project-base/server-skynet-source-3rd-jemalloc
2
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Remove medium size classes.

Browse Source 
Remove medium size classes, because concurrent dirty page purging is
no longer capable of purging inactive dirty pages inside active runs
(due to recent arena/bin locking changes).

Enhance tcache to support caching large objects, so that the same range
of size classes is still cached, despite the removal of medium size
class support.
This commit is contained in:
Jason Evans
2010-03-17 16:27:39 -07:00
parent e69bee01de
commit dafde14e08
12 changed files with 514 additions and 483 deletions
Download Patch File Download Diff File Expand all files Collapse all files

131
jemalloc/doc/jemalloc.3.in
View File

@@ -38,7 +38,7 @@
.\" @(#)malloc.3 8.1 (Berkeley) 6/4/93
.\" $FreeBSD: head/lib/libc/stdlib/malloc.3 182225 2008-08-27 02:00:53Z jasone $
.\"
.Dd March 14, 2010
.Dd March 17, 2010
.Dt JEMALLOC 3
.Os
.Sh NAME
@@ -378,13 +378,15 @@ will disable dirty page purging.
@roff_tcache@.It H
@roff_tcache@Enable/disable thread-specific caching.
@roff_tcache@When there are multiple threads, each thread uses a
@roff_tcache@thread-specific cache for small and medium objects.
@roff_tcache@thread-specific cache for objects up to a certain size.
@roff_tcache@Thread-specific caching allows many allocations to be satisfied
@roff_tcache@without performing any thread synchronization, at the cost of
@roff_tcache@increased memory use.
@roff_tcache@See the
@roff_tcache@.Dq G
@roff_tcache@option for related tuning information.
@roff_tcache@and
@roff_tcache@.Dq M
@roff_tcache@options for related tuning information.
@roff_tcache@This option is enabled by default.
@roff_prof@.It I
@roff_prof@Double/halve the average interval between memory profile dumps, as
@@ -426,16 +428,15 @@ The default chunk size is 4 MiB.
@roff_prof@See the
@roff_prof@.Dq F option for information on analyzing heap profile output.
@roff_prof@This option is disabled by default.
.It M
Double/halve the size of the maximum medium size class.
The valid range is from one page to one half chunk.
The default value is 32 KiB.
@roff_tcache@.It M
@roff_tcache@Double/halve the maximum size class to cache.
@roff_tcache@At a minimum, all small size classes are cached, and at a maximum
@roff_tcache@all large size classes are cached.
@roff_tcache@The default maximum is 32 KiB.
.It N
Double/halve the number of arenas.
The default number of arenas is
@roff_tcache@two
@roff_no_tcache@four
times the number of CPUs, or one if there is a single CPU.
The default number of arenas is four times the number of CPUs, or one if there
is a single CPU.
@roff_swap@.It O
@roff_swap@Over-commit memory as a side effect of using anonymous
@roff_swap@.Xr mmap 2
@@ -550,9 +551,9 @@ However, it may make sense to reduce the number of arenas if an application
does not make much use of the allocation functions.
.Pp
@roff_tcache@In addition to multiple arenas, this allocator supports
@roff_tcache@thread-specific caching for small and medium objects, in order to
@roff_tcache@make it possible to completely avoid synchronization for most small
@roff_tcache@and medium allocation requests.
@roff_tcache@thread-specific caching for small objects, in order to make it
@roff_tcache@possible to completely avoid synchronization for most small
@roff_tcache@allocation requests.
@roff_tcache@Such caching allows very fast allocation in the common case, but it
@roff_tcache@increases memory usage and fragmentation, since a bounded number of
@roff_tcache@objects can remain allocated in each thread cache.
@@ -563,27 +564,23 @@ Chunks are always aligned to multiples of the chunk size.
This alignment makes it possible to find metadata for user objects very
quickly.
.Pp
User objects are broken into four categories according to size: small, medium,
large, and huge.
User objects are broken into three categories according to size: small, large,
and huge.
Small objects are smaller than one page.
Medium objects range from one page to an upper limit determined at run time (see
the
.Dq M
option).
Large objects are smaller than the chunk size.
Huge objects are a multiple of the chunk size.
Small, medium, and large objects are managed by arenas; huge objects are managed
Small and large objects are managed by arenas; huge objects are managed
separately in a single data structure that is shared by all threads.
Huge objects are used by applications infrequently enough that this single
data structure is not a scalability issue.
.Pp
Each chunk that is managed by an arena tracks its contents as runs of
contiguous pages (unused, backing a set of small or medium objects, or backing
one large object).
contiguous pages (unused, backing a set of small objects, or backing one large
object).
The combination of chunk alignment and chunk page maps makes it possible to
determine all metadata regarding small and large allocations in constant time.
.Pp
Small and medium objects are managed in groups by page runs.
Small objects are managed in groups by page runs.
Each run maintains a bitmap that tracks which regions are in use.
@roff_tiny@Allocation requests that are no more than half the quantum (8 or 16,
@roff_tiny@depending on architecture) are rounded up to the nearest power of
@@ -603,13 +600,7 @@ Allocation requests that are more than the minimum subpage-multiple size class,
but no more than the maximum subpage-multiple size class are rounded up to the
nearest multiple of the subpage size (256).
Allocation requests that are more than the maximum subpage-multiple size class,
but no more than the maximum medium size class (see the
.Dq M
option) are rounded up to the nearest medium size class; spacing is an
automatically determined power of two and ranges from the subpage size to the
page size.
Allocation requests that are more than the maximum medium size class, but small
enough to fit in an arena-managed chunk (see the
but small enough to fit in an arena-managed chunk (see the
.Dq K
option), are rounded up to the nearest run size.
Allocation requests that are too large to fit in an arena-managed chunk are
@@ -838,13 +829,6 @@ See the
option.
.Ed
.\"-----------------------------------------------------------------------------
.It Sy "opt.lg_medium_max (size_t) r-"
.Bd -ragged -offset indent -compact
See the
.Dq M
option.
.Ed
.\"-----------------------------------------------------------------------------
.It Sy "opt.lg_dirty_mult (ssize_t) r-"
.Bd -ragged -offset indent -compact
See the
@@ -900,11 +884,6 @@ Subpage size class interval.
Page size.
.Ed
.\"-----------------------------------------------------------------------------
.It Sy "arenas.medium (size_t) r-"
.Bd -ragged -offset indent -compact
Medium size class interval.
.Ed
.\"-----------------------------------------------------------------------------
.It Sy "arenas.chunksize (size_t) r-"
.Bd -ragged -offset indent -compact
Chunk size.
@@ -952,15 +931,10 @@ Minimum subpage-spaced size class.
Maximum subpage-spaced size class.
.Ed
.\"-----------------------------------------------------------------------------
.It Sy "arenas.medium_min (size_t) r-"
.Bd -ragged -offset indent -compact
Minimum medium-spaced size class.
.Ed
.\"-----------------------------------------------------------------------------
.It Sy "arenas.medium_max (size_t) r-"
.Bd -ragged -offset indent -compact
Maximum medium-spaced size class.
.Ed
@roff_tcache@.It Sy "arenas.tcache_max (size_t) r-"
@roff_tcache@.Bd -ragged -offset indent -compact
@roff_tcache@Maximum thread-cached size class.
@roff_tcache@.Ed
.\"-----------------------------------------------------------------------------
.It Sy "arenas.ntbins (unsigned) r-"
.Bd -ragged -offset indent -compact
@@ -982,16 +956,16 @@ Number of cacheline-spaced bin size classes.
Number of subpage-spaced bin size classes.
.Ed
.\"-----------------------------------------------------------------------------
.It Sy "arenas.nmbins (unsigned) r-"
.Bd -ragged -offset indent -compact
Number of medium-spaced bin size classes.
.Ed
.\"-----------------------------------------------------------------------------
.It Sy "arenas.nbins (unsigned) r-"
.Bd -ragged -offset indent -compact
Total number of bin size classes.
.Ed
.\"-----------------------------------------------------------------------------
@roff_tcache@.It Sy "arenas.nhbins (unsigned) r-"
@roff_tcache@.Bd -ragged -offset indent -compact
@roff_tcache@Total number of thread cache bin size classes.
@roff_tcache@.Ed
.\"-----------------------------------------------------------------------------
.It Sy "arenas.bin.<i>.size (size_t) r-"
.Bd -ragged -offset indent -compact
Maximum size supported by size class.
@@ -1147,26 +1121,6 @@ has not been called.
@roff_stats@Cumulative number of small allocation requests.
@roff_stats@.Ed
.\"-----------------------------------------------------------------------------
@roff_stats@.It Sy "stats.arenas.<i>.medium.allocated (size_t) r-"
@roff_stats@.Bd -ragged -offset indent -compact
@roff_stats@Number of bytes currently allocated by medium objects.
@roff_stats@.Ed
.\"-----------------------------------------------------------------------------
@roff_stats@.It Sy "stats.arenas.<i>.medium.nmalloc (uint64_t) r-"
@roff_stats@.Bd -ragged -offset indent -compact
@roff_stats@Cumulative number of allocation requests served by medium bins.
@roff_stats@.Ed
.\"-----------------------------------------------------------------------------
@roff_stats@.It Sy "stats.arenas.<i>.medium.ndalloc (uint64_t) r-"
@roff_stats@.Bd -ragged -offset indent -compact
@roff_stats@Cumulative number of medium objects returned to bins.
@roff_stats@.Ed
.\"-----------------------------------------------------------------------------
@roff_stats@.It Sy "stats.arenas.<i>.medium.nrequests (uint64_t) r-"
@roff_stats@.Bd -ragged -offset indent -compact
@roff_stats@Cumulative number of medium allocation requests.
@roff_stats@.Ed
.\"-----------------------------------------------------------------------------
@roff_stats@.It Sy "stats.arenas.<i>.large.allocated (size_t) r-"
@roff_stats@.Bd -ragged -offset indent -compact
@roff_stats@Number of bytes currently allocated by large objects.
@@ -1174,12 +1128,19 @@ has not been called.
.\"-----------------------------------------------------------------------------
@roff_stats@.It Sy "stats.arenas.<i>.large.nmalloc (uint64_t) r-"
@roff_stats@.Bd -ragged -offset indent -compact
@roff_stats@Cumulative number of large allocation requests.
@roff_stats@Cumulative number of large allocation requests served directly by
@roff_stats@the arena.
@roff_stats@.Ed
.\"-----------------------------------------------------------------------------
@roff_stats@.It Sy "stats.arenas.<i>.large.ndalloc (uint64_t) r-"
@roff_stats@.Bd -ragged -offset indent -compact
@roff_stats@Cumulative number of large deallocation requests.
@roff_stats@Cumulative number of large deallocation requests served directly by
@roff_stats@the arena.
@roff_stats@.Ed
.\"-----------------------------------------------------------------------------
@roff_stats@.It Sy "stats.arenas.<i>.large.nrequests (uint64_t) r-"
@roff_stats@.Bd -ragged -offset indent -compact
@roff_stats@Cumulative number of large allocation requests.
@roff_stats@.Ed
.\"-----------------------------------------------------------------------------
@roff_stats@.It Sy "stats.arenas.<i>.bins.<j>.allocated (size_t) r-"
@@ -1233,6 +1194,18 @@ has not been called.
@roff_stats@Current number of runs.
@roff_stats@.Ed
.\"-----------------------------------------------------------------------------
@roff_stats@.It Sy "stats.arenas.<i>.lruns.<j>.nmalloc (uint64_t) r-"
@roff_stats@.Bd -ragged -offset indent -compact
@roff_stats@Cumulative number of allocation requests for this size class served
@roff_stats@directly by the arena.
@roff_stats@.Ed
.\"-----------------------------------------------------------------------------
@roff_stats@.It Sy "stats.arenas.<i>.lruns.<j>.ndalloc (uint64_t) r-"
@roff_stats@.Bd -ragged -offset indent -compact
@roff_stats@Cumulative number of deallocation requests for this size class
@roff_stats@served directly by the arena.
@roff_stats@.Ed
.\"-----------------------------------------------------------------------------
@roff_stats@.It Sy "stats.arenas.<i>.lruns.<j>.nrequests (uint64_t) r-"
@roff_stats@.Bd -ragged -offset indent -compact
@roff_stats@Cumulative number of allocation requests for this size class.