Commit Graph

498 Commits

Author SHA1 Message Date
David Goldblatt
46a9d7fc0b PA: Move in rest of purging. 2020-04-10 13:12:47 -07:00
David Goldblatt
2d6eec7b5c PA: Move in decay-all pathway. 2020-04-10 13:12:47 -07:00
David Goldblatt
f012c43be0 PA: Move in decay_to_limit 2020-04-10 13:12:47 -07:00
David Goldblatt
103f5feda5 Move bg thread activity check out of purging core. 2020-04-10 13:12:47 -07:00
David Goldblatt
3034f4a508 PA: Move in decay_stashed. 2020-04-10 13:12:47 -07:00
David Goldblatt
aef28b2f8f PA: Move in stash_decayed. 2020-04-10 13:12:47 -07:00
David Goldblatt
655a096343 Move bg inactivity check out of purge inner loop.
I.e. do it once per call to arena_decay_stashed instead of once per muzzy purge.
2020-04-10 13:12:47 -07:00
David Goldblatt
71fc0dc968 PA: Move in remaining page allocation functions. 2020-04-10 13:12:47 -07:00
David Goldblatt
7be3dea82c PA: Have slab allocations use it. 2020-04-10 13:12:47 -07:00
David Goldblatt
9f93625c14 PA: Move in arena large allocation functionality. 2020-04-10 13:12:47 -07:00
David Goldblatt
eba35e2e48 Remove extent knowledge of arena. 2020-04-10 13:12:47 -07:00
David Goldblatt
e77f47a85a Move arena decay getters to PA. 2020-04-10 13:12:47 -07:00
David Goldblatt
f77cec311e Decay: Take current time as an argument.
This better facilitates testing.
2020-04-10 13:12:47 -07:00
David Goldblatt
8f2193dc8d Decay: Move in arena decay functions. 2020-04-10 13:12:47 -07:00
David Goldblatt
7b62885476 Introduce decay module and put decay objects in PA 2020-04-10 13:12:47 -07:00
David Goldblatt
70d12ffa05 PA: Move mapped into pa stats. 2020-04-10 13:12:47 -07:00
David Goldblatt
ce8c0d6c09 PA: Move in arena extent_sn counter.
Just another step towards making PA self-contained.
2020-04-10 13:12:47 -07:00
David Goldblatt
1ad368c8b7 PA: Move in decay stats. 2020-04-10 13:12:47 -07:00
David Goldblatt
356aaa7dc6 Introduce lockedint module.
This pulls out the various abstractions where some stats counter is sometimes an
atomic, sometimes a plain variable, sometimes always protected by a lock,
sometimes protected by reads but not writes, etc.  With this change, these cases
are treated consistently, and access patterns tagged.

In the process, we fix a few missed-update bugs (where one caller assumes
"protected-by-a-lock" semantics and another does not).
2020-04-10 13:12:47 -07:00
David Goldblatt
acd0bf6a26 PA: move in ecache_grow. 2020-04-10 13:12:47 -07:00
David Goldblatt
32cb7c2f0b PA: Add a stats type. 2020-04-10 13:12:47 -07:00
David Goldblatt
688fb3eb89 PA: Move in the arena edata_cache. 2020-04-10 13:12:47 -07:00
David Goldblatt
8433ad84ea PA: move in shard initialization. 2020-04-10 13:12:47 -07:00
David Goldblatt
a24faed569 PA: Move in the ecache_t objects. 2020-04-10 13:12:47 -07:00
David Goldblatt
585f925055 Move cache index randomization out of extent.
This is logically at a higher level of the stack; extent should just allocate
things at the page-level; it shouldn't care exactly why the callers wants a
given number of pages.
2020-04-10 13:12:47 -07:00
David Goldblatt
2e5899c129 Stats: Fix tcache_bytes reporting.
Previously, large allocations in tcaches would have their sizes reduced during
stats estimation.  Added a test, which fails before this change but passes now.

This fixes a bug introduced in 5934846612, which
was itself fixing a bug introduced in 9c0549007d.
2020-03-13 07:53:34 -07:00
David Goldblatt
ff6acc6ed5 Cache bin: simplify names and argument ordering.
We always start with the cache bin, then its info (if necessary).
2020-03-12 11:54:19 -07:00
David Goldblatt
e1dcc557d6 Cache bin: Only take the relevant cache_bin_info_t
Previously, we took an array of cache_bin_info_ts and an index, and dereferenced
ourselves.  But infos for other cache_bins aren't relevant to any particular
cache bin, so that should be the caller's job.
2020-03-12 11:54:19 -07:00
David Goldblatt
1b00d808d7 cache_bin: Don't let arena see empty position. 2020-03-12 11:54:19 -07:00
David Goldblatt
74d36d78ef Cache bin: Make ncached_max a query on the info_t. 2020-03-12 11:54:19 -07:00
David Goldblatt
909c501b07 Cache_bin: Shouldn't know about tcache.
Instead, have it take the cache_bin_info_ts to use by pointer.  While we're
here, add a src file for the cache bin.
2020-03-12 11:54:19 -07:00
David Goldblatt
79f1ee2fc0 Move junking out of arena/tcache code.
This is debug only and we keep it off the fast path.  Moving it here simplifies
the internal logic.

This never tries to junk on regions that were shrunk via xallocx.  I think this
is fine for two reasons:
- The shrunk-with-xallocx case is rare.
- We don't always do that anyway before this diff (it depends on the opt
  settings and extent hooks in effect).
2020-03-12 11:54:19 -07:00
David Goldblatt
7e6c8a7286 Emap: Standardize naming.
Namespace everything under emap_, always specify what it is we're looking up
(emap_lookup -> emap_edata_lookup), and use "ctx" over "info".
2020-02-17 10:50:51 -08:00
David Goldblatt
ac50c1e44b Emap: Remove direct access to emap internals.
In the process, we do a few local cleanups and optimizations.  In particular,
the size safety check on tcache flush no longer does a redundant load.
2020-02-17 10:50:51 -08:00
David Goldblatt
65a54d7714 Emap: Move in szind and slab modifications. 2020-02-17 10:50:51 -08:00
David Goldblatt
9b5d105fc3 Emap: Move in iealloc.
This is logically scoped to the emap.
2020-02-17 10:50:51 -08:00
David Goldblatt
01f255161c Add emap, for tracking extent locking. 2020-02-17 10:50:51 -08:00
Qi Wang
ba0e35411c Rework the bin locking around tcache refill / flush.
Previously, tcache fill/flush (as well as small alloc/dalloc on the arena) may
potentially drop the bin lock for slab_alloc and slab_dalloc.  This commit
refactors the logic so that the slab calls happen in the same function / level
as the bin lock / unlock.  The main purpose is to be able to use flat combining
without having to keep track of stack state.

In the meantime, this change reduces the locking, especially for slab_dalloc
calls, where nothing happens after the call.
2020-02-13 23:31:54 -08:00
Qi Wang
d71a145ec1 Chagne prof_accum_t to counter_accum_t for general purpose. 2020-01-29 09:57:55 -08:00
David Goldblatt
bd3be8e0b1 Remove commit parameter to ecache functions.
No caller ever wants uncommitted memory.
2020-01-17 10:54:56 -08:00
David Goldblatt
2f4fa80414 Rename extents -> ecache. 2019-12-20 10:18:40 -08:00
David Goldblatt
576d7047ab Ecache: Should know its arena_ind.
What we call an arena_ind is really the index associated with some particular
set of ehooks; the arena is just the user-visible portion of that.  Making this
explicit, and reframing checks in terms of that, makes the code simpler and
cleaner, and helps us avoid passing the arena itself all throughout extent code.

This lets us put back an arena-specific assert.
2019-12-20 10:18:40 -08:00
David Goldblatt
c792f3e4ab edata_cache: Remember the associated base_t.
This will save us some trouble down the line when we stop passing arena pointers
everywhere; we won't have to pass around a base_t pointer either.
2019-12-20 10:18:40 -08:00
David Goldblatt
ae23e5f426 Unify extent_alloc_wrapper with the other wrappers.
Previously, it was really more like extents_alloc (it looks in an ecache for an
extent to reuse as its primary allocation pathway).  Make that pathway more
explciitly like extents_alloc, and rename extent_alloc_wrapper_hard accordingly.
2019-12-20 10:18:40 -08:00
David Goldblatt
d8b0b66c6c Put extent_state_t into ecache as well as eset. 2019-12-20 10:18:40 -08:00
David Goldblatt
bb70df8e5b Extent refactor: Introduce ecache module.
This will eventually completely wrap the eset, and handle concurrency,
allocation, and deallocation.  For now, we only pull out the mutex from the
eset.
2019-12-20 10:18:40 -08:00
David Goldblatt
7859184179 Pull out edata_t caching into its own module. 2019-12-20 10:18:40 -08:00
David Goldblatt
a7862df616 Rename extent_t to edata_t.
This frees us up from the unfortunate extent/extent2 naming collision.
2019-12-20 10:18:40 -08:00
David Goldblatt
d0f187ad3b Arena: Loosen arena_may_have_muzzy restrictions.
If there are custom extent hooks, pages_can_purge_lazy is not necessarily the
right guard.  We could check ehooks_are_default too, but the case where
purge_lazy is unsupported is rare and getting rarer.  Just checking the decay
interval captures most of the benefit.
2019-12-20 10:18:40 -08:00
David Goldblatt
ae0d8e8591 Move extent ehook calls into ehooks 2019-12-20 10:18:40 -08:00
David Goldblatt
9f6eb09585 Extents: Eagerly initialize extent hooks.
When deferred initialization was added, initializing required copying
sizeof(extent_hooks_t) bytes after a pointer chase. Today, it's just a single
pointer loaded from the base_t. In subsequent diffs, we'll get rid of even that.
2019-12-20 10:18:40 -08:00
David Goldblatt
4278f84603 Move extent hook getters/setters to arena.c
This is where they're logically scoped; they access arena data.
2019-12-20 10:18:40 -08:00
Yinan Zhang
1d01e4c770 Initialization utilities for nstime 2019-12-16 16:08:56 -08:00
Qi Wang
9a3c738009 Refactor arena_bin_malloc_hard(). 2019-11-21 11:41:26 -08:00
Qi Wang
9a7ae3c97f Reduce footprint of bin_t.
Avoid storing mutex_prof_data_t in bin_t.  Added bin_stats_data_t which is used
for reporting bin stats.
2019-11-21 11:08:36 -08:00
Qi Wang
04cb7d4d6b Bail out early for muzzy decay.
This avoids taking the muzzy decay mutex with the default setting.
2019-11-15 16:24:15 -08:00
Qi Wang
19a51abf33 Avoid arena->offset_state when tsd not available for prng.
Use stack locals and remove the offset_state in arena.
2019-11-11 10:35:37 -08:00
Nick Desaulniers
d01b425e5d Add -Wimplicit-fallthrough checks if supported
Clang since r369414 (clang-10) can now check -Wimplicit-fallthrough for
C code, and use the GNU C style attribute to denote fallthrough.

Move the test from header only to autoconf. The previous test used
brittle version detection which did not work for newer clang that
supported this feature.

The attribute has to be its own statement, hence the added `;`. It also
can only precede case statements, so the final cases should be
explicitly terminated with break statements.

Fixes commit 3d29d11ac2 ("Clean compilation -Wextra")
Link: 1e0affb6e5
Signed-off-by: Nick Desaulniers <ndesaulniers@google.com>
2019-11-08 13:03:03 -08:00
Yinan Zhang
198f02e797 Pull prof_accumbytes into thread event handler 2019-11-04 15:21:16 -08:00
David T. Goldblatt
3d84bd57f4 Arena: Add helper function arena_get_from_extent. 2019-09-23 23:06:27 -07:00
David T. Goldblatt
821dd53a1d Extent -> Eset: Rename arena members. 2019-09-23 23:06:27 -07:00
David T. Goldblatt
e144b21e4b Extent -> Eset: Move fork handling. 2019-09-23 23:06:27 -07:00
David T. Goldblatt
a42861540e Extents -> Eset: Convert some stats getters. 2019-09-23 23:06:27 -07:00
David T. Goldblatt
63d1b7a7a7 Extents -> Eset: move extents_state_get. 2019-09-23 23:06:27 -07:00
David T. Goldblatt
b416b96a39 Extents -> Eset: rename/move extents_init. 2019-09-23 23:06:27 -07:00
David T. Goldblatt
4e5e43f22e Rename extents_t -> eset_t. 2019-09-23 23:06:27 -07:00
David T. Goldblatt
41187bdfb0 Extents: Break extent-struct/arena interactions
Specifically, the extent_arena_[g|s]et functions and the address randomization.

These are the only things that tie the extent struct itself to the arena code.
2019-09-23 23:06:27 -07:00
David T. Goldblatt
e7cf84a8dd Rearrange slab data and constants
The constants logically belong in the sc module. The slab data bitmap isn't
really scoped to an arena; move it to its own module.
2019-09-23 23:06:27 -07:00
Qi Wang
0043e68d4c Track low_water == -1 case explicitly.
The -1 value of low_water indicates if the cache has been depleted and
refilled.  Track the status explicitly in the tcache struct.

This allows the fast path to check if (cur_ptr > low_water), instead of >=,
which avoids reaching slow path when the last item is allocated.
2019-08-21 16:00:38 -07:00
Qi Wang
937ca1db9f Store ncached_max * ptr_size in tcache_bin_info.
With the cache bin metadata switched to pointers, ncached_max is usually
accessed and timed by sizeof(ptr). Store the results in tcache_bin_info for
direct access, and add a helper function for the ncached_max value.
2019-08-19 12:23:24 -07:00
Qi Wang
7599c82d48 Redesign the cache bin metadata for fast path.
Implement the pointer-based metadata for tcache bins --
- 3 pointers are maintained to represent each bin;
- 2 of the pointers are compressed on 64-bit;
- is_full / is_empty done through pointer comparison;

Comparing to the previous counter based design --
- fast-path speed up ~15% in benchmarks
- direct pointer comparison and de-reference
- no need to access tcache_bin_info in common case
2019-08-19 12:21:44 -07:00
Qi Wang
5934846612 Fix large bin index accessed through cache bin descriptor. 2019-08-11 16:31:12 -07:00
Qi Wang
bc0998a905 Invoke arena_dalloc_promoted() properly w/o tcache.
When tcache was disabled, the dalloc promoted case was missing.
2019-07-24 18:30:54 -07:00
Qi Wang
4e36ce34c1 Track the leaked VM space via the abandoned_vm counter.
The counter is 0 unless metadata allocation failed (indicates OOM), and is
mainly for sanity checking.
2019-07-24 11:24:22 -07:00
Qi Wang
07c44847c2 Track nfills and nflushes for arenas.i.small / large.
Small is added purely for convenience.  Large flushes wasn't tracked before and
can be useful in analysis.  Large fill simply reports nmalloc, since there is no
batch fill for large currently.
2019-05-15 10:05:09 -07:00
Doron Roberts-Kedes
7fc4f2a32c Add nonfull_slabs to bin_stats_t.
When config_stats is enabled track the size of bin->slabs_nonfull in
the new nonfull_slabs counter in bin_stats_t. This metric should be
useful for establishing an upper ceiling on the savings possible by
meshing.
2019-04-29 13:35:02 -07:00
David Goldblatt
33e1dad680 Safety checks: Add a redzoning feature. 2019-04-15 16:48:12 -07:00
Qi Wang
e3db480f6f Rename huge_threshold to oversize_threshold.
The keyword huge tend to remind people of huge pages which is not relevent to
the feature.
2019-01-25 13:15:45 -08:00
Qi Wang
bbe8e6a909 Avoid creating bg thds for huge arena lone.
For low arena count settings, the huge threshold feature may trigger an unwanted
bg thd creation.  Given that the huge arena does eager purging by default,
bypass bg thd creation when initializing the huge arena.
2019-01-15 16:00:34 -08:00
Qi Wang
98b56ab23d Store the bin shard selection in TSD.
This avoids having to choose bin shard on the fly, also will allow flexible bin
binding for each thread.
2018-12-03 17:17:03 -08:00
Qi Wang
3f9f2833f6 Add opt.bin_shards to specify number of bin shards.
The option uses the same format as "slab_sizes" to specify number of shards for
each bin size.
2018-12-03 17:17:03 -08:00
Qi Wang
37b8913925 Add support for sharded bins within an arena.
This makes it possible to have multiple set of bins in an arena, which improves
arena scalability because the bins (especially the small ones) are always the
limiting factor in production workload.

A bin shard is picked on allocation; each extent tracks the bin shard id for
deallocation.  The shard size will be determined using runtime options.
2018-12-03 17:17:03 -08:00
Dave Watson
13c237c7ef Add a fastpath for arena_slab_reg_alloc_batch
Also adds a configure.ac check for __builtin_popcount, which is used
in the new fastpath.
2018-11-14 07:09:11 -08:00
Dave Watson
17aa470760 add extent_nfree_sub 2018-11-14 07:09:11 -08:00
Dave Watson
4b82872ebf arena: Refactor tcache_fill to batch fill from slab
Refactor tcache_fill, introducing a new function arena_slab_reg_alloc_batch,
which will fill multiple pointers from a slab.

There should be no functional changes here, but allows future optimization
on reg_alloc_batch.
2018-11-14 07:09:11 -08:00
Tyler Etzel
126252a7e6 Add stats for the size of extent_avail heap 2018-08-02 10:16:06 -07:00
Tyler Etzel
c14e6c0819 Add extents information to mallocstats output
- Show number/bytes of extents of each size that are dirty, muzzy, retained.
2018-08-02 10:16:06 -07:00
David Goldblatt
3aba072cef SC: Remove global data.
The global data is mostly only used at initialization, or for easy access to
values we could compute statically.  Instead of consuming that space (and
risking TLB misses), we can just pass around a pointer to stack data during
bootstrapping.
2018-07-23 13:37:08 -07:00
David Goldblatt
55e5cc1341 SC: Make some key size classes static.
The largest small class, smallest large class, and largest large class may all
be needed down fast paths; to avoid the risk of touching another cache line, we
can make them available as constants.
2018-07-12 20:53:06 -07:00
David Goldblatt
e904f813b4 Hide size class computation behind a layer of indirection.
This class removes almost all the dependencies on size_classes.h, accessing the
data there only via the new module sc.h, which does not depend on any
configuration options.

In a subsequent commit, we'll remove the configure-time size class computations,
doing them at boot time, instead.
2018-07-12 20:53:06 -07:00
gnzlbg
3d29d11ac2 Clean compilation -Wextra
Before this commit jemalloc produced many warnings when compiled with -Wextra
with both Clang and GCC. This commit fixes the issues raised by these warnings
or suppresses them if they were spurious at least for the Clang and GCC
versions covered by CI.

This commit:

* adds `JEMALLOC_DIAGNOSTIC` macros: `JEMALLOC_DIAGNOSTIC_{PUSH,POP}` are
  used to modify the stack of enabled diagnostics. The
  `JEMALLOC_DIAGNOSTIC_IGNORE_...` macros are used to ignore a concrete
  diagnostic.

* adds `JEMALLOC_FALLTHROUGH` macro to explicitly state that falling
  through `case` labels in a `switch` statement is intended

* Removes all UNUSED annotations on function parameters. The warning
  -Wunused-parameter is now disabled globally in
  `jemalloc_internal_macros.h` for all translation units that include
  that header. It is never re-enabled since that header cannot be
  included by users.

* locally suppresses some -Wextra diagnostics:

  * `-Wmissing-field-initializer` is buggy in older Clang and GCC versions,
    where it does not understanding that, in C, `= {0}` is a common C idiom
    to initialize a struct to zero

  * `-Wtype-bounds` is suppressed in a particular situation where a generic
    macro, used in multiple different places, compares an unsigned integer for
    smaller than zero, which is always true.

  * `-Walloc-larger-than-size=` diagnostics warn when an allocation function is
    called with a size that is too large (out-of-range). These are suppressed in
    the parts of the tests where `jemalloc` explicitly does this to test that the
    allocation functions fail properly.

* adds a new CI build bot that runs the log unit test on CI.

Closes #1196 .
2018-07-09 21:40:42 -07:00
Qi Wang
94a88c26f4 Implement huge arena: opt.huge_threshold.
The feature allows using a dedicated arena for huge allocations.  We want the
addtional arena to separate huge allocation because: 1) mixing small extents
with huge ones causes fragmentation over the long run (this feature reduces VM
size significantly); 2) with many arenas, huge extents rarely get reused across
threads; and 3) huge allocations happen way less frequently, therefore no
concerns for lock contention.
2018-06-29 10:35:02 -07:00
Qi Wang
0ff7ff3ec7 Optimize ixalloc by avoiding a size lookup. 2018-06-05 21:03:51 -07:00
Qi Wang
d22e150320 Avoid taking extents_muzzy mutex when muzzy is disabled.
When muzzy decay is disabled, no need to allocate from extents_muzzy.  This
saves us a couple of mutex operations down the extents_alloc path.
2018-05-24 14:40:56 -07:00
David Goldblatt
cb0707c0fc Hooks: hook the realloc pathways that move/expand. 2018-05-18 11:43:03 -07:00
David Goldblatt
c7a87e0e0b Rename hooks module to test_hooks.
"Hooks" is really the best name for the module that will contain the publicly
exposed hooks.  So lets rename the current "hooks" module (that hook external
dependencies, for reentrancy testing) to "test_hooks".
2018-05-18 11:43:03 -07:00
Qi Wang
0fadf4a2e3 Add UNUSED to avoid compiler warnings. 2018-04-16 13:50:21 -07:00
Jason Evans
4937309620 Silence a compiler warning. 2018-04-10 17:59:00 -07:00
David Goldblatt
d41b19f9c7 Implement arena regind computation using div_info_t.
This eliminates the need to generate an enormous switch statement in
arena_slab_regind.
2017-12-21 14:25:43 -08:00
David T. Goldblatt
7f1b02e3fa Split up and standardize naming of stats code.
The arena-associated stats are now all prefixed with arena_stats_, and live in
their own file.  Likewise, malloc_bin_stats_t -> bin_stats_t, also in its own
file.
2017-12-18 16:29:10 -08:00