc_ctx);
if (unlikely(alloc_ctx.szind !=
sz_size2index(usize))) {
safety_check_fail_sized_dealloc(
/* current_dealloc */ true, ptr,
/* true_size */ sz_index2size(
alloc_ctx.szind),
/* true_size */ true_size,
/* input_size */ usize);
}
}
} else {
alloc_ctx.szind = sz_size2index(usize);
alloc_ctx.slab = (alloc_ctx.szind < SC_NBINS);
emap_alloc_ctx_init(&alloc_ctx, szind,
(szind < SC_NBINS), usize);
}
}
bool fail = maybe_check_alloc_ctx(tsd, ptr, &alloc_ctx);
@@ -3486,7 +3516,7 @@ do_rallocx(void *ptr, size_t size, int flags, bool is_realloc) {
emap_alloc_ctx_lookup(tsd_tsdn(tsd), &arena_emap_global, ptr,
&alloc_ctx);
assert(alloc_ctx.szind != SC_NSIZES);
old_usize = sz_index2size(alloc_ctx.szind);
old_usize = emap_alloc_ctx_usize_get(&alloc_ctx);
assert(old_usize == isalloc(tsd_tsdn(tsd), ptr));
if (aligned_usize_get(size, alignment, &usize, NULL, false)) {
goto label_oom;
@@ -3756,7 +3786,7 @@ je_xallocx(void *ptr, size_t size, size_t extra, int flags) {
emap_alloc_ctx_lookup(tsd_tsdn(tsd), &arena_emap_global, ptr,
&alloc_ctx);
assert(alloc_ctx.szind != SC_NSIZES);
old_usize = sz_index2size(alloc_ctx.szind);
old_usize = emap_alloc_ctx_usize_get(&alloc_ctx);
assert(old_usize == isalloc(tsd_tsdn(tsd), ptr));
/*
* The API explicitly absolves itself of protecting against (size +