Import jemalloc 9ef7f5dc34ff02f50d401e41c8d9a4a928e7c2aa (dev branch,

prior to 3.0.0 release) as contrib/jemalloc, and integrate it into libc.
The code being imported by this commit diverged from
lib/libc/stdlib/malloc.c in March 2010, which means that a portion of
the jemalloc 1.0.0 ChangeLog entries are relevant, as are the entries
for all subsequent releases.

contrib/jemalloc/COPYING

@@ -0,0 +1,27 @@
+Unless otherwise specified, files in the jemalloc source distribution are
+subject to the following license:
+--------------------------------------------------------------------------------
+Copyright (C) 2002-2012 Jason Evans <jasone@canonware.com>.
+All rights reserved.
+Copyright (C) 2007-2012 Mozilla Foundation.  All rights reserved.
+Copyright (C) 2009-2012 Facebook, Inc.  All rights reserved.
+
+Redistribution and use in source and binary forms, with or without
+modification, are permitted provided that the following conditions are met:
+1. Redistributions of source code must retain the above copyright notice(s),
+   this list of conditions and the following disclaimer.
+2. Redistributions in binary form must reproduce the above copyright notice(s),
+   this list of conditions and the following disclaimer in the documentation
+   and/or other materials provided with the distribution.
+
+THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY EXPRESS
+OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO
+EVENT SHALL THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY DIRECT, INDIRECT,
+INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
+LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
+PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
+OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF
+ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+--------------------------------------------------------------------------------

contrib/jemalloc/ChangeLog

@@ -0,0 +1,322 @@
+Following are change highlights associated with official releases.  Important
+bug fixes are all mentioned, but internal enhancements are omitted here for
+brevity (even though they are more fun to write about).  Much more detail can be
+found in the git revision history:
+
+    http://www.canonware.com/cgi-bin/gitweb.cgi?p=jemalloc.git
+    git://canonware.com/jemalloc.git
+
+* 3.0.0 (XXX not yet released)
+
+  Although this version adds some major new features, the primary focus is on
+  internal code cleanup that facilitates maintainability and portability, most
+  of which is not reflected in the ChangeLog.  This is the first release to
+  incorporate substantial contributions from numerous other developers, and the
+  result is a more broadly useful allocator (see the git revision history for
+  contribution details).  Note that the license has been unified, thanks to
+  Facebook granting a license under the same terms as the other copyright
+  holders (see COPYING).
+
+  New features:
+  - Implement Valgrind support, redzones, and quarantine.
+  - Add support for additional operating systems:
+    + FreeBSD
+    + Mac OS X Lion
+  - Add support for additional architectures:
+    + MIPS
+    + SH4
+    + Tilera
+  - Add support for cross compiling.
+  - Add nallocm(), which rounds a request size up to the nearest size class
+    without actually allocating.
+  - Implement aligned_alloc() (blame C11).
+  - Add the --disable-munmap option, and make it the default on Linux.
+  - Add the --with-mangling option.
+  - Add the --disable-experimental option.
+  - Add the "thread.tcache.enabled" mallctl.
+
+  Incompatible changes:
+  - Enable stats by default.
+  - Enable fill by default.
+  - Disable lazy locking by default.
+  - Rename the "tcache.flush" mallctl to "thread.tcache.flush".
+  - Rename the "arenas.pagesize" mallctl to "arenas.page".
+
+  Removed features:
+  - Remove the swap feature, including the "config.swap", "swap.avail",
+    "swap.prezeroed", "swap.nfds", and "swap.fds" mallctls.
+  - Remove highruns statistics, including the
+    "stats.arenas.<i>.bins.<j>.highruns" and
+    "stats.arenas.<i>.lruns.<j>.highruns" mallctls.
+  - As part of small size class refactoring, remove the "opt.lg_[qc]space_max",
+    "arenas.cacheline", "arenas.subpage", "arenas.[tqcs]space_{min,max}", and
+    "arenas.[tqcs]bins" mallctls.
+  - Remove the "arenas.chunksize" mallctl.
+  - Remove the "opt.lg_prof_tcmax" option.
+  - Remove the "opt.lg_prof_bt_max" option.
+  - Remove the "opt.lg_tcache_gc_sweep" option.
+  - Remove the --disable-tiny option, including the "config.tiny" mallctl.
+  - Remove the --enable-dynamic-page-shift configure option.
+  - Remove the --enable-sysv configure option.
+
+  Bug fixes:
+  - Fix fork-related bugs that could cause deadlock in children between fork
+    and exec.
+  - Fix a statistics-related bug in the "thread.arena" mallctl that could cause
+    invalid statistics and crashes.
+  - Work around TLS dallocation via free() on Linux.  This bug could cause
+    write-after-free memory corruption.
+  - Fix malloc_stats_print() to honor 'b' and 'l' in the opts parameter.
+  - Fix realloc(p, 0) to act like free(p).
+  - Do not enforce minimum alignment in memalign().
+  - Check for NULL pointer in malloc_usable_size().
+  - Fix bin->runcur management to fix a layout policy bug.  This bug did not
+    affect correctness.
+  - Fix a bug in choose_arena_hard() that potentially caused more arenas to be
+    initialized than necessary.
+  - Add missing "opt.lg_tcache_max" mallctl implementation.
+  - Use glibc allocator hooks to make mixed allocator usage less likely.
+  - Fix build issues for --disable-tcache.
+
+* 2.2.5 (November 14, 2011)
+
+  Bug fixes:
+  - Fix huge_ralloc() race when using mremap(2).  This is a serious bug that
+    could cause memory corruption and/or crashes.
+  - Fix huge_ralloc() to maintain chunk statistics.
+  - Fix malloc_stats_print(..., "a") output.
+
+* 2.2.4 (November 5, 2011)
+
+  Bug fixes:
+  - Initialize arenas_tsd before using it.  This bug existed for 2.2.[0-3], as
+    well as for --disable-tls builds in earlier releases.
+  - Do not assume a 4 KiB page size in test/rallocm.c.
+
+* 2.2.3 (August 31, 2011)
+
+  This version fixes numerous bugs related to heap profiling.
+
+  Bug fixes:
+  - Fix a prof-related race condition.  This bug could cause memory corruption,
+    but only occurred in non-default configurations (prof_accum:false).
+  - Fix off-by-one backtracing issues (make sure that prof_alloc_prep() is
+    excluded from backtraces).
+  - Fix a prof-related bug in realloc() (only triggered by OOM errors).
+  - Fix prof-related bugs in allocm() and rallocm().
+  - Fix prof_tdata_cleanup() for --disable-tls builds.
+  - Fix a relative include path, to fix objdir builds.
+
+* 2.2.2 (July 30, 2011)
+
+  Bug fixes:
+  - Fix a build error for --disable-tcache.
+  - Fix assertions in arena_purge() (for real this time).
+  - Add the --with-private-namespace option.  This is a workaround for symbol
+    conflicts that can inadvertently arise when using static libraries.
+
+* 2.2.1 (March 30, 2011)
+
+  Bug fixes:
+  - Implement atomic operations for x86/x64.  This fixes compilation failures
+    for versions of gcc that are still in wide use.
+  - Fix an assertion in arena_purge().
+
+* 2.2.0 (March 22, 2011)
+
+  This version incorporates several improvements to algorithms and data
+  structures that tend to reduce fragmentation and increase speed.
+
+  New features:
+  - Add the "stats.cactive" mallctl.
+  - Update pprof (from google-perftools 1.7).
+  - Improve backtracing-related configuration logic, and add the
+    --disable-prof-libgcc option.
+
+  Bug fixes:
+  - Change default symbol visibility from "internal", to "hidden", which
+    decreases the overhead of library-internal function calls.
+  - Fix symbol visibility so that it is also set on OS X.
+  - Fix a build dependency regression caused by the introduction of the .pic.o
+    suffix for PIC object files.
+  - Add missing checks for mutex initialization failures.
+  - Don't use libgcc-based backtracing except on x64, where it is known to work.
+  - Fix deadlocks on OS X that were due to memory allocation in
+    pthread_mutex_lock().
+  - Heap profiling-specific fixes:
+    + Fix memory corruption due to integer overflow in small region index
+      computation, when using a small enough sample interval that profiling
+      context pointers are stored in small run headers.
+    + Fix a bootstrap ordering bug that only occurred with TLS disabled.
+    + Fix a rallocm() rsize bug.
+    + Fix error detection bugs for aligned memory allocation.
+
+* 2.1.3 (March 14, 2011)
+
+  Bug fixes:
+  - Fix a cpp logic regression (due to the "thread.{de,}allocatedp" mallctl fix
+    for OS X in 2.1.2).
+  - Fix a "thread.arena" mallctl bug.
+  - Fix a thread cache stats merging bug.
+
+* 2.1.2 (March 2, 2011)
+
+  Bug fixes:
+  - Fix "thread.{de,}allocatedp" mallctl for OS X.
+  - Add missing jemalloc.a to build system.
+
+* 2.1.1 (January 31, 2011)
+
+  Bug fixes:
+  - Fix aligned huge reallocation (affected allocm()).
+  - Fix the ALLOCM_LG_ALIGN macro definition.
+  - Fix a heap dumping deadlock.
+  - Fix a "thread.arena" mallctl bug.
+
+* 2.1.0 (December 3, 2010)
+
+  This version incorporates some optimizations that can't quite be considered
+  bug fixes.
+
+  New features:
+  - Use Linux's mremap(2) for huge object reallocation when possible.
+  - Avoid locking in mallctl*() when possible.
+  - Add the "thread.[de]allocatedp" mallctl's.
+  - Convert the manual page source from roff to DocBook, and generate both roff
+    and HTML manuals.
+
+  Bug fixes:
+  - Fix a crash due to incorrect bootstrap ordering.  This only impacted
+    --enable-debug --enable-dss configurations.
+  - Fix a minor statistics bug for mallctl("swap.avail", ...).
+
+* 2.0.1 (October 29, 2010)
+
+  Bug fixes:
+  - Fix a race condition in heap profiling that could cause undefined behavior
+    if "opt.prof_accum" were disabled.
+  - Add missing mutex unlocks for some OOM error paths in the heap profiling
+    code.
+  - Fix a compilation error for non-C99 builds.
+
+* 2.0.0 (October 24, 2010)
+
+  This version focuses on the experimental *allocm() API, and on improved
+  run-time configuration/introspection.  Nonetheless, numerous performance
+  improvements are also included.
+
+  New features:
+  - Implement the experimental {,r,s,d}allocm() API, which provides a superset
+    of the functionality available via malloc(), calloc(), posix_memalign(),
+    realloc(), malloc_usable_size(), and free().  These functions can be used to
+    allocate/reallocate aligned zeroed memory, ask for optional extra memory
+    during reallocation, prevent object movement during reallocation, etc.
+  - Replace JEMALLOC_OPTIONS/JEMALLOC_PROF_PREFIX with MALLOC_CONF, which is
+    more human-readable, and more flexible.  For example:
+      JEMALLOC_OPTIONS=AJP
+    is now:
+      MALLOC_CONF=abort:true,fill:true,stats_print:true
+  - Port to Apple OS X.  Sponsored by Mozilla.
+  - Make it possible for the application to control thread-->arena mappings via
+    the "thread.arena" mallctl.
+  - Add compile-time support for all TLS-related functionality via pthreads TSD.
+    This is mainly of interest for OS X, which does not support TLS, but has a
+    TSD implementation with similar performance.
+  - Override memalign() and valloc() if they are provided by the system.
+  - Add the "arenas.purge" mallctl, which can be used to synchronously purge all
+    dirty unused pages.
+  - Make cumulative heap profiling data optional, so that it is possible to
+    limit the amount of memory consumed by heap profiling data structures.
+  - Add per thread allocation counters that can be accessed via the
+    "thread.allocated" and "thread.deallocated" mallctls.
+
+  Incompatible changes:
+  - Remove JEMALLOC_OPTIONS and malloc_options (see MALLOC_CONF above).
+  - Increase default backtrace depth from 4 to 128 for heap profiling.
+  - Disable interval-based profile dumps by default.
+
+  Bug fixes:
+  - Remove bad assertions in fork handler functions.  These assertions could
+    cause aborts for some combinations of configure settings.
+  - Fix strerror_r() usage to deal with non-standard semantics in GNU libc.
+  - Fix leak context reporting.  This bug tended to cause the number of contexts
+    to be underreported (though the reported number of objects and bytes were
+    correct).
+  - Fix a realloc() bug for large in-place growing reallocation.  This bug could
+    cause memory corruption, but it was hard to trigger.
+  - Fix an allocation bug for small allocations that could be triggered if
+    multiple threads raced to create a new run of backing pages.
+  - Enhance the heap profiler to trigger samples based on usable size, rather
+    than request size.
+  - Fix a heap profiling bug due to sometimes losing track of requested object
+    size for sampled objects.
+
+* 1.0.3 (August 12, 2010)
+
+  Bug fixes:
+  - Fix the libunwind-based implementation of stack backtracing (used for heap
+    profiling).  This bug could cause zero-length backtraces to be reported.
+  - Add a missing mutex unlock in library initialization code.  If multiple
+    threads raced to initialize malloc, some of them could end up permanently
+    blocked.
+
+* 1.0.2 (May 11, 2010)
+
+  Bug fixes:
+  - Fix junk filling of large objects, which could cause memory corruption.
+  - Add MAP_NORESERVE support for chunk mapping, because otherwise virtual
+    memory limits could cause swap file configuration to fail.  Contributed by
+    Jordan DeLong.
+
+* 1.0.1 (April 14, 2010)
+
+  Bug fixes:
+  - Fix compilation when --enable-fill is specified.
+  - Fix threads-related profiling bugs that affected accuracy and caused memory
+    to be leaked during thread exit.
+  - Fix dirty page purging race conditions that could cause crashes.
+  - Fix crash in tcache flushing code during thread destruction.
+
+* 1.0.0 (April 11, 2010)
+
+  This release focuses on speed and run-time introspection.  Numerous
+  algorithmic improvements make this release substantially faster than its
+  predecessors.
+
+  New features:
+  - Implement autoconf-based configuration system.
+  - Add mallctl*(), for the purposes of introspection and run-time
+    configuration.
+  - Make it possible for the application to manually flush a thread's cache, via
+    the "tcache.flush" mallctl.
+  - Base maximum dirty page count on proportion of active memory.
+  - Compute various addtional run-time statistics, including per size class
+    statistics for large objects.
+  - Expose malloc_stats_print(), which can be called repeatedly by the
+    application.
+  - Simplify the malloc_message() signature to only take one string argument,
+    and incorporate an opaque data pointer argument for use by the application
+    in combination with malloc_stats_print().
+  - Add support for allocation backed by one or more swap files, and allow the
+    application to disable over-commit if swap files are in use.
+  - Implement allocation profiling and leak checking.
+
+  Removed features:
+  - Remove the dynamic arena rebalancing code, since thread-specific caching
+    reduces its utility.
+
+  Bug fixes:
+  - Modify chunk allocation to work when address space layout randomization
+    (ASLR) is in use.
+  - Fix thread cleanup bugs related to TLS destruction.
+  - Handle 0-size allocation requests in posix_memalign().
+  - Fix a chunk leak.  The leaked chunks were never touched, so this impacted
+    virtual memory usage, but not physical memory usage.
+
+* linux_2008082[78]a (August 27/28, 2008)
+
+  These snapshot releases are the simple result of incorporating Linux-specific
+  support into the FreeBSD malloc sources.
+
+--------------------------------------------------------------------------------
+vim:filetype=text:textwidth=80

contrib/jemalloc/FREEBSD-Xlist

@@ -0,0 +1,23 @@
+$FreeBSD$
+.git
+.gitignore
+FREEBSD-*
+INSTALL
+Makefile*
+README
+autogen.sh
+autom4te.cache/
+bin/
+config.*
+configure*
+doc/*.in
+doc/*.xml
+doc/*.xsl
+doc/*.html
+include/jemalloc/internal/jemalloc_internal.h.in
+include/jemalloc/internal/size_classes.sh
+include/jemalloc/jemalloc.h.in
+include/jemalloc/jemalloc_defs.h.in
+install-sh
+src/zone.c
+test/

contrib/jemalloc/FREEBSD-diffs

@@ -0,0 +1,247 @@
+diff --git a/doc/jemalloc.xml.in b/doc/jemalloc.xml.in
+index 98d0ba4..23d2152 100644
+--- a/doc/jemalloc.xml.in
++++ b/doc/jemalloc.xml.in
+@@ -51,12 +51,23 @@
+     <para>This manual describes jemalloc @jemalloc_version@.  More information
+     can be found at the <ulink
+     url="http://www.canonware.com/jemalloc/">jemalloc website</ulink>.</para>
++
++    <para>The following configuration options are enabled in libc's built-in
++    jemalloc: <option>--enable-dss</option>,
++    <option>--enable-experimental</option>, <option>--enable-fill</option>,
++    <option>--enable-lazy-lock</option>, <option>--enable-munmap</option>,
++    <option>--enable-stats</option>, <option>--enable-tcache</option>,
++    <option>--enable-tls</option>, <option>--enable-utrace</option>, and
++    <option>--enable-xmalloc</option>.  Additionally,
++    <option>--enable-debug</option> is enabled in development versions of
++    FreeBSD (controlled by the <constant>MALLOC_PRODUCTION</constant> make
++    variable).</para>
+   </refsect1>
+   <refsynopsisdiv>
+     <title>SYNOPSIS</title>
+     <funcsynopsis>
+       <funcsynopsisinfo>#include &lt;<filename class="headerfile">stdlib.h</filename>&gt;
+-#include &lt;<filename class="headerfile">jemalloc/jemalloc.h</filename>&gt;</funcsynopsisinfo>
++#include &lt;<filename class="headerfile">malloc_np.h</filename>&gt;</funcsynopsisinfo>
+       <refsect2>
+         <title>Standard API</title>
+         <funcprototype>
+@@ -2080,4 +2091,16 @@ malloc_conf = "lg_chunk:24";]]></programlisting></para>
+     <para>The <function>posix_memalign<parameter/></function> function conforms
+     to IEEE Std 1003.1-2001 (&ldquo;POSIX.1&rdquo;).</para>
+   </refsect1>
++  <refsect1 id="history">
++    <title>HISTORY</title>
++    <para>The <function>malloc_usable_size<parameter/></function> and
++    <function>posix_memalign<parameter/></function> functions first appeared in
++    FreeBSD 7.0.</para>
++
++    <para>The <function>aligned_alloc<parameter/></function>,
++    <function>malloc_stats_print<parameter/></function>,
++    <function>mallctl*<parameter/></function>, and
++    <function>*allocm<parameter/></function> functions first appeared in
++    FreeBSD 10.0.</para>
++  </refsect1>
+ </refentry>
+diff --git a/include/jemalloc/internal/jemalloc_internal.h.in b/include/jemalloc/internal/jemalloc_internal.h.in
+index aa21aa5..e0f5fed 100644
+--- a/include/jemalloc/internal/jemalloc_internal.h.in
++++ b/include/jemalloc/internal/jemalloc_internal.h.in
+@@ -1,3 +1,6 @@
++#include "libc_private.h"
++#include "namespace.h"
++
+ #include <sys/mman.h>
+ #include <sys/param.h>
+ #include <sys/syscall.h>
+@@ -33,6 +36,9 @@
+ #include <pthread.h>
+ #include <math.h>
+ 
++#include "un-namespace.h"
++#include "libc_private.h"
++
+ #define	JEMALLOC_NO_DEMANGLE
+ #include "../jemalloc@install_suffix@.h"
+ 
+diff --git a/include/jemalloc/internal/mutex.h b/include/jemalloc/internal/mutex.h
+index c46feee..d7133f4 100644
+--- a/include/jemalloc/internal/mutex.h
++++ b/include/jemalloc/internal/mutex.h
+@@ -39,8 +39,6 @@ struct malloc_mutex_s {
+ 
+ #ifdef JEMALLOC_LAZY_LOCK
+ extern bool isthreaded;
+-#else
+-#  define isthreaded true
+ #endif
+ 
+ bool	malloc_mutex_init(malloc_mutex_t *mutex);
+diff --git a/include/jemalloc/jemalloc.h.in b/include/jemalloc/jemalloc.h.in
+index f0581db..f26d8bc 100644
+--- a/include/jemalloc/jemalloc.h.in
++++ b/include/jemalloc/jemalloc.h.in
+@@ -15,6 +15,7 @@ extern "C" {
+ #define	JEMALLOC_VERSION_GID "@jemalloc_version_gid@"
+ 
+ #include "jemalloc_defs@install_suffix@.h"
++#include "jemalloc_FreeBSD.h"
+ 
+ #ifdef JEMALLOC_EXPERIMENTAL
+ #define	ALLOCM_LG_ALIGN(la)	(la)
+diff --git a/include/jemalloc/jemalloc_FreeBSD.h b/include/jemalloc/jemalloc_FreeBSD.h
+new file mode 100644
+index 0000000..2c5797f
+--- /dev/null
++++ b/include/jemalloc/jemalloc_FreeBSD.h
+@@ -0,0 +1,76 @@
++/*
++ * Override settings that were generated in jemalloc_defs.h as necessary.
++ */
++
++#undef JEMALLOC_OVERRIDE_VALLOC
++
++#ifndef MALLOC_PRODUCTION
++#define	JEMALLOC_DEBUG
++#endif
++
++/*
++ * The following are architecture-dependent, so conditionally define them for
++ * each supported architecture.
++ */
++#undef CPU_SPINWAIT
++#undef JEMALLOC_TLS_MODEL
++#undef STATIC_PAGE_SHIFT
++#undef LG_SIZEOF_PTR
++#undef LG_SIZEOF_INT
++#undef LG_SIZEOF_LONG
++#undef LG_SIZEOF_INTMAX_T
++
++#ifdef __i386__
++#  define LG_SIZEOF_PTR		2
++#  define CPU_SPINWAIT		__asm__ volatile("pause")
++#  define JEMALLOC_TLS_MODEL	__attribute__((tls_model("initial-exec")))
++#endif
++#ifdef __ia64__
++#  define LG_SIZEOF_PTR		3
++#endif
++#ifdef __sparc64__
++#  define LG_SIZEOF_PTR		3
++#  define JEMALLOC_TLS_MODEL	__attribute__((tls_model("initial-exec")))
++#endif
++#ifdef __amd64__
++#  define LG_SIZEOF_PTR		3
++#  define CPU_SPINWAIT		__asm__ volatile("pause")
++#  define JEMALLOC_TLS_MODEL	__attribute__((tls_model("initial-exec")))
++#endif
++#ifdef __arm__
++#  define LG_SIZEOF_PTR		2
++#endif
++#ifdef __mips__
++#  define LG_SIZEOF_PTR		2
++#endif
++#ifdef __powerpc64__
++#  define LG_SIZEOF_PTR		3
++#elif defined(__powerpc__)
++#  define LG_SIZEOF_PTR		2
++#endif
++
++#ifndef JEMALLOC_TLS_MODEL
++#  define JEMALLOC_TLS_MODEL	/* Default. */
++#endif
++#ifdef __clang__
++#  undef JEMALLOC_TLS_MODEL
++#  define JEMALLOC_TLS_MODEL	/* clang does not support tls_model yet. */
++#endif
++
++#define	STATIC_PAGE_SHIFT	PAGE_SHIFT
++#define	LG_SIZEOF_INT		2
++#define	LG_SIZEOF_LONG		LG_SIZEOF_PTR
++#define	LG_SIZEOF_INTMAX_T	3
++
++/* Disable lazy-lock machinery, mangle isthreaded, and adjust its type. */
++#undef JEMALLOC_LAZY_LOCK
++extern int __isthreaded;
++#define	isthreaded		((bool)__isthreaded)
++
++/* Mangle. */
++#define	open			_open
++#define	read			_read
++#define	write			_write
++#define	close			_close
++#define	pthread_mutex_lock	_pthread_mutex_lock
++#define	pthread_mutex_unlock	_pthread_mutex_unlock
+diff --git a/src/jemalloc.c b/src/jemalloc.c
+index 0decd8a..73fad29 100644
+--- a/src/jemalloc.c
++++ b/src/jemalloc.c
+@@ -8,6 +8,9 @@ malloc_tsd_data(, arenas, arena_t *, NULL)
+ malloc_tsd_data(, thread_allocated, thread_allocated_t,
+     THREAD_ALLOCATED_INITIALIZER)
+ 
++const char	*__malloc_options_1_0;
++__sym_compat(_malloc_options, __malloc_options_1_0, FBSD_1.0);
++
+ /* Runtime configuration options. */
+ const char	*je_malloc_conf JEMALLOC_ATTR(visibility("default"));
+ #ifdef JEMALLOC_DEBUG
+@@ -401,7 +404,8 @@ malloc_conf_init(void)
+ #endif
+ 			    ;
+ 
+-			if ((opts = getenv(envname)) != NULL) {
++			if (issetugid() == 0 && (opts = getenv(envname)) !=
++			    NULL) {
+ 				/*
+ 				 * Do nothing; opts is already initialized to
+ 				 * the value of the MALLOC_CONF environment
+diff --git a/src/mutex.c b/src/mutex.c
+index 4b8ce57..7be5fc9 100644
+--- a/src/mutex.c
++++ b/src/mutex.c
+@@ -63,6 +63,17 @@ pthread_create(pthread_t *__restrict thread,
+ #ifdef JEMALLOC_MUTEX_INIT_CB
+ int	_pthread_mutex_init_calloc_cb(pthread_mutex_t *mutex,
+     void *(calloc_cb)(size_t, size_t));
++
++__weak_reference(_pthread_mutex_init_calloc_cb_stub,
++    _pthread_mutex_init_calloc_cb);
++
++int
++_pthread_mutex_init_calloc_cb_stub(pthread_mutex_t *mutex,
++    void *(calloc_cb)(size_t, size_t))
++{
++
++	return (0);
++}
+ #endif
+ 
+ bool
+diff --git a/src/util.c b/src/util.c
+index 2aab61f..8b05042 100644
+--- a/src/util.c
++++ b/src/util.c
+@@ -60,6 +60,22 @@ wrtmessage(void *cbopaque, const char *s)
+ void	(*je_malloc_message)(void *, const char *s)
+     JEMALLOC_ATTR(visibility("default")) = wrtmessage;
+ 
++JEMALLOC_CATTR(visibility("hidden"), static)
++void
++wrtmessage_1_0(const char *s1, const char *s2, const char *s3,
++    const char *s4)
++{
++
++	wrtmessage(NULL, s1);
++	wrtmessage(NULL, s2);
++	wrtmessage(NULL, s3);
++	wrtmessage(NULL, s4);
++}
++
++void	(*__malloc_message_1_0)(const char *s1, const char *s2, const char *s3,
++    const char *s4) = wrtmessage_1_0;
++__sym_compat(_malloc_message, __malloc_message_1_0, FBSD_1.0);
++
+ /*
+  * glibc provides a non-standard strerror_r() when _GNU_SOURCE is defined, so
+  * provide a wrapper.

contrib/jemalloc/FREEBSD-upgrade

@@ -0,0 +1,122 @@
+#!/bin/sh
+# $FreeBSD$
+#
+# Usage: cd /usr/src/contrib/jemalloc
+#        ./FREEBSD-upgrade <command> [args]
+#
+# At least the following ports are required when importing jemalloc:
+# - devel/autoconf
+# - devel/git
+# - devel/gmake
+# - textproc/docbook-xsl
+#
+# The normal workflow for importing a new release is:
+#
+#   cd /usr/src/contrib/jemalloc
+#
+# Merge local changes that were made since the previous import:
+#
+#   ./FREEBSD-upgrade merge-changes
+#   ./FREEBSD-upgrade rediff
+#
+# Extract latest jemalloc release.
+#
+#   ./FREEBSD-upgrade extract
+#
+# Fix patch conflicts as necessary, then regenerate diffs to update line
+# offsets:
+#
+#   ./FREEBSD-upgrade rediff
+#   ./FREEBSD-upgrade extract
+#
+# Do multiple buildworld/installworld rounds.  If problems arise and patches
+# are needed, edit the code in ${work} as necessary, then:
+#
+#   ./FREEBSD-upgrade rediff
+#   ./FREEBSD-upgrade extract
+#
+# The rediff/extract order is important because rediff saves the local
+# changes, then extract blows away the work tree and re-creates it with the
+# diffs applied.
+#
+# Finally, to clean up:
+#
+#  ./FREEBSD-upgrade clean
+
+set -e
+
+if [ ! -x "FREEBSD-upgrade" ] ; then
+  echo "Run from within src/contrib/jemalloc/" >&2
+  exit 1
+fi
+
+src=`pwd`
+workname="jemalloc.git"
+work="${src}/../${workname}" # merge-changes expects ${workname} in "..".
+changes="${src}/FREEBSD-changes"
+
+do_extract() {
+  local rev=$1
+  # Clone.
+  rm -rf ${work}
+  git clone git://canonware.com/jemalloc.git ${work}
+  (
+    cd ${work}
+    if [ "x${rev}" != "x" ] ; then
+      # Use optional rev argument to check out a revision other than HEAD on
+      # master.
+      git checkout ${rev}
+    fi
+    # Apply diffs before generating files.
+    patch -p1 < "${src}/FREEBSD-diffs"
+    find . -name '*.orig' -delete
+    # Generate various files.
+    ./autogen.sh --enable-cc-silence --enable-dss --enable-xmalloc \
+      --enable-utrace --with-xslroot=/usr/local/share/xsl/docbook
+    gmake dist
+  )
+}
+
+do_diff() {
+  (cd ${work}; git add -A; git diff --cached) > FREEBSD-diffs
+}
+
+command=$1
+shift
+case "${command}" in
+  merge-changes) # Merge local changes that were made since the previous import.
+    rev=`cat VERSION |tr 'g' ' ' |awk '{print $2}'`
+    # Extract code corresponding to most recent import.
+    do_extract ${rev}
+    # Compute local differences to the upstream+patches and apply them.
+    (
+      cd ..
+      diff -ru -X ${src}/FREEBSD-Xlist ${workname} jemalloc > ${changes} || true
+    )
+    (
+      cd ${work}
+      patch -p1 < ${changes}
+      find . -name '*.orig' -delete
+    )
+    # Update diff.
+    do_diff
+    ;;
+  extract) # Extract upstream sources, apply patches, copy to contrib/jemalloc.
+    rev=$1
+    do_extract ${rev}
+    # Delete existing files so that cruft doesn't silently remain.
+    rm -rf ChangeLog COPYING VERSION doc include src
+    # Copy files over.
+    tar cf - -C ${work} -X FREEBSD-Xlist . |tar xvf -
+    ;;
+  rediff) # Regenerate diffs based on working tree.
+    do_diff
+    ;;
+  clean) # Remove working tree and temporary files.
+    rm -rf ${work} ${changes}
+    ;;
+  *)
+    echo "Unsupported command: \"${command}\"" >&2
+    exit 1
+    ;;
+esac

contrib/jemalloc/VERSION

@@ -0,0 +1 @@
+1.0.0-258-g9ef7f5dc34ff02f50d401e41c8d9a4a928e7c2aa

contrib/jemalloc/include/jemalloc/internal/arena.h

@@ -0,0 +1,685 @@
+/******************************************************************************/
+#ifdef JEMALLOC_H_TYPES
+
+/*
+ * RUN_MAX_OVRHD indicates maximum desired run header overhead.  Runs are sized
+ * as small as possible such that this setting is still honored, without
+ * violating other constraints.  The goal is to make runs as small as possible
+ * without exceeding a per run external fragmentation threshold.
+ *
+ * We use binary fixed point math for overhead computations, where the binary
+ * point is implicitly RUN_BFP bits to the left.
+ *
+ * Note that it is possible to set RUN_MAX_OVRHD low enough that it cannot be
+ * honored for some/all object sizes, since when heap profiling is enabled
+ * there is one pointer of header overhead per object (plus a constant).  This
+ * constraint is relaxed (ignored) for runs that are so small that the
+ * per-region overhead is greater than:
+ *
+ *   (RUN_MAX_OVRHD / (reg_interval << (3+RUN_BFP))
+ */
+#define	RUN_BFP			12
+/*                                    \/   Implicit binary fixed point. */
+#define	RUN_MAX_OVRHD		0x0000003dU
+#define	RUN_MAX_OVRHD_RELAX	0x00001800U
+
+/* Maximum number of regions in one run. */
+#define	LG_RUN_MAXREGS		11
+#define	RUN_MAXREGS		(1U << LG_RUN_MAXREGS)
+
+/*
+ * Minimum redzone size.  Redzones may be larger than this if necessary to
+ * preserve region alignment.
+ */
+#define	REDZONE_MINSIZE		16
+
+/*
+ * The minimum ratio of active:dirty pages per arena is computed as:
+ *
+ *   (nactive >> opt_lg_dirty_mult) >= ndirty
+ *
+ * So, supposing that opt_lg_dirty_mult is 5, there can be no less than 32
+ * times as many active pages as dirty pages.
+ */
+#define	LG_DIRTY_MULT_DEFAULT	5
+
+typedef struct arena_chunk_map_s arena_chunk_map_t;
+typedef struct arena_chunk_s arena_chunk_t;
+typedef struct arena_run_s arena_run_t;
+typedef struct arena_bin_info_s arena_bin_info_t;
+typedef struct arena_bin_s arena_bin_t;
+typedef struct arena_s arena_t;
+
+#endif /* JEMALLOC_H_TYPES */
+/******************************************************************************/
+#ifdef JEMALLOC_H_STRUCTS
+
+/* Each element of the chunk map corresponds to one page within the chunk. */
+struct arena_chunk_map_s {
+#ifndef JEMALLOC_PROF
+	/*
+	 * Overlay prof_ctx in order to allow it to be referenced by dead code.
+	 * Such antics aren't warranted for per arena data structures, but
+	 * chunk map overhead accounts for a percentage of memory, rather than
+	 * being just a fixed cost.
+	 */
+	union {
+#endif
+	union {
+		/*
+		 * Linkage for run trees.  There are two disjoint uses:
+		 *
+		 * 1) arena_t's runs_avail_{clean,dirty} trees.
+		 * 2) arena_run_t conceptually uses this linkage for in-use
+		 *    non-full runs, rather than directly embedding linkage.
+		 */
+		rb_node(arena_chunk_map_t)	rb_link;
+		/*
+		 * List of runs currently in purgatory.  arena_chunk_purge()
+		 * temporarily allocates runs that contain dirty pages while
+		 * purging, so that other threads cannot use the runs while the
+		 * purging thread is operating without the arena lock held.
+		 */
+		ql_elm(arena_chunk_map_t)	ql_link;
+	}				u;
+
+	/* Profile counters, used for large object runs. */
+	prof_ctx_t			*prof_ctx;
+#ifndef JEMALLOC_PROF
+	}; /* union { ... }; */
+#endif
+
+	/*
+	 * Run address (or size) and various flags are stored together.  The bit
+	 * layout looks like (assuming 32-bit system):
+	 *
+	 *   ???????? ???????? ????---- ----dula
+	 *
+	 * ? : Unallocated: Run address for first/last pages, unset for internal
+	 *                  pages.
+	 *     Small: Run page offset.
+	 *     Large: Run size for first page, unset for trailing pages.
+	 * - : Unused.
+	 * d : dirty?
+	 * u : unzeroed?
+	 * l : large?
+	 * a : allocated?
+	 *
+	 * Following are example bit patterns for the three types of runs.
+	 *
+	 * p : run page offset
+	 * s : run size
+	 * c : (binind+1) for size class (used only if prof_promote is true)
+	 * x : don't care
+	 * - : 0
+	 * + : 1
+	 * [DULA] : bit set
+	 * [dula] : bit unset
+	 *
+	 *   Unallocated (clean):
+	 *     ssssssss ssssssss ssss---- ----du-a
+	 *     xxxxxxxx xxxxxxxx xxxx---- -----Uxx
+	 *     ssssssss ssssssss ssss---- ----dU-a
+	 *
+	 *   Unallocated (dirty):
+	 *     ssssssss ssssssss ssss---- ----D--a
+	 *     xxxxxxxx xxxxxxxx xxxx---- ----xxxx
+	 *     ssssssss ssssssss ssss---- ----D--a
+	 *
+	 *   Small:
+	 *     pppppppp pppppppp pppp---- ----d--A
+	 *     pppppppp pppppppp pppp---- -------A
+	 *     pppppppp pppppppp pppp---- ----d--A
+	 *
+	 *   Large:
+	 *     ssssssss ssssssss ssss---- ----D-LA
+	 *     xxxxxxxx xxxxxxxx xxxx---- ----xxxx
+	 *     -------- -------- -------- ----D-LA
+	 *
+	 *   Large (sampled, size <= PAGE):
+	 *     ssssssss ssssssss sssscccc ccccD-LA
+	 *
+	 *   Large (not sampled, size == PAGE):
+	 *     ssssssss ssssssss ssss---- ----D-LA
+	 */
+	size_t				bits;
+#define	CHUNK_MAP_CLASS_SHIFT	4
+#define	CHUNK_MAP_CLASS_MASK	((size_t)0xff0U)
+#define	CHUNK_MAP_FLAGS_MASK	((size_t)0xfU)
+#define	CHUNK_MAP_DIRTY		((size_t)0x8U)
+#define	CHUNK_MAP_UNZEROED	((size_t)0x4U)
+#define	CHUNK_MAP_LARGE		((size_t)0x2U)
+#define	CHUNK_MAP_ALLOCATED	((size_t)0x1U)
+#define	CHUNK_MAP_KEY		CHUNK_MAP_ALLOCATED
+};
+typedef rb_tree(arena_chunk_map_t) arena_avail_tree_t;
+typedef rb_tree(arena_chunk_map_t) arena_run_tree_t;
+
+/* Arena chunk header. */
+struct arena_chunk_s {
+	/* Arena that owns the chunk. */
+	arena_t		*arena;
+
+	/* Linkage for the arena's chunks_dirty list. */
+	ql_elm(arena_chunk_t) link_dirty;
+
+	/*
+	 * True if the chunk is currently in the chunks_dirty list, due to
+	 * having at some point contained one or more dirty pages.  Removal
+	 * from chunks_dirty is lazy, so (dirtied && ndirty == 0) is possible.
+	 */
+	bool		dirtied;
+
+	/* Number of dirty pages. */
+	size_t		ndirty;
+
+	/*
+	 * Map of pages within chunk that keeps track of free/large/small.  The
+	 * first map_bias entries are omitted, since the chunk header does not
+	 * need to be tracked in the map.  This omission saves a header page
+	 * for common chunk sizes (e.g. 4 MiB).
+	 */
+	arena_chunk_map_t map[1]; /* Dynamically sized. */
+};
+typedef rb_tree(arena_chunk_t) arena_chunk_tree_t;
+
+struct arena_run_s {
+	/* Bin this run is associated with. */
+	arena_bin_t	*bin;
+
+	/* Index of next region that has never been allocated, or nregs. */
+	uint32_t	nextind;
+
+	/* Number of free regions in run. */
+	unsigned	nfree;
+};
+
+/*
+ * Read-only information associated with each element of arena_t's bins array
+ * is stored separately, partly to reduce memory usage (only one copy, rather
+ * than one per arena), but mainly to avoid false cacheline sharing.
+ *
+ * Each run has the following layout:
+ *
+ *               /--------------------\
+ *               | arena_run_t header |
+ *               | ...                |
+ * bitmap_offset | bitmap             |
+ *               | ...                |
+ *   ctx0_offset | ctx map            |
+ *               | ...                |
+ *               |--------------------|
+ *               | redzone            |
+ *   reg0_offset | region 0           |
+ *               | redzone            |
+ *               |--------------------| \
+ *               | redzone            | |
+ *               | region 1           |  > reg_interval
+ *               | redzone            | /
+ *               |--------------------|
+ *               | ...                |
+ *               | ...                |
+ *               | ...                |
+ *               |--------------------|
+ *               | redzone            |
+ *               | region nregs-1     |
+ *               | redzone            |
+ *               |--------------------|
+ *               | alignment pad?     |
+ *               \--------------------/
+ *
+ * reg_interval has at least the same minimum alignment as reg_size; this
+ * preserves the alignment constraint that sa2u() depends on.  Alignment pad is
+ * either 0 or redzone_size; it is present only if needed to align reg0_offset.
+ */
+struct arena_bin_info_s {
+	/* Size of regions in a run for this bin's size class. */
+	size_t		reg_size;
+
+	/* Redzone size. */
+	size_t		redzone_size;
+
+	/* Interval between regions (reg_size + (redzone_size << 1)). */
+	size_t		reg_interval;
+
+	/* Total size of a run for this bin's size class. */
+	size_t		run_size;
+
+	/* Total number of regions in a run for this bin's size class. */
+	uint32_t	nregs;
+
+	/*
+	 * Offset of first bitmap_t element in a run header for this bin's size
+	 * class.
+	 */
+	uint32_t	bitmap_offset;
+
+	/*
+	 * Metadata used to manipulate bitmaps for runs associated with this
+	 * bin.
+	 */
+	bitmap_info_t	bitmap_info;
+
+	/*
+	 * Offset of first (prof_ctx_t *) in a run header for this bin's size
+	 * class, or 0 if (config_prof == false || opt_prof == false).
+	 */
+	uint32_t	ctx0_offset;
+
+	/* Offset of first region in a run for this bin's size class. */
+	uint32_t	reg0_offset;
+};
+
+struct arena_bin_s {
+	/*
+	 * All operations on runcur, runs, and stats require that lock be
+	 * locked.  Run allocation/deallocation are protected by the arena lock,
+	 * which may be acquired while holding one or more bin locks, but not
+	 * vise versa.
+	 */
+	malloc_mutex_t	lock;
+
+	/*
+	 * Current run being used to service allocations of this bin's size
+	 * class.
+	 */
+	arena_run_t	*runcur;
+
+	/*
+	 * Tree of non-full runs.  This tree is used when looking for an
+	 * existing run when runcur is no longer usable.  We choose the
+	 * non-full run that is lowest in memory; this policy tends to keep
+	 * objects packed well, and it can also help reduce the number of
+	 * almost-empty chunks.
+	 */
+	arena_run_tree_t runs;
+
+	/* Bin statistics. */
+	malloc_bin_stats_t stats;
+};
+
+struct arena_s {
+	/* This arena's index within the arenas array. */
+	unsigned		ind;
+
+	/*
+	 * Number of threads currently assigned to this arena.  This field is
+	 * protected by arenas_lock.
+	 */
+	unsigned		nthreads;
+
+	/*
+	 * There are three classes of arena operations from a locking
+	 * perspective:
+	 * 1) Thread asssignment (modifies nthreads) is protected by
+	 *    arenas_lock.
+	 * 2) Bin-related operations are protected by bin locks.
+	 * 3) Chunk- and run-related operations are protected by this mutex.
+	 */
+	malloc_mutex_t		lock;
+
+	arena_stats_t		stats;
+	/*
+	 * List of tcaches for extant threads associated with this arena.
+	 * Stats from these are merged incrementally, and at exit.
+	 */
+	ql_head(tcache_t)	tcache_ql;
+
+	uint64_t		prof_accumbytes;
+
+	/* List of dirty-page-containing chunks this arena manages. */
+	ql_head(arena_chunk_t)	chunks_dirty;
+
+	/*
+	 * In order to avoid rapid chunk allocation/deallocation when an arena
+	 * oscillates right on the cusp of needing a new chunk, cache the most
+	 * recently freed chunk.  The spare is left in the arena's chunk trees
+	 * until it is deleted.
+	 *
+	 * There is one spare chunk per arena, rather than one spare total, in
+	 * order to avoid interactions between multiple threads that could make
+	 * a single spare inadequate.
+	 */
+	arena_chunk_t		*spare;
+
+	/* Number of pages in active runs. */
+	size_t			nactive;
+
+	/*
+	 * Current count of pages within unused runs that are potentially
+	 * dirty, and for which madvise(... MADV_DONTNEED) has not been called.
+	 * By tracking this, we can institute a limit on how much dirty unused
+	 * memory is mapped for each arena.
+	 */
+	size_t			ndirty;
+
+	/*
+	 * Approximate number of pages being purged.  It is possible for
+	 * multiple threads to purge dirty pages concurrently, and they use
+	 * npurgatory to indicate the total number of pages all threads are
+	 * attempting to purge.
+	 */
+	size_t			npurgatory;
+
+	/*
+	 * Size/address-ordered trees of this arena's available runs.  The trees
+	 * are used for first-best-fit run allocation.  The dirty tree contains
+	 * runs with dirty pages (i.e. very likely to have been touched and
+	 * therefore have associated physical pages), whereas the clean tree
+	 * contains runs with pages that either have no associated physical
+	 * pages, or have pages that the kernel may recycle at any time due to
+	 * previous madvise(2) calls.  The dirty tree is used in preference to
+	 * the clean tree for allocations, because using dirty pages reduces
+	 * the amount of dirty purging necessary to keep the active:dirty page
+	 * ratio below the purge threshold.
+	 */
+	arena_avail_tree_t	runs_avail_clean;
+	arena_avail_tree_t	runs_avail_dirty;
+
+	/* bins is used to store trees of free regions. */
+	arena_bin_t		bins[NBINS];
+};
+
+#endif /* JEMALLOC_H_STRUCTS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_EXTERNS
+
+extern ssize_t	opt_lg_dirty_mult;
+/*
+ * small_size2bin is a compact lookup table that rounds request sizes up to
+ * size classes.  In order to reduce cache footprint, the table is compressed,
+ * and all accesses are via the SMALL_SIZE2BIN macro.
+ */
+extern uint8_t const	small_size2bin[];
+#define	SMALL_SIZE2BIN(s)	(small_size2bin[(s-1) >> LG_TINY_MIN])
+
+extern arena_bin_info_t	arena_bin_info[NBINS];
+
+/* Number of large size classes. */
+#define			nlclasses (chunk_npages - map_bias)
+
+void	arena_purge_all(arena_t *arena);
+void	arena_prof_accum(arena_t *arena, uint64_t accumbytes);
+void	arena_tcache_fill_small(arena_t *arena, tcache_bin_t *tbin,
+    size_t binind, uint64_t prof_accumbytes);
+void	arena_alloc_junk_small(void *ptr, arena_bin_info_t *bin_info,
+    bool zero);
+void	arena_dalloc_junk_small(void *ptr, arena_bin_info_t *bin_info);
+void	*arena_malloc_small(arena_t *arena, size_t size, bool zero);
+void	*arena_malloc_large(arena_t *arena, size_t size, bool zero);
+void	*arena_palloc(arena_t *arena, size_t size, size_t alignment, bool zero);
+size_t	arena_salloc(const void *ptr, bool demote);
+void	arena_prof_promoted(const void *ptr, size_t size);
+void	arena_dalloc_bin(arena_t *arena, arena_chunk_t *chunk, void *ptr,
+    arena_chunk_map_t *mapelm);
+void	arena_dalloc_large(arena_t *arena, arena_chunk_t *chunk, void *ptr);
+void	arena_stats_merge(arena_t *arena, size_t *nactive, size_t *ndirty,
+    arena_stats_t *astats, malloc_bin_stats_t *bstats,
+    malloc_large_stats_t *lstats);
+void	*arena_ralloc_no_move(void *ptr, size_t oldsize, size_t size,
+    size_t extra, bool zero);
+void	*arena_ralloc(void *ptr, size_t oldsize, size_t size, size_t extra,
+    size_t alignment, bool zero, bool try_tcache);
+bool	arena_new(arena_t *arena, unsigned ind);
+void	arena_boot(void);
+void	arena_prefork(arena_t *arena);
+void	arena_postfork_parent(arena_t *arena);
+void	arena_postfork_child(arena_t *arena);
+
+#endif /* JEMALLOC_H_EXTERNS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_INLINES
+
+#ifndef JEMALLOC_ENABLE_INLINE
+size_t	arena_bin_index(arena_t *arena, arena_bin_t *bin);
+unsigned	arena_run_regind(arena_run_t *run, arena_bin_info_t *bin_info,
+    const void *ptr);
+prof_ctx_t	*arena_prof_ctx_get(const void *ptr);
+void	arena_prof_ctx_set(const void *ptr, prof_ctx_t *ctx);
+void	*arena_malloc(arena_t *arena, size_t size, bool zero, bool try_tcache);
+void	arena_dalloc(arena_t *arena, arena_chunk_t *chunk, void *ptr,
+    bool try_tcache);
+#endif
+
+#if (defined(JEMALLOC_ENABLE_INLINE) || defined(JEMALLOC_ARENA_C_))
+JEMALLOC_INLINE size_t
+arena_bin_index(arena_t *arena, arena_bin_t *bin)
+{
+	size_t binind = bin - arena->bins;
+	assert(binind < NBINS);
+	return (binind);
+}
+
+JEMALLOC_INLINE unsigned
+arena_run_regind(arena_run_t *run, arena_bin_info_t *bin_info, const void *ptr)
+{
+	unsigned shift, diff, regind;
+	size_t interval;
+
+	/*
+	 * Freeing a pointer lower than region zero can cause assertion
+	 * failure.
+	 */
+	assert((uintptr_t)ptr >= (uintptr_t)run +
+	    (uintptr_t)bin_info->reg0_offset);
+
+	/*
+	 * Avoid doing division with a variable divisor if possible.  Using
+	 * actual division here can reduce allocator throughput by over 20%!
+	 */
+	diff = (unsigned)((uintptr_t)ptr - (uintptr_t)run -
+	    bin_info->reg0_offset);
+
+	/* Rescale (factor powers of 2 out of the numerator and denominator). */
+	interval = bin_info->reg_interval;
+	shift = ffs(interval) - 1;
+	diff >>= shift;
+	interval >>= shift;
+
+	if (interval == 1) {
+		/* The divisor was a power of 2. */
+		regind = diff;
+	} else {
+		/*
+		 * To divide by a number D that is not a power of two we
+		 * multiply by (2^21 / D) and then right shift by 21 positions.
+		 *
+		 *   X / D
+		 *
+		 * becomes
+		 *
+		 *   (X * interval_invs[D - 3]) >> SIZE_INV_SHIFT
+		 *
+		 * We can omit the first three elements, because we never
+		 * divide by 0, and 1 and 2 are both powers of two, which are
+		 * handled above.
+		 */
+#define	SIZE_INV_SHIFT	((sizeof(unsigned) << 3) - LG_RUN_MAXREGS)
+#define	SIZE_INV(s)	(((1U << SIZE_INV_SHIFT) / (s)) + 1)
+		static const unsigned interval_invs[] = {
+		    SIZE_INV(3),
+		    SIZE_INV(4), SIZE_INV(5), SIZE_INV(6), SIZE_INV(7),
+		    SIZE_INV(8), SIZE_INV(9), SIZE_INV(10), SIZE_INV(11),
+		    SIZE_INV(12), SIZE_INV(13), SIZE_INV(14), SIZE_INV(15),
+		    SIZE_INV(16), SIZE_INV(17), SIZE_INV(18), SIZE_INV(19),
+		    SIZE_INV(20), SIZE_INV(21), SIZE_INV(22), SIZE_INV(23),
+		    SIZE_INV(24), SIZE_INV(25), SIZE_INV(26), SIZE_INV(27),
+		    SIZE_INV(28), SIZE_INV(29), SIZE_INV(30), SIZE_INV(31)
+		};
+
+		if (interval <= ((sizeof(interval_invs) / sizeof(unsigned)) +
+		    2)) {
+			regind = (diff * interval_invs[interval - 3]) >>
+			    SIZE_INV_SHIFT;
+		} else
+			regind = diff / interval;
+#undef SIZE_INV
+#undef SIZE_INV_SHIFT
+	}
+	assert(diff == regind * interval);
+	assert(regind < bin_info->nregs);
+
+	return (regind);
+}
+
+JEMALLOC_INLINE prof_ctx_t *
+arena_prof_ctx_get(const void *ptr)
+{
+	prof_ctx_t *ret;
+	arena_chunk_t *chunk;
+	size_t pageind, mapbits;
+
+	cassert(config_prof);
+	assert(ptr != NULL);
+	assert(CHUNK_ADDR2BASE(ptr) != ptr);
+
+	chunk = (arena_chunk_t *)CHUNK_ADDR2BASE(ptr);
+	pageind = ((uintptr_t)ptr - (uintptr_t)chunk) >> LG_PAGE;
+	mapbits = chunk->map[pageind-map_bias].bits;
+	assert((mapbits & CHUNK_MAP_ALLOCATED) != 0);
+	if ((mapbits & CHUNK_MAP_LARGE) == 0) {
+		if (prof_promote)
+			ret = (prof_ctx_t *)(uintptr_t)1U;
+		else {
+			arena_run_t *run = (arena_run_t *)((uintptr_t)chunk +
+			    (uintptr_t)((pageind - (mapbits >> LG_PAGE)) <<
+			    LG_PAGE));
+			size_t binind = arena_bin_index(chunk->arena, run->bin);
+			arena_bin_info_t *bin_info = &arena_bin_info[binind];
+			unsigned regind;
+
+			regind = arena_run_regind(run, bin_info, ptr);
+			ret = *(prof_ctx_t **)((uintptr_t)run +
+			    bin_info->ctx0_offset + (regind *
+			    sizeof(prof_ctx_t *)));
+		}
+	} else
+		ret = chunk->map[pageind-map_bias].prof_ctx;
+
+	return (ret);
+}
+
+JEMALLOC_INLINE void
+arena_prof_ctx_set(const void *ptr, prof_ctx_t *ctx)
+{
+	arena_chunk_t *chunk;
+	size_t pageind, mapbits;
+
+	cassert(config_prof);
+	assert(ptr != NULL);
+	assert(CHUNK_ADDR2BASE(ptr) != ptr);
+
+	chunk = (arena_chunk_t *)CHUNK_ADDR2BASE(ptr);
+	pageind = ((uintptr_t)ptr - (uintptr_t)chunk) >> LG_PAGE;
+	mapbits = chunk->map[pageind-map_bias].bits;
+	assert((mapbits & CHUNK_MAP_ALLOCATED) != 0);
+	if ((mapbits & CHUNK_MAP_LARGE) == 0) {
+		if (prof_promote == false) {
+			arena_run_t *run = (arena_run_t *)((uintptr_t)chunk +
+			    (uintptr_t)((pageind - (mapbits >> LG_PAGE)) <<
+			    LG_PAGE));
+			arena_bin_t *bin = run->bin;
+			size_t binind;
+			arena_bin_info_t *bin_info;
+			unsigned regind;
+
+			binind = arena_bin_index(chunk->arena, bin);
+			bin_info = &arena_bin_info[binind];
+			regind = arena_run_regind(run, bin_info, ptr);
+
+			*((prof_ctx_t **)((uintptr_t)run + bin_info->ctx0_offset
+			    + (regind * sizeof(prof_ctx_t *)))) = ctx;
+		} else
+			assert((uintptr_t)ctx == (uintptr_t)1U);
+	} else
+		chunk->map[pageind-map_bias].prof_ctx = ctx;
+}
+
+JEMALLOC_INLINE void *
+arena_malloc(arena_t *arena, size_t size, bool zero, bool try_tcache)
+{
+	tcache_t *tcache;
+
+	assert(size != 0);
+	assert(size <= arena_maxclass);
+
+	if (size <= SMALL_MAXCLASS) {
+		if (try_tcache && (tcache = tcache_get(true)) != NULL)
+			return (tcache_alloc_small(tcache, size, zero));
+		else {
+			return (arena_malloc_small(choose_arena(arena), size,
+			    zero));
+		}
+	} else {
+		/*
+		 * Initialize tcache after checking size in order to avoid
+		 * infinite recursion during tcache initialization.
+		 */
+		if (try_tcache && size <= tcache_maxclass && (tcache =
+		    tcache_get(true)) != NULL)
+			return (tcache_alloc_large(tcache, size, zero));
+		else {
+			return (arena_malloc_large(choose_arena(arena), size,
+			    zero));
+		}
+	}
+}
+
+JEMALLOC_INLINE void
+arena_dalloc(arena_t *arena, arena_chunk_t *chunk, void *ptr, bool try_tcache)
+{
+	size_t pageind;
+	arena_chunk_map_t *mapelm;
+	tcache_t *tcache;
+
+	assert(arena != NULL);
+	assert(chunk->arena == arena);
+	assert(ptr != NULL);
+	assert(CHUNK_ADDR2BASE(ptr) != ptr);
+
+	pageind = ((uintptr_t)ptr - (uintptr_t)chunk) >> LG_PAGE;
+	mapelm = &chunk->map[pageind-map_bias];
+	assert((mapelm->bits & CHUNK_MAP_ALLOCATED) != 0);
+	if ((mapelm->bits & CHUNK_MAP_LARGE) == 0) {
+		/* Small allocation. */
+		if (try_tcache && (tcache = tcache_get(false)) != NULL)
+			tcache_dalloc_small(tcache, ptr);
+		else {
+			arena_run_t *run;
+			arena_bin_t *bin;
+
+			run = (arena_run_t *)((uintptr_t)chunk +
+			    (uintptr_t)((pageind - (mapelm->bits >> LG_PAGE)) <<
+			    LG_PAGE));
+			bin = run->bin;
+			if (config_debug) {
+				size_t binind = arena_bin_index(arena, bin);
+				UNUSED arena_bin_info_t *bin_info =
+				    &arena_bin_info[binind];
+				assert(((uintptr_t)ptr - ((uintptr_t)run +
+				    (uintptr_t)bin_info->reg0_offset)) %
+				    bin_info->reg_interval == 0);
+			}
+			malloc_mutex_lock(&bin->lock);
+			arena_dalloc_bin(arena, chunk, ptr, mapelm);
+			malloc_mutex_unlock(&bin->lock);
+		}
+	} else {
+		size_t size = mapelm->bits & ~PAGE_MASK;
+
+		assert(((uintptr_t)ptr & PAGE_MASK) == 0);
+
+		if (try_tcache && size <= tcache_maxclass && (tcache =
+		    tcache_get(false)) != NULL) {
+			tcache_dalloc_large(tcache, ptr, size);
+		} else {
+			malloc_mutex_lock(&arena->lock);
+			arena_dalloc_large(arena, chunk, ptr);
+			malloc_mutex_unlock(&arena->lock);
+		}
+	}
+}
+#endif
+
+#endif /* JEMALLOC_H_INLINES */
+/******************************************************************************/

contrib/jemalloc/include/jemalloc/internal/atomic.h

@@ -0,0 +1,240 @@
+/******************************************************************************/
+#ifdef JEMALLOC_H_TYPES
+
+#endif /* JEMALLOC_H_TYPES */
+/******************************************************************************/
+#ifdef JEMALLOC_H_STRUCTS
+
+#endif /* JEMALLOC_H_STRUCTS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_EXTERNS
+
+#define	atomic_read_uint64(p)	atomic_add_uint64(p, 0)
+#define	atomic_read_uint32(p)	atomic_add_uint32(p, 0)
+#define	atomic_read_z(p)	atomic_add_z(p, 0)
+#define	atomic_read_u(p)	atomic_add_u(p, 0)
+
+#endif /* JEMALLOC_H_EXTERNS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_INLINES
+
+#ifndef JEMALLOC_ENABLE_INLINE
+uint64_t	atomic_add_uint64(uint64_t *p, uint64_t x);
+uint64_t	atomic_sub_uint64(uint64_t *p, uint64_t x);
+uint32_t	atomic_add_uint32(uint32_t *p, uint32_t x);
+uint32_t	atomic_sub_uint32(uint32_t *p, uint32_t x);
+size_t	atomic_add_z(size_t *p, size_t x);
+size_t	atomic_sub_z(size_t *p, size_t x);
+unsigned	atomic_add_u(unsigned *p, unsigned x);
+unsigned	atomic_sub_u(unsigned *p, unsigned x);
+#endif
+
+#if (defined(JEMALLOC_ENABLE_INLINE) || defined(JEMALLOC_ATOMIC_C_))
+/******************************************************************************/
+/* 64-bit operations. */
+#ifdef __GCC_HAVE_SYNC_COMPARE_AND_SWAP_8
+JEMALLOC_INLINE uint64_t
+atomic_add_uint64(uint64_t *p, uint64_t x)
+{
+
+	return (__sync_add_and_fetch(p, x));
+}
+
+JEMALLOC_INLINE uint64_t
+atomic_sub_uint64(uint64_t *p, uint64_t x)
+{
+
+	return (__sync_sub_and_fetch(p, x));
+}
+#elif (defined(JEMALLOC_OSATOMIC))
+JEMALLOC_INLINE uint64_t
+atomic_add_uint64(uint64_t *p, uint64_t x)
+{
+
+	return (OSAtomicAdd64((int64_t)x, (int64_t *)p));
+}
+
+JEMALLOC_INLINE uint64_t
+atomic_sub_uint64(uint64_t *p, uint64_t x)
+{
+
+	return (OSAtomicAdd64(-((int64_t)x), (int64_t *)p));
+}
+#elif (defined(__amd64__) || defined(__x86_64__))
+JEMALLOC_INLINE uint64_t
+atomic_add_uint64(uint64_t *p, uint64_t x)
+{
+
+	asm volatile (
+	    "lock; xaddq %0, %1;"
+	    : "+r" (x), "=m" (*p) /* Outputs. */
+	    : "m" (*p) /* Inputs. */
+	    );
+
+	return (x);
+}
+
+JEMALLOC_INLINE uint64_t
+atomic_sub_uint64(uint64_t *p, uint64_t x)
+{
+
+	x = (uint64_t)(-(int64_t)x);
+	asm volatile (
+	    "lock; xaddq %0, %1;"
+	    : "+r" (x), "=m" (*p) /* Outputs. */
+	    : "m" (*p) /* Inputs. */
+	    );
+
+	return (x);
+}
+#elif (defined(JE_FORCE_SYNC_COMPARE_AND_SWAP_8))
+JEMALLOC_INLINE uint64_t
+atomic_add_uint64(uint64_t *p, uint64_t x)
+{
+
+	return (__sync_add_and_fetch(p, x));
+}
+
+JEMALLOC_INLINE uint64_t
+atomic_sub_uint64(uint64_t *p, uint64_t x)
+{
+
+	return (__sync_sub_and_fetch(p, x));
+}
+#else
+#  if (LG_SIZEOF_PTR == 3)
+#    error "Missing implementation for 64-bit atomic operations"
+#  endif
+#endif
+
+/******************************************************************************/
+/* 32-bit operations. */
+#ifdef __GCC_HAVE_SYNC_COMPARE_AND_SWAP_4
+JEMALLOC_INLINE uint32_t
+atomic_add_uint32(uint32_t *p, uint32_t x)
+{
+
+	return (__sync_add_and_fetch(p, x));
+}
+
+JEMALLOC_INLINE uint32_t
+atomic_sub_uint32(uint32_t *p, uint32_t x)
+{
+
+	return (__sync_sub_and_fetch(p, x));
+}
+#elif (defined(JEMALLOC_OSATOMIC))
+JEMALLOC_INLINE uint32_t
+atomic_add_uint32(uint32_t *p, uint32_t x)
+{
+
+	return (OSAtomicAdd32((int32_t)x, (int32_t *)p));
+}
+
+JEMALLOC_INLINE uint32_t
+atomic_sub_uint32(uint32_t *p, uint32_t x)
+{
+
+	return (OSAtomicAdd32(-((int32_t)x), (int32_t *)p));
+}
+#elif (defined(__i386__) || defined(__amd64__) || defined(__x86_64__))
+JEMALLOC_INLINE uint32_t
+atomic_add_uint32(uint32_t *p, uint32_t x)
+{
+
+	asm volatile (
+	    "lock; xaddl %0, %1;"
+	    : "+r" (x), "=m" (*p) /* Outputs. */
+	    : "m" (*p) /* Inputs. */
+	    );
+
+	return (x);
+}
+
+JEMALLOC_INLINE uint32_t
+atomic_sub_uint32(uint32_t *p, uint32_t x)
+{
+
+	x = (uint32_t)(-(int32_t)x);
+	asm volatile (
+	    "lock; xaddl %0, %1;"
+	    : "+r" (x), "=m" (*p) /* Outputs. */
+	    : "m" (*p) /* Inputs. */
+	    );
+
+	return (x);
+}
+#elif (defined(JE_FORCE_SYNC_COMPARE_AND_SWAP_4))
+JEMALLOC_INLINE uint32_t
+atomic_add_uint32(uint32_t *p, uint32_t x)
+{
+
+	return (__sync_add_and_fetch(p, x));
+}
+
+JEMALLOC_INLINE uint32_t
+atomic_sub_uint32(uint32_t *p, uint32_t x)
+{
+
+	return (__sync_sub_and_fetch(p, x));
+}
+#else
+#  error "Missing implementation for 32-bit atomic operations"
+#endif
+
+/******************************************************************************/
+/* size_t operations. */
+JEMALLOC_INLINE size_t
+atomic_add_z(size_t *p, size_t x)
+{
+
+#if (LG_SIZEOF_PTR == 3)
+	return ((size_t)atomic_add_uint64((uint64_t *)p, (uint64_t)x));
+#elif (LG_SIZEOF_PTR == 2)
+	return ((size_t)atomic_add_uint32((uint32_t *)p, (uint32_t)x));
+#endif
+}
+
+JEMALLOC_INLINE size_t
+atomic_sub_z(size_t *p, size_t x)
+{
+
+#if (LG_SIZEOF_PTR == 3)
+	return ((size_t)atomic_add_uint64((uint64_t *)p,
+	    (uint64_t)-((int64_t)x)));
+#elif (LG_SIZEOF_PTR == 2)
+	return ((size_t)atomic_add_uint32((uint32_t *)p,
+	    (uint32_t)-((int32_t)x)));
+#endif
+}
+
+/******************************************************************************/
+/* unsigned operations. */
+JEMALLOC_INLINE unsigned
+atomic_add_u(unsigned *p, unsigned x)
+{
+
+#if (LG_SIZEOF_INT == 3)
+	return ((unsigned)atomic_add_uint64((uint64_t *)p, (uint64_t)x));
+#elif (LG_SIZEOF_INT == 2)
+	return ((unsigned)atomic_add_uint32((uint32_t *)p, (uint32_t)x));
+#endif
+}
+
+JEMALLOC_INLINE unsigned
+atomic_sub_u(unsigned *p, unsigned x)
+{
+
+#if (LG_SIZEOF_INT == 3)
+	return ((unsigned)atomic_add_uint64((uint64_t *)p,
+	    (uint64_t)-((int64_t)x)));
+#elif (LG_SIZEOF_INT == 2)
+	return ((unsigned)atomic_add_uint32((uint32_t *)p,
+	    (uint32_t)-((int32_t)x)));
+#endif
+}
+/******************************************************************************/
+#endif
+
+#endif /* JEMALLOC_H_INLINES */
+/******************************************************************************/

contrib/jemalloc/include/jemalloc/internal/base.h

@@ -0,0 +1,26 @@
+/******************************************************************************/
+#ifdef JEMALLOC_H_TYPES
+
+#endif /* JEMALLOC_H_TYPES */
+/******************************************************************************/
+#ifdef JEMALLOC_H_STRUCTS
+
+#endif /* JEMALLOC_H_STRUCTS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_EXTERNS
+
+void	*base_alloc(size_t size);
+void	*base_calloc(size_t number, size_t size);
+extent_node_t *base_node_alloc(void);
+void	base_node_dealloc(extent_node_t *node);
+bool	base_boot(void);
+void	base_prefork(void);
+void	base_postfork_parent(void);
+void	base_postfork_child(void);
+
+#endif /* JEMALLOC_H_EXTERNS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_INLINES
+
+#endif /* JEMALLOC_H_INLINES */
+/******************************************************************************/

contrib/jemalloc/include/jemalloc/internal/bitmap.h

@@ -0,0 +1,184 @@
+/******************************************************************************/
+#ifdef JEMALLOC_H_TYPES
+
+/* Maximum bitmap bit count is 2^LG_BITMAP_MAXBITS. */
+#define	LG_BITMAP_MAXBITS	LG_RUN_MAXREGS
+
+typedef struct bitmap_level_s bitmap_level_t;
+typedef struct bitmap_info_s bitmap_info_t;
+typedef unsigned long bitmap_t;
+#define	LG_SIZEOF_BITMAP	LG_SIZEOF_LONG
+
+/* Number of bits per group. */
+#define	LG_BITMAP_GROUP_NBITS		(LG_SIZEOF_BITMAP + 3)
+#define	BITMAP_GROUP_NBITS		(ZU(1) << LG_BITMAP_GROUP_NBITS)
+#define	BITMAP_GROUP_NBITS_MASK		(BITMAP_GROUP_NBITS-1)
+
+/* Maximum number of levels possible. */
+#define	BITMAP_MAX_LEVELS						\
+    (LG_BITMAP_MAXBITS / LG_SIZEOF_BITMAP)				\
+    + !!(LG_BITMAP_MAXBITS % LG_SIZEOF_BITMAP)
+
+#endif /* JEMALLOC_H_TYPES */
+/******************************************************************************/
+#ifdef JEMALLOC_H_STRUCTS
+
+struct bitmap_level_s {
+	/* Offset of this level's groups within the array of groups. */
+	size_t group_offset;
+};
+
+struct bitmap_info_s {
+	/* Logical number of bits in bitmap (stored at bottom level). */
+	size_t nbits;
+
+	/* Number of levels necessary for nbits. */
+	unsigned nlevels;
+
+	/*
+	 * Only the first (nlevels+1) elements are used, and levels are ordered
+	 * bottom to top (e.g. the bottom level is stored in levels[0]).
+	 */
+	bitmap_level_t levels[BITMAP_MAX_LEVELS+1];
+};
+
+#endif /* JEMALLOC_H_STRUCTS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_EXTERNS
+
+void	bitmap_info_init(bitmap_info_t *binfo, size_t nbits);
+size_t	bitmap_info_ngroups(const bitmap_info_t *binfo);
+size_t	bitmap_size(size_t nbits);
+void	bitmap_init(bitmap_t *bitmap, const bitmap_info_t *binfo);
+
+#endif /* JEMALLOC_H_EXTERNS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_INLINES
+
+#ifndef JEMALLOC_ENABLE_INLINE
+bool	bitmap_full(bitmap_t *bitmap, const bitmap_info_t *binfo);
+bool	bitmap_get(bitmap_t *bitmap, const bitmap_info_t *binfo, size_t bit);
+void	bitmap_set(bitmap_t *bitmap, const bitmap_info_t *binfo, size_t bit);
+size_t	bitmap_sfu(bitmap_t *bitmap, const bitmap_info_t *binfo);
+void	bitmap_unset(bitmap_t *bitmap, const bitmap_info_t *binfo, size_t bit);
+#endif
+
+#if (defined(JEMALLOC_ENABLE_INLINE) || defined(JEMALLOC_BITMAP_C_))
+JEMALLOC_INLINE bool
+bitmap_full(bitmap_t *bitmap, const bitmap_info_t *binfo)
+{
+	unsigned rgoff = binfo->levels[binfo->nlevels].group_offset - 1;
+	bitmap_t rg = bitmap[rgoff];
+	/* The bitmap is full iff the root group is 0. */
+	return (rg == 0);
+}
+
+JEMALLOC_INLINE bool
+bitmap_get(bitmap_t *bitmap, const bitmap_info_t *binfo, size_t bit)
+{
+	size_t goff;
+	bitmap_t g;
+
+	assert(bit < binfo->nbits);
+	goff = bit >> LG_BITMAP_GROUP_NBITS;
+	g = bitmap[goff];
+	return (!(g & (1LU << (bit & BITMAP_GROUP_NBITS_MASK))));
+}
+
+JEMALLOC_INLINE void
+bitmap_set(bitmap_t *bitmap, const bitmap_info_t *binfo, size_t bit)
+{
+	size_t goff;
+	bitmap_t *gp;
+	bitmap_t g;
+
+	assert(bit < binfo->nbits);
+	assert(bitmap_get(bitmap, binfo, bit) == false);
+	goff = bit >> LG_BITMAP_GROUP_NBITS;
+	gp = &bitmap[goff];
+	g = *gp;
+	assert(g & (1LU << (bit & BITMAP_GROUP_NBITS_MASK)));
+	g ^= 1LU << (bit & BITMAP_GROUP_NBITS_MASK);
+	*gp = g;
+	assert(bitmap_get(bitmap, binfo, bit));
+	/* Propagate group state transitions up the tree. */
+	if (g == 0) {
+		unsigned i;
+		for (i = 1; i < binfo->nlevels; i++) {
+			bit = goff;
+			goff = bit >> LG_BITMAP_GROUP_NBITS;
+			gp = &bitmap[binfo->levels[i].group_offset + goff];
+			g = *gp;
+			assert(g & (1LU << (bit & BITMAP_GROUP_NBITS_MASK)));
+			g ^= 1LU << (bit & BITMAP_GROUP_NBITS_MASK);
+			*gp = g;
+			if (g != 0)
+				break;
+		}
+	}
+}
+
+/* sfu: set first unset. */
+JEMALLOC_INLINE size_t
+bitmap_sfu(bitmap_t *bitmap, const bitmap_info_t *binfo)
+{
+	size_t bit;
+	bitmap_t g;
+	unsigned i;
+
+	assert(bitmap_full(bitmap, binfo) == false);
+
+	i = binfo->nlevels - 1;
+	g = bitmap[binfo->levels[i].group_offset];
+	bit = ffsl(g) - 1;
+	while (i > 0) {
+		i--;
+		g = bitmap[binfo->levels[i].group_offset + bit];
+		bit = (bit << LG_BITMAP_GROUP_NBITS) + (ffsl(g) - 1);
+	}
+
+	bitmap_set(bitmap, binfo, bit);
+	return (bit);
+}
+
+JEMALLOC_INLINE void
+bitmap_unset(bitmap_t *bitmap, const bitmap_info_t *binfo, size_t bit)
+{
+	size_t goff;
+	bitmap_t *gp;
+	bitmap_t g;
+	bool propagate;
+
+	assert(bit < binfo->nbits);
+	assert(bitmap_get(bitmap, binfo, bit));
+	goff = bit >> LG_BITMAP_GROUP_NBITS;
+	gp = &bitmap[goff];
+	g = *gp;
+	propagate = (g == 0);
+	assert((g & (1LU << (bit & BITMAP_GROUP_NBITS_MASK))) == 0);
+	g ^= 1LU << (bit & BITMAP_GROUP_NBITS_MASK);
+	*gp = g;
+	assert(bitmap_get(bitmap, binfo, bit) == false);
+	/* Propagate group state transitions up the tree. */
+	if (propagate) {
+		unsigned i;
+		for (i = 1; i < binfo->nlevels; i++) {
+			bit = goff;
+			goff = bit >> LG_BITMAP_GROUP_NBITS;
+			gp = &bitmap[binfo->levels[i].group_offset + goff];
+			g = *gp;
+			propagate = (g == 0);
+			assert((g & (1LU << (bit & BITMAP_GROUP_NBITS_MASK)))
+			    == 0);
+			g ^= 1LU << (bit & BITMAP_GROUP_NBITS_MASK);
+			*gp = g;
+			if (propagate == false)
+				break;
+		}
+	}
+}
+
+#endif
+
+#endif /* JEMALLOC_H_INLINES */
+/******************************************************************************/

contrib/jemalloc/include/jemalloc/internal/chunk.h

@@ -0,0 +1,58 @@
+/******************************************************************************/
+#ifdef JEMALLOC_H_TYPES
+
+/*
+ * Size and alignment of memory chunks that are allocated by the OS's virtual
+ * memory system.
+ */
+#define	LG_CHUNK_DEFAULT	22
+
+/* Return the chunk address for allocation address a. */
+#define	CHUNK_ADDR2BASE(a)						\
+	((void *)((uintptr_t)(a) & ~chunksize_mask))
+
+/* Return the chunk offset of address a. */
+#define	CHUNK_ADDR2OFFSET(a)						\
+	((size_t)((uintptr_t)(a) & chunksize_mask))
+
+/* Return the smallest chunk multiple that is >= s. */
+#define	CHUNK_CEILING(s)						\
+	(((s) + chunksize_mask) & ~chunksize_mask)
+
+#endif /* JEMALLOC_H_TYPES */
+/******************************************************************************/
+#ifdef JEMALLOC_H_STRUCTS
+
+#endif /* JEMALLOC_H_STRUCTS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_EXTERNS
+
+extern size_t		opt_lg_chunk;
+
+/* Protects stats_chunks; currently not used for any other purpose. */
+extern malloc_mutex_t	chunks_mtx;
+/* Chunk statistics. */
+extern chunk_stats_t	stats_chunks;
+
+extern rtree_t		*chunks_rtree;
+
+extern size_t		chunksize;
+extern size_t		chunksize_mask; /* (chunksize - 1). */
+extern size_t		chunk_npages;
+extern size_t		map_bias; /* Number of arena chunk header pages. */
+extern size_t		arena_maxclass; /* Max size class for arenas. */
+
+void	*chunk_alloc(size_t size, size_t alignment, bool base, bool *zero);
+void	chunk_dealloc(void *chunk, size_t size, bool unmap);
+bool	chunk_boot0(void);
+bool	chunk_boot1(void);
+
+#endif /* JEMALLOC_H_EXTERNS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_INLINES
+
+#endif /* JEMALLOC_H_INLINES */
+/******************************************************************************/
+
+#include "jemalloc/internal/chunk_dss.h"
+#include "jemalloc/internal/chunk_mmap.h"

contrib/jemalloc/include/jemalloc/internal/chunk_dss.h

@@ -0,0 +1,24 @@
+/******************************************************************************/
+#ifdef JEMALLOC_H_TYPES
+
+#endif /* JEMALLOC_H_TYPES */
+/******************************************************************************/
+#ifdef JEMALLOC_H_STRUCTS
+
+#endif /* JEMALLOC_H_STRUCTS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_EXTERNS
+
+void	*chunk_alloc_dss(size_t size, size_t alignment, bool *zero);
+bool	chunk_in_dss(void *chunk);
+bool	chunk_dss_boot(void);
+void	chunk_dss_prefork(void);
+void	chunk_dss_postfork_parent(void);
+void	chunk_dss_postfork_child(void);
+
+#endif /* JEMALLOC_H_EXTERNS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_INLINES
+
+#endif /* JEMALLOC_H_INLINES */
+/******************************************************************************/

contrib/jemalloc/include/jemalloc/internal/chunk_mmap.h

@@ -0,0 +1,22 @@
+/******************************************************************************/
+#ifdef JEMALLOC_H_TYPES
+
+#endif /* JEMALLOC_H_TYPES */
+/******************************************************************************/
+#ifdef JEMALLOC_H_STRUCTS
+
+#endif /* JEMALLOC_H_STRUCTS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_EXTERNS
+
+void	*chunk_alloc_mmap(size_t size, size_t alignment);
+bool	chunk_dealloc_mmap(void *chunk, size_t size);
+
+bool	chunk_mmap_boot(void);
+
+#endif /* JEMALLOC_H_EXTERNS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_INLINES
+
+#endif /* JEMALLOC_H_INLINES */
+/******************************************************************************/

contrib/jemalloc/include/jemalloc/internal/ckh.h

@@ -0,0 +1,90 @@
+/******************************************************************************/
+#ifdef JEMALLOC_H_TYPES
+
+typedef struct ckh_s ckh_t;
+typedef struct ckhc_s ckhc_t;
+
+/* Typedefs to allow easy function pointer passing. */
+typedef void ckh_hash_t (const void *, unsigned, size_t *, size_t *);
+typedef bool ckh_keycomp_t (const void *, const void *);
+
+/* Maintain counters used to get an idea of performance. */
+/* #define	CKH_COUNT */
+/* Print counter values in ckh_delete() (requires CKH_COUNT). */
+/* #define	CKH_VERBOSE */
+
+/*
+ * There are 2^LG_CKH_BUCKET_CELLS cells in each hash table bucket.  Try to fit
+ * one bucket per L1 cache line.
+ */
+#define LG_CKH_BUCKET_CELLS (LG_CACHELINE - LG_SIZEOF_PTR - 1)
+
+#endif /* JEMALLOC_H_TYPES */
+/******************************************************************************/
+#ifdef JEMALLOC_H_STRUCTS
+
+/* Hash table cell. */
+struct ckhc_s {
+	const void	*key;
+	const void	*data;
+};
+
+struct ckh_s {
+#ifdef CKH_COUNT
+	/* Counters used to get an idea of performance. */
+	uint64_t	ngrows;
+	uint64_t	nshrinks;
+	uint64_t	nshrinkfails;
+	uint64_t	ninserts;
+	uint64_t	nrelocs;
+#endif
+
+	/* Used for pseudo-random number generation. */
+#define	CKH_A		1103515241
+#define	CKH_C		12347
+	uint32_t	prng_state;
+
+	/* Total number of items. */
+	size_t		count;
+
+	/*
+	 * Minimum and current number of hash table buckets.  There are
+	 * 2^LG_CKH_BUCKET_CELLS cells per bucket.
+	 */
+	unsigned	lg_minbuckets;
+	unsigned	lg_curbuckets;
+
+	/* Hash and comparison functions. */
+	ckh_hash_t	*hash;
+	ckh_keycomp_t	*keycomp;
+
+	/* Hash table with 2^lg_curbuckets buckets. */
+	ckhc_t		*tab;
+};
+
+#endif /* JEMALLOC_H_STRUCTS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_EXTERNS
+
+bool	ckh_new(ckh_t *ckh, size_t minitems, ckh_hash_t *hash,
+    ckh_keycomp_t *keycomp);
+void	ckh_delete(ckh_t *ckh);
+size_t	ckh_count(ckh_t *ckh);
+bool	ckh_iter(ckh_t *ckh, size_t *tabind, void **key, void **data);
+bool	ckh_insert(ckh_t *ckh, const void *key, const void *data);
+bool	ckh_remove(ckh_t *ckh, const void *searchkey, void **key,
+    void **data);
+bool	ckh_search(ckh_t *ckh, const void *seachkey, void **key, void **data);
+void	ckh_string_hash(const void *key, unsigned minbits, size_t *hash1,
+    size_t *hash2);
+bool	ckh_string_keycomp(const void *k1, const void *k2);
+void	ckh_pointer_hash(const void *key, unsigned minbits, size_t *hash1,
+    size_t *hash2);
+bool	ckh_pointer_keycomp(const void *k1, const void *k2);
+
+#endif /* JEMALLOC_H_EXTERNS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_INLINES
+
+#endif /* JEMALLOC_H_INLINES */
+/******************************************************************************/

contrib/jemalloc/include/jemalloc/internal/ctl.h

@@ -0,0 +1,109 @@
+/******************************************************************************/
+#ifdef JEMALLOC_H_TYPES
+
+typedef struct ctl_node_s ctl_node_t;
+typedef struct ctl_arena_stats_s ctl_arena_stats_t;
+typedef struct ctl_stats_s ctl_stats_t;
+
+#endif /* JEMALLOC_H_TYPES */
+/******************************************************************************/
+#ifdef JEMALLOC_H_STRUCTS
+
+struct ctl_node_s {
+	bool			named;
+	union {
+		struct {
+			const char	*name;
+			/* If (nchildren == 0), this is a terminal node. */
+			unsigned	nchildren;
+			const	ctl_node_t *children;
+		} named;
+		struct {
+			const ctl_node_t *(*index)(const size_t *, size_t,
+			    size_t);
+		} indexed;
+	} u;
+	int	(*ctl)(const size_t *, size_t, void *, size_t *, void *,
+	    size_t);
+};
+
+struct ctl_arena_stats_s {
+	bool			initialized;
+	unsigned		nthreads;
+	size_t			pactive;
+	size_t			pdirty;
+	arena_stats_t		astats;
+
+	/* Aggregate stats for small size classes, based on bin stats. */
+	size_t			allocated_small;
+	uint64_t		nmalloc_small;
+	uint64_t		ndalloc_small;
+	uint64_t		nrequests_small;
+
+	malloc_bin_stats_t	bstats[NBINS];
+	malloc_large_stats_t	*lstats;	/* nlclasses elements. */
+};
+
+struct ctl_stats_s {
+	size_t			allocated;
+	size_t			active;
+	size_t			mapped;
+	struct {
+		size_t		current;	/* stats_chunks.curchunks */
+		uint64_t	total;		/* stats_chunks.nchunks */
+		size_t		high;		/* stats_chunks.highchunks */
+	} chunks;
+	struct {
+		size_t		allocated;	/* huge_allocated */
+		uint64_t	nmalloc;	/* huge_nmalloc */
+		uint64_t	ndalloc;	/* huge_ndalloc */
+	} huge;
+	ctl_arena_stats_t	*arenas;	/* (narenas + 1) elements. */
+};
+
+#endif /* JEMALLOC_H_STRUCTS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_EXTERNS
+
+int	ctl_byname(const char *name, void *oldp, size_t *oldlenp, void *newp,
+    size_t newlen);
+int	ctl_nametomib(const char *name, size_t *mibp, size_t *miblenp);
+
+int	ctl_bymib(const size_t *mib, size_t miblen, void *oldp, size_t *oldlenp,
+    void *newp, size_t newlen);
+bool	ctl_boot(void);
+
+#define	xmallctl(name, oldp, oldlenp, newp, newlen) do {		\
+	if (je_mallctl(name, oldp, oldlenp, newp, newlen)		\
+	    != 0) {							\
+		malloc_printf(						\
+		    "<jemalloc>: Failure in xmallctl(\"%s\", ...)\n",	\
+		    name);						\
+		abort();						\
+	}								\
+} while (0)
+
+#define	xmallctlnametomib(name, mibp, miblenp) do {			\
+	if (je_mallctlnametomib(name, mibp, miblenp) != 0) {		\
+		malloc_printf("<jemalloc>: Failure in "			\
+		    "xmallctlnametomib(\"%s\", ...)\n", name);		\
+		abort();						\
+	}								\
+} while (0)
+
+#define	xmallctlbymib(mib, miblen, oldp, oldlenp, newp, newlen) do {	\
+	if (je_mallctlbymib(mib, miblen, oldp, oldlenp, newp,		\
+	    newlen) != 0) {						\
+		malloc_write(						\
+		    "<jemalloc>: Failure in xmallctlbymib()\n");	\
+		abort();						\
+	}								\
+} while (0)
+
+#endif /* JEMALLOC_H_EXTERNS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_INLINES
+
+#endif /* JEMALLOC_H_INLINES */
+/******************************************************************************/
+

contrib/jemalloc/include/jemalloc/internal/extent.h

@@ -0,0 +1,43 @@
+/******************************************************************************/
+#ifdef JEMALLOC_H_TYPES
+
+typedef struct extent_node_s extent_node_t;
+
+#endif /* JEMALLOC_H_TYPES */
+/******************************************************************************/
+#ifdef JEMALLOC_H_STRUCTS
+
+/* Tree of extents. */
+struct extent_node_s {
+	/* Linkage for the size/address-ordered tree. */
+	rb_node(extent_node_t)	link_szad;
+
+	/* Linkage for the address-ordered tree. */
+	rb_node(extent_node_t)	link_ad;
+
+	/* Profile counters, used for huge objects. */
+	prof_ctx_t		*prof_ctx;
+
+	/* Pointer to the extent that this tree node is responsible for. */
+	void			*addr;
+
+	/* Total region size. */
+	size_t			size;
+};
+typedef rb_tree(extent_node_t) extent_tree_t;
+
+#endif /* JEMALLOC_H_STRUCTS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_EXTERNS
+
+rb_proto(, extent_tree_szad_, extent_tree_t, extent_node_t)
+
+rb_proto(, extent_tree_ad_, extent_tree_t, extent_node_t)
+
+#endif /* JEMALLOC_H_EXTERNS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_INLINES
+
+#endif /* JEMALLOC_H_INLINES */
+/******************************************************************************/
+

contrib/jemalloc/include/jemalloc/internal/hash.h

@@ -0,0 +1,70 @@
+/******************************************************************************/
+#ifdef JEMALLOC_H_TYPES
+
+#endif /* JEMALLOC_H_TYPES */
+/******************************************************************************/
+#ifdef JEMALLOC_H_STRUCTS
+
+#endif /* JEMALLOC_H_STRUCTS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_EXTERNS
+
+#endif /* JEMALLOC_H_EXTERNS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_INLINES
+
+#ifndef JEMALLOC_ENABLE_INLINE
+uint64_t	hash(const void *key, size_t len, uint64_t seed);
+#endif
+
+#if (defined(JEMALLOC_ENABLE_INLINE) || defined(JEMALLOC_HASH_C_))
+/*
+ * The following hash function is based on MurmurHash64A(), placed into the
+ * public domain by Austin Appleby.  See http://murmurhash.googlepages.com/ for
+ * details.
+ */
+JEMALLOC_INLINE uint64_t
+hash(const void *key, size_t len, uint64_t seed)
+{
+	const uint64_t m = UINT64_C(0xc6a4a7935bd1e995);
+	const int r = 47;
+	uint64_t h = seed ^ (len * m);
+	const uint64_t *data = (const uint64_t *)key;
+	const uint64_t *end = data + (len/8);
+	const unsigned char *data2;
+
+	assert(((uintptr_t)key & 0x7) == 0);
+
+	while(data != end) {
+		uint64_t k = *data++;
+
+		k *= m;
+		k ^= k >> r;
+		k *= m;
+
+		h ^= k;
+		h *= m;
+	}
+
+	data2 = (const unsigned char *)data;
+	switch(len & 7) {
+	case 7: h ^= ((uint64_t)(data2[6])) << 48;
+	case 6: h ^= ((uint64_t)(data2[5])) << 40;
+	case 5: h ^= ((uint64_t)(data2[4])) << 32;
+	case 4: h ^= ((uint64_t)(data2[3])) << 24;
+	case 3: h ^= ((uint64_t)(data2[2])) << 16;
+	case 2: h ^= ((uint64_t)(data2[1])) << 8;
+	case 1: h ^= ((uint64_t)(data2[0]));
+		h *= m;
+	}
+
+	h ^= h >> r;
+	h *= m;
+	h ^= h >> r;
+
+	return (h);
+}
+#endif
+
+#endif /* JEMALLOC_H_INLINES */
+/******************************************************************************/

contrib/jemalloc/include/jemalloc/internal/huge.h

@@ -0,0 +1,40 @@
+/******************************************************************************/
+#ifdef JEMALLOC_H_TYPES
+
+#endif /* JEMALLOC_H_TYPES */
+/******************************************************************************/
+#ifdef JEMALLOC_H_STRUCTS
+
+#endif /* JEMALLOC_H_STRUCTS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_EXTERNS
+
+/* Huge allocation statistics. */
+extern uint64_t		huge_nmalloc;
+extern uint64_t		huge_ndalloc;
+extern size_t		huge_allocated;
+
+/* Protects chunk-related data structures. */
+extern malloc_mutex_t	huge_mtx;
+
+void	*huge_malloc(size_t size, bool zero);
+void	*huge_palloc(size_t size, size_t alignment, bool zero);
+void	*huge_ralloc_no_move(void *ptr, size_t oldsize, size_t size,
+    size_t extra);
+void	*huge_ralloc(void *ptr, size_t oldsize, size_t size, size_t extra,
+    size_t alignment, bool zero);
+void	huge_dalloc(void *ptr, bool unmap);
+size_t	huge_salloc(const void *ptr);
+prof_ctx_t	*huge_prof_ctx_get(const void *ptr);
+void	huge_prof_ctx_set(const void *ptr, prof_ctx_t *ctx);
+bool	huge_boot(void);
+void	huge_prefork(void);
+void	huge_postfork_parent(void);
+void	huge_postfork_child(void);
+
+#endif /* JEMALLOC_H_EXTERNS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_INLINES
+
+#endif /* JEMALLOC_H_INLINES */
+/******************************************************************************/

contrib/jemalloc/include/jemalloc/internal/jemalloc_internal.h

@@ -0,0 +1,876 @@
+#include "libc_private.h"
+#include "namespace.h"
+
+#include <sys/mman.h>
+#include <sys/param.h>
+#include <sys/syscall.h>
+#if !defined(SYS_write) && defined(__NR_write)
+#define	SYS_write __NR_write
+#endif
+#include <sys/time.h>
+#include <sys/types.h>
+#include <sys/uio.h>
+
+#include <errno.h>
+#include <limits.h>
+#ifndef SIZE_T_MAX
+#  define SIZE_T_MAX	SIZE_MAX
+#endif
+#include <pthread.h>
+#include <sched.h>
+#include <stdarg.h>
+#include <stdbool.h>
+#include <stdio.h>
+#include <stdlib.h>
+#include <stdint.h>
+#include <stddef.h>
+#ifndef offsetof
+#  define offsetof(type, member)	((size_t)&(((type *)NULL)->member))
+#endif
+#include <inttypes.h>
+#include <string.h>
+#include <strings.h>
+#include <ctype.h>
+#include <unistd.h>
+#include <fcntl.h>
+#include <pthread.h>
+#include <math.h>
+
+#include "un-namespace.h"
+#include "libc_private.h"
+
+#define	JEMALLOC_NO_DEMANGLE
+#include "../jemalloc.h"
+
+#ifdef JEMALLOC_UTRACE
+#include <sys/ktrace.h>
+#endif
+
+#ifdef JEMALLOC_VALGRIND
+#include <valgrind/valgrind.h>
+#include <valgrind/memcheck.h>
+#endif
+
+#include "jemalloc/internal/private_namespace.h"
+
+#ifdef JEMALLOC_CC_SILENCE
+#define	UNUSED JEMALLOC_ATTR(unused)
+#else
+#define	UNUSED
+#endif
+
+static const bool config_debug =
+#ifdef JEMALLOC_DEBUG
+    true
+#else
+    false
+#endif
+    ;
+static const bool config_dss =
+#ifdef JEMALLOC_DSS
+    true
+#else
+    false
+#endif
+    ;
+static const bool config_fill =
+#ifdef JEMALLOC_FILL
+    true
+#else
+    false
+#endif
+    ;
+static const bool config_lazy_lock =
+#ifdef JEMALLOC_LAZY_LOCK
+    true
+#else
+    false
+#endif
+    ;
+static const bool config_prof =
+#ifdef JEMALLOC_PROF
+    true
+#else
+    false
+#endif
+    ;
+static const bool config_prof_libgcc =
+#ifdef JEMALLOC_PROF_LIBGCC
+    true
+#else
+    false
+#endif
+    ;
+static const bool config_prof_libunwind =
+#ifdef JEMALLOC_PROF_LIBUNWIND
+    true
+#else
+    false
+#endif
+    ;
+static const bool config_munmap =
+#ifdef JEMALLOC_MUNMAP
+    true
+#else
+    false
+#endif
+    ;
+static const bool config_stats =
+#ifdef JEMALLOC_STATS
+    true
+#else
+    false
+#endif
+    ;
+static const bool config_tcache =
+#ifdef JEMALLOC_TCACHE
+    true
+#else
+    false
+#endif
+    ;
+static const bool config_tls =
+#ifdef JEMALLOC_TLS
+    true
+#else
+    false
+#endif
+    ;
+static const bool config_utrace =
+#ifdef JEMALLOC_UTRACE
+    true
+#else
+    false
+#endif
+    ;
+static const bool config_valgrind =
+#ifdef JEMALLOC_VALGRIND
+    true
+#else
+    false
+#endif
+    ;
+static const bool config_xmalloc =
+#ifdef JEMALLOC_XMALLOC
+    true
+#else
+    false
+#endif
+    ;
+static const bool config_ivsalloc =
+#ifdef JEMALLOC_IVSALLOC
+    true
+#else
+    false
+#endif
+    ;
+
+#if (defined(JEMALLOC_OSATOMIC) || defined(JEMALLOC_OSSPIN))
+#include <libkern/OSAtomic.h>
+#endif
+
+#ifdef JEMALLOC_ZONE
+#include <mach/mach_error.h>
+#include <mach/mach_init.h>
+#include <mach/vm_map.h>
+#include <malloc/malloc.h>
+#endif
+
+#define	RB_COMPACT
+#include "jemalloc/internal/rb.h"
+#include "jemalloc/internal/qr.h"
+#include "jemalloc/internal/ql.h"
+
+/*
+ * jemalloc can conceptually be broken into components (arena, tcache, etc.),
+ * but there are circular dependencies that cannot be broken without
+ * substantial performance degradation.  In order to reduce the effect on
+ * visual code flow, read the header files in multiple passes, with one of the
+ * following cpp variables defined during each pass:
+ *
+ *   JEMALLOC_H_TYPES   : Preprocessor-defined constants and psuedo-opaque data
+ *                        types.
+ *   JEMALLOC_H_STRUCTS : Data structures.
+ *   JEMALLOC_H_EXTERNS : Extern data declarations and function prototypes.
+ *   JEMALLOC_H_INLINES : Inline functions.
+ */
+/******************************************************************************/
+#define JEMALLOC_H_TYPES
+
+#define	ALLOCM_LG_ALIGN_MASK	((int)0x3f)
+
+#define	ZU(z)	((size_t)z)
+
+#ifndef __DECONST
+#  define	__DECONST(type, var)	((type)(uintptr_t)(const void *)(var))
+#endif
+
+#ifdef JEMALLOC_DEBUG
+   /* Disable inlining to make debugging easier. */
+#  define JEMALLOC_INLINE
+#  define inline
+#else
+#  define JEMALLOC_ENABLE_INLINE
+#  define JEMALLOC_INLINE static inline
+#endif
+
+/* Smallest size class to support. */
+#define	LG_TINY_MIN		3
+#define	TINY_MIN		(1U << LG_TINY_MIN)
+
+/*
+ * Minimum alignment of allocations is 2^LG_QUANTUM bytes (ignoring tiny size
+ * classes).
+ */
+#ifndef LG_QUANTUM
+#  ifdef __i386__
+#    define LG_QUANTUM		4
+#  endif
+#  ifdef __ia64__
+#    define LG_QUANTUM		4
+#  endif
+#  ifdef __alpha__
+#    define LG_QUANTUM		4
+#  endif
+#  ifdef __sparc64__
+#    define LG_QUANTUM		4
+#  endif
+#  if (defined(__amd64__) || defined(__x86_64__))
+#    define LG_QUANTUM		4
+#  endif
+#  ifdef __arm__
+#    define LG_QUANTUM		3
+#  endif
+#  ifdef __mips__
+#    define LG_QUANTUM		3
+#  endif
+#  ifdef __powerpc__
+#    define LG_QUANTUM		4
+#  endif
+#  ifdef __s390x__
+#    define LG_QUANTUM		4
+#  endif
+#  ifdef __SH4__
+#    define LG_QUANTUM		4
+#  endif
+#  ifdef __tile__
+#    define LG_QUANTUM		4
+#  endif
+#  ifndef LG_QUANTUM
+#    error "No LG_QUANTUM definition for architecture; specify via CPPFLAGS"
+#  endif
+#endif
+
+#define	QUANTUM			((size_t)(1U << LG_QUANTUM))
+#define	QUANTUM_MASK		(QUANTUM - 1)
+
+/* Return the smallest quantum multiple that is >= a. */
+#define	QUANTUM_CEILING(a)						\
+	(((a) + QUANTUM_MASK) & ~QUANTUM_MASK)
+
+#define	LONG			((size_t)(1U << LG_SIZEOF_LONG))
+#define	LONG_MASK		(LONG - 1)
+
+/* Return the smallest long multiple that is >= a. */
+#define	LONG_CEILING(a)							\
+	(((a) + LONG_MASK) & ~LONG_MASK)
+
+#define	SIZEOF_PTR		(1U << LG_SIZEOF_PTR)
+#define	PTR_MASK		(SIZEOF_PTR - 1)
+
+/* Return the smallest (void *) multiple that is >= a. */
+#define	PTR_CEILING(a)							\
+	(((a) + PTR_MASK) & ~PTR_MASK)
+
+/*
+ * Maximum size of L1 cache line.  This is used to avoid cache line aliasing.
+ * In addition, this controls the spacing of cacheline-spaced size classes.
+ */
+#define	LG_CACHELINE		6
+#define	CACHELINE		((size_t)(1U << LG_CACHELINE))
+#define	CACHELINE_MASK		(CACHELINE - 1)
+
+/* Return the smallest cacheline multiple that is >= s. */
+#define	CACHELINE_CEILING(s)						\
+	(((s) + CACHELINE_MASK) & ~CACHELINE_MASK)
+
+/* Page size.  STATIC_PAGE_SHIFT is determined by the configure script. */
+#ifdef PAGE_MASK
+#  undef PAGE_MASK
+#endif
+#define	LG_PAGE		STATIC_PAGE_SHIFT
+#define	PAGE		((size_t)(1U << STATIC_PAGE_SHIFT))
+#define	PAGE_MASK	((size_t)(PAGE - 1))
+
+/* Return the smallest pagesize multiple that is >= s. */
+#define	PAGE_CEILING(s)							\
+	(((s) + PAGE_MASK) & ~PAGE_MASK)
+
+/* Return the nearest aligned address at or below a. */
+#define	ALIGNMENT_ADDR2BASE(a, alignment)				\
+	((void *)((uintptr_t)(a) & (-(alignment))))
+
+/* Return the offset between a and the nearest aligned address at or below a. */
+#define	ALIGNMENT_ADDR2OFFSET(a, alignment)				\
+	((size_t)((uintptr_t)(a) & (alignment - 1)))
+
+/* Return the smallest alignment multiple that is >= s. */
+#define	ALIGNMENT_CEILING(s, alignment)					\
+	(((s) + (alignment - 1)) & (-(alignment)))
+
+#ifdef JEMALLOC_VALGRIND
+/*
+ * The JEMALLOC_VALGRIND_*() macros must be macros rather than functions
+ * so that when Valgrind reports errors, there are no extra stack frames
+ * in the backtraces.
+ *
+ * The size that is reported to valgrind must be consistent through a chain of
+ * malloc..realloc..realloc calls.  Request size isn't recorded anywhere in
+ * jemalloc, so it is critical that all callers of these macros provide usize
+ * rather than request size.  As a result, buffer overflow detection is
+ * technically weakened for the standard API, though it is generally accepted
+ * practice to consider any extra bytes reported by malloc_usable_size() as
+ * usable space.
+ */
+#define	JEMALLOC_VALGRIND_MALLOC(cond, ptr, usize, zero) do {		\
+	if (config_valgrind && opt_valgrind && cond)			\
+		VALGRIND_MALLOCLIKE_BLOCK(ptr, usize, p2rz(ptr), zero);	\
+} while (0)
+#define	JEMALLOC_VALGRIND_REALLOC(ptr, usize, old_ptr, old_usize,	\
+    old_rzsize, zero)  do {						\
+	if (config_valgrind && opt_valgrind) {				\
+		size_t rzsize = p2rz(ptr);				\
+									\
+		if (ptr == old_ptr) {					\
+			VALGRIND_RESIZEINPLACE_BLOCK(ptr, old_usize,	\
+			    usize, rzsize);				\
+			if (zero && old_usize < usize) {		\
+				VALGRIND_MAKE_MEM_DEFINED(		\
+				    (void *)((uintptr_t)ptr +		\
+				    old_usize), usize - old_usize);	\
+			}						\
+		} else {						\
+			if (old_ptr != NULL) {				\
+				VALGRIND_FREELIKE_BLOCK(old_ptr,	\
+				    old_rzsize);			\
+			}						\
+			if (ptr != NULL) {				\
+				size_t copy_size = (old_usize < usize)	\
+				    ?  old_usize : usize;		\
+				size_t tail_size = usize - copy_size;	\
+				VALGRIND_MALLOCLIKE_BLOCK(ptr, usize,	\
+				    rzsize, false);			\
+				if (copy_size > 0) {			\
+					VALGRIND_MAKE_MEM_DEFINED(ptr,	\
+					    copy_size);			\
+				}					\
+				if (zero && tail_size > 0) {		\
+					VALGRIND_MAKE_MEM_DEFINED(	\
+					    (void *)((uintptr_t)ptr +	\
+					    copy_size), tail_size);	\
+				}					\
+			}						\
+		}							\
+	}								\
+} while (0)
+#define	JEMALLOC_VALGRIND_FREE(ptr, rzsize) do {			\
+	if (config_valgrind && opt_valgrind)				\
+		VALGRIND_FREELIKE_BLOCK(ptr, rzsize);			\
+} while (0)
+#else
+#define	VALGRIND_MALLOCLIKE_BLOCK(addr, sizeB, rzB, is_zeroed)
+#define	VALGRIND_RESIZEINPLACE_BLOCK(addr, oldSizeB, newSizeB, rzB)
+#define	VALGRIND_FREELIKE_BLOCK(addr, rzB)
+#define	VALGRIND_MAKE_MEM_UNDEFINED(_qzz_addr, _qzz_len)
+#define	VALGRIND_MAKE_MEM_DEFINED(_qzz_addr, _qzz_len)
+#define	JEMALLOC_VALGRIND_MALLOC(cond, ptr, usize, zero)
+#define	JEMALLOC_VALGRIND_REALLOC(ptr, usize, old_ptr, old_usize,	\
+    old_rzsize, zero)
+#define	JEMALLOC_VALGRIND_FREE(ptr, rzsize)
+#endif
+
+#include "jemalloc/internal/util.h"
+#include "jemalloc/internal/atomic.h"
+#include "jemalloc/internal/prng.h"
+#include "jemalloc/internal/ckh.h"
+#include "jemalloc/internal/size_classes.h"
+#include "jemalloc/internal/stats.h"
+#include "jemalloc/internal/ctl.h"
+#include "jemalloc/internal/mutex.h"
+#include "jemalloc/internal/tsd.h"
+#include "jemalloc/internal/mb.h"
+#include "jemalloc/internal/extent.h"
+#include "jemalloc/internal/arena.h"
+#include "jemalloc/internal/bitmap.h"
+#include "jemalloc/internal/base.h"
+#include "jemalloc/internal/chunk.h"
+#include "jemalloc/internal/huge.h"
+#include "jemalloc/internal/rtree.h"
+#include "jemalloc/internal/tcache.h"
+#include "jemalloc/internal/hash.h"
+#include "jemalloc/internal/quarantine.h"
+#include "jemalloc/internal/prof.h"
+
+#undef JEMALLOC_H_TYPES
+/******************************************************************************/
+#define JEMALLOC_H_STRUCTS
+
+#include "jemalloc/internal/util.h"
+#include "jemalloc/internal/atomic.h"
+#include "jemalloc/internal/prng.h"
+#include "jemalloc/internal/ckh.h"
+#include "jemalloc/internal/size_classes.h"
+#include "jemalloc/internal/stats.h"
+#include "jemalloc/internal/ctl.h"
+#include "jemalloc/internal/mutex.h"
+#include "jemalloc/internal/tsd.h"
+#include "jemalloc/internal/mb.h"
+#include "jemalloc/internal/bitmap.h"
+#include "jemalloc/internal/extent.h"
+#include "jemalloc/internal/arena.h"
+#include "jemalloc/internal/base.h"
+#include "jemalloc/internal/chunk.h"
+#include "jemalloc/internal/huge.h"
+#include "jemalloc/internal/rtree.h"
+#include "jemalloc/internal/tcache.h"
+#include "jemalloc/internal/hash.h"
+#include "jemalloc/internal/quarantine.h"
+#include "jemalloc/internal/prof.h"
+
+typedef struct {
+	uint64_t	allocated;
+	uint64_t	deallocated;
+} thread_allocated_t;
+/*
+ * The JEMALLOC_CONCAT() wrapper is necessary to pass {0, 0} via a cpp macro
+ * argument.
+ */
+#define	THREAD_ALLOCATED_INITIALIZER	JEMALLOC_CONCAT({0, 0})
+
+#undef JEMALLOC_H_STRUCTS
+/******************************************************************************/
+#define JEMALLOC_H_EXTERNS
+
+extern bool	opt_abort;
+extern bool	opt_junk;
+extern size_t	opt_quarantine;
+extern bool	opt_redzone;
+extern bool	opt_utrace;
+extern bool	opt_valgrind;
+extern bool	opt_xmalloc;
+extern bool	opt_zero;
+extern size_t	opt_narenas;
+
+/* Number of CPUs. */
+extern unsigned		ncpus;
+
+extern malloc_mutex_t	arenas_lock; /* Protects arenas initialization. */
+/*
+ * Arenas that are used to service external requests.  Not all elements of the
+ * arenas array are necessarily used; arenas are created lazily as needed.
+ */
+extern arena_t		**arenas;
+extern unsigned		narenas;
+
+arena_t	*arenas_extend(unsigned ind);
+void	arenas_cleanup(void *arg);
+arena_t	*choose_arena_hard(void);
+void	jemalloc_prefork(void);
+void	jemalloc_postfork_parent(void);
+void	jemalloc_postfork_child(void);
+
+#include "jemalloc/internal/util.h"
+#include "jemalloc/internal/atomic.h"
+#include "jemalloc/internal/prng.h"
+#include "jemalloc/internal/ckh.h"
+#include "jemalloc/internal/size_classes.h"
+#include "jemalloc/internal/stats.h"
+#include "jemalloc/internal/ctl.h"
+#include "jemalloc/internal/mutex.h"
+#include "jemalloc/internal/tsd.h"
+#include "jemalloc/internal/mb.h"
+#include "jemalloc/internal/bitmap.h"
+#include "jemalloc/internal/extent.h"
+#include "jemalloc/internal/arena.h"
+#include "jemalloc/internal/base.h"
+#include "jemalloc/internal/chunk.h"
+#include "jemalloc/internal/huge.h"
+#include "jemalloc/internal/rtree.h"
+#include "jemalloc/internal/tcache.h"
+#include "jemalloc/internal/hash.h"
+#include "jemalloc/internal/quarantine.h"
+#include "jemalloc/internal/prof.h"
+
+#undef JEMALLOC_H_EXTERNS
+/******************************************************************************/
+#define JEMALLOC_H_INLINES
+
+#include "jemalloc/internal/util.h"
+#include "jemalloc/internal/atomic.h"
+#include "jemalloc/internal/prng.h"
+#include "jemalloc/internal/ckh.h"
+#include "jemalloc/internal/size_classes.h"
+#include "jemalloc/internal/stats.h"
+#include "jemalloc/internal/ctl.h"
+#include "jemalloc/internal/mutex.h"
+#include "jemalloc/internal/tsd.h"
+#include "jemalloc/internal/mb.h"
+#include "jemalloc/internal/extent.h"
+#include "jemalloc/internal/base.h"
+#include "jemalloc/internal/chunk.h"
+#include "jemalloc/internal/huge.h"
+
+#ifndef JEMALLOC_ENABLE_INLINE
+malloc_tsd_protos(JEMALLOC_ATTR(unused), arenas, arena_t *)
+
+size_t	s2u(size_t size);
+size_t	sa2u(size_t size, size_t alignment);
+arena_t	*choose_arena(arena_t *arena);
+#endif
+
+#if (defined(JEMALLOC_ENABLE_INLINE) || defined(JEMALLOC_C_))
+/*
+ * Map of pthread_self() --> arenas[???], used for selecting an arena to use
+ * for allocations.
+ */
+malloc_tsd_externs(arenas, arena_t *)
+malloc_tsd_funcs(JEMALLOC_INLINE, arenas, arena_t *, NULL, arenas_cleanup)
+
+/*
+ * Compute usable size that would result from allocating an object with the
+ * specified size.
+ */
+JEMALLOC_INLINE size_t
+s2u(size_t size)
+{
+
+	if (size <= SMALL_MAXCLASS)
+		return (arena_bin_info[SMALL_SIZE2BIN(size)].reg_size);
+	if (size <= arena_maxclass)
+		return (PAGE_CEILING(size));
+	return (CHUNK_CEILING(size));
+}
+
+/*
+ * Compute usable size that would result from allocating an object with the
+ * specified size and alignment.
+ */
+JEMALLOC_INLINE size_t
+sa2u(size_t size, size_t alignment)
+{
+	size_t usize;
+
+	assert(alignment != 0 && ((alignment - 1) & alignment) == 0);
+
+	/*
+	 * Round size up to the nearest multiple of alignment.
+	 *
+	 * This done, we can take advantage of the fact that for each small
+	 * size class, every object is aligned at the smallest power of two
+	 * that is non-zero in the base two representation of the size.  For
+	 * example:
+	 *
+	 *   Size |   Base 2 | Minimum alignment
+	 *   -----+----------+------------------
+	 *     96 |  1100000 |  32
+	 *    144 | 10100000 |  32
+	 *    192 | 11000000 |  64
+	 */
+	usize = ALIGNMENT_CEILING(size, alignment);
+	/*
+	 * (usize < size) protects against the combination of maximal
+	 * alignment and size greater than maximal alignment.
+	 */
+	if (usize < size) {
+		/* size_t overflow. */
+		return (0);
+	}
+
+	if (usize <= arena_maxclass && alignment <= PAGE) {
+		if (usize <= SMALL_MAXCLASS)
+			return (arena_bin_info[SMALL_SIZE2BIN(usize)].reg_size);
+		return (PAGE_CEILING(usize));
+	} else {
+		size_t run_size;
+
+		/*
+		 * We can't achieve subpage alignment, so round up alignment
+		 * permanently; it makes later calculations simpler.
+		 */
+		alignment = PAGE_CEILING(alignment);
+		usize = PAGE_CEILING(size);
+		/*
+		 * (usize < size) protects against very large sizes within
+		 * PAGE of SIZE_T_MAX.
+		 *
+		 * (usize + alignment < usize) protects against the
+		 * combination of maximal alignment and usize large enough
+		 * to cause overflow.  This is similar to the first overflow
+		 * check above, but it needs to be repeated due to the new
+		 * usize value, which may now be *equal* to maximal
+		 * alignment, whereas before we only detected overflow if the
+		 * original size was *greater* than maximal alignment.
+		 */
+		if (usize < size || usize + alignment < usize) {
+			/* size_t overflow. */
+			return (0);
+		}
+
+		/*
+		 * Calculate the size of the over-size run that arena_palloc()
+		 * would need to allocate in order to guarantee the alignment.
+		 * If the run wouldn't fit within a chunk, round up to a huge
+		 * allocation size.
+		 */
+		run_size = usize + alignment - PAGE;
+		if (run_size <= arena_maxclass)
+			return (PAGE_CEILING(usize));
+		return (CHUNK_CEILING(usize));
+	}
+}
+
+/* Choose an arena based on a per-thread value. */
+JEMALLOC_INLINE arena_t *
+choose_arena(arena_t *arena)
+{
+	arena_t *ret;
+
+	if (arena != NULL)
+		return (arena);
+
+	if ((ret = *arenas_tsd_get()) == NULL) {
+		ret = choose_arena_hard();
+		assert(ret != NULL);
+	}
+
+	return (ret);
+}
+#endif
+
+#include "jemalloc/internal/bitmap.h"
+#include "jemalloc/internal/rtree.h"
+#include "jemalloc/internal/tcache.h"
+#include "jemalloc/internal/arena.h"
+#include "jemalloc/internal/hash.h"
+#include "jemalloc/internal/quarantine.h"
+
+#ifndef JEMALLOC_ENABLE_INLINE
+void	*imalloc(size_t size);
+void	*icalloc(size_t size);
+void	*ipalloc(size_t usize, size_t alignment, bool zero);
+size_t	isalloc(const void *ptr, bool demote);
+size_t	ivsalloc(const void *ptr, bool demote);
+size_t	u2rz(size_t usize);
+size_t	p2rz(const void *ptr);
+void	idalloc(void *ptr);
+void	iqalloc(void *ptr);
+void	*iralloc(void *ptr, size_t size, size_t extra, size_t alignment,
+    bool zero, bool no_move);
+malloc_tsd_protos(JEMALLOC_ATTR(unused), thread_allocated, thread_allocated_t)
+#endif
+
+#if (defined(JEMALLOC_ENABLE_INLINE) || defined(JEMALLOC_C_))
+JEMALLOC_INLINE void *
+imalloc(size_t size)
+{
+
+	assert(size != 0);
+
+	if (size <= arena_maxclass)
+		return (arena_malloc(NULL, size, false, true));
+	else
+		return (huge_malloc(size, false));
+}
+
+JEMALLOC_INLINE void *
+icalloc(size_t size)
+{
+
+	if (size <= arena_maxclass)
+		return (arena_malloc(NULL, size, true, true));
+	else
+		return (huge_malloc(size, true));
+}
+
+JEMALLOC_INLINE void *
+ipalloc(size_t usize, size_t alignment, bool zero)
+{
+	void *ret;
+
+	assert(usize != 0);
+	assert(usize == sa2u(usize, alignment));
+
+	if (usize <= arena_maxclass && alignment <= PAGE)
+		ret = arena_malloc(NULL, usize, zero, true);
+	else {
+		if (usize <= arena_maxclass) {
+			ret = arena_palloc(choose_arena(NULL), usize, alignment,
+			    zero);
+		} else if (alignment <= chunksize)
+			ret = huge_malloc(usize, zero);
+		else
+			ret = huge_palloc(usize, alignment, zero);
+	}
+
+	assert(ALIGNMENT_ADDR2BASE(ret, alignment) == ret);
+	return (ret);
+}
+
+/*
+ * Typical usage:
+ *   void *ptr = [...]
+ *   size_t sz = isalloc(ptr, config_prof);
+ */
+JEMALLOC_INLINE size_t
+isalloc(const void *ptr, bool demote)
+{
+	size_t ret;
+	arena_chunk_t *chunk;
+
+	assert(ptr != NULL);
+	/* Demotion only makes sense if config_prof is true. */
+	assert(config_prof || demote == false);
+
+	chunk = (arena_chunk_t *)CHUNK_ADDR2BASE(ptr);
+	if (chunk != ptr) {
+		/* Region. */
+		ret = arena_salloc(ptr, demote);
+	} else
+		ret = huge_salloc(ptr);
+
+	return (ret);
+}
+
+JEMALLOC_INLINE size_t
+ivsalloc(const void *ptr, bool demote)
+{
+
+	/* Return 0 if ptr is not within a chunk managed by jemalloc. */
+	if (rtree_get(chunks_rtree, (uintptr_t)CHUNK_ADDR2BASE(ptr)) == NULL)
+		return (0);
+
+	return (isalloc(ptr, demote));
+}
+
+JEMALLOC_INLINE size_t
+u2rz(size_t usize)
+{
+	size_t ret;
+
+	if (usize <= SMALL_MAXCLASS) {
+		size_t binind = SMALL_SIZE2BIN(usize);
+		ret = arena_bin_info[binind].redzone_size;
+	} else
+		ret = 0;
+
+	return (ret);
+}
+
+JEMALLOC_INLINE size_t
+p2rz(const void *ptr)
+{
+	size_t usize = isalloc(ptr, false);
+
+	return (u2rz(usize));
+}
+
+JEMALLOC_INLINE void
+idalloc(void *ptr)
+{
+	arena_chunk_t *chunk;
+
+	assert(ptr != NULL);
+
+	chunk = (arena_chunk_t *)CHUNK_ADDR2BASE(ptr);
+	if (chunk != ptr)
+		arena_dalloc(chunk->arena, chunk, ptr, true);
+	else
+		huge_dalloc(ptr, true);
+}
+
+JEMALLOC_INLINE void
+iqalloc(void *ptr)
+{
+
+	if (config_fill && opt_quarantine)
+		quarantine(ptr);
+	else
+		idalloc(ptr);
+}
+
+JEMALLOC_INLINE void *
+iralloc(void *ptr, size_t size, size_t extra, size_t alignment, bool zero,
+    bool no_move)
+{
+	void *ret;
+	size_t oldsize;
+
+	assert(ptr != NULL);
+	assert(size != 0);
+
+	oldsize = isalloc(ptr, config_prof);
+
+	if (alignment != 0 && ((uintptr_t)ptr & ((uintptr_t)alignment-1))
+	    != 0) {
+		size_t usize, copysize;
+
+		/*
+		 * Existing object alignment is inadequate; allocate new space
+		 * and copy.
+		 */
+		if (no_move)
+			return (NULL);
+		usize = sa2u(size + extra, alignment);
+		if (usize == 0)
+			return (NULL);
+		ret = ipalloc(usize, alignment, zero);
+		if (ret == NULL) {
+			if (extra == 0)
+				return (NULL);
+			/* Try again, without extra this time. */
+			usize = sa2u(size, alignment);
+			if (usize == 0)
+				return (NULL);
+			ret = ipalloc(usize, alignment, zero);
+			if (ret == NULL)
+				return (NULL);
+		}
+		/*
+		 * Copy at most size bytes (not size+extra), since the caller
+		 * has no expectation that the extra bytes will be reliably
+		 * preserved.
+		 */
+		copysize = (size < oldsize) ? size : oldsize;
+		memcpy(ret, ptr, copysize);
+		iqalloc(ptr);
+		return (ret);
+	}
+
+	if (no_move) {
+		if (size <= arena_maxclass) {
+			return (arena_ralloc_no_move(ptr, oldsize, size,
+			    extra, zero));
+		} else {
+			return (huge_ralloc_no_move(ptr, oldsize, size,
+			    extra));
+		}
+	} else {
+		if (size + extra <= arena_maxclass) {
+			return (arena_ralloc(ptr, oldsize, size, extra,
+			    alignment, zero, true));
+		} else {
+			return (huge_ralloc(ptr, oldsize, size, extra,
+			    alignment, zero));
+		}
+	}
+}
+
+malloc_tsd_externs(thread_allocated, thread_allocated_t)
+malloc_tsd_funcs(JEMALLOC_INLINE, thread_allocated, thread_allocated_t,
+    THREAD_ALLOCATED_INITIALIZER, malloc_tsd_no_cleanup)
+#endif
+
+#include "jemalloc/internal/prof.h"
+
+#undef JEMALLOC_H_INLINES
+/******************************************************************************/

contrib/jemalloc/include/jemalloc/internal/mb.h

@@ -0,0 +1,115 @@
+/******************************************************************************/
+#ifdef JEMALLOC_H_TYPES
+
+#endif /* JEMALLOC_H_TYPES */
+/******************************************************************************/
+#ifdef JEMALLOC_H_STRUCTS
+
+#endif /* JEMALLOC_H_STRUCTS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_EXTERNS
+
+#endif /* JEMALLOC_H_EXTERNS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_INLINES
+
+#ifndef JEMALLOC_ENABLE_INLINE
+void	mb_write(void);
+#endif
+
+#if (defined(JEMALLOC_ENABLE_INLINE) || defined(JEMALLOC_MB_C_))
+#ifdef __i386__
+/*
+ * According to the Intel Architecture Software Developer's Manual, current
+ * processors execute instructions in order from the perspective of other
+ * processors in a multiprocessor system, but 1) Intel reserves the right to
+ * change that, and 2) the compiler's optimizer could re-order instructions if
+ * there weren't some form of barrier.  Therefore, even if running on an
+ * architecture that does not need memory barriers (everything through at least
+ * i686), an "optimizer barrier" is necessary.
+ */
+JEMALLOC_INLINE void
+mb_write(void)
+{
+
+#  if 0
+	/* This is a true memory barrier. */
+	asm volatile ("pusha;"
+	    "xor  %%eax,%%eax;"
+	    "cpuid;"
+	    "popa;"
+	    : /* Outputs. */
+	    : /* Inputs. */
+	    : "memory" /* Clobbers. */
+	    );
+#else
+	/*
+	 * This is hopefully enough to keep the compiler from reordering
+	 * instructions around this one.
+	 */
+	asm volatile ("nop;"
+	    : /* Outputs. */
+	    : /* Inputs. */
+	    : "memory" /* Clobbers. */
+	    );
+#endif
+}
+#elif (defined(__amd64__) || defined(__x86_64__))
+JEMALLOC_INLINE void
+mb_write(void)
+{
+
+	asm volatile ("sfence"
+	    : /* Outputs. */
+	    : /* Inputs. */
+	    : "memory" /* Clobbers. */
+	    );
+}
+#elif defined(__powerpc__)
+JEMALLOC_INLINE void
+mb_write(void)
+{
+
+	asm volatile ("eieio"
+	    : /* Outputs. */
+	    : /* Inputs. */
+	    : "memory" /* Clobbers. */
+	    );
+}
+#elif defined(__sparc64__)
+JEMALLOC_INLINE void
+mb_write(void)
+{
+
+	asm volatile ("membar #StoreStore"
+	    : /* Outputs. */
+	    : /* Inputs. */
+	    : "memory" /* Clobbers. */
+	    );
+}
+#elif defined(__tile__)
+JEMALLOC_INLINE void
+mb_write(void)
+{
+
+	__sync_synchronize();
+}
+#else
+/*
+ * This is much slower than a simple memory barrier, but the semantics of mutex
+ * unlock make this work.
+ */
+JEMALLOC_INLINE void
+mb_write(void)
+{
+	malloc_mutex_t mtx;
+
+	malloc_mutex_init(&mtx);
+	malloc_mutex_lock(&mtx);
+	malloc_mutex_unlock(&mtx);
+}
+#endif
+#endif
+
+#endif /* JEMALLOC_H_INLINES */
+/******************************************************************************/

contrib/jemalloc/include/jemalloc/internal/mutex.h

@@ -0,0 +1,88 @@
+/******************************************************************************/
+#ifdef JEMALLOC_H_TYPES
+
+typedef struct malloc_mutex_s malloc_mutex_t;
+
+#ifdef JEMALLOC_OSSPIN
+#define	MALLOC_MUTEX_INITIALIZER {0}
+#elif (defined(JEMALLOC_MUTEX_INIT_CB))
+#define	MALLOC_MUTEX_INITIALIZER {PTHREAD_MUTEX_INITIALIZER, NULL}
+#else
+#  if (defined(PTHREAD_MUTEX_ADAPTIVE_NP) &&				\
+       defined(PTHREAD_ADAPTIVE_MUTEX_INITIALIZER_NP))
+#    define MALLOC_MUTEX_TYPE PTHREAD_MUTEX_ADAPTIVE_NP
+#    define MALLOC_MUTEX_INITIALIZER {PTHREAD_ADAPTIVE_MUTEX_INITIALIZER_NP}
+#  else
+#    define MALLOC_MUTEX_TYPE PTHREAD_MUTEX_DEFAULT
+#    define MALLOC_MUTEX_INITIALIZER {PTHREAD_MUTEX_INITIALIZER}
+#  endif
+#endif
+
+#endif /* JEMALLOC_H_TYPES */
+/******************************************************************************/
+#ifdef JEMALLOC_H_STRUCTS
+
+struct malloc_mutex_s {
+#ifdef JEMALLOC_OSSPIN
+	OSSpinLock		lock;
+#elif (defined(JEMALLOC_MUTEX_INIT_CB))
+	pthread_mutex_t		lock;
+	malloc_mutex_t		*postponed_next;
+#else
+	pthread_mutex_t		lock;
+#endif
+};
+
+#endif /* JEMALLOC_H_STRUCTS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_EXTERNS
+
+#ifdef JEMALLOC_LAZY_LOCK
+extern bool isthreaded;
+#endif
+
+bool	malloc_mutex_init(malloc_mutex_t *mutex);
+void	malloc_mutex_prefork(malloc_mutex_t *mutex);
+void	malloc_mutex_postfork_parent(malloc_mutex_t *mutex);
+void	malloc_mutex_postfork_child(malloc_mutex_t *mutex);
+bool	mutex_boot(void);
+
+#endif /* JEMALLOC_H_EXTERNS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_INLINES
+
+#ifndef JEMALLOC_ENABLE_INLINE
+void	malloc_mutex_lock(malloc_mutex_t *mutex);
+void	malloc_mutex_unlock(malloc_mutex_t *mutex);
+#endif
+
+#if (defined(JEMALLOC_ENABLE_INLINE) || defined(JEMALLOC_MUTEX_C_))
+JEMALLOC_INLINE void
+malloc_mutex_lock(malloc_mutex_t *mutex)
+{
+
+	if (isthreaded) {
+#ifdef JEMALLOC_OSSPIN
+		OSSpinLockLock(&mutex->lock);
+#else
+		pthread_mutex_lock(&mutex->lock);
+#endif
+	}
+}
+
+JEMALLOC_INLINE void
+malloc_mutex_unlock(malloc_mutex_t *mutex)
+{
+
+	if (isthreaded) {
+#ifdef JEMALLOC_OSSPIN
+		OSSpinLockUnlock(&mutex->lock);
+#else
+		pthread_mutex_unlock(&mutex->lock);
+#endif
+	}
+}
+#endif
+
+#endif /* JEMALLOC_H_INLINES */
+/******************************************************************************/

contrib/jemalloc/include/jemalloc/internal/private_namespace.h

@@ -0,0 +1,274 @@
+#define	arena_alloc_junk_small JEMALLOC_N(arena_alloc_junk_small)
+#define	arena_bin_index JEMALLOC_N(arena_bin_index)
+#define	arena_boot JEMALLOC_N(arena_boot)
+#define	arena_dalloc JEMALLOC_N(arena_dalloc)
+#define	arena_dalloc_bin JEMALLOC_N(arena_dalloc_bin)
+#define	arena_dalloc_junk_small JEMALLOC_N(arena_dalloc_junk_small)
+#define	arena_dalloc_large JEMALLOC_N(arena_dalloc_large)
+#define	arena_malloc JEMALLOC_N(arena_malloc)
+#define	arena_malloc_large JEMALLOC_N(arena_malloc_large)
+#define	arena_malloc_small JEMALLOC_N(arena_malloc_small)
+#define	arena_new JEMALLOC_N(arena_new)
+#define	arena_palloc JEMALLOC_N(arena_palloc)
+#define	arena_postfork_child JEMALLOC_N(arena_postfork_child)
+#define	arena_postfork_parent JEMALLOC_N(arena_postfork_parent)
+#define	arena_prefork JEMALLOC_N(arena_prefork)
+#define	arena_prof_accum JEMALLOC_N(arena_prof_accum)
+#define	arena_prof_ctx_get JEMALLOC_N(arena_prof_ctx_get)
+#define	arena_prof_ctx_set JEMALLOC_N(arena_prof_ctx_set)
+#define	arena_prof_promoted JEMALLOC_N(arena_prof_promoted)
+#define	arena_purge_all JEMALLOC_N(arena_purge_all)
+#define	arena_ralloc JEMALLOC_N(arena_ralloc)
+#define	arena_ralloc_no_move JEMALLOC_N(arena_ralloc_no_move)
+#define	arena_run_regind JEMALLOC_N(arena_run_regind)
+#define	arena_salloc JEMALLOC_N(arena_salloc)
+#define	arena_stats_merge JEMALLOC_N(arena_stats_merge)
+#define	arena_tcache_fill_small JEMALLOC_N(arena_tcache_fill_small)
+#define	arenas_bin_i_index JEMALLOC_N(arenas_bin_i_index)
+#define	arenas_cleanup JEMALLOC_N(arenas_cleanup)
+#define	arenas_extend JEMALLOC_N(arenas_extend)
+#define	arenas_lrun_i_index JEMALLOC_N(arenas_lrun_i_index)
+#define	arenas_tls JEMALLOC_N(arenas_tls)
+#define	arenas_tsd_boot JEMALLOC_N(arenas_tsd_boot)
+#define	arenas_tsd_cleanup_wrapper JEMALLOC_N(arenas_tsd_cleanup_wrapper)
+#define	arenas_tsd_get JEMALLOC_N(arenas_tsd_get)
+#define	arenas_tsd_set JEMALLOC_N(arenas_tsd_set)
+#define	atomic_add_u JEMALLOC_N(atomic_add_u)
+#define	atomic_add_uint32 JEMALLOC_N(atomic_add_uint32)
+#define	atomic_add_uint64 JEMALLOC_N(atomic_add_uint64)
+#define	atomic_add_z JEMALLOC_N(atomic_add_z)
+#define	atomic_sub_u JEMALLOC_N(atomic_sub_u)
+#define	atomic_sub_uint32 JEMALLOC_N(atomic_sub_uint32)
+#define	atomic_sub_uint64 JEMALLOC_N(atomic_sub_uint64)
+#define	atomic_sub_z JEMALLOC_N(atomic_sub_z)
+#define	base_alloc JEMALLOC_N(base_alloc)
+#define	base_boot JEMALLOC_N(base_boot)
+#define	base_calloc JEMALLOC_N(base_calloc)
+#define	base_node_alloc JEMALLOC_N(base_node_alloc)
+#define	base_node_dealloc JEMALLOC_N(base_node_dealloc)
+#define	base_postfork_child JEMALLOC_N(base_postfork_child)
+#define	base_postfork_parent JEMALLOC_N(base_postfork_parent)
+#define	base_prefork JEMALLOC_N(base_prefork)
+#define	bitmap_full JEMALLOC_N(bitmap_full)
+#define	bitmap_get JEMALLOC_N(bitmap_get)
+#define	bitmap_info_init JEMALLOC_N(bitmap_info_init)
+#define	bitmap_info_ngroups JEMALLOC_N(bitmap_info_ngroups)
+#define	bitmap_init JEMALLOC_N(bitmap_init)
+#define	bitmap_set JEMALLOC_N(bitmap_set)
+#define	bitmap_sfu JEMALLOC_N(bitmap_sfu)
+#define	bitmap_size JEMALLOC_N(bitmap_size)
+#define	bitmap_unset JEMALLOC_N(bitmap_unset)
+#define	bt_init JEMALLOC_N(bt_init)
+#define	buferror JEMALLOC_N(buferror)
+#define	choose_arena JEMALLOC_N(choose_arena)
+#define	choose_arena_hard JEMALLOC_N(choose_arena_hard)
+#define	chunk_alloc JEMALLOC_N(chunk_alloc)
+#define	chunk_alloc_dss JEMALLOC_N(chunk_alloc_dss)
+#define	chunk_alloc_mmap JEMALLOC_N(chunk_alloc_mmap)
+#define	chunk_boot0 JEMALLOC_N(chunk_boot0)
+#define	chunk_boot1 JEMALLOC_N(chunk_boot1)
+#define	chunk_dealloc JEMALLOC_N(chunk_dealloc)
+#define	chunk_dealloc_mmap JEMALLOC_N(chunk_dealloc_mmap)
+#define	chunk_dss_boot JEMALLOC_N(chunk_dss_boot)
+#define	chunk_dss_postfork_child JEMALLOC_N(chunk_dss_postfork_child)
+#define	chunk_dss_postfork_parent JEMALLOC_N(chunk_dss_postfork_parent)
+#define	chunk_dss_prefork JEMALLOC_N(chunk_dss_prefork)
+#define	chunk_in_dss JEMALLOC_N(chunk_in_dss)
+#define	chunk_mmap_boot JEMALLOC_N(chunk_mmap_boot)
+#define	ckh_bucket_search JEMALLOC_N(ckh_bucket_search)
+#define	ckh_count JEMALLOC_N(ckh_count)
+#define	ckh_delete JEMALLOC_N(ckh_delete)
+#define	ckh_evict_reloc_insert JEMALLOC_N(ckh_evict_reloc_insert)
+#define	ckh_insert JEMALLOC_N(ckh_insert)
+#define	ckh_isearch JEMALLOC_N(ckh_isearch)
+#define	ckh_iter JEMALLOC_N(ckh_iter)
+#define	ckh_new JEMALLOC_N(ckh_new)
+#define	ckh_pointer_hash JEMALLOC_N(ckh_pointer_hash)
+#define	ckh_pointer_keycomp JEMALLOC_N(ckh_pointer_keycomp)
+#define	ckh_rebuild JEMALLOC_N(ckh_rebuild)
+#define	ckh_remove JEMALLOC_N(ckh_remove)
+#define	ckh_search JEMALLOC_N(ckh_search)
+#define	ckh_string_hash JEMALLOC_N(ckh_string_hash)
+#define	ckh_string_keycomp JEMALLOC_N(ckh_string_keycomp)
+#define	ckh_try_bucket_insert JEMALLOC_N(ckh_try_bucket_insert)
+#define	ckh_try_insert JEMALLOC_N(ckh_try_insert)
+#define	ctl_boot JEMALLOC_N(ctl_boot)
+#define	ctl_bymib JEMALLOC_N(ctl_bymib)
+#define	ctl_byname JEMALLOC_N(ctl_byname)
+#define	ctl_nametomib JEMALLOC_N(ctl_nametomib)
+#define	extent_tree_ad_first JEMALLOC_N(extent_tree_ad_first)
+#define	extent_tree_ad_insert JEMALLOC_N(extent_tree_ad_insert)
+#define	extent_tree_ad_iter JEMALLOC_N(extent_tree_ad_iter)
+#define	extent_tree_ad_iter_recurse JEMALLOC_N(extent_tree_ad_iter_recurse)
+#define	extent_tree_ad_iter_start JEMALLOC_N(extent_tree_ad_iter_start)
+#define	extent_tree_ad_last JEMALLOC_N(extent_tree_ad_last)
+#define	extent_tree_ad_new JEMALLOC_N(extent_tree_ad_new)
+#define	extent_tree_ad_next JEMALLOC_N(extent_tree_ad_next)
+#define	extent_tree_ad_nsearch JEMALLOC_N(extent_tree_ad_nsearch)
+#define	extent_tree_ad_prev JEMALLOC_N(extent_tree_ad_prev)
+#define	extent_tree_ad_psearch JEMALLOC_N(extent_tree_ad_psearch)
+#define	extent_tree_ad_remove JEMALLOC_N(extent_tree_ad_remove)
+#define	extent_tree_ad_reverse_iter JEMALLOC_N(extent_tree_ad_reverse_iter)
+#define	extent_tree_ad_reverse_iter_recurse JEMALLOC_N(extent_tree_ad_reverse_iter_recurse)
+#define	extent_tree_ad_reverse_iter_start JEMALLOC_N(extent_tree_ad_reverse_iter_start)
+#define	extent_tree_ad_search JEMALLOC_N(extent_tree_ad_search)
+#define	extent_tree_szad_first JEMALLOC_N(extent_tree_szad_first)
+#define	extent_tree_szad_insert JEMALLOC_N(extent_tree_szad_insert)
+#define	extent_tree_szad_iter JEMALLOC_N(extent_tree_szad_iter)
+#define	extent_tree_szad_iter_recurse JEMALLOC_N(extent_tree_szad_iter_recurse)
+#define	extent_tree_szad_iter_start JEMALLOC_N(extent_tree_szad_iter_start)
+#define	extent_tree_szad_last JEMALLOC_N(extent_tree_szad_last)
+#define	extent_tree_szad_new JEMALLOC_N(extent_tree_szad_new)
+#define	extent_tree_szad_next JEMALLOC_N(extent_tree_szad_next)
+#define	extent_tree_szad_nsearch JEMALLOC_N(extent_tree_szad_nsearch)
+#define	extent_tree_szad_prev JEMALLOC_N(extent_tree_szad_prev)
+#define	extent_tree_szad_psearch JEMALLOC_N(extent_tree_szad_psearch)
+#define	extent_tree_szad_remove JEMALLOC_N(extent_tree_szad_remove)
+#define	extent_tree_szad_reverse_iter JEMALLOC_N(extent_tree_szad_reverse_iter)
+#define	extent_tree_szad_reverse_iter_recurse JEMALLOC_N(extent_tree_szad_reverse_iter_recurse)
+#define	extent_tree_szad_reverse_iter_start JEMALLOC_N(extent_tree_szad_reverse_iter_start)
+#define	extent_tree_szad_search JEMALLOC_N(extent_tree_szad_search)
+#define	hash JEMALLOC_N(hash)
+#define	huge_boot JEMALLOC_N(huge_boot)
+#define	huge_dalloc JEMALLOC_N(huge_dalloc)
+#define	huge_malloc JEMALLOC_N(huge_malloc)
+#define	huge_palloc JEMALLOC_N(huge_palloc)
+#define	huge_postfork_child JEMALLOC_N(huge_postfork_child)
+#define	huge_postfork_parent JEMALLOC_N(huge_postfork_parent)
+#define	huge_prefork JEMALLOC_N(huge_prefork)
+#define	huge_prof_ctx_get JEMALLOC_N(huge_prof_ctx_get)
+#define	huge_prof_ctx_set JEMALLOC_N(huge_prof_ctx_set)
+#define	huge_ralloc JEMALLOC_N(huge_ralloc)
+#define	huge_ralloc_no_move JEMALLOC_N(huge_ralloc_no_move)
+#define	huge_salloc JEMALLOC_N(huge_salloc)
+#define	iallocm JEMALLOC_N(iallocm)
+#define	icalloc JEMALLOC_N(icalloc)
+#define	idalloc JEMALLOC_N(idalloc)
+#define	imalloc JEMALLOC_N(imalloc)
+#define	ipalloc JEMALLOC_N(ipalloc)
+#define	iqalloc JEMALLOC_N(iqalloc)
+#define	iralloc JEMALLOC_N(iralloc)
+#define	isalloc JEMALLOC_N(isalloc)
+#define	ivsalloc JEMALLOC_N(ivsalloc)
+#define	jemalloc_postfork_child JEMALLOC_N(jemalloc_postfork_child)
+#define	jemalloc_postfork_parent JEMALLOC_N(jemalloc_postfork_parent)
+#define	jemalloc_prefork JEMALLOC_N(jemalloc_prefork)
+#define	malloc_cprintf JEMALLOC_N(malloc_cprintf)
+#define	malloc_mutex_init JEMALLOC_N(malloc_mutex_init)
+#define	malloc_mutex_lock JEMALLOC_N(malloc_mutex_lock)
+#define	malloc_mutex_postfork_child JEMALLOC_N(malloc_mutex_postfork_child)
+#define	malloc_mutex_postfork_parent JEMALLOC_N(malloc_mutex_postfork_parent)
+#define	malloc_mutex_prefork JEMALLOC_N(malloc_mutex_prefork)
+#define	malloc_mutex_unlock JEMALLOC_N(malloc_mutex_unlock)
+#define	malloc_printf JEMALLOC_N(malloc_printf)
+#define	malloc_snprintf JEMALLOC_N(malloc_snprintf)
+#define	malloc_strtoumax JEMALLOC_N(malloc_strtoumax)
+#define	malloc_tsd_boot JEMALLOC_N(malloc_tsd_boot)
+#define	malloc_tsd_cleanup_register JEMALLOC_N(malloc_tsd_cleanup_register)
+#define	malloc_tsd_dalloc JEMALLOC_N(malloc_tsd_dalloc)
+#define	malloc_tsd_malloc JEMALLOC_N(malloc_tsd_malloc)
+#define	malloc_tsd_no_cleanup JEMALLOC_N(malloc_tsd_no_cleanup)
+#define	malloc_vcprintf JEMALLOC_N(malloc_vcprintf)
+#define	malloc_vsnprintf JEMALLOC_N(malloc_vsnprintf)
+#define	malloc_write JEMALLOC_N(malloc_write)
+#define	mb_write JEMALLOC_N(mb_write)
+#define	mmap_unaligned_tsd_boot JEMALLOC_N(mmap_unaligned_tsd_boot)
+#define	mmap_unaligned_tsd_cleanup_wrapper JEMALLOC_N(mmap_unaligned_tsd_cleanup_wrapper)
+#define	mmap_unaligned_tsd_get JEMALLOC_N(mmap_unaligned_tsd_get)
+#define	mmap_unaligned_tsd_set JEMALLOC_N(mmap_unaligned_tsd_set)
+#define	mutex_boot JEMALLOC_N(mutex_boot)
+#define	opt_abort JEMALLOC_N(opt_abort)
+#define	opt_junk JEMALLOC_N(opt_junk)
+#define	opt_lg_chunk JEMALLOC_N(opt_lg_chunk)
+#define	opt_lg_dirty_mult JEMALLOC_N(opt_lg_dirty_mult)
+#define	opt_lg_prof_interval JEMALLOC_N(opt_lg_prof_interval)
+#define	opt_lg_prof_sample JEMALLOC_N(opt_lg_prof_sample)
+#define	opt_lg_tcache_max JEMALLOC_N(opt_lg_tcache_max)
+#define	opt_narenas JEMALLOC_N(opt_narenas)
+#define	opt_prof JEMALLOC_N(opt_prof)
+#define	opt_prof_accum JEMALLOC_N(opt_prof_accum)
+#define	opt_prof_active JEMALLOC_N(opt_prof_active)
+#define	opt_prof_gdump JEMALLOC_N(opt_prof_gdump)
+#define	opt_prof_leak JEMALLOC_N(opt_prof_leak)
+#define	opt_stats_print JEMALLOC_N(opt_stats_print)
+#define	opt_tcache JEMALLOC_N(opt_tcache)
+#define	opt_utrace JEMALLOC_N(opt_utrace)
+#define	opt_xmalloc JEMALLOC_N(opt_xmalloc)
+#define	opt_zero JEMALLOC_N(opt_zero)
+#define	p2rz JEMALLOC_N(p2rz)
+#define	pow2_ceil JEMALLOC_N(pow2_ceil)
+#define	prof_backtrace JEMALLOC_N(prof_backtrace)
+#define	prof_boot0 JEMALLOC_N(prof_boot0)
+#define	prof_boot1 JEMALLOC_N(prof_boot1)
+#define	prof_boot2 JEMALLOC_N(prof_boot2)
+#define	prof_ctx_get JEMALLOC_N(prof_ctx_get)
+#define	prof_ctx_set JEMALLOC_N(prof_ctx_set)
+#define	prof_free JEMALLOC_N(prof_free)
+#define	prof_gdump JEMALLOC_N(prof_gdump)
+#define	prof_idump JEMALLOC_N(prof_idump)
+#define	prof_lookup JEMALLOC_N(prof_lookup)
+#define	prof_malloc JEMALLOC_N(prof_malloc)
+#define	prof_mdump JEMALLOC_N(prof_mdump)
+#define	prof_realloc JEMALLOC_N(prof_realloc)
+#define	prof_sample_accum_update JEMALLOC_N(prof_sample_accum_update)
+#define	prof_sample_threshold_update JEMALLOC_N(prof_sample_threshold_update)
+#define	prof_tdata_cleanup JEMALLOC_N(prof_tdata_cleanup)
+#define	prof_tdata_tsd_boot JEMALLOC_N(prof_tdata_tsd_boot)
+#define	prof_tdata_tsd_cleanup_wrapper JEMALLOC_N(prof_tdata_tsd_cleanup_wrapper)
+#define	prof_tdata_tsd_get JEMALLOC_N(prof_tdata_tsd_get)
+#define	prof_tdata_tsd_set JEMALLOC_N(prof_tdata_tsd_set)
+#define	pthread_create JEMALLOC_N(pthread_create)
+#define	quarantine JEMALLOC_N(quarantine)
+#define	quarantine_boot JEMALLOC_N(quarantine_boot)
+#define	quarantine_tsd_boot JEMALLOC_N(quarantine_tsd_boot)
+#define	quarantine_tsd_cleanup_wrapper JEMALLOC_N(quarantine_tsd_cleanup_wrapper)
+#define	quarantine_tsd_get JEMALLOC_N(quarantine_tsd_get)
+#define	quarantine_tsd_set JEMALLOC_N(quarantine_tsd_set)
+#define	register_zone JEMALLOC_N(register_zone)
+#define	rtree_get JEMALLOC_N(rtree_get)
+#define	rtree_get_locked JEMALLOC_N(rtree_get_locked)
+#define	rtree_new JEMALLOC_N(rtree_new)
+#define	rtree_set JEMALLOC_N(rtree_set)
+#define	s2u JEMALLOC_N(s2u)
+#define	sa2u JEMALLOC_N(sa2u)
+#define	stats_arenas_i_bins_j_index JEMALLOC_N(stats_arenas_i_bins_j_index)
+#define	stats_arenas_i_index JEMALLOC_N(stats_arenas_i_index)
+#define	stats_arenas_i_lruns_j_index JEMALLOC_N(stats_arenas_i_lruns_j_index)
+#define	stats_cactive JEMALLOC_N(stats_cactive)
+#define	stats_cactive_add JEMALLOC_N(stats_cactive_add)
+#define	stats_cactive_get JEMALLOC_N(stats_cactive_get)
+#define	stats_cactive_sub JEMALLOC_N(stats_cactive_sub)
+#define	stats_print JEMALLOC_N(stats_print)
+#define	tcache_alloc_easy JEMALLOC_N(tcache_alloc_easy)
+#define	tcache_alloc_large JEMALLOC_N(tcache_alloc_large)
+#define	tcache_alloc_small JEMALLOC_N(tcache_alloc_small)
+#define	tcache_alloc_small_hard JEMALLOC_N(tcache_alloc_small_hard)
+#define	tcache_arena_associate JEMALLOC_N(tcache_arena_associate)
+#define	tcache_arena_dissociate JEMALLOC_N(tcache_arena_dissociate)
+#define	tcache_bin_flush_large JEMALLOC_N(tcache_bin_flush_large)
+#define	tcache_bin_flush_small JEMALLOC_N(tcache_bin_flush_small)
+#define	tcache_boot0 JEMALLOC_N(tcache_boot0)
+#define	tcache_boot1 JEMALLOC_N(tcache_boot1)
+#define	tcache_create JEMALLOC_N(tcache_create)
+#define	tcache_dalloc_large JEMALLOC_N(tcache_dalloc_large)
+#define	tcache_dalloc_small JEMALLOC_N(tcache_dalloc_small)
+#define	tcache_destroy JEMALLOC_N(tcache_destroy)
+#define	tcache_enabled_get JEMALLOC_N(tcache_enabled_get)
+#define	tcache_enabled_set JEMALLOC_N(tcache_enabled_set)
+#define	tcache_enabled_tsd_boot JEMALLOC_N(tcache_enabled_tsd_boot)
+#define	tcache_enabled_tsd_cleanup_wrapper JEMALLOC_N(tcache_enabled_tsd_cleanup_wrapper)
+#define	tcache_enabled_tsd_get JEMALLOC_N(tcache_enabled_tsd_get)
+#define	tcache_enabled_tsd_set JEMALLOC_N(tcache_enabled_tsd_set)
+#define	tcache_event JEMALLOC_N(tcache_event)
+#define	tcache_flush JEMALLOC_N(tcache_flush)
+#define	tcache_stats_merge JEMALLOC_N(tcache_stats_merge)
+#define	tcache_thread_cleanup JEMALLOC_N(tcache_thread_cleanup)
+#define	tcache_tsd_boot JEMALLOC_N(tcache_tsd_boot)
+#define	tcache_tsd_cleanup_wrapper JEMALLOC_N(tcache_tsd_cleanup_wrapper)
+#define	tcache_tsd_get JEMALLOC_N(tcache_tsd_get)
+#define	tcache_tsd_set JEMALLOC_N(tcache_tsd_set)
+#define	thread_allocated_tsd_boot JEMALLOC_N(thread_allocated_tsd_boot)
+#define	thread_allocated_tsd_cleanup_wrapper JEMALLOC_N(thread_allocated_tsd_cleanup_wrapper)
+#define	thread_allocated_tsd_get JEMALLOC_N(thread_allocated_tsd_get)
+#define	thread_allocated_tsd_set JEMALLOC_N(thread_allocated_tsd_set)
+#define	u2rz JEMALLOC_N(u2rz)

contrib/jemalloc/include/jemalloc/internal/prng.h

@@ -0,0 +1,60 @@
+/******************************************************************************/
+#ifdef JEMALLOC_H_TYPES
+
+/*
+ * Simple linear congruential pseudo-random number generator:
+ *
+ *   prng(y) = (a*x + c) % m
+ *
+ * where the following constants ensure maximal period:
+ *
+ *   a == Odd number (relatively prime to 2^n), and (a-1) is a multiple of 4.
+ *   c == Odd number (relatively prime to 2^n).
+ *   m == 2^32
+ *
+ * See Knuth's TAOCP 3rd Ed., Vol. 2, pg. 17 for details on these constraints.
+ *
+ * This choice of m has the disadvantage that the quality of the bits is
+ * proportional to bit position.  For example. the lowest bit has a cycle of 2,
+ * the next has a cycle of 4, etc.  For this reason, we prefer to use the upper
+ * bits.
+ *
+ * Macro parameters:
+ *   uint32_t r          : Result.
+ *   unsigned lg_range   : (0..32], number of least significant bits to return.
+ *   uint32_t state      : Seed value.
+ *   const uint32_t a, c : See above discussion.
+ */
+#define prng32(r, lg_range, state, a, c) do {				\
+	assert(lg_range > 0);						\
+	assert(lg_range <= 32);						\
+									\
+	r = (state * (a)) + (c);					\
+	state = r;							\
+	r >>= (32 - lg_range);						\
+} while (false)
+
+/* Same as prng32(), but 64 bits of pseudo-randomness, using uint64_t. */
+#define prng64(r, lg_range, state, a, c) do {				\
+	assert(lg_range > 0);						\
+	assert(lg_range <= 64);						\
+									\
+	r = (state * (a)) + (c);					\
+	state = r;							\
+	r >>= (64 - lg_range);						\
+} while (false)
+
+#endif /* JEMALLOC_H_TYPES */
+/******************************************************************************/
+#ifdef JEMALLOC_H_STRUCTS
+
+#endif /* JEMALLOC_H_STRUCTS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_EXTERNS
+
+#endif /* JEMALLOC_H_EXTERNS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_INLINES
+
+#endif /* JEMALLOC_H_INLINES */
+/******************************************************************************/

contrib/jemalloc/include/jemalloc/internal/prof.h

@@ -0,0 +1,535 @@
+/******************************************************************************/
+#ifdef JEMALLOC_H_TYPES
+
+typedef struct prof_bt_s prof_bt_t;
+typedef struct prof_cnt_s prof_cnt_t;
+typedef struct prof_thr_cnt_s prof_thr_cnt_t;
+typedef struct prof_ctx_s prof_ctx_t;
+typedef struct prof_tdata_s prof_tdata_t;
+
+/* Option defaults. */
+#define	PROF_PREFIX_DEFAULT		"jeprof"
+#define	LG_PROF_SAMPLE_DEFAULT		0
+#define	LG_PROF_INTERVAL_DEFAULT	-1
+
+/*
+ * Hard limit on stack backtrace depth.  The version of prof_backtrace() that
+ * is based on __builtin_return_address() necessarily has a hard-coded number
+ * of backtrace frame handlers, and should be kept in sync with this setting.
+ */
+#define	PROF_BT_MAX			128
+
+/* Maximum number of backtraces to store in each per thread LRU cache. */
+#define	PROF_TCMAX			1024
+
+/* Initial hash table size. */
+#define	PROF_CKH_MINITEMS		64
+
+/* Size of memory buffer to use when writing dump files. */
+#define	PROF_DUMP_BUFSIZE		65536
+
+/* Size of stack-allocated buffer used by prof_printf(). */
+#define	PROF_PRINTF_BUFSIZE		128
+
+/*
+ * Number of mutexes shared among all ctx's.  No space is allocated for these
+ * unless profiling is enabled, so it's okay to over-provision.
+ */
+#define	PROF_NCTX_LOCKS			1024
+
+#endif /* JEMALLOC_H_TYPES */
+/******************************************************************************/
+#ifdef JEMALLOC_H_STRUCTS
+
+struct prof_bt_s {
+	/* Backtrace, stored as len program counters. */
+	void		**vec;
+	unsigned	len;
+};
+
+#ifdef JEMALLOC_PROF_LIBGCC
+/* Data structure passed to libgcc _Unwind_Backtrace() callback functions. */
+typedef struct {
+	prof_bt_t	*bt;
+	unsigned	nignore;
+	unsigned	max;
+} prof_unwind_data_t;
+#endif
+
+struct prof_cnt_s {
+	/*
+	 * Profiling counters.  An allocation/deallocation pair can operate on
+	 * different prof_thr_cnt_t objects that are linked into the same
+	 * prof_ctx_t cnts_ql, so it is possible for the cur* counters to go
+	 * negative.  In principle it is possible for the *bytes counters to
+	 * overflow/underflow, but a general solution would require something
+	 * like 128-bit counters; this implementation doesn't bother to solve
+	 * that problem.
+	 */
+	int64_t		curobjs;
+	int64_t		curbytes;
+	uint64_t	accumobjs;
+	uint64_t	accumbytes;
+};
+
+struct prof_thr_cnt_s {
+	/* Linkage into prof_ctx_t's cnts_ql. */
+	ql_elm(prof_thr_cnt_t)	cnts_link;
+
+	/* Linkage into thread's LRU. */
+	ql_elm(prof_thr_cnt_t)	lru_link;
+
+	/*
+	 * Associated context.  If a thread frees an object that it did not
+	 * allocate, it is possible that the context is not cached in the
+	 * thread's hash table, in which case it must be able to look up the
+	 * context, insert a new prof_thr_cnt_t into the thread's hash table,
+	 * and link it into the prof_ctx_t's cnts_ql.
+	 */
+	prof_ctx_t		*ctx;
+
+	/*
+	 * Threads use memory barriers to update the counters.  Since there is
+	 * only ever one writer, the only challenge is for the reader to get a
+	 * consistent read of the counters.
+	 *
+	 * The writer uses this series of operations:
+	 *
+	 * 1) Increment epoch to an odd number.
+	 * 2) Update counters.
+	 * 3) Increment epoch to an even number.
+	 *
+	 * The reader must assure 1) that the epoch is even while it reads the
+	 * counters, and 2) that the epoch doesn't change between the time it
+	 * starts and finishes reading the counters.
+	 */
+	unsigned		epoch;
+
+	/* Profiling counters. */
+	prof_cnt_t		cnts;
+};
+
+struct prof_ctx_s {
+	/* Associated backtrace. */
+	prof_bt_t		*bt;
+
+	/* Protects cnt_merged and cnts_ql. */
+	malloc_mutex_t		*lock;
+
+	/* Temporary storage for summation during dump. */
+	prof_cnt_t		cnt_summed;
+
+	/* When threads exit, they merge their stats into cnt_merged. */
+	prof_cnt_t		cnt_merged;
+
+	/*
+	 * List of profile counters, one for each thread that has allocated in
+	 * this context.
+	 */
+	ql_head(prof_thr_cnt_t)	cnts_ql;
+};
+
+struct prof_tdata_s {
+	/*
+	 * Hash of (prof_bt_t *)-->(prof_thr_cnt_t *).  Each thread keeps a
+	 * cache of backtraces, with associated thread-specific prof_thr_cnt_t
+	 * objects.  Other threads may read the prof_thr_cnt_t contents, but no
+	 * others will ever write them.
+	 *
+	 * Upon thread exit, the thread must merge all the prof_thr_cnt_t
+	 * counter data into the associated prof_ctx_t objects, and unlink/free
+	 * the prof_thr_cnt_t objects.
+	 */
+	ckh_t			bt2cnt;
+
+	/* LRU for contents of bt2cnt. */
+	ql_head(prof_thr_cnt_t)	lru_ql;
+
+	/* Backtrace vector, used for calls to prof_backtrace(). */
+	void			**vec;
+
+	/* Sampling state. */
+	uint64_t		prng_state;
+	uint64_t		threshold;
+	uint64_t		accum;
+};
+
+#endif /* JEMALLOC_H_STRUCTS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_EXTERNS
+
+extern bool	opt_prof;
+/*
+ * Even if opt_prof is true, sampling can be temporarily disabled by setting
+ * opt_prof_active to false.  No locking is used when updating opt_prof_active,
+ * so there are no guarantees regarding how long it will take for all threads
+ * to notice state changes.
+ */
+extern bool	opt_prof_active;
+extern size_t	opt_lg_prof_sample;   /* Mean bytes between samples. */
+extern ssize_t	opt_lg_prof_interval; /* lg(prof_interval). */
+extern bool	opt_prof_gdump;       /* High-water memory dumping. */
+extern bool	opt_prof_leak;        /* Dump leak summary at exit. */
+extern bool	opt_prof_accum;       /* Report cumulative bytes. */
+extern char	opt_prof_prefix[PATH_MAX + 1];
+
+/*
+ * Profile dump interval, measured in bytes allocated.  Each arena triggers a
+ * profile dump when it reaches this threshold.  The effect is that the
+ * interval between profile dumps averages prof_interval, though the actual
+ * interval between dumps will tend to be sporadic, and the interval will be a
+ * maximum of approximately (prof_interval * narenas).
+ */
+extern uint64_t	prof_interval;
+
+/*
+ * If true, promote small sampled objects to large objects, since small run
+ * headers do not have embedded profile context pointers.
+ */
+extern bool	prof_promote;
+
+void	bt_init(prof_bt_t *bt, void **vec);
+void	prof_backtrace(prof_bt_t *bt, unsigned nignore);
+prof_thr_cnt_t	*prof_lookup(prof_bt_t *bt);
+void	prof_idump(void);
+bool	prof_mdump(const char *filename);
+void	prof_gdump(void);
+prof_tdata_t	*prof_tdata_init(void);
+void	prof_tdata_cleanup(void *arg);
+void	prof_boot0(void);
+void	prof_boot1(void);
+bool	prof_boot2(void);
+
+#endif /* JEMALLOC_H_EXTERNS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_INLINES
+
+#define	PROF_ALLOC_PREP(nignore, size, ret) do {			\
+	prof_tdata_t *prof_tdata;					\
+	prof_bt_t bt;							\
+									\
+	assert(size == s2u(size));					\
+									\
+	prof_tdata = *prof_tdata_tsd_get();				\
+	if (prof_tdata == NULL) {					\
+		prof_tdata = prof_tdata_init();				\
+		if (prof_tdata == NULL) {				\
+			ret = NULL;					\
+			break;						\
+		}							\
+	}								\
+									\
+	if (opt_prof_active == false) {					\
+		/* Sampling is currently inactive, so avoid sampling. */\
+		ret = (prof_thr_cnt_t *)(uintptr_t)1U;			\
+	} else if (opt_lg_prof_sample == 0) {				\
+		/* Don't bother with sampling logic, since sampling   */\
+		/* interval is 1.                                     */\
+		bt_init(&bt, prof_tdata->vec);				\
+		prof_backtrace(&bt, nignore);				\
+		ret = prof_lookup(&bt);					\
+	} else {							\
+		if (prof_tdata->threshold == 0) {			\
+			/* Initialize.  Seed the prng differently for */\
+			/* each thread.                               */\
+			prof_tdata->prng_state =			\
+			    (uint64_t)(uintptr_t)&size;			\
+			prof_sample_threshold_update(prof_tdata);	\
+		}							\
+									\
+		/* Determine whether to capture a backtrace based on  */\
+		/* whether size is enough for prof_accum to reach     */\
+		/* prof_tdata->threshold.  However, delay updating    */\
+		/* these variables until prof_{m,re}alloc(), because  */\
+		/* we don't know for sure that the allocation will    */\
+		/* succeed.                                           */\
+		/*                                                    */\
+		/* Use subtraction rather than addition to avoid      */\
+		/* potential integer overflow.                        */\
+		if (size >= prof_tdata->threshold -			\
+		    prof_tdata->accum) {				\
+			bt_init(&bt, prof_tdata->vec);			\
+			prof_backtrace(&bt, nignore);			\
+			ret = prof_lookup(&bt);				\
+		} else							\
+			ret = (prof_thr_cnt_t *)(uintptr_t)1U;		\
+	}								\
+} while (0)
+
+#ifndef JEMALLOC_ENABLE_INLINE
+malloc_tsd_protos(JEMALLOC_ATTR(unused), prof_tdata, prof_tdata_t *)
+
+void	prof_sample_threshold_update(prof_tdata_t *prof_tdata);
+prof_ctx_t	*prof_ctx_get(const void *ptr);
+void	prof_ctx_set(const void *ptr, prof_ctx_t *ctx);
+bool	prof_sample_accum_update(size_t size);
+void	prof_malloc(const void *ptr, size_t size, prof_thr_cnt_t *cnt);
+void	prof_realloc(const void *ptr, size_t size, prof_thr_cnt_t *cnt,
+    size_t old_size, prof_ctx_t *old_ctx);
+void	prof_free(const void *ptr, size_t size);
+#endif
+
+#if (defined(JEMALLOC_ENABLE_INLINE) || defined(JEMALLOC_PROF_C_))
+/* Thread-specific backtrace cache, used to reduce bt2ctx contention. */
+malloc_tsd_externs(prof_tdata, prof_tdata_t *)
+malloc_tsd_funcs(JEMALLOC_INLINE, prof_tdata, prof_tdata_t *, NULL,
+    prof_tdata_cleanup)
+
+JEMALLOC_INLINE void
+prof_sample_threshold_update(prof_tdata_t *prof_tdata)
+{
+	uint64_t r;
+	double u;
+
+	cassert(config_prof);
+
+	/*
+	 * Compute sample threshold as a geometrically distributed random
+	 * variable with mean (2^opt_lg_prof_sample).
+	 *
+	 *                         __        __
+	 *                         |  log(u)  |                     1
+	 * prof_tdata->threshold = | -------- |, where p = -------------------
+	 *                         | log(1-p) |             opt_lg_prof_sample
+	 *                                                 2
+	 *
+	 * For more information on the math, see:
+	 *
+	 *   Non-Uniform Random Variate Generation
+	 *   Luc Devroye
+	 *   Springer-Verlag, New York, 1986
+	 *   pp 500
+	 *   (http://cg.scs.carleton.ca/~luc/rnbookindex.html)
+	 */
+	prng64(r, 53, prof_tdata->prng_state,
+	    UINT64_C(6364136223846793005), UINT64_C(1442695040888963407));
+	u = (double)r * (1.0/9007199254740992.0L);
+	prof_tdata->threshold = (uint64_t)(log(u) /
+	    log(1.0 - (1.0 / (double)((uint64_t)1U << opt_lg_prof_sample))))
+	    + (uint64_t)1U;
+}
+
+JEMALLOC_INLINE prof_ctx_t *
+prof_ctx_get(const void *ptr)
+{
+	prof_ctx_t *ret;
+	arena_chunk_t *chunk;
+
+	cassert(config_prof);
+	assert(ptr != NULL);
+
+	chunk = (arena_chunk_t *)CHUNK_ADDR2BASE(ptr);
+	if (chunk != ptr) {
+		/* Region. */
+		ret = arena_prof_ctx_get(ptr);
+	} else
+		ret = huge_prof_ctx_get(ptr);
+
+	return (ret);
+}
+
+JEMALLOC_INLINE void
+prof_ctx_set(const void *ptr, prof_ctx_t *ctx)
+{
+	arena_chunk_t *chunk;
+
+	cassert(config_prof);
+	assert(ptr != NULL);
+
+	chunk = (arena_chunk_t *)CHUNK_ADDR2BASE(ptr);
+	if (chunk != ptr) {
+		/* Region. */
+		arena_prof_ctx_set(ptr, ctx);
+	} else
+		huge_prof_ctx_set(ptr, ctx);
+}
+
+JEMALLOC_INLINE bool
+prof_sample_accum_update(size_t size)
+{
+	prof_tdata_t *prof_tdata;
+
+	cassert(config_prof);
+	/* Sampling logic is unnecessary if the interval is 1. */
+	assert(opt_lg_prof_sample != 0);
+
+	prof_tdata = *prof_tdata_tsd_get();
+	assert(prof_tdata != NULL);
+
+	/* Take care to avoid integer overflow. */
+	if (size >= prof_tdata->threshold - prof_tdata->accum) {
+		prof_tdata->accum -= (prof_tdata->threshold - size);
+		/* Compute new sample threshold. */
+		prof_sample_threshold_update(prof_tdata);
+		while (prof_tdata->accum >= prof_tdata->threshold) {
+			prof_tdata->accum -= prof_tdata->threshold;
+			prof_sample_threshold_update(prof_tdata);
+		}
+		return (false);
+	} else {
+		prof_tdata->accum += size;
+		return (true);
+	}
+}
+
+JEMALLOC_INLINE void
+prof_malloc(const void *ptr, size_t size, prof_thr_cnt_t *cnt)
+{
+
+	cassert(config_prof);
+	assert(ptr != NULL);
+	assert(size == isalloc(ptr, true));
+
+	if (opt_lg_prof_sample != 0) {
+		if (prof_sample_accum_update(size)) {
+			/*
+			 * Don't sample.  For malloc()-like allocation, it is
+			 * always possible to tell in advance how large an
+			 * object's usable size will be, so there should never
+			 * be a difference between the size passed to
+			 * PROF_ALLOC_PREP() and prof_malloc().
+			 */
+			assert((uintptr_t)cnt == (uintptr_t)1U);
+		}
+	}
+
+	if ((uintptr_t)cnt > (uintptr_t)1U) {
+		prof_ctx_set(ptr, cnt->ctx);
+
+		cnt->epoch++;
+		/*********/
+		mb_write();
+		/*********/
+		cnt->cnts.curobjs++;
+		cnt->cnts.curbytes += size;
+		if (opt_prof_accum) {
+			cnt->cnts.accumobjs++;
+			cnt->cnts.accumbytes += size;
+		}
+		/*********/
+		mb_write();
+		/*********/
+		cnt->epoch++;
+		/*********/
+		mb_write();
+		/*********/
+	} else
+		prof_ctx_set(ptr, (prof_ctx_t *)(uintptr_t)1U);
+}
+
+JEMALLOC_INLINE void
+prof_realloc(const void *ptr, size_t size, prof_thr_cnt_t *cnt,
+    size_t old_size, prof_ctx_t *old_ctx)
+{
+	prof_thr_cnt_t *told_cnt;
+
+	cassert(config_prof);
+	assert(ptr != NULL || (uintptr_t)cnt <= (uintptr_t)1U);
+
+	if (ptr != NULL) {
+		assert(size == isalloc(ptr, true));
+		if (opt_lg_prof_sample != 0) {
+			if (prof_sample_accum_update(size)) {
+				/*
+				 * Don't sample.  The size passed to
+				 * PROF_ALLOC_PREP() was larger than what
+				 * actually got allocated, so a backtrace was
+				 * captured for this allocation, even though
+				 * its actual size was insufficient to cross
+				 * the sample threshold.
+				 */
+				cnt = (prof_thr_cnt_t *)(uintptr_t)1U;
+			}
+		}
+	}
+
+	if ((uintptr_t)old_ctx > (uintptr_t)1U) {
+		told_cnt = prof_lookup(old_ctx->bt);
+		if (told_cnt == NULL) {
+			/*
+			 * It's too late to propagate OOM for this realloc(),
+			 * so operate directly on old_cnt->ctx->cnt_merged.
+			 */
+			malloc_mutex_lock(old_ctx->lock);
+			old_ctx->cnt_merged.curobjs--;
+			old_ctx->cnt_merged.curbytes -= old_size;
+			malloc_mutex_unlock(old_ctx->lock);
+			told_cnt = (prof_thr_cnt_t *)(uintptr_t)1U;
+		}
+	} else
+		told_cnt = (prof_thr_cnt_t *)(uintptr_t)1U;
+
+	if ((uintptr_t)told_cnt > (uintptr_t)1U)
+		told_cnt->epoch++;
+	if ((uintptr_t)cnt > (uintptr_t)1U) {
+		prof_ctx_set(ptr, cnt->ctx);
+		cnt->epoch++;
+	} else
+		prof_ctx_set(ptr, (prof_ctx_t *)(uintptr_t)1U);
+	/*********/
+	mb_write();
+	/*********/
+	if ((uintptr_t)told_cnt > (uintptr_t)1U) {
+		told_cnt->cnts.curobjs--;
+		told_cnt->cnts.curbytes -= old_size;
+	}
+	if ((uintptr_t)cnt > (uintptr_t)1U) {
+		cnt->cnts.curobjs++;
+		cnt->cnts.curbytes += size;
+		if (opt_prof_accum) {
+			cnt->cnts.accumobjs++;
+			cnt->cnts.accumbytes += size;
+		}
+	}
+	/*********/
+	mb_write();
+	/*********/
+	if ((uintptr_t)told_cnt > (uintptr_t)1U)
+		told_cnt->epoch++;
+	if ((uintptr_t)cnt > (uintptr_t)1U)
+		cnt->epoch++;
+	/*********/
+	mb_write(); /* Not strictly necessary. */
+}
+
+JEMALLOC_INLINE void
+prof_free(const void *ptr, size_t size)
+{
+	prof_ctx_t *ctx = prof_ctx_get(ptr);
+
+	cassert(config_prof);
+
+	if ((uintptr_t)ctx > (uintptr_t)1) {
+		assert(size == isalloc(ptr, true));
+		prof_thr_cnt_t *tcnt = prof_lookup(ctx->bt);
+
+		if (tcnt != NULL) {
+			tcnt->epoch++;
+			/*********/
+			mb_write();
+			/*********/
+			tcnt->cnts.curobjs--;
+			tcnt->cnts.curbytes -= size;
+			/*********/
+			mb_write();
+			/*********/
+			tcnt->epoch++;
+			/*********/
+			mb_write();
+			/*********/
+		} else {
+			/*
+			 * OOM during free() cannot be propagated, so operate
+			 * directly on cnt->ctx->cnt_merged.
+			 */
+			malloc_mutex_lock(ctx->lock);
+			ctx->cnt_merged.curobjs--;
+			ctx->cnt_merged.curbytes -= size;
+			malloc_mutex_unlock(ctx->lock);
+		}
+	}
+}
+#endif
+
+#endif /* JEMALLOC_H_INLINES */
+/******************************************************************************/

contrib/jemalloc/include/jemalloc/internal/ql.h

@@ -1,40 +1,3 @@
-/******************************************************************************
- *
- * Copyright (C) 2002 Jason Evans <jasone@FreeBSD.org>.
- * All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions
- * are met:
- * 1. Redistributions of source code must retain the above copyright
- *    notice(s), this list of conditions and the following disclaimer
- *    unmodified other than the allowable addition of one or more
- *    copyright notices.
- * 2. Redistributions in binary form must reproduce the above copyright
- *    notice(s), this list of conditions and the following disclaimer in
- *    the documentation and/or other materials provided with the
- *    distribution.
- *
- * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY
- * EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
- * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
- * PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE COPYRIGHT HOLDER(S) BE
- * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
- * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
- * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR
- * BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
- * WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
- * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE,
- * EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
- *
- ******************************************************************************/
-
-#ifndef QL_H_
-#define	QL_H_
-
-#include <sys/cdefs.h>
-__FBSDID("$FreeBSD$");
-
 /*
  * List definitions.
  */
@@ -118,5 +81,3 @@ struct {								\
 
 #define ql_reverse_foreach(a_var, a_head, a_field)			\
 	qr_reverse_foreach((a_var), ql_first(a_head), a_field)
-
-#endif /* QL_H_ */

contrib/jemalloc/include/jemalloc/internal/qr.h

@@ -1,40 +1,3 @@
-/******************************************************************************
- *
- * Copyright (C) 2002 Jason Evans <jasone@FreeBSD.org>.
- * All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions
- * are met:
- * 1. Redistributions of source code must retain the above copyright
- *    notice(s), this list of conditions and the following disclaimer
- *    unmodified other than the allowable addition of one or more
- *    copyright notices.
- * 2. Redistributions in binary form must reproduce the above copyright
- *    notice(s), this list of conditions and the following disclaimer in
- *    the documentation and/or other materials provided with the
- *    distribution.
- *
- * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY
- * EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
- * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
- * PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE COPYRIGHT HOLDER(S) BE
- * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
- * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
- * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR
- * BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
- * WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
- * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE,
- * EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
- *
- ******************************************************************************/
-
-#ifndef QR_H_
-#define	QR_H_
-
-#include <sys/cdefs.h>
-__FBSDID("$FreeBSD$");
-
 /* Ring definitions. */
 #define qr(a_type)							\
 struct {								\
@@ -102,5 +65,3 @@ struct {								\
 	    (var) != NULL;						\
 	    (var) = (((var) != (a_qr))					\
 	    ? (var)->a_field.qre_prev : NULL))
-
-#endif /* QR_H_ */

contrib/jemalloc/include/jemalloc/internal/quarantine.h

@@ -0,0 +1,24 @@
+/******************************************************************************/
+#ifdef JEMALLOC_H_TYPES
+
+/* Default per thread quarantine size if valgrind is enabled. */
+#define	JEMALLOC_VALGRIND_QUARANTINE_DEFAULT	(ZU(1) << 24)
+
+#endif /* JEMALLOC_H_TYPES */
+/******************************************************************************/
+#ifdef JEMALLOC_H_STRUCTS
+
+#endif /* JEMALLOC_H_STRUCTS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_EXTERNS
+
+void	quarantine(void *ptr);
+bool	quarantine_boot(void);
+
+#endif /* JEMALLOC_H_EXTERNS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_INLINES
+
+#endif /* JEMALLOC_H_INLINES */
+/******************************************************************************/
+

contrib/jemalloc/include/jemalloc/internal/rb.h

@@ -1,35 +1,6 @@
 /*-
  *******************************************************************************
  *
- * Copyright (C) 2008-2010 Jason Evans <jasone@FreeBSD.org>.
- * All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions
- * are met:
- * 1. Redistributions of source code must retain the above copyright
- *    notice(s), this list of conditions and the following disclaimer
- *    unmodified other than the allowable addition of one or more
- *    copyright notices.
- * 2. Redistributions in binary form must reproduce the above copyright
- *    notice(s), this list of conditions and the following disclaimer in
- *    the documentation and/or other materials provided with the
- *    distribution.
- *
- * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY
- * EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
- * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
- * PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE COPYRIGHT HOLDER(S) BE
- * LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
- * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
- * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR
- * BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
- * WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
- * OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE,
- * EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
- *
- ******************************************************************************
- *
  * cpp macro implementation of left-leaning 2-3 red-black trees.  Parent
  * pointers are not used, and color bits are stored in the least significant
  * bit of right-child pointers (if RB_COMPACT is defined), thus making node
@@ -51,8 +22,9 @@
 #ifndef RB_H_
 #define	RB_H_
 
-#include <sys/cdefs.h>
+#if 0
 __FBSDID("$FreeBSD$");
+#endif
 
 #ifdef RB_COMPACT
 /* Node structure. */
@@ -222,11 +194,10 @@ a_prefix##reverse_iter(a_rbt_type *rbtree, a_type *start,		\
  * Arguments:
  *
  *   a_attr    : Function attribute for generated functions (ex: static).
- *   a_prefix  : Prefix for generated functions (ex: extree_).
- *   a_rb_type : Type for red-black tree data structure (ex: extree_t).
- *   a_type    : Type for red-black tree node data structure (ex:
- *               extree_node_t).
- *   a_field   : Name of red-black tree node linkage (ex: extree_link).
+ *   a_prefix  : Prefix for generated functions (ex: ex_).
+ *   a_rb_type : Type for red-black tree data structure (ex: ex_t).
+ *   a_type    : Type for red-black tree node data structure (ex: ex_node_t).
+ *   a_field   : Name of red-black tree node linkage (ex: ex_link).
  *   a_cmp     : Node comparison function name, with the following prototype:
  *                 int (a_cmp *)(a_type *a_node, a_type *a_other);
  *                                       ^^^^^^
@@ -246,94 +217,94 @@ a_prefix##reverse_iter(a_rbt_type *rbtree, a_type *start,		\
  *   struct ex_node_s {
  *       rb_node(ex_node_t) ex_link;
  *   };
- *   typedef rb(ex_node_t) ex_t;
- *   rb_gen(static, ex_, ex_t, ex_node_t, ex_link, ex_cmp, 1297, 1301)
+ *   typedef rb_tree(ex_node_t) ex_t;
+ *   rb_gen(static, ex_, ex_t, ex_node_t, ex_link, ex_cmp)
  *
  * The following API is generated:
  *
  *   static void
- *   ex_new(ex_t *extree);
+ *   ex_new(ex_t *tree);
  *       Description: Initialize a red-black tree structure.
  *       Args:
- *         extree: Pointer to an uninitialized red-black tree object.
+ *         tree: Pointer to an uninitialized red-black tree object.
  *
  *   static ex_node_t *
- *   ex_first(ex_t *extree);
+ *   ex_first(ex_t *tree);
  *   static ex_node_t *
- *   ex_last(ex_t *extree);
- *       Description: Get the first/last node in extree.
+ *   ex_last(ex_t *tree);
+ *       Description: Get the first/last node in tree.
  *       Args:
- *         extree: Pointer to an initialized red-black tree object.
- *       Ret: First/last node in extree, or NULL if extree is empty.
+ *         tree: Pointer to an initialized red-black tree object.
+ *       Ret: First/last node in tree, or NULL if tree is empty.
  *
  *   static ex_node_t *
- *   ex_next(ex_t *extree, ex_node_t *node);
+ *   ex_next(ex_t *tree, ex_node_t *node);
  *   static ex_node_t *
- *   ex_prev(ex_t *extree, ex_node_t *node);
+ *   ex_prev(ex_t *tree, ex_node_t *node);
  *       Description: Get node's successor/predecessor.
  *       Args:
- *         extree: Pointer to an initialized red-black tree object.
- *         node : A node in extree.
- *       Ret: node's successor/predecessor in extree, or NULL if node is
+ *         tree: Pointer to an initialized red-black tree object.
+ *         node: A node in tree.
+ *       Ret: node's successor/predecessor in tree, or NULL if node is
  *            last/first.
  *
  *   static ex_node_t *
- *   ex_search(ex_t *extree, ex_node_t *key);
+ *   ex_search(ex_t *tree, ex_node_t *key);
  *       Description: Search for node that matches key.
  *       Args:
- *         extree: Pointer to an initialized red-black tree object.
- *         key  : Search key.
- *       Ret: Node in extree that matches key, or NULL if no match.
+ *         tree: Pointer to an initialized red-black tree object.
+ *         key : Search key.
+ *       Ret: Node in tree that matches key, or NULL if no match.
  *
  *   static ex_node_t *
- *   ex_nsearch(ex_t *extree, ex_node_t *key);
+ *   ex_nsearch(ex_t *tree, ex_node_t *key);
  *   static ex_node_t *
- *   ex_psearch(ex_t *extree, ex_node_t *key);
+ *   ex_psearch(ex_t *tree, ex_node_t *key);
  *       Description: Search for node that matches key.  If no match is found,
  *                    return what would be key's successor/predecessor, were
- *                    key in extree.
+ *                    key in tree.
  *       Args:
- *         extree: Pointer to an initialized red-black tree object.
- *         key   : Search key.
- *       Ret: Node in extree that matches key, or if no match, hypothetical
- *            node's successor/predecessor (NULL if no successor/predecessor).
+ *         tree: Pointer to an initialized red-black tree object.
+ *         key : Search key.
+ *       Ret: Node in tree that matches key, or if no match, hypothetical node's
+ *            successor/predecessor (NULL if no successor/predecessor).
  *
  *   static void
- *   ex_insert(ex_t *extree, ex_node_t *node);
- *       Description: Insert node into extree.
+ *   ex_insert(ex_t *tree, ex_node_t *node);
+ *       Description: Insert node into tree.
  *       Args:
- *         extree: Pointer to an initialized red-black tree object.
- *         node  : Node to be inserted into extree.
+ *         tree: Pointer to an initialized red-black tree object.
+ *         node: Node to be inserted into tree.
  *
  *   static void
- *   ex_remove(ex_t *extree, ex_node_t *node);
- *       Description: Remove node from extree.
+ *   ex_remove(ex_t *tree, ex_node_t *node);
+ *       Description: Remove node from tree.
  *       Args:
- *         extree: Pointer to an initialized red-black tree object.
- *         node  : Node in extree to be removed.
+ *         tree: Pointer to an initialized red-black tree object.
+ *         node: Node in tree to be removed.
  *
  *   static ex_node_t *
- *   ex_iter(ex_t *extree, ex_node_t *start, ex_node_t *(*cb)(ex_t *,
+ *   ex_iter(ex_t *tree, ex_node_t *start, ex_node_t *(*cb)(ex_t *,
  *     ex_node_t *, void *), void *arg);
  *   static ex_node_t *
- *   ex_reverse_iter(ex_t *extree, ex_node_t *start, ex_node *(*cb)(ex_t *,
+ *   ex_reverse_iter(ex_t *tree, ex_node_t *start, ex_node *(*cb)(ex_t *,
  *     ex_node_t *, void *), void *arg);
- *       Description: Iterate forward/backward over extree, starting at node.
- *                    If extree is modified, iteration must be immediately
+ *       Description: Iterate forward/backward over tree, starting at node.  If
+ *                    tree is modified, iteration must be immediately
  *                    terminated by the callback function that causes the
  *                    modification.
  *       Args:
- *         extree: Pointer to an initialized red-black tree object.
- *         start : Node at which to start iteration, or NULL to start at
- *                 first/last node.
- *         cb    : Callback function, which is called for each node during
- *                 iteration.  Under normal circumstances the callback function
- *                 should return NULL, which causes iteration to continue.  If a
- *                 callback function returns non-NULL, iteration is immediately
- *                 terminated and the non-NULL return value is returned by the
- *                 iterator.  This is useful for re-starting iteration after
- *                 modifying extree.
- *         arg   : Opaque pointer passed to cb().
+ *         tree : Pointer to an initialized red-black tree object.
+ *         start: Node at which to start iteration, or NULL to start at
+ *                first/last node.
+ *         cb   : Callback function, which is called for each node during
+ *                iteration.  Under normal circumstances the callback function
+ *                should return NULL, which causes iteration to continue.  If a
+ *                callback function returns non-NULL, iteration is immediately
+ *                terminated and the non-NULL return value is returned by the
+ *                iterator.  This is useful for re-starting iteration after
+ *                modifying tree.
+ *         arg  : Opaque pointer passed to cb().
  *       Ret: NULL if iteration completed, or the non-NULL callback return value
  *            that caused termination of the iteration.
  */

contrib/jemalloc/include/jemalloc/internal/rtree.h

@@ -0,0 +1,161 @@
+/*
+ * This radix tree implementation is tailored to the singular purpose of
+ * tracking which chunks are currently owned by jemalloc.  This functionality
+ * is mandatory for OS X, where jemalloc must be able to respond to object
+ * ownership queries.
+ *
+ *******************************************************************************
+ */
+#ifdef JEMALLOC_H_TYPES
+
+typedef struct rtree_s rtree_t;
+
+/*
+ * Size of each radix tree node (must be a power of 2).  This impacts tree
+ * depth.
+ */
+#if (LG_SIZEOF_PTR == 2)
+#  define RTREE_NODESIZE (1U << 14)
+#else
+#  define RTREE_NODESIZE CACHELINE
+#endif
+
+#endif /* JEMALLOC_H_TYPES */
+/******************************************************************************/
+#ifdef JEMALLOC_H_STRUCTS
+
+struct rtree_s {
+	malloc_mutex_t	mutex;
+	void		**root;
+	unsigned	height;
+	unsigned	level2bits[1]; /* Dynamically sized. */
+};
+
+#endif /* JEMALLOC_H_STRUCTS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_EXTERNS
+
+rtree_t	*rtree_new(unsigned bits);
+
+#endif /* JEMALLOC_H_EXTERNS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_INLINES
+
+#ifndef JEMALLOC_ENABLE_INLINE
+#ifndef JEMALLOC_DEBUG
+void	*rtree_get_locked(rtree_t *rtree, uintptr_t key);
+#endif
+void	*rtree_get(rtree_t *rtree, uintptr_t key);
+bool	rtree_set(rtree_t *rtree, uintptr_t key, void *val);
+#endif
+
+#if (defined(JEMALLOC_ENABLE_INLINE) || defined(JEMALLOC_RTREE_C_))
+#define	RTREE_GET_GENERATE(f)						\
+/* The least significant bits of the key are ignored. */		\
+JEMALLOC_INLINE void *							\
+f(rtree_t *rtree, uintptr_t key)					\
+{									\
+	void *ret;							\
+	uintptr_t subkey;						\
+	unsigned i, lshift, height, bits;				\
+	void **node, **child;						\
+									\
+	RTREE_LOCK(&rtree->mutex);					\
+	for (i = lshift = 0, height = rtree->height, node = rtree->root;\
+	    i < height - 1;						\
+	    i++, lshift += bits, node = child) {			\
+		bits = rtree->level2bits[i];				\
+		subkey = (key << lshift) >> ((ZU(1) << (LG_SIZEOF_PTR + \
+		    3)) - bits);					\
+		child = (void**)node[subkey];				\
+		if (child == NULL) {					\
+			RTREE_UNLOCK(&rtree->mutex);			\
+			return (NULL);					\
+		}							\
+	}								\
+									\
+	/*								\
+	 * node is a leaf, so it contains values rather than node	\
+	 * pointers.							\
+	 */								\
+	bits = rtree->level2bits[i];					\
+	subkey = (key << lshift) >> ((ZU(1) << (LG_SIZEOF_PTR+3)) -	\
+	    bits);							\
+	ret = node[subkey];						\
+	RTREE_UNLOCK(&rtree->mutex);					\
+									\
+	RTREE_GET_VALIDATE						\
+	return (ret);							\
+}
+
+#ifdef JEMALLOC_DEBUG
+#  define RTREE_LOCK(l)		malloc_mutex_lock(l)
+#  define RTREE_UNLOCK(l)	malloc_mutex_unlock(l)
+#  define RTREE_GET_VALIDATE
+RTREE_GET_GENERATE(rtree_get_locked)
+#  undef RTREE_LOCK
+#  undef RTREE_UNLOCK
+#  undef RTREE_GET_VALIDATE
+#endif
+
+#define	RTREE_LOCK(l)
+#define	RTREE_UNLOCK(l)
+#ifdef JEMALLOC_DEBUG
+   /*
+    * Suppose that it were possible for a jemalloc-allocated chunk to be
+    * munmap()ped, followed by a different allocator in another thread re-using
+    * overlapping virtual memory, all without invalidating the cached rtree
+    * value.  The result would be a false positive (the rtree would claim that
+    * jemalloc owns memory that it had actually discarded).  This scenario
+    * seems impossible, but the following assertion is a prudent sanity check.
+    */
+#  define RTREE_GET_VALIDATE						\
+	assert(rtree_get_locked(rtree, key) == ret);
+#else
+#  define RTREE_GET_VALIDATE
+#endif
+RTREE_GET_GENERATE(rtree_get)
+#undef RTREE_LOCK
+#undef RTREE_UNLOCK
+#undef RTREE_GET_VALIDATE
+
+JEMALLOC_INLINE bool
+rtree_set(rtree_t *rtree, uintptr_t key, void *val)
+{
+	uintptr_t subkey;
+	unsigned i, lshift, height, bits;
+	void **node, **child;
+
+	malloc_mutex_lock(&rtree->mutex);
+	for (i = lshift = 0, height = rtree->height, node = rtree->root;
+	    i < height - 1;
+	    i++, lshift += bits, node = child) {
+		bits = rtree->level2bits[i];
+		subkey = (key << lshift) >> ((ZU(1) << (LG_SIZEOF_PTR+3)) -
+		    bits);
+		child = (void**)node[subkey];
+		if (child == NULL) {
+			child = (void**)base_alloc(sizeof(void *) <<
+			    rtree->level2bits[i+1]);
+			if (child == NULL) {
+				malloc_mutex_unlock(&rtree->mutex);
+				return (true);
+			}
+			memset(child, 0, sizeof(void *) <<
+			    rtree->level2bits[i+1]);
+			node[subkey] = child;
+		}
+	}
+
+	/* node is a leaf, so it contains values rather than node pointers. */
+	bits = rtree->level2bits[i];
+	subkey = (key << lshift) >> ((ZU(1) << (LG_SIZEOF_PTR+3)) - bits);
+	node[subkey] = val;
+	malloc_mutex_unlock(&rtree->mutex);
+
+	return (false);
+}
+#endif
+
+#endif /* JEMALLOC_H_INLINES */
+/******************************************************************************/

contrib/jemalloc/include/jemalloc/internal/size_classes.h

@@ -0,0 +1,721 @@
+/* This file was automatically generated by size_classes.sh. */
+/******************************************************************************/
+#ifdef JEMALLOC_H_TYPES
+
+#if (LG_TINY_MIN == 3 && LG_QUANTUM == 3 && LG_PAGE == 12)
+#define	SIZE_CLASSES_DEFINED
+/*  SIZE_CLASS(bin,	delta,	sz) */
+#define	SIZE_CLASSES							\
+    SIZE_CLASS(0,	8,	8)					\
+    SIZE_CLASS(1,	8,	16)					\
+    SIZE_CLASS(2,	8,	24)					\
+    SIZE_CLASS(3,	8,	32)					\
+    SIZE_CLASS(4,	8,	40)					\
+    SIZE_CLASS(5,	8,	48)					\
+    SIZE_CLASS(6,	8,	56)					\
+    SIZE_CLASS(7,	8,	64)					\
+    SIZE_CLASS(8,	16,	80)					\
+    SIZE_CLASS(9,	16,	96)					\
+    SIZE_CLASS(10,	16,	112)					\
+    SIZE_CLASS(11,	16,	128)					\
+    SIZE_CLASS(12,	32,	160)					\
+    SIZE_CLASS(13,	32,	192)					\
+    SIZE_CLASS(14,	32,	224)					\
+    SIZE_CLASS(15,	32,	256)					\
+    SIZE_CLASS(16,	64,	320)					\
+    SIZE_CLASS(17,	64,	384)					\
+    SIZE_CLASS(18,	64,	448)					\
+    SIZE_CLASS(19,	64,	512)					\
+    SIZE_CLASS(20,	128,	640)					\
+    SIZE_CLASS(21,	128,	768)					\
+    SIZE_CLASS(22,	128,	896)					\
+    SIZE_CLASS(23,	128,	1024)					\
+    SIZE_CLASS(24,	256,	1280)					\
+    SIZE_CLASS(25,	256,	1536)					\
+    SIZE_CLASS(26,	256,	1792)					\
+    SIZE_CLASS(27,	256,	2048)					\
+    SIZE_CLASS(28,	512,	2560)					\
+    SIZE_CLASS(29,	512,	3072)					\
+    SIZE_CLASS(30,	512,	3584)					\
+
+#define	NBINS		31
+#define	SMALL_MAXCLASS	3584
+#endif
+
+#if (LG_TINY_MIN == 3 && LG_QUANTUM == 3 && LG_PAGE == 13)
+#define	SIZE_CLASSES_DEFINED
+/*  SIZE_CLASS(bin,	delta,	sz) */
+#define	SIZE_CLASSES							\
+    SIZE_CLASS(0,	8,	8)					\
+    SIZE_CLASS(1,	8,	16)					\
+    SIZE_CLASS(2,	8,	24)					\
+    SIZE_CLASS(3,	8,	32)					\
+    SIZE_CLASS(4,	8,	40)					\
+    SIZE_CLASS(5,	8,	48)					\
+    SIZE_CLASS(6,	8,	56)					\
+    SIZE_CLASS(7,	8,	64)					\
+    SIZE_CLASS(8,	16,	80)					\
+    SIZE_CLASS(9,	16,	96)					\
+    SIZE_CLASS(10,	16,	112)					\
+    SIZE_CLASS(11,	16,	128)					\
+    SIZE_CLASS(12,	32,	160)					\
+    SIZE_CLASS(13,	32,	192)					\
+    SIZE_CLASS(14,	32,	224)					\
+    SIZE_CLASS(15,	32,	256)					\
+    SIZE_CLASS(16,	64,	320)					\
+    SIZE_CLASS(17,	64,	384)					\
+    SIZE_CLASS(18,	64,	448)					\
+    SIZE_CLASS(19,	64,	512)					\
+    SIZE_CLASS(20,	128,	640)					\
+    SIZE_CLASS(21,	128,	768)					\
+    SIZE_CLASS(22,	128,	896)					\
+    SIZE_CLASS(23,	128,	1024)					\
+    SIZE_CLASS(24,	256,	1280)					\
+    SIZE_CLASS(25,	256,	1536)					\
+    SIZE_CLASS(26,	256,	1792)					\
+    SIZE_CLASS(27,	256,	2048)					\
+    SIZE_CLASS(28,	512,	2560)					\
+    SIZE_CLASS(29,	512,	3072)					\
+    SIZE_CLASS(30,	512,	3584)					\
+    SIZE_CLASS(31,	512,	4096)					\
+    SIZE_CLASS(32,	1024,	5120)					\
+    SIZE_CLASS(33,	1024,	6144)					\
+    SIZE_CLASS(34,	1024,	7168)					\
+
+#define	NBINS		35
+#define	SMALL_MAXCLASS	7168
+#endif
+
+#if (LG_TINY_MIN == 3 && LG_QUANTUM == 3 && LG_PAGE == 14)
+#define	SIZE_CLASSES_DEFINED
+/*  SIZE_CLASS(bin,	delta,	sz) */
+#define	SIZE_CLASSES							\
+    SIZE_CLASS(0,	8,	8)					\
+    SIZE_CLASS(1,	8,	16)					\
+    SIZE_CLASS(2,	8,	24)					\
+    SIZE_CLASS(3,	8,	32)					\
+    SIZE_CLASS(4,	8,	40)					\
+    SIZE_CLASS(5,	8,	48)					\
+    SIZE_CLASS(6,	8,	56)					\
+    SIZE_CLASS(7,	8,	64)					\
+    SIZE_CLASS(8,	16,	80)					\
+    SIZE_CLASS(9,	16,	96)					\
+    SIZE_CLASS(10,	16,	112)					\
+    SIZE_CLASS(11,	16,	128)					\
+    SIZE_CLASS(12,	32,	160)					\
+    SIZE_CLASS(13,	32,	192)					\
+    SIZE_CLASS(14,	32,	224)					\
+    SIZE_CLASS(15,	32,	256)					\
+    SIZE_CLASS(16,	64,	320)					\
+    SIZE_CLASS(17,	64,	384)					\
+    SIZE_CLASS(18,	64,	448)					\
+    SIZE_CLASS(19,	64,	512)					\
+    SIZE_CLASS(20,	128,	640)					\
+    SIZE_CLASS(21,	128,	768)					\
+    SIZE_CLASS(22,	128,	896)					\
+    SIZE_CLASS(23,	128,	1024)					\
+    SIZE_CLASS(24,	256,	1280)					\
+    SIZE_CLASS(25,	256,	1536)					\
+    SIZE_CLASS(26,	256,	1792)					\
+    SIZE_CLASS(27,	256,	2048)					\
+    SIZE_CLASS(28,	512,	2560)					\
+    SIZE_CLASS(29,	512,	3072)					\
+    SIZE_CLASS(30,	512,	3584)					\
+    SIZE_CLASS(31,	512,	4096)					\
+    SIZE_CLASS(32,	1024,	5120)					\
+    SIZE_CLASS(33,	1024,	6144)					\
+    SIZE_CLASS(34,	1024,	7168)					\
+    SIZE_CLASS(35,	1024,	8192)					\
+    SIZE_CLASS(36,	2048,	10240)					\
+    SIZE_CLASS(37,	2048,	12288)					\
+    SIZE_CLASS(38,	2048,	14336)					\
+
+#define	NBINS		39
+#define	SMALL_MAXCLASS	14336
+#endif
+
+#if (LG_TINY_MIN == 3 && LG_QUANTUM == 3 && LG_PAGE == 15)
+#define	SIZE_CLASSES_DEFINED
+/*  SIZE_CLASS(bin,	delta,	sz) */
+#define	SIZE_CLASSES							\
+    SIZE_CLASS(0,	8,	8)					\
+    SIZE_CLASS(1,	8,	16)					\
+    SIZE_CLASS(2,	8,	24)					\
+    SIZE_CLASS(3,	8,	32)					\
+    SIZE_CLASS(4,	8,	40)					\
+    SIZE_CLASS(5,	8,	48)					\
+    SIZE_CLASS(6,	8,	56)					\
+    SIZE_CLASS(7,	8,	64)					\
+    SIZE_CLASS(8,	16,	80)					\
+    SIZE_CLASS(9,	16,	96)					\
+    SIZE_CLASS(10,	16,	112)					\
+    SIZE_CLASS(11,	16,	128)					\
+    SIZE_CLASS(12,	32,	160)					\
+    SIZE_CLASS(13,	32,	192)					\
+    SIZE_CLASS(14,	32,	224)					\
+    SIZE_CLASS(15,	32,	256)					\
+    SIZE_CLASS(16,	64,	320)					\
+    SIZE_CLASS(17,	64,	384)					\
+    SIZE_CLASS(18,	64,	448)					\
+    SIZE_CLASS(19,	64,	512)					\
+    SIZE_CLASS(20,	128,	640)					\
+    SIZE_CLASS(21,	128,	768)					\
+    SIZE_CLASS(22,	128,	896)					\
+    SIZE_CLASS(23,	128,	1024)					\
+    SIZE_CLASS(24,	256,	1280)					\
+    SIZE_CLASS(25,	256,	1536)					\
+    SIZE_CLASS(26,	256,	1792)					\
+    SIZE_CLASS(27,	256,	2048)					\
+    SIZE_CLASS(28,	512,	2560)					\
+    SIZE_CLASS(29,	512,	3072)					\
+    SIZE_CLASS(30,	512,	3584)					\
+    SIZE_CLASS(31,	512,	4096)					\
+    SIZE_CLASS(32,	1024,	5120)					\
+    SIZE_CLASS(33,	1024,	6144)					\
+    SIZE_CLASS(34,	1024,	7168)					\
+    SIZE_CLASS(35,	1024,	8192)					\
+    SIZE_CLASS(36,	2048,	10240)					\
+    SIZE_CLASS(37,	2048,	12288)					\
+    SIZE_CLASS(38,	2048,	14336)					\
+    SIZE_CLASS(39,	2048,	16384)					\
+    SIZE_CLASS(40,	4096,	20480)					\
+    SIZE_CLASS(41,	4096,	24576)					\
+    SIZE_CLASS(42,	4096,	28672)					\
+
+#define	NBINS		43
+#define	SMALL_MAXCLASS	28672
+#endif
+
+#if (LG_TINY_MIN == 3 && LG_QUANTUM == 3 && LG_PAGE == 16)
+#define	SIZE_CLASSES_DEFINED
+/*  SIZE_CLASS(bin,	delta,	sz) */
+#define	SIZE_CLASSES							\
+    SIZE_CLASS(0,	8,	8)					\
+    SIZE_CLASS(1,	8,	16)					\
+    SIZE_CLASS(2,	8,	24)					\
+    SIZE_CLASS(3,	8,	32)					\
+    SIZE_CLASS(4,	8,	40)					\
+    SIZE_CLASS(5,	8,	48)					\
+    SIZE_CLASS(6,	8,	56)					\
+    SIZE_CLASS(7,	8,	64)					\
+    SIZE_CLASS(8,	16,	80)					\
+    SIZE_CLASS(9,	16,	96)					\
+    SIZE_CLASS(10,	16,	112)					\
+    SIZE_CLASS(11,	16,	128)					\
+    SIZE_CLASS(12,	32,	160)					\
+    SIZE_CLASS(13,	32,	192)					\
+    SIZE_CLASS(14,	32,	224)					\
+    SIZE_CLASS(15,	32,	256)					\
+    SIZE_CLASS(16,	64,	320)					\
+    SIZE_CLASS(17,	64,	384)					\
+    SIZE_CLASS(18,	64,	448)					\
+    SIZE_CLASS(19,	64,	512)					\
+    SIZE_CLASS(20,	128,	640)					\
+    SIZE_CLASS(21,	128,	768)					\
+    SIZE_CLASS(22,	128,	896)					\
+    SIZE_CLASS(23,	128,	1024)					\
+    SIZE_CLASS(24,	256,	1280)					\
+    SIZE_CLASS(25,	256,	1536)					\
+    SIZE_CLASS(26,	256,	1792)					\
+    SIZE_CLASS(27,	256,	2048)					\
+    SIZE_CLASS(28,	512,	2560)					\
+    SIZE_CLASS(29,	512,	3072)					\
+    SIZE_CLASS(30,	512,	3584)					\
+    SIZE_CLASS(31,	512,	4096)					\
+    SIZE_CLASS(32,	1024,	5120)					\
+    SIZE_CLASS(33,	1024,	6144)					\
+    SIZE_CLASS(34,	1024,	7168)					\
+    SIZE_CLASS(35,	1024,	8192)					\
+    SIZE_CLASS(36,	2048,	10240)					\
+    SIZE_CLASS(37,	2048,	12288)					\
+    SIZE_CLASS(38,	2048,	14336)					\
+    SIZE_CLASS(39,	2048,	16384)					\
+    SIZE_CLASS(40,	4096,	20480)					\
+    SIZE_CLASS(41,	4096,	24576)					\
+    SIZE_CLASS(42,	4096,	28672)					\
+    SIZE_CLASS(43,	4096,	32768)					\
+    SIZE_CLASS(44,	8192,	40960)					\
+    SIZE_CLASS(45,	8192,	49152)					\
+    SIZE_CLASS(46,	8192,	57344)					\
+
+#define	NBINS		47
+#define	SMALL_MAXCLASS	57344
+#endif
+
+#if (LG_TINY_MIN == 3 && LG_QUANTUM == 4 && LG_PAGE == 12)
+#define	SIZE_CLASSES_DEFINED
+/*  SIZE_CLASS(bin,	delta,	sz) */
+#define	SIZE_CLASSES							\
+    SIZE_CLASS(0,	8,	8)					\
+    SIZE_CLASS(1,	8,	16)					\
+    SIZE_CLASS(2,	16,	32)					\
+    SIZE_CLASS(3,	16,	48)					\
+    SIZE_CLASS(4,	16,	64)					\
+    SIZE_CLASS(5,	16,	80)					\
+    SIZE_CLASS(6,	16,	96)					\
+    SIZE_CLASS(7,	16,	112)					\
+    SIZE_CLASS(8,	16,	128)					\
+    SIZE_CLASS(9,	32,	160)					\
+    SIZE_CLASS(10,	32,	192)					\
+    SIZE_CLASS(11,	32,	224)					\
+    SIZE_CLASS(12,	32,	256)					\
+    SIZE_CLASS(13,	64,	320)					\
+    SIZE_CLASS(14,	64,	384)					\
+    SIZE_CLASS(15,	64,	448)					\
+    SIZE_CLASS(16,	64,	512)					\
+    SIZE_CLASS(17,	128,	640)					\
+    SIZE_CLASS(18,	128,	768)					\
+    SIZE_CLASS(19,	128,	896)					\
+    SIZE_CLASS(20,	128,	1024)					\
+    SIZE_CLASS(21,	256,	1280)					\
+    SIZE_CLASS(22,	256,	1536)					\
+    SIZE_CLASS(23,	256,	1792)					\
+    SIZE_CLASS(24,	256,	2048)					\
+    SIZE_CLASS(25,	512,	2560)					\
+    SIZE_CLASS(26,	512,	3072)					\
+    SIZE_CLASS(27,	512,	3584)					\
+
+#define	NBINS		28
+#define	SMALL_MAXCLASS	3584
+#endif
+
+#if (LG_TINY_MIN == 3 && LG_QUANTUM == 4 && LG_PAGE == 13)
+#define	SIZE_CLASSES_DEFINED
+/*  SIZE_CLASS(bin,	delta,	sz) */
+#define	SIZE_CLASSES							\
+    SIZE_CLASS(0,	8,	8)					\
+    SIZE_CLASS(1,	8,	16)					\
+    SIZE_CLASS(2,	16,	32)					\
+    SIZE_CLASS(3,	16,	48)					\
+    SIZE_CLASS(4,	16,	64)					\
+    SIZE_CLASS(5,	16,	80)					\
+    SIZE_CLASS(6,	16,	96)					\
+    SIZE_CLASS(7,	16,	112)					\
+    SIZE_CLASS(8,	16,	128)					\
+    SIZE_CLASS(9,	32,	160)					\
+    SIZE_CLASS(10,	32,	192)					\
+    SIZE_CLASS(11,	32,	224)					\
+    SIZE_CLASS(12,	32,	256)					\
+    SIZE_CLASS(13,	64,	320)					\
+    SIZE_CLASS(14,	64,	384)					\
+    SIZE_CLASS(15,	64,	448)					\
+    SIZE_CLASS(16,	64,	512)					\
+    SIZE_CLASS(17,	128,	640)					\
+    SIZE_CLASS(18,	128,	768)					\
+    SIZE_CLASS(19,	128,	896)					\
+    SIZE_CLASS(20,	128,	1024)					\
+    SIZE_CLASS(21,	256,	1280)					\
+    SIZE_CLASS(22,	256,	1536)					\
+    SIZE_CLASS(23,	256,	1792)					\
+    SIZE_CLASS(24,	256,	2048)					\
+    SIZE_CLASS(25,	512,	2560)					\
+    SIZE_CLASS(26,	512,	3072)					\
+    SIZE_CLASS(27,	512,	3584)					\
+    SIZE_CLASS(28,	512,	4096)					\
+    SIZE_CLASS(29,	1024,	5120)					\
+    SIZE_CLASS(30,	1024,	6144)					\
+    SIZE_CLASS(31,	1024,	7168)					\
+
+#define	NBINS		32
+#define	SMALL_MAXCLASS	7168
+#endif
+
+#if (LG_TINY_MIN == 3 && LG_QUANTUM == 4 && LG_PAGE == 14)
+#define	SIZE_CLASSES_DEFINED
+/*  SIZE_CLASS(bin,	delta,	sz) */
+#define	SIZE_CLASSES							\
+    SIZE_CLASS(0,	8,	8)					\
+    SIZE_CLASS(1,	8,	16)					\
+    SIZE_CLASS(2,	16,	32)					\
+    SIZE_CLASS(3,	16,	48)					\
+    SIZE_CLASS(4,	16,	64)					\
+    SIZE_CLASS(5,	16,	80)					\
+    SIZE_CLASS(6,	16,	96)					\
+    SIZE_CLASS(7,	16,	112)					\
+    SIZE_CLASS(8,	16,	128)					\
+    SIZE_CLASS(9,	32,	160)					\
+    SIZE_CLASS(10,	32,	192)					\
+    SIZE_CLASS(11,	32,	224)					\
+    SIZE_CLASS(12,	32,	256)					\
+    SIZE_CLASS(13,	64,	320)					\
+    SIZE_CLASS(14,	64,	384)					\
+    SIZE_CLASS(15,	64,	448)					\
+    SIZE_CLASS(16,	64,	512)					\
+    SIZE_CLASS(17,	128,	640)					\
+    SIZE_CLASS(18,	128,	768)					\
+    SIZE_CLASS(19,	128,	896)					\
+    SIZE_CLASS(20,	128,	1024)					\
+    SIZE_CLASS(21,	256,	1280)					\
+    SIZE_CLASS(22,	256,	1536)					\
+    SIZE_CLASS(23,	256,	1792)					\
+    SIZE_CLASS(24,	256,	2048)					\
+    SIZE_CLASS(25,	512,	2560)					\
+    SIZE_CLASS(26,	512,	3072)					\
+    SIZE_CLASS(27,	512,	3584)					\
+    SIZE_CLASS(28,	512,	4096)					\
+    SIZE_CLASS(29,	1024,	5120)					\
+    SIZE_CLASS(30,	1024,	6144)					\
+    SIZE_CLASS(31,	1024,	7168)					\
+    SIZE_CLASS(32,	1024,	8192)					\
+    SIZE_CLASS(33,	2048,	10240)					\
+    SIZE_CLASS(34,	2048,	12288)					\
+    SIZE_CLASS(35,	2048,	14336)					\
+
+#define	NBINS		36
+#define	SMALL_MAXCLASS	14336
+#endif
+
+#if (LG_TINY_MIN == 3 && LG_QUANTUM == 4 && LG_PAGE == 15)
+#define	SIZE_CLASSES_DEFINED
+/*  SIZE_CLASS(bin,	delta,	sz) */
+#define	SIZE_CLASSES							\
+    SIZE_CLASS(0,	8,	8)					\
+    SIZE_CLASS(1,	8,	16)					\
+    SIZE_CLASS(2,	16,	32)					\
+    SIZE_CLASS(3,	16,	48)					\
+    SIZE_CLASS(4,	16,	64)					\
+    SIZE_CLASS(5,	16,	80)					\
+    SIZE_CLASS(6,	16,	96)					\
+    SIZE_CLASS(7,	16,	112)					\
+    SIZE_CLASS(8,	16,	128)					\
+    SIZE_CLASS(9,	32,	160)					\
+    SIZE_CLASS(10,	32,	192)					\
+    SIZE_CLASS(11,	32,	224)					\
+    SIZE_CLASS(12,	32,	256)					\
+    SIZE_CLASS(13,	64,	320)					\
+    SIZE_CLASS(14,	64,	384)					\
+    SIZE_CLASS(15,	64,	448)					\
+    SIZE_CLASS(16,	64,	512)					\
+    SIZE_CLASS(17,	128,	640)					\
+    SIZE_CLASS(18,	128,	768)					\
+    SIZE_CLASS(19,	128,	896)					\
+    SIZE_CLASS(20,	128,	1024)					\
+    SIZE_CLASS(21,	256,	1280)					\
+    SIZE_CLASS(22,	256,	1536)					\
+    SIZE_CLASS(23,	256,	1792)					\
+    SIZE_CLASS(24,	256,	2048)					\
+    SIZE_CLASS(25,	512,	2560)					\
+    SIZE_CLASS(26,	512,	3072)					\
+    SIZE_CLASS(27,	512,	3584)					\
+    SIZE_CLASS(28,	512,	4096)					\
+    SIZE_CLASS(29,	1024,	5120)					\
+    SIZE_CLASS(30,	1024,	6144)					\
+    SIZE_CLASS(31,	1024,	7168)					\
+    SIZE_CLASS(32,	1024,	8192)					\
+    SIZE_CLASS(33,	2048,	10240)					\
+    SIZE_CLASS(34,	2048,	12288)					\
+    SIZE_CLASS(35,	2048,	14336)					\
+    SIZE_CLASS(36,	2048,	16384)					\
+    SIZE_CLASS(37,	4096,	20480)					\
+    SIZE_CLASS(38,	4096,	24576)					\
+    SIZE_CLASS(39,	4096,	28672)					\
+
+#define	NBINS		40
+#define	SMALL_MAXCLASS	28672
+#endif
+
+#if (LG_TINY_MIN == 3 && LG_QUANTUM == 4 && LG_PAGE == 16)
+#define	SIZE_CLASSES_DEFINED
+/*  SIZE_CLASS(bin,	delta,	sz) */
+#define	SIZE_CLASSES							\
+    SIZE_CLASS(0,	8,	8)					\
+    SIZE_CLASS(1,	8,	16)					\
+    SIZE_CLASS(2,	16,	32)					\
+    SIZE_CLASS(3,	16,	48)					\
+    SIZE_CLASS(4,	16,	64)					\
+    SIZE_CLASS(5,	16,	80)					\
+    SIZE_CLASS(6,	16,	96)					\
+    SIZE_CLASS(7,	16,	112)					\
+    SIZE_CLASS(8,	16,	128)					\
+    SIZE_CLASS(9,	32,	160)					\
+    SIZE_CLASS(10,	32,	192)					\
+    SIZE_CLASS(11,	32,	224)					\
+    SIZE_CLASS(12,	32,	256)					\
+    SIZE_CLASS(13,	64,	320)					\
+    SIZE_CLASS(14,	64,	384)					\
+    SIZE_CLASS(15,	64,	448)					\
+    SIZE_CLASS(16,	64,	512)					\
+    SIZE_CLASS(17,	128,	640)					\
+    SIZE_CLASS(18,	128,	768)					\
+    SIZE_CLASS(19,	128,	896)					\
+    SIZE_CLASS(20,	128,	1024)					\
+    SIZE_CLASS(21,	256,	1280)					\
+    SIZE_CLASS(22,	256,	1536)					\
+    SIZE_CLASS(23,	256,	1792)					\
+    SIZE_CLASS(24,	256,	2048)					\
+    SIZE_CLASS(25,	512,	2560)					\
+    SIZE_CLASS(26,	512,	3072)					\
+    SIZE_CLASS(27,	512,	3584)					\
+    SIZE_CLASS(28,	512,	4096)					\
+    SIZE_CLASS(29,	1024,	5120)					\
+    SIZE_CLASS(30,	1024,	6144)					\
+    SIZE_CLASS(31,	1024,	7168)					\
+    SIZE_CLASS(32,	1024,	8192)					\
+    SIZE_CLASS(33,	2048,	10240)					\
+    SIZE_CLASS(34,	2048,	12288)					\
+    SIZE_CLASS(35,	2048,	14336)					\
+    SIZE_CLASS(36,	2048,	16384)					\
+    SIZE_CLASS(37,	4096,	20480)					\
+    SIZE_CLASS(38,	4096,	24576)					\
+    SIZE_CLASS(39,	4096,	28672)					\
+    SIZE_CLASS(40,	4096,	32768)					\
+    SIZE_CLASS(41,	8192,	40960)					\
+    SIZE_CLASS(42,	8192,	49152)					\
+    SIZE_CLASS(43,	8192,	57344)					\
+
+#define	NBINS		44
+#define	SMALL_MAXCLASS	57344
+#endif
+
+#if (LG_TINY_MIN == 4 && LG_QUANTUM == 4 && LG_PAGE == 12)
+#define	SIZE_CLASSES_DEFINED
+/*  SIZE_CLASS(bin,	delta,	sz) */
+#define	SIZE_CLASSES							\
+    SIZE_CLASS(0,	16,	16)					\
+    SIZE_CLASS(1,	16,	32)					\
+    SIZE_CLASS(2,	16,	48)					\
+    SIZE_CLASS(3,	16,	64)					\
+    SIZE_CLASS(4,	16,	80)					\
+    SIZE_CLASS(5,	16,	96)					\
+    SIZE_CLASS(6,	16,	112)					\
+    SIZE_CLASS(7,	16,	128)					\
+    SIZE_CLASS(8,	32,	160)					\
+    SIZE_CLASS(9,	32,	192)					\
+    SIZE_CLASS(10,	32,	224)					\
+    SIZE_CLASS(11,	32,	256)					\
+    SIZE_CLASS(12,	64,	320)					\
+    SIZE_CLASS(13,	64,	384)					\
+    SIZE_CLASS(14,	64,	448)					\
+    SIZE_CLASS(15,	64,	512)					\
+    SIZE_CLASS(16,	128,	640)					\
+    SIZE_CLASS(17,	128,	768)					\
+    SIZE_CLASS(18,	128,	896)					\
+    SIZE_CLASS(19,	128,	1024)					\
+    SIZE_CLASS(20,	256,	1280)					\
+    SIZE_CLASS(21,	256,	1536)					\
+    SIZE_CLASS(22,	256,	1792)					\
+    SIZE_CLASS(23,	256,	2048)					\
+    SIZE_CLASS(24,	512,	2560)					\
+    SIZE_CLASS(25,	512,	3072)					\
+    SIZE_CLASS(26,	512,	3584)					\
+
+#define	NBINS		27
+#define	SMALL_MAXCLASS	3584
+#endif
+
+#if (LG_TINY_MIN == 4 && LG_QUANTUM == 4 && LG_PAGE == 13)
+#define	SIZE_CLASSES_DEFINED
+/*  SIZE_CLASS(bin,	delta,	sz) */
+#define	SIZE_CLASSES							\
+    SIZE_CLASS(0,	16,	16)					\
+    SIZE_CLASS(1,	16,	32)					\
+    SIZE_CLASS(2,	16,	48)					\
+    SIZE_CLASS(3,	16,	64)					\
+    SIZE_CLASS(4,	16,	80)					\
+    SIZE_CLASS(5,	16,	96)					\
+    SIZE_CLASS(6,	16,	112)					\
+    SIZE_CLASS(7,	16,	128)					\
+    SIZE_CLASS(8,	32,	160)					\
+    SIZE_CLASS(9,	32,	192)					\
+    SIZE_CLASS(10,	32,	224)					\
+    SIZE_CLASS(11,	32,	256)					\
+    SIZE_CLASS(12,	64,	320)					\
+    SIZE_CLASS(13,	64,	384)					\
+    SIZE_CLASS(14,	64,	448)					\
+    SIZE_CLASS(15,	64,	512)					\
+    SIZE_CLASS(16,	128,	640)					\
+    SIZE_CLASS(17,	128,	768)					\
+    SIZE_CLASS(18,	128,	896)					\
+    SIZE_CLASS(19,	128,	1024)					\
+    SIZE_CLASS(20,	256,	1280)					\
+    SIZE_CLASS(21,	256,	1536)					\
+    SIZE_CLASS(22,	256,	1792)					\
+    SIZE_CLASS(23,	256,	2048)					\
+    SIZE_CLASS(24,	512,	2560)					\
+    SIZE_CLASS(25,	512,	3072)					\
+    SIZE_CLASS(26,	512,	3584)					\
+    SIZE_CLASS(27,	512,	4096)					\
+    SIZE_CLASS(28,	1024,	5120)					\
+    SIZE_CLASS(29,	1024,	6144)					\
+    SIZE_CLASS(30,	1024,	7168)					\
+
+#define	NBINS		31
+#define	SMALL_MAXCLASS	7168
+#endif
+
+#if (LG_TINY_MIN == 4 && LG_QUANTUM == 4 && LG_PAGE == 14)
+#define	SIZE_CLASSES_DEFINED
+/*  SIZE_CLASS(bin,	delta,	sz) */
+#define	SIZE_CLASSES							\
+    SIZE_CLASS(0,	16,	16)					\
+    SIZE_CLASS(1,	16,	32)					\
+    SIZE_CLASS(2,	16,	48)					\
+    SIZE_CLASS(3,	16,	64)					\
+    SIZE_CLASS(4,	16,	80)					\
+    SIZE_CLASS(5,	16,	96)					\
+    SIZE_CLASS(6,	16,	112)					\
+    SIZE_CLASS(7,	16,	128)					\
+    SIZE_CLASS(8,	32,	160)					\
+    SIZE_CLASS(9,	32,	192)					\
+    SIZE_CLASS(10,	32,	224)					\
+    SIZE_CLASS(11,	32,	256)					\
+    SIZE_CLASS(12,	64,	320)					\
+    SIZE_CLASS(13,	64,	384)					\
+    SIZE_CLASS(14,	64,	448)					\
+    SIZE_CLASS(15,	64,	512)					\
+    SIZE_CLASS(16,	128,	640)					\
+    SIZE_CLASS(17,	128,	768)					\
+    SIZE_CLASS(18,	128,	896)					\
+    SIZE_CLASS(19,	128,	1024)					\
+    SIZE_CLASS(20,	256,	1280)					\
+    SIZE_CLASS(21,	256,	1536)					\
+    SIZE_CLASS(22,	256,	1792)					\
+    SIZE_CLASS(23,	256,	2048)					\
+    SIZE_CLASS(24,	512,	2560)					\
+    SIZE_CLASS(25,	512,	3072)					\
+    SIZE_CLASS(26,	512,	3584)					\
+    SIZE_CLASS(27,	512,	4096)					\
+    SIZE_CLASS(28,	1024,	5120)					\
+    SIZE_CLASS(29,	1024,	6144)					\
+    SIZE_CLASS(30,	1024,	7168)					\
+    SIZE_CLASS(31,	1024,	8192)					\
+    SIZE_CLASS(32,	2048,	10240)					\
+    SIZE_CLASS(33,	2048,	12288)					\
+    SIZE_CLASS(34,	2048,	14336)					\
+
+#define	NBINS		35
+#define	SMALL_MAXCLASS	14336
+#endif
+
+#if (LG_TINY_MIN == 4 && LG_QUANTUM == 4 && LG_PAGE == 15)
+#define	SIZE_CLASSES_DEFINED
+/*  SIZE_CLASS(bin,	delta,	sz) */
+#define	SIZE_CLASSES							\
+    SIZE_CLASS(0,	16,	16)					\
+    SIZE_CLASS(1,	16,	32)					\
+    SIZE_CLASS(2,	16,	48)					\
+    SIZE_CLASS(3,	16,	64)					\
+    SIZE_CLASS(4,	16,	80)					\
+    SIZE_CLASS(5,	16,	96)					\
+    SIZE_CLASS(6,	16,	112)					\
+    SIZE_CLASS(7,	16,	128)					\
+    SIZE_CLASS(8,	32,	160)					\
+    SIZE_CLASS(9,	32,	192)					\
+    SIZE_CLASS(10,	32,	224)					\
+    SIZE_CLASS(11,	32,	256)					\
+    SIZE_CLASS(12,	64,	320)					\
+    SIZE_CLASS(13,	64,	384)					\
+    SIZE_CLASS(14,	64,	448)					\
+    SIZE_CLASS(15,	64,	512)					\
+    SIZE_CLASS(16,	128,	640)					\
+    SIZE_CLASS(17,	128,	768)					\
+    SIZE_CLASS(18,	128,	896)					\
+    SIZE_CLASS(19,	128,	1024)					\
+    SIZE_CLASS(20,	256,	1280)					\
+    SIZE_CLASS(21,	256,	1536)					\
+    SIZE_CLASS(22,	256,	1792)					\
+    SIZE_CLASS(23,	256,	2048)					\
+    SIZE_CLASS(24,	512,	2560)					\
+    SIZE_CLASS(25,	512,	3072)					\
+    SIZE_CLASS(26,	512,	3584)					\
+    SIZE_CLASS(27,	512,	4096)					\
+    SIZE_CLASS(28,	1024,	5120)					\
+    SIZE_CLASS(29,	1024,	6144)					\
+    SIZE_CLASS(30,	1024,	7168)					\
+    SIZE_CLASS(31,	1024,	8192)					\
+    SIZE_CLASS(32,	2048,	10240)					\
+    SIZE_CLASS(33,	2048,	12288)					\
+    SIZE_CLASS(34,	2048,	14336)					\
+    SIZE_CLASS(35,	2048,	16384)					\
+    SIZE_CLASS(36,	4096,	20480)					\
+    SIZE_CLASS(37,	4096,	24576)					\
+    SIZE_CLASS(38,	4096,	28672)					\
+
+#define	NBINS		39
+#define	SMALL_MAXCLASS	28672
+#endif
+
+#if (LG_TINY_MIN == 4 && LG_QUANTUM == 4 && LG_PAGE == 16)
+#define	SIZE_CLASSES_DEFINED
+/*  SIZE_CLASS(bin,	delta,	sz) */
+#define	SIZE_CLASSES							\
+    SIZE_CLASS(0,	16,	16)					\
+    SIZE_CLASS(1,	16,	32)					\
+    SIZE_CLASS(2,	16,	48)					\
+    SIZE_CLASS(3,	16,	64)					\
+    SIZE_CLASS(4,	16,	80)					\
+    SIZE_CLASS(5,	16,	96)					\
+    SIZE_CLASS(6,	16,	112)					\
+    SIZE_CLASS(7,	16,	128)					\
+    SIZE_CLASS(8,	32,	160)					\
+    SIZE_CLASS(9,	32,	192)					\
+    SIZE_CLASS(10,	32,	224)					\
+    SIZE_CLASS(11,	32,	256)					\
+    SIZE_CLASS(12,	64,	320)					\
+    SIZE_CLASS(13,	64,	384)					\
+    SIZE_CLASS(14,	64,	448)					\
+    SIZE_CLASS(15,	64,	512)					\
+    SIZE_CLASS(16,	128,	640)					\
+    SIZE_CLASS(17,	128,	768)					\
+    SIZE_CLASS(18,	128,	896)					\
+    SIZE_CLASS(19,	128,	1024)					\
+    SIZE_CLASS(20,	256,	1280)					\
+    SIZE_CLASS(21,	256,	1536)					\
+    SIZE_CLASS(22,	256,	1792)					\
+    SIZE_CLASS(23,	256,	2048)					\
+    SIZE_CLASS(24,	512,	2560)					\
+    SIZE_CLASS(25,	512,	3072)					\
+    SIZE_CLASS(26,	512,	3584)					\
+    SIZE_CLASS(27,	512,	4096)					\
+    SIZE_CLASS(28,	1024,	5120)					\
+    SIZE_CLASS(29,	1024,	6144)					\
+    SIZE_CLASS(30,	1024,	7168)					\
+    SIZE_CLASS(31,	1024,	8192)					\
+    SIZE_CLASS(32,	2048,	10240)					\
+    SIZE_CLASS(33,	2048,	12288)					\
+    SIZE_CLASS(34,	2048,	14336)					\
+    SIZE_CLASS(35,	2048,	16384)					\
+    SIZE_CLASS(36,	4096,	20480)					\
+    SIZE_CLASS(37,	4096,	24576)					\
+    SIZE_CLASS(38,	4096,	28672)					\
+    SIZE_CLASS(39,	4096,	32768)					\
+    SIZE_CLASS(40,	8192,	40960)					\
+    SIZE_CLASS(41,	8192,	49152)					\
+    SIZE_CLASS(42,	8192,	57344)					\
+
+#define	NBINS		43
+#define	SMALL_MAXCLASS	57344
+#endif
+
+#ifndef SIZE_CLASSES_DEFINED
+#  error "No size class definitions match configuration"
+#endif
+#undef SIZE_CLASSES_DEFINED
+/*
+ * The small_size2bin lookup table uses uint8_t to encode each bin index, so we
+ * cannot support more than 256 small size classes.  Further constrain NBINS to
+ * 255 to support prof_promote, since all small size classes, plus a "not
+ * small" size class must be stored in 8 bits of arena_chunk_map_t's bits
+ * field.
+ */
+#if (NBINS > 255)
+#  error "Too many small size classes"
+#endif
+
+#endif /* JEMALLOC_H_TYPES */
+/******************************************************************************/
+#ifdef JEMALLOC_H_STRUCTS
+
+
+#endif /* JEMALLOC_H_STRUCTS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_EXTERNS
+
+
+#endif /* JEMALLOC_H_EXTERNS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_INLINES
+
+
+#endif /* JEMALLOC_H_INLINES */
+/******************************************************************************/

contrib/jemalloc/include/jemalloc/internal/stats.h

@@ -0,0 +1,173 @@
+/******************************************************************************/
+#ifdef JEMALLOC_H_TYPES
+
+typedef struct tcache_bin_stats_s tcache_bin_stats_t;
+typedef struct malloc_bin_stats_s malloc_bin_stats_t;
+typedef struct malloc_large_stats_s malloc_large_stats_t;
+typedef struct arena_stats_s arena_stats_t;
+typedef struct chunk_stats_s chunk_stats_t;
+
+#endif /* JEMALLOC_H_TYPES */
+/******************************************************************************/
+#ifdef JEMALLOC_H_STRUCTS
+
+struct tcache_bin_stats_s {
+	/*
+	 * Number of allocation requests that corresponded to the size of this
+	 * bin.
+	 */
+	uint64_t	nrequests;
+};
+
+struct malloc_bin_stats_s {
+	/*
+	 * Current number of bytes allocated, including objects currently
+	 * cached by tcache.
+	 */
+	size_t		allocated;
+
+	/*
+	 * Total number of allocation/deallocation requests served directly by
+	 * the bin.  Note that tcache may allocate an object, then recycle it
+	 * many times, resulting many increments to nrequests, but only one
+	 * each to nmalloc and ndalloc.
+	 */
+	uint64_t	nmalloc;
+	uint64_t	ndalloc;
+
+	/*
+	 * Number of allocation requests that correspond to the size of this
+	 * bin.  This includes requests served by tcache, though tcache only
+	 * periodically merges into this counter.
+	 */
+	uint64_t	nrequests;
+
+	/* Number of tcache fills from this bin. */
+	uint64_t	nfills;
+
+	/* Number of tcache flushes to this bin. */
+	uint64_t	nflushes;
+
+	/* Total number of runs created for this bin's size class. */
+	uint64_t	nruns;
+
+	/*
+	 * Total number of runs reused by extracting them from the runs tree for
+	 * this bin's size class.
+	 */
+	uint64_t	reruns;
+
+	/* Current number of runs in this bin. */
+	size_t		curruns;
+};
+
+struct malloc_large_stats_s {
+	/*
+	 * Total number of allocation/deallocation requests served directly by
+	 * the arena.  Note that tcache may allocate an object, then recycle it
+	 * many times, resulting many increments to nrequests, but only one
+	 * each to nmalloc and ndalloc.
+	 */
+	uint64_t	nmalloc;
+	uint64_t	ndalloc;
+
+	/*
+	 * Number of allocation requests that correspond to this size class.
+	 * This includes requests served by tcache, though tcache only
+	 * periodically merges into this counter.
+	 */
+	uint64_t	nrequests;
+
+	/* Current number of runs of this size class. */
+	size_t		curruns;
+};
+
+struct arena_stats_s {
+	/* Number of bytes currently mapped. */
+	size_t		mapped;
+
+	/*
+	 * Total number of purge sweeps, total number of madvise calls made,
+	 * and total pages purged in order to keep dirty unused memory under
+	 * control.
+	 */
+	uint64_t	npurge;
+	uint64_t	nmadvise;
+	uint64_t	purged;
+
+	/* Per-size-category statistics. */
+	size_t		allocated_large;
+	uint64_t	nmalloc_large;
+	uint64_t	ndalloc_large;
+	uint64_t	nrequests_large;
+
+	/*
+	 * One element for each possible size class, including sizes that
+	 * overlap with bin size classes.  This is necessary because ipalloc()
+	 * sometimes has to use such large objects in order to assure proper
+	 * alignment.
+	 */
+	malloc_large_stats_t	*lstats;
+};
+
+struct chunk_stats_s {
+	/* Number of chunks that were allocated. */
+	uint64_t	nchunks;
+
+	/* High-water mark for number of chunks allocated. */
+	size_t		highchunks;
+
+	/*
+	 * Current number of chunks allocated.  This value isn't maintained for
+	 * any other purpose, so keep track of it in order to be able to set
+	 * highchunks.
+	 */
+	size_t		curchunks;
+};
+
+#endif /* JEMALLOC_H_STRUCTS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_EXTERNS
+
+extern bool	opt_stats_print;
+
+extern size_t	stats_cactive;
+
+void	stats_print(void (*write)(void *, const char *), void *cbopaque,
+    const char *opts);
+
+#endif /* JEMALLOC_H_EXTERNS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_INLINES
+
+#ifndef JEMALLOC_ENABLE_INLINE
+size_t	stats_cactive_get(void);
+void	stats_cactive_add(size_t size);
+void	stats_cactive_sub(size_t size);
+#endif
+
+#if (defined(JEMALLOC_ENABLE_INLINE) || defined(JEMALLOC_STATS_C_))
+JEMALLOC_INLINE size_t
+stats_cactive_get(void)
+{
+
+	return (atomic_read_z(&stats_cactive));
+}
+
+JEMALLOC_INLINE void
+stats_cactive_add(size_t size)
+{
+
+	atomic_add_z(&stats_cactive, size);
+}
+
+JEMALLOC_INLINE void
+stats_cactive_sub(size_t size)
+{
+
+	atomic_sub_z(&stats_cactive, size);
+}
+#endif
+
+#endif /* JEMALLOC_H_INLINES */
+/******************************************************************************/

contrib/jemalloc/include/jemalloc/internal/tcache.h

@@ -0,0 +1,494 @@
+/******************************************************************************/
+#ifdef JEMALLOC_H_TYPES
+
+typedef struct tcache_bin_info_s tcache_bin_info_t;
+typedef struct tcache_bin_s tcache_bin_t;
+typedef struct tcache_s tcache_t;
+
+/*
+ * tcache pointers close to NULL are used to encode state information that is
+ * used for two purposes: preventing thread caching on a per thread basis and
+ * cleaning up during thread shutdown.
+ */
+#define	TCACHE_STATE_DISABLED		((tcache_t *)(uintptr_t)1)
+#define	TCACHE_STATE_REINCARNATED	((tcache_t *)(uintptr_t)2)
+#define	TCACHE_STATE_PURGATORY		((tcache_t *)(uintptr_t)3)
+#define	TCACHE_STATE_MAX		TCACHE_STATE_PURGATORY
+
+/*
+ * Absolute maximum number of cache slots for each small bin in the thread
+ * cache.  This is an additional constraint beyond that imposed as: twice the
+ * number of regions per run for this size class.
+ *
+ * This constant must be an even number.
+ */
+#define	TCACHE_NSLOTS_SMALL_MAX		200
+
+/* Number of cache slots for large size classes. */
+#define	TCACHE_NSLOTS_LARGE		20
+
+/* (1U << opt_lg_tcache_max) is used to compute tcache_maxclass. */
+#define	LG_TCACHE_MAXCLASS_DEFAULT	15
+
+/*
+ * TCACHE_GC_SWEEP is the approximate number of allocation events between
+ * full GC sweeps.  Integer rounding may cause the actual number to be
+ * slightly higher, since GC is performed incrementally.
+ */
+#define	TCACHE_GC_SWEEP			8192
+
+/* Number of tcache allocation/deallocation events between incremental GCs. */
+#define	TCACHE_GC_INCR							\
+    ((TCACHE_GC_SWEEP / NBINS) + ((TCACHE_GC_SWEEP / NBINS == 0) ? 0 : 1))
+
+#endif /* JEMALLOC_H_TYPES */
+/******************************************************************************/
+#ifdef JEMALLOC_H_STRUCTS
+
+typedef enum {
+	tcache_enabled_false   = 0, /* Enable cast to/from bool. */
+	tcache_enabled_true    = 1,
+	tcache_enabled_default = 2
+} tcache_enabled_t;
+
+/*
+ * Read-only information associated with each element of tcache_t's tbins array
+ * is stored separately, mainly to reduce memory usage.
+ */
+struct tcache_bin_info_s {
+	unsigned	ncached_max;	/* Upper limit on ncached. */
+};
+
+struct tcache_bin_s {
+	tcache_bin_stats_t tstats;
+	int		low_water;	/* Min # cached since last GC. */
+	unsigned	lg_fill_div;	/* Fill (ncached_max >> lg_fill_div). */
+	unsigned	ncached;	/* # of cached objects. */
+	void		**avail;	/* Stack of available objects. */
+};
+
+struct tcache_s {
+	ql_elm(tcache_t) link;		/* Used for aggregating stats. */
+	uint64_t	prof_accumbytes;/* Cleared after arena_prof_accum() */
+	arena_t		*arena;		/* This thread's arena. */
+	unsigned	ev_cnt;		/* Event count since incremental GC. */
+	unsigned	next_gc_bin;	/* Next bin to GC. */
+	tcache_bin_t	tbins[1];	/* Dynamically sized. */
+	/*
+	 * The pointer stacks associated with tbins follow as a contiguous
+	 * array.  During tcache initialization, the avail pointer in each
+	 * element of tbins is initialized to point to the proper offset within
+	 * this array.
+	 */
+};
+
+#endif /* JEMALLOC_H_STRUCTS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_EXTERNS
+
+extern bool	opt_tcache;
+extern ssize_t	opt_lg_tcache_max;
+
+extern tcache_bin_info_t	*tcache_bin_info;
+
+/*
+ * Number of tcache bins.  There are NBINS small-object bins, plus 0 or more
+ * large-object bins.
+ */
+extern size_t			nhbins;
+
+/* Maximum cached size class. */
+extern size_t			tcache_maxclass;
+
+void	tcache_bin_flush_small(tcache_bin_t *tbin, size_t binind, unsigned rem,
+    tcache_t *tcache);
+void	tcache_bin_flush_large(tcache_bin_t *tbin, size_t binind, unsigned rem,
+    tcache_t *tcache);
+void	tcache_arena_associate(tcache_t *tcache, arena_t *arena);
+void	tcache_arena_dissociate(tcache_t *tcache);
+tcache_t *tcache_create(arena_t *arena);
+void	*tcache_alloc_small_hard(tcache_t *tcache, tcache_bin_t *tbin,
+    size_t binind);
+void	tcache_destroy(tcache_t *tcache);
+void	tcache_thread_cleanup(void *arg);
+void	tcache_stats_merge(tcache_t *tcache, arena_t *arena);
+bool	tcache_boot0(void);
+bool	tcache_boot1(void);
+
+#endif /* JEMALLOC_H_EXTERNS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_INLINES
+
+#ifndef JEMALLOC_ENABLE_INLINE
+malloc_tsd_protos(JEMALLOC_ATTR(unused), tcache, tcache_t *)
+malloc_tsd_protos(JEMALLOC_ATTR(unused), tcache_enabled, tcache_enabled_t)
+
+void	tcache_event(tcache_t *tcache);
+void	tcache_flush(void);
+bool	tcache_enabled_get(void);
+tcache_t *tcache_get(bool create);
+void	tcache_enabled_set(bool enabled);
+void	*tcache_alloc_easy(tcache_bin_t *tbin);
+void	*tcache_alloc_small(tcache_t *tcache, size_t size, bool zero);
+void	*tcache_alloc_large(tcache_t *tcache, size_t size, bool zero);
+void	tcache_dalloc_small(tcache_t *tcache, void *ptr);
+void	tcache_dalloc_large(tcache_t *tcache, void *ptr, size_t size);
+#endif
+
+#if (defined(JEMALLOC_ENABLE_INLINE) || defined(JEMALLOC_TCACHE_C_))
+/* Map of thread-specific caches. */
+malloc_tsd_externs(tcache, tcache_t *)
+malloc_tsd_funcs(JEMALLOC_INLINE, tcache, tcache_t *, NULL,
+    tcache_thread_cleanup)
+/* Per thread flag that allows thread caches to be disabled. */
+malloc_tsd_externs(tcache_enabled, tcache_enabled_t)
+malloc_tsd_funcs(JEMALLOC_INLINE, tcache_enabled, tcache_enabled_t,
+    tcache_enabled_default, malloc_tsd_no_cleanup)
+
+JEMALLOC_INLINE void
+tcache_flush(void)
+{
+	tcache_t *tcache;
+
+	cassert(config_tcache);
+
+	tcache = *tcache_tsd_get();
+	if ((uintptr_t)tcache <= (uintptr_t)TCACHE_STATE_MAX)
+		return;
+	tcache_destroy(tcache);
+	tcache = NULL;
+	tcache_tsd_set(&tcache);
+}
+
+JEMALLOC_INLINE bool
+tcache_enabled_get(void)
+{
+	tcache_enabled_t tcache_enabled;
+
+	cassert(config_tcache);
+
+	tcache_enabled = *tcache_enabled_tsd_get();
+	if (tcache_enabled == tcache_enabled_default) {
+		tcache_enabled = (tcache_enabled_t)opt_tcache;
+		tcache_enabled_tsd_set(&tcache_enabled);
+	}
+
+	return ((bool)tcache_enabled);
+}
+
+JEMALLOC_INLINE void
+tcache_enabled_set(bool enabled)
+{
+	tcache_enabled_t tcache_enabled;
+	tcache_t *tcache;
+
+	cassert(config_tcache);
+
+	tcache_enabled = (tcache_enabled_t)enabled;
+	tcache_enabled_tsd_set(&tcache_enabled);
+	tcache = *tcache_tsd_get();
+	if (enabled) {
+		if (tcache == TCACHE_STATE_DISABLED) {
+			tcache = NULL;
+			tcache_tsd_set(&tcache);
+		}
+	} else /* disabled */ {
+		if (tcache > TCACHE_STATE_MAX) {
+			tcache_destroy(tcache);
+			tcache = NULL;
+		}
+		if (tcache == NULL) {
+			tcache = TCACHE_STATE_DISABLED;
+			tcache_tsd_set(&tcache);
+		}
+	}
+}
+
+JEMALLOC_INLINE tcache_t *
+tcache_get(bool create)
+{
+	tcache_t *tcache;
+
+	if (config_tcache == false)
+		return (NULL);
+	if (config_lazy_lock && isthreaded == false)
+		return (NULL);
+
+	tcache = *tcache_tsd_get();
+	if ((uintptr_t)tcache <= (uintptr_t)TCACHE_STATE_MAX) {
+		if (tcache == TCACHE_STATE_DISABLED)
+			return (NULL);
+		if (tcache == NULL) {
+			if (create == false) {
+				/*
+				 * Creating a tcache here would cause
+				 * allocation as a side effect of free().
+				 * Ordinarily that would be okay since
+				 * tcache_create() failure is a soft failure
+				 * that doesn't propagate.  However, if TLS
+				 * data are freed via free() as in glibc,
+				 * subtle corruption could result from setting
+				 * a TLS variable after its backing memory is
+				 * freed.
+				 */
+				return (NULL);
+			}
+			if (tcache_enabled_get() == false) {
+				tcache_enabled_set(false); /* Memoize. */
+				return (NULL);
+			}
+			return (tcache_create(choose_arena(NULL)));
+		}
+		if (tcache == TCACHE_STATE_PURGATORY) {
+			/*
+			 * Make a note that an allocator function was called
+			 * after tcache_thread_cleanup() was called.
+			 */
+			tcache = TCACHE_STATE_REINCARNATED;
+			tcache_tsd_set(&tcache);
+			return (NULL);
+		}
+		if (tcache == TCACHE_STATE_REINCARNATED)
+			return (NULL);
+		not_reached();
+	}
+
+	return (tcache);
+}
+
+JEMALLOC_INLINE void
+tcache_event(tcache_t *tcache)
+{
+
+	if (TCACHE_GC_INCR == 0)
+		return;
+
+	tcache->ev_cnt++;
+	assert(tcache->ev_cnt <= TCACHE_GC_INCR);
+	if (tcache->ev_cnt == TCACHE_GC_INCR) {
+		size_t binind = tcache->next_gc_bin;
+		tcache_bin_t *tbin = &tcache->tbins[binind];
+		tcache_bin_info_t *tbin_info = &tcache_bin_info[binind];
+
+		if (tbin->low_water > 0) {
+			/*
+			 * Flush (ceiling) 3/4 of the objects below the low
+			 * water mark.
+			 */
+			if (binind < NBINS) {
+				tcache_bin_flush_small(tbin, binind,
+				    tbin->ncached - tbin->low_water +
+				    (tbin->low_water >> 2), tcache);
+			} else {
+				tcache_bin_flush_large(tbin, binind,
+				    tbin->ncached - tbin->low_water +
+				    (tbin->low_water >> 2), tcache);
+			}
+			/*
+			 * Reduce fill count by 2X.  Limit lg_fill_div such that
+			 * the fill count is always at least 1.
+			 */
+			if ((tbin_info->ncached_max >> (tbin->lg_fill_div+1))
+			    >= 1)
+				tbin->lg_fill_div++;
+		} else if (tbin->low_water < 0) {
+			/*
+			 * Increase fill count by 2X.  Make sure lg_fill_div
+			 * stays greater than 0.
+			 */
+			if (tbin->lg_fill_div > 1)
+				tbin->lg_fill_div--;
+		}
+		tbin->low_water = tbin->ncached;
+
+		tcache->next_gc_bin++;
+		if (tcache->next_gc_bin == nhbins)
+			tcache->next_gc_bin = 0;
+		tcache->ev_cnt = 0;
+	}
+}
+
+JEMALLOC_INLINE void *
+tcache_alloc_easy(tcache_bin_t *tbin)
+{
+	void *ret;
+
+	if (tbin->ncached == 0) {
+		tbin->low_water = -1;
+		return (NULL);
+	}
+	tbin->ncached--;
+	if ((int)tbin->ncached < tbin->low_water)
+		tbin->low_water = tbin->ncached;
+	ret = tbin->avail[tbin->ncached];
+	return (ret);
+}
+
+JEMALLOC_INLINE void *
+tcache_alloc_small(tcache_t *tcache, size_t size, bool zero)
+{
+	void *ret;
+	size_t binind;
+	tcache_bin_t *tbin;
+
+	binind = SMALL_SIZE2BIN(size);
+	assert(binind < NBINS);
+	tbin = &tcache->tbins[binind];
+	ret = tcache_alloc_easy(tbin);
+	if (ret == NULL) {
+		ret = tcache_alloc_small_hard(tcache, tbin, binind);
+		if (ret == NULL)
+			return (NULL);
+	}
+	assert(arena_salloc(ret, false) == arena_bin_info[binind].reg_size);
+
+	if (zero == false) {
+		if (config_fill) {
+			if (opt_junk) {
+				arena_alloc_junk_small(ret,
+				    &arena_bin_info[binind], false);
+			} else if (opt_zero)
+				memset(ret, 0, size);
+		}
+	} else {
+		if (config_fill && opt_junk) {
+			arena_alloc_junk_small(ret, &arena_bin_info[binind],
+			    true);
+		}
+		VALGRIND_MAKE_MEM_UNDEFINED(ret, size);
+		memset(ret, 0, size);
+	}
+
+	if (config_stats)
+		tbin->tstats.nrequests++;
+	if (config_prof)
+		tcache->prof_accumbytes += arena_bin_info[binind].reg_size;
+	tcache_event(tcache);
+	return (ret);
+}
+
+JEMALLOC_INLINE void *
+tcache_alloc_large(tcache_t *tcache, size_t size, bool zero)
+{
+	void *ret;
+	size_t binind;
+	tcache_bin_t *tbin;
+
+	size = PAGE_CEILING(size);
+	assert(size <= tcache_maxclass);
+	binind = NBINS + (size >> LG_PAGE) - 1;
+	assert(binind < nhbins);
+	tbin = &tcache->tbins[binind];
+	ret = tcache_alloc_easy(tbin);
+	if (ret == NULL) {
+		/*
+		 * Only allocate one large object at a time, because it's quite
+		 * expensive to create one and not use it.
+		 */
+		ret = arena_malloc_large(tcache->arena, size, zero);
+		if (ret == NULL)
+			return (NULL);
+	} else {
+		if (config_prof) {
+			arena_chunk_t *chunk =
+			    (arena_chunk_t *)CHUNK_ADDR2BASE(ret);
+			size_t pageind = (((uintptr_t)ret - (uintptr_t)chunk) >>
+			    LG_PAGE);
+			chunk->map[pageind-map_bias].bits &=
+			    ~CHUNK_MAP_CLASS_MASK;
+		}
+		if (zero == false) {
+			if (config_fill) {
+				if (opt_junk)
+					memset(ret, 0xa5, size);
+				else if (opt_zero)
+					memset(ret, 0, size);
+			}
+		} else {
+			VALGRIND_MAKE_MEM_UNDEFINED(ret, size);
+			memset(ret, 0, size);
+		}
+
+		if (config_stats)
+			tbin->tstats.nrequests++;
+		if (config_prof)
+			tcache->prof_accumbytes += size;
+	}
+
+	tcache_event(tcache);
+	return (ret);
+}
+
+JEMALLOC_INLINE void
+tcache_dalloc_small(tcache_t *tcache, void *ptr)
+{
+	arena_t *arena;
+	arena_chunk_t *chunk;
+	arena_run_t *run;
+	arena_bin_t *bin;
+	tcache_bin_t *tbin;
+	tcache_bin_info_t *tbin_info;
+	size_t pageind, binind;
+	arena_chunk_map_t *mapelm;
+
+	assert(arena_salloc(ptr, false) <= SMALL_MAXCLASS);
+
+	chunk = (arena_chunk_t *)CHUNK_ADDR2BASE(ptr);
+	arena = chunk->arena;
+	pageind = ((uintptr_t)ptr - (uintptr_t)chunk) >> LG_PAGE;
+	mapelm = &chunk->map[pageind-map_bias];
+	run = (arena_run_t *)((uintptr_t)chunk + (uintptr_t)((pageind -
+	    (mapelm->bits >> LG_PAGE)) << LG_PAGE));
+	bin = run->bin;
+	binind = ((uintptr_t)bin - (uintptr_t)&arena->bins) /
+	    sizeof(arena_bin_t);
+	assert(binind < NBINS);
+
+	if (config_fill && opt_junk)
+		arena_dalloc_junk_small(ptr, &arena_bin_info[binind]);
+
+	tbin = &tcache->tbins[binind];
+	tbin_info = &tcache_bin_info[binind];
+	if (tbin->ncached == tbin_info->ncached_max) {
+		tcache_bin_flush_small(tbin, binind, (tbin_info->ncached_max >>
+		    1), tcache);
+	}
+	assert(tbin->ncached < tbin_info->ncached_max);
+	tbin->avail[tbin->ncached] = ptr;
+	tbin->ncached++;
+
+	tcache_event(tcache);
+}
+
+JEMALLOC_INLINE void
+tcache_dalloc_large(tcache_t *tcache, void *ptr, size_t size)
+{
+	size_t binind;
+	tcache_bin_t *tbin;
+	tcache_bin_info_t *tbin_info;
+
+	assert((size & PAGE_MASK) == 0);
+	assert(arena_salloc(ptr, false) > SMALL_MAXCLASS);
+	assert(arena_salloc(ptr, false) <= tcache_maxclass);
+
+	binind = NBINS + (size >> LG_PAGE) - 1;
+
+	if (config_fill && opt_junk)
+		memset(ptr, 0x5a, size);
+
+	tbin = &tcache->tbins[binind];
+	tbin_info = &tcache_bin_info[binind];
+	if (tbin->ncached == tbin_info->ncached_max) {
+		tcache_bin_flush_large(tbin, binind, (tbin_info->ncached_max >>
+		    1), tcache);
+	}
+	assert(tbin->ncached < tbin_info->ncached_max);
+	tbin->avail[tbin->ncached] = ptr;
+	tbin->ncached++;
+
+	tcache_event(tcache);
+}
+#endif
+
+#endif /* JEMALLOC_H_INLINES */
+/******************************************************************************/

contrib/jemalloc/include/jemalloc/internal/tsd.h

@@ -0,0 +1,309 @@
+/******************************************************************************/
+#ifdef JEMALLOC_H_TYPES
+
+/* Maximum number of malloc_tsd users with cleanup functions. */
+#define	MALLOC_TSD_CLEANUPS_MAX	8
+
+typedef struct malloc_tsd_cleanup_s malloc_tsd_cleanup_t;
+struct malloc_tsd_cleanup_s {
+	bool	(*f)(void *);
+	void	*arg;
+};
+
+/*
+ * TLS/TSD-agnostic macro-based implementation of thread-specific data.  There
+ * are four macros that support (at least) three use cases: file-private,
+ * library-private, and library-private inlined.  Following is an example
+ * library-private tsd variable:
+ *
+ * In example.h:
+ *   typedef struct {
+ *           int x;
+ *           int y;
+ *   } example_t;
+ *   #define EX_INITIALIZER JEMALLOC_CONCAT({0, 0})
+ *   malloc_tsd_protos(, example, example_t *)
+ *   malloc_tsd_externs(example, example_t *)
+ * In example.c:
+ *   malloc_tsd_data(, example, example_t *, EX_INITIALIZER)
+ *   malloc_tsd_funcs(, example, example_t *, EX_INITIALIZER,
+ *       example_tsd_cleanup)
+ *
+ * The result is a set of generated functions, e.g.:
+ *
+ *   bool example_tsd_boot(void) {...}
+ *   example_t **example_tsd_get() {...}
+ *   void example_tsd_set(example_t **val) {...}
+ *
+ * Note that all of the functions deal in terms of (a_type *) rather than
+ * (a_type)  so that it is possible to support non-pointer types (unlike
+ * pthreads TSD).  example_tsd_cleanup() is passed an (a_type *) pointer that is
+ * cast to (void *).  This means that the cleanup function needs to cast *and*
+ * dereference the function argument, e.g.:
+ *
+ *   void
+ *   example_tsd_cleanup(void *arg)
+ *   {
+ *           example_t *example = *(example_t **)arg;
+ *
+ *           [...]
+ *           if ([want the cleanup function to be called again]) {
+ *                   example_tsd_set(&example);
+ *           }
+ *   }
+ *
+ * If example_tsd_set() is called within example_tsd_cleanup(), it will be
+ * called again.  This is similar to how pthreads TSD destruction works, except
+ * that pthreads only calls the cleanup function again if the value was set to
+ * non-NULL.
+ */
+
+/* malloc_tsd_protos(). */
+#define	malloc_tsd_protos(a_attr, a_name, a_type)			\
+a_attr bool								\
+a_name##_tsd_boot(void);						\
+a_attr a_type *								\
+a_name##_tsd_get(void);							\
+a_attr void								\
+a_name##_tsd_set(a_type *val);
+
+/* malloc_tsd_externs(). */
+#ifdef JEMALLOC_MALLOC_THREAD_CLEANUP
+#define	malloc_tsd_externs(a_name, a_type)				\
+extern __thread a_type	a_name##_tls;					\
+extern __thread bool	a_name##_initialized;				\
+extern bool		a_name##_booted;
+#elif (defined(JEMALLOC_TLS))
+#define	malloc_tsd_externs(a_name, a_type)				\
+extern __thread a_type	a_name##_tls;					\
+extern pthread_key_t	a_name##_tsd;					\
+extern bool		a_name##_booted;
+#else
+#define	malloc_tsd_externs(a_name, a_type)				\
+extern pthread_key_t	a_name##_tsd;					\
+extern bool		a_name##_booted;
+#endif
+
+/* malloc_tsd_data(). */
+#ifdef JEMALLOC_MALLOC_THREAD_CLEANUP
+#define	malloc_tsd_data(a_attr, a_name, a_type, a_initializer)		\
+a_attr __thread a_type JEMALLOC_TLS_MODEL				\
+    a_name##_tls = a_initializer;					\
+a_attr __thread bool JEMALLOC_TLS_MODEL					\
+    a_name##_initialized = false;					\
+a_attr bool		a_name##_booted = false;
+#elif (defined(JEMALLOC_TLS))
+#define	malloc_tsd_data(a_attr, a_name, a_type, a_initializer)		\
+a_attr __thread a_type JEMALLOC_TLS_MODEL				\
+    a_name##_tls = a_initializer;					\
+a_attr pthread_key_t	a_name##_tsd;					\
+a_attr bool		a_name##_booted = false;
+#else
+#define	malloc_tsd_data(a_attr, a_name, a_type, a_initializer)		\
+a_attr pthread_key_t	a_name##_tsd;					\
+a_attr bool		a_name##_booted = false;
+#endif
+
+/* malloc_tsd_funcs(). */
+#ifdef JEMALLOC_MALLOC_THREAD_CLEANUP
+#define	malloc_tsd_funcs(a_attr, a_name, a_type, a_initializer,		\
+    a_cleanup)								\
+/* Initialization/cleanup. */						\
+a_attr bool								\
+a_name##_tsd_cleanup_wrapper(void *arg)					\
+{									\
+	bool (*cleanup)(void *) = arg;					\
+									\
+	if (a_name##_initialized) {					\
+		a_name##_initialized = false;				\
+		cleanup(&a_name##_tls);					\
+	}								\
+	return (a_name##_initialized);					\
+}									\
+a_attr bool								\
+a_name##_tsd_boot(void)							\
+{									\
+									\
+	if (a_cleanup != malloc_tsd_no_cleanup) {			\
+		malloc_tsd_cleanup_register(				\
+		    &a_name##_tsd_cleanup_wrapper, a_cleanup);		\
+	}								\
+	a_name##_booted = true;						\
+	return (false);							\
+}									\
+/* Get/set. */								\
+a_attr a_type *								\
+a_name##_tsd_get(void)							\
+{									\
+									\
+	assert(a_name##_booted);					\
+	return (&a_name##_tls);						\
+}									\
+a_attr void								\
+a_name##_tsd_set(a_type *val)						\
+{									\
+									\
+	assert(a_name##_booted);					\
+	a_name##_tls = (*val);						\
+	if (a_cleanup != malloc_tsd_no_cleanup)				\
+		a_name##_initialized = true;				\
+}
+#elif (defined(JEMALLOC_TLS))
+#define	malloc_tsd_funcs(a_attr, a_name, a_type, a_initializer,		\
+    a_cleanup)								\
+/* Initialization/cleanup. */						\
+a_attr bool								\
+a_name##_tsd_boot(void)							\
+{									\
+									\
+	if (a_cleanup != malloc_tsd_no_cleanup) {			\
+		if (pthread_key_create(&a_name##_tsd, a_cleanup) != 0)	\
+			return (true);					\
+	}								\
+	a_name##_booted = true;						\
+	return (false);							\
+}									\
+/* Get/set. */								\
+a_attr a_type *								\
+a_name##_tsd_get(void)							\
+{									\
+									\
+	assert(a_name##_booted);					\
+	return (&a_name##_tls);						\
+}									\
+a_attr void								\
+a_name##_tsd_set(a_type *val)						\
+{									\
+									\
+	assert(a_name##_booted);					\
+	a_name##_tls = (*val);						\
+	if (a_cleanup != malloc_tsd_no_cleanup) {			\
+		if (pthread_setspecific(a_name##_tsd,			\
+		    (void *)(&a_name##_tls))) {				\
+			malloc_write("<jemalloc>: Error"		\
+			    " setting TSD for "#a_name"\n");		\
+			if (opt_abort)					\
+				abort();				\
+		}							\
+	}								\
+}
+#else
+#define	malloc_tsd_funcs(a_attr, a_name, a_type, a_initializer,		\
+    a_cleanup)								\
+/* Data structure. */							\
+typedef struct {							\
+	bool	isstatic;						\
+	bool	initialized;						\
+	a_type	val;							\
+} a_name##_tsd_wrapper_t;						\
+/* Initialization/cleanup. */						\
+a_attr void								\
+a_name##_tsd_cleanup_wrapper(void *arg)					\
+{									\
+	a_name##_tsd_wrapper_t *wrapper = (a_name##_tsd_wrapper_t *)arg;\
+									\
+	if (a_cleanup != malloc_tsd_no_cleanup &&			\
+	    wrapper->initialized) {					\
+		wrapper->initialized = false;				\
+		a_cleanup(&wrapper->val);				\
+		if (wrapper->initialized) {				\
+			/* Trigger another cleanup round. */		\
+			if (pthread_setspecific(a_name##_tsd,		\
+			    (void *)wrapper)) {				\
+				malloc_write("<jemalloc>: Error"	\
+				    " setting TSD for "#a_name"\n");	\
+				if (opt_abort)				\
+					abort();			\
+			}						\
+			return;						\
+		}							\
+	}								\
+	if (wrapper->isstatic == false)					\
+		malloc_tsd_dalloc(wrapper);				\
+}									\
+a_attr bool								\
+a_name##_tsd_boot(void)							\
+{									\
+									\
+	if (pthread_key_create(&a_name##_tsd,				\
+	    a_name##_tsd_cleanup_wrapper) != 0)				\
+		return (true);						\
+	a_name##_booted = true;						\
+	return (false);							\
+}									\
+/* Get/set. */								\
+a_attr a_name##_tsd_wrapper_t *						\
+a_name##_tsd_get_wrapper(void)						\
+{									\
+	a_name##_tsd_wrapper_t *wrapper = (a_name##_tsd_wrapper_t *)	\
+	    pthread_getspecific(a_name##_tsd);				\
+									\
+	if (wrapper == NULL) {						\
+		wrapper = (a_name##_tsd_wrapper_t *)			\
+		    malloc_tsd_malloc(sizeof(a_name##_tsd_wrapper_t));	\
+		if (wrapper == NULL) {					\
+			static a_name##_tsd_wrapper_t			\
+			    a_name##_tsd_static_data =			\
+			    {true, false, a_initializer};		\
+			malloc_write("<jemalloc>: Error allocating"	\
+			    " TSD for "#a_name"\n");			\
+			if (opt_abort)					\
+				abort();				\
+			wrapper = &a_name##_tsd_static_data;		\
+		} else {						\
+			static a_type tsd_static_data = a_initializer;	\
+			wrapper->isstatic = false;			\
+			wrapper->val = tsd_static_data;			\
+		}							\
+		if (pthread_setspecific(a_name##_tsd,			\
+		    (void *)wrapper)) {					\
+			malloc_write("<jemalloc>: Error setting"	\
+			    " TSD for "#a_name"\n");			\
+			if (opt_abort)					\
+				abort();				\
+		}							\
+	}								\
+	return (wrapper);						\
+}									\
+a_attr a_type *								\
+a_name##_tsd_get(void)							\
+{									\
+	a_name##_tsd_wrapper_t *wrapper;				\
+									\
+	assert(a_name##_booted);					\
+	wrapper = a_name##_tsd_get_wrapper();				\
+	return (&wrapper->val);						\
+}									\
+a_attr void								\
+a_name##_tsd_set(a_type *val)						\
+{									\
+	a_name##_tsd_wrapper_t *wrapper;				\
+									\
+	assert(a_name##_booted);					\
+	wrapper = a_name##_tsd_get_wrapper();				\
+	wrapper->val = *(val);						\
+	if (a_cleanup != malloc_tsd_no_cleanup)				\
+		wrapper->initialized = true;				\
+}
+#endif
+
+#endif /* JEMALLOC_H_TYPES */
+/******************************************************************************/
+#ifdef JEMALLOC_H_STRUCTS
+
+#endif /* JEMALLOC_H_STRUCTS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_EXTERNS
+
+void	*malloc_tsd_malloc(size_t size);
+void	malloc_tsd_dalloc(void *wrapper);
+void	malloc_tsd_no_cleanup(void *);
+void	malloc_tsd_cleanup_register(bool (*f)(void *), void *arg);
+void	malloc_tsd_boot(void);
+
+#endif /* JEMALLOC_H_EXTERNS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_INLINES
+
+#endif /* JEMALLOC_H_INLINES */
+/******************************************************************************/

contrib/jemalloc/include/jemalloc/internal/util.h

@@ -0,0 +1,146 @@
+/******************************************************************************/
+#ifdef JEMALLOC_H_TYPES
+
+/* Size of stack-allocated buffer passed to buferror(). */
+#define	BUFERROR_BUF		64
+
+/*
+ * Size of stack-allocated buffer used by malloc_{,v,vc}printf().  This must be
+ * large enough for all possible uses within jemalloc.
+ */
+#define	MALLOC_PRINTF_BUFSIZE	4096
+
+/*
+ * Wrap a cpp argument that contains commas such that it isn't broken up into
+ * multiple arguments.
+ */
+#define JEMALLOC_CONCAT(...) __VA_ARGS__
+
+/*
+ * Silence compiler warnings due to uninitialized values.  This is used
+ * wherever the compiler fails to recognize that the variable is never used
+ * uninitialized.
+ */
+#ifdef JEMALLOC_CC_SILENCE
+#  define JEMALLOC_CC_SILENCE_INIT(v) = v
+#else
+#  define JEMALLOC_CC_SILENCE_INIT(v)
+#endif
+
+/*
+ * Define a custom assert() in order to reduce the chances of deadlock during
+ * assertion failure.
+ */
+#ifndef assert
+#define	assert(e) do {							\
+	if (config_debug && !(e)) {					\
+		malloc_printf(						\
+		    "<jemalloc>: %s:%d: Failed assertion: \"%s\"\n",	\
+		    __FILE__, __LINE__, #e);				\
+		abort();						\
+	}								\
+} while (0)
+#endif
+
+/* Use to assert a particular configuration, e.g., cassert(config_debug). */
+#define	cassert(c) do {							\
+	if ((c) == false)						\
+		assert(false);						\
+} while (0)
+
+#ifndef not_reached
+#define	not_reached() do {						\
+	if (config_debug) {						\
+		malloc_printf(						\
+		    "<jemalloc>: %s:%d: Unreachable code reached\n",	\
+		    __FILE__, __LINE__);				\
+		abort();						\
+	}								\
+} while (0)
+#endif
+
+#ifndef not_implemented
+#define	not_implemented() do {						\
+	if (config_debug) {						\
+		malloc_printf("<jemalloc>: %s:%d: Not implemented\n",	\
+		    __FILE__, __LINE__);				\
+		abort();						\
+	}								\
+} while (0)
+#endif
+
+#define	assert_not_implemented(e) do {					\
+	if (config_debug && !(e))					\
+		not_implemented();					\
+} while (0)
+
+#endif /* JEMALLOC_H_TYPES */
+/******************************************************************************/
+#ifdef JEMALLOC_H_STRUCTS
+
+#endif /* JEMALLOC_H_STRUCTS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_EXTERNS
+
+extern void	(*je_malloc_message)(void *wcbopaque, const char *s);
+
+int	buferror(int errnum, char *buf, size_t buflen);
+uintmax_t	malloc_strtoumax(const char *nptr, char **endptr, int base);
+
+/*
+ * malloc_vsnprintf() supports a subset of snprintf(3) that avoids floating
+ * point math.
+ */
+int	malloc_vsnprintf(char *str, size_t size, const char *format,
+    va_list ap);
+int	malloc_snprintf(char *str, size_t size, const char *format, ...)
+    JEMALLOC_ATTR(format(printf, 3, 4));
+void	malloc_vcprintf(void (*write_cb)(void *, const char *), void *cbopaque,
+    const char *format, va_list ap);
+void malloc_cprintf(void (*write)(void *, const char *), void *cbopaque,
+    const char *format, ...) JEMALLOC_ATTR(format(printf, 3, 4));
+void	malloc_printf(const char *format, ...)
+    JEMALLOC_ATTR(format(printf, 1, 2));
+
+#endif /* JEMALLOC_H_EXTERNS */
+/******************************************************************************/
+#ifdef JEMALLOC_H_INLINES
+
+#ifndef JEMALLOC_ENABLE_INLINE
+size_t	pow2_ceil(size_t x);
+void	malloc_write(const char *s);
+#endif
+
+#if (defined(JEMALLOC_ENABLE_INLINE) || defined(JEMALLOC_UTIL_C_))
+/* Compute the smallest power of 2 that is >= x. */
+JEMALLOC_INLINE size_t
+pow2_ceil(size_t x)
+{
+
+	x--;
+	x |= x >> 1;
+	x |= x >> 2;
+	x |= x >> 4;
+	x |= x >> 8;
+	x |= x >> 16;
+#if (LG_SIZEOF_PTR == 3)
+	x |= x >> 32;
+#endif
+	x++;
+	return (x);
+}
+
+/*
+ * Wrapper around malloc_message() that avoids the need for
+ * je_malloc_message(...) throughout the code.
+ */
+JEMALLOC_INLINE void
+malloc_write(const char *s)
+{
+
+	je_malloc_message(NULL, s);
+}
+#endif
+
+#endif /* JEMALLOC_H_INLINES */
+/******************************************************************************/

contrib/jemalloc/include/jemalloc/jemalloc.h

@@ -0,0 +1,141 @@
+#ifndef JEMALLOC_H_
+#define	JEMALLOC_H_
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#include <limits.h>
+#include <strings.h>
+
+#define	JEMALLOC_VERSION "1.0.0-258-g9ef7f5dc34ff02f50d401e41c8d9a4a928e7c2aa"
+#define	JEMALLOC_VERSION_MAJOR 1
+#define	JEMALLOC_VERSION_MINOR 0
+#define	JEMALLOC_VERSION_BUGFIX 0
+#define	JEMALLOC_VERSION_NREV 258
+#define	JEMALLOC_VERSION_GID "9ef7f5dc34ff02f50d401e41c8d9a4a928e7c2aa"
+
+#include "jemalloc_defs.h"
+#include "jemalloc_FreeBSD.h"
+
+#ifdef JEMALLOC_EXPERIMENTAL
+#define	ALLOCM_LG_ALIGN(la)	(la)
+#if LG_SIZEOF_PTR == 2
+#define	ALLOCM_ALIGN(a)	(ffs(a)-1)
+#else
+#define	ALLOCM_ALIGN(a)	((a < (size_t)INT_MAX) ? ffs(a)-1 : ffs(a>>32)+31)
+#endif
+#define	ALLOCM_ZERO	((int)0x40)
+#define	ALLOCM_NO_MOVE	((int)0x80)
+
+#define	ALLOCM_SUCCESS		0
+#define	ALLOCM_ERR_OOM		1
+#define	ALLOCM_ERR_NOT_MOVED	2
+#endif
+
+/*
+ * The je_ prefix on the following public symbol declarations is an artifact of
+ * namespace management, and should be omitted in application code unless
+ * JEMALLOC_NO_DEMANGLE is defined (see below).
+ */
+extern const char	*je_malloc_conf;
+extern void		(*je_malloc_message)(void *, const char *);
+
+void	*je_malloc(size_t size) JEMALLOC_ATTR(malloc);
+void	*je_calloc(size_t num, size_t size) JEMALLOC_ATTR(malloc);
+int	je_posix_memalign(void **memptr, size_t alignment, size_t size)
+    JEMALLOC_ATTR(nonnull(1));
+void	*je_aligned_alloc(size_t alignment, size_t size) JEMALLOC_ATTR(malloc);
+void	*je_realloc(void *ptr, size_t size);
+void	je_free(void *ptr);
+
+size_t	je_malloc_usable_size(const void *ptr);
+void	je_malloc_stats_print(void (*write_cb)(void *, const char *),
+    void *je_cbopaque, const char *opts);
+int	je_mallctl(const char *name, void *oldp, size_t *oldlenp, void *newp,
+    size_t newlen);
+int	je_mallctlnametomib(const char *name, size_t *mibp, size_t *miblenp);
+int	je_mallctlbymib(const size_t *mib, size_t miblen, void *oldp,
+    size_t *oldlenp, void *newp, size_t newlen);
+
+#ifdef JEMALLOC_EXPERIMENTAL
+int	je_allocm(void **ptr, size_t *rsize, size_t size, int flags)
+    JEMALLOC_ATTR(nonnull(1));
+int	je_rallocm(void **ptr, size_t *rsize, size_t size, size_t extra,
+    int flags) JEMALLOC_ATTR(nonnull(1));
+int	je_sallocm(const void *ptr, size_t *rsize, int flags)
+    JEMALLOC_ATTR(nonnull(1));
+int	je_dallocm(void *ptr, int flags) JEMALLOC_ATTR(nonnull(1));
+int	je_nallocm(size_t *rsize, size_t size, int flags);
+#endif
+
+/*
+ * By default application code must explicitly refer to mangled symbol names,
+ * so that it is possible to use jemalloc in conjunction with another allocator
+ * in the same application.  Define JEMALLOC_MANGLE in order to cause automatic
+ * name mangling that matches the API prefixing that happened as a result of
+ * --with-mangling and/or --with-jemalloc-prefix configuration settings.
+ */
+#ifdef JEMALLOC_MANGLE
+#ifndef JEMALLOC_NO_DEMANGLE
+#define	JEMALLOC_NO_DEMANGLE
+#endif
+#define	malloc_conf je_malloc_conf
+#define	malloc_message je_malloc_message
+#define	malloc je_malloc
+#define	calloc je_calloc
+#define	posix_memalign je_posix_memalign
+#define	aligned_alloc je_aligned_alloc
+#define	realloc je_realloc
+#define	free je_free
+#define	malloc_usable_size je_malloc_usable_size
+#define	malloc_stats_print je_malloc_stats_print
+#define	mallctl je_mallctl
+#define	mallctlnametomib je_mallctlnametomib
+#define	mallctlbymib je_mallctlbymib
+#define	memalign je_memalign
+#define	valloc je_valloc
+#ifdef JEMALLOC_EXPERIMENTAL
+#define	allocm je_allocm
+#define	rallocm je_rallocm
+#define	sallocm je_sallocm
+#define	dallocm je_dallocm
+#define	nallocm je_nallocm
+#endif
+#endif
+
+/*
+ * The je_* macros can be used as stable alternative names for the public
+ * jemalloc API if JEMALLOC_NO_DEMANGLE is defined.  This is primarily meant
+ * for use in jemalloc itself, but it can be used by application code to
+ * provide isolation from the name mangling specified via --with-mangling
+ * and/or --with-jemalloc-prefix.
+ */
+#ifndef JEMALLOC_NO_DEMANGLE
+#undef je_malloc_conf
+#undef je_malloc_message
+#undef je_malloc
+#undef je_calloc
+#undef je_posix_memalign
+#undef je_aligned_alloc
+#undef je_realloc
+#undef je_free
+#undef je_malloc_usable_size
+#undef je_malloc_stats_print
+#undef je_mallctl
+#undef je_mallctlnametomib
+#undef je_mallctlbymib
+#undef je_memalign
+#undef je_valloc
+#ifdef JEMALLOC_EXPERIMENTAL
+#undef je_allocm
+#undef je_rallocm
+#undef je_sallocm
+#undef je_dallocm
+#undef je_nallocm
+#endif
+#endif
+
+#ifdef __cplusplus
+};
+#endif
+#endif /* JEMALLOC_H_ */

contrib/jemalloc/include/jemalloc/jemalloc_FreeBSD.h

@@ -0,0 +1,76 @@
+/*
+ * Override settings that were generated in jemalloc_defs.h as necessary.
+ */
+
+#undef JEMALLOC_OVERRIDE_VALLOC
+
+#ifndef MALLOC_PRODUCTION
+#define	JEMALLOC_DEBUG
+#endif
+
+/*
+ * The following are architecture-dependent, so conditionally define them for
+ * each supported architecture.
+ */
+#undef CPU_SPINWAIT
+#undef JEMALLOC_TLS_MODEL
+#undef STATIC_PAGE_SHIFT
+#undef LG_SIZEOF_PTR
+#undef LG_SIZEOF_INT
+#undef LG_SIZEOF_LONG
+#undef LG_SIZEOF_INTMAX_T
+
+#ifdef __i386__
+#  define LG_SIZEOF_PTR		2
+#  define CPU_SPINWAIT		__asm__ volatile("pause")
+#  define JEMALLOC_TLS_MODEL	__attribute__((tls_model("initial-exec")))
+#endif
+#ifdef __ia64__
+#  define LG_SIZEOF_PTR		3
+#endif
+#ifdef __sparc64__
+#  define LG_SIZEOF_PTR		3
+#  define JEMALLOC_TLS_MODEL	__attribute__((tls_model("initial-exec")))
+#endif
+#ifdef __amd64__
+#  define LG_SIZEOF_PTR		3
+#  define CPU_SPINWAIT		__asm__ volatile("pause")
+#  define JEMALLOC_TLS_MODEL	__attribute__((tls_model("initial-exec")))
+#endif
+#ifdef __arm__
+#  define LG_SIZEOF_PTR		2
+#endif
+#ifdef __mips__
+#  define LG_SIZEOF_PTR		2
+#endif
+#ifdef __powerpc64__
+#  define LG_SIZEOF_PTR		3
+#elif defined(__powerpc__)
+#  define LG_SIZEOF_PTR		2
+#endif
+
+#ifndef JEMALLOC_TLS_MODEL
+#  define JEMALLOC_TLS_MODEL	/* Default. */
+#endif
+#ifdef __clang__
+#  undef JEMALLOC_TLS_MODEL
+#  define JEMALLOC_TLS_MODEL	/* clang does not support tls_model yet. */
+#endif
+
+#define	STATIC_PAGE_SHIFT	PAGE_SHIFT
+#define	LG_SIZEOF_INT		2
+#define	LG_SIZEOF_LONG		LG_SIZEOF_PTR
+#define	LG_SIZEOF_INTMAX_T	3
+
+/* Disable lazy-lock machinery, mangle isthreaded, and adjust its type. */
+#undef JEMALLOC_LAZY_LOCK
+extern int __isthreaded;
+#define	isthreaded		((bool)__isthreaded)
+
+/* Mangle. */
+#define	open			_open
+#define	read			_read
+#define	write			_write
+#define	close			_close
+#define	pthread_mutex_lock	_pthread_mutex_lock
+#define	pthread_mutex_unlock	_pthread_mutex_unlock

contrib/jemalloc/include/jemalloc/jemalloc_defs.h

@@ -0,0 +1,239 @@
+/* include/jemalloc/jemalloc_defs.h.  Generated from jemalloc_defs.h.in by configure.  */
+/*
+ * If JEMALLOC_PREFIX is defined via --with-jemalloc-prefix, it will cause all
+ * public APIs to be prefixed.  This makes it possible, with some care, to use
+ * multiple allocators simultaneously.
+ */
+/* #undef JEMALLOC_PREFIX */
+/* #undef JEMALLOC_CPREFIX */
+
+/*
+ * Name mangling for public symbols is controlled by --with-mangling and
+ * --with-jemalloc-prefix.  With default settings the je_ prefix is stripped by
+ * these macro definitions.
+ */
+#define je_malloc_conf malloc_conf
+#define je_malloc_message malloc_message
+#define je_malloc malloc
+#define je_calloc calloc
+#define je_posix_memalign posix_memalign
+#define je_aligned_alloc aligned_alloc
+#define je_realloc realloc
+#define je_free free
+#define je_malloc_usable_size malloc_usable_size
+#define je_malloc_stats_print malloc_stats_print
+#define je_mallctl mallctl
+#define je_mallctlnametomib mallctlnametomib
+#define je_mallctlbymib mallctlbymib
+/* #undef je_memalign */
+#define je_valloc valloc
+#define je_allocm allocm
+#define je_rallocm rallocm
+#define je_sallocm sallocm
+#define je_dallocm dallocm
+#define je_nallocm nallocm
+
+/*
+ * JEMALLOC_PRIVATE_NAMESPACE is used as a prefix for all library-private APIs.
+ * For shared libraries, symbol visibility mechanisms prevent these symbols
+ * from being exported, but for static libraries, naming collisions are a real
+ * possibility.
+ */
+#define JEMALLOC_PRIVATE_NAMESPACE ""
+#define JEMALLOC_N(string_that_no_one_should_want_to_use_as_a_jemalloc_private_namespace_prefix) string_that_no_one_should_want_to_use_as_a_jemalloc_private_namespace_prefix
+
+/*
+ * Hyper-threaded CPUs may need a special instruction inside spin loops in
+ * order to yield to another virtual CPU.
+ */
+#define CPU_SPINWAIT __asm__ volatile("pause")
+
+/*
+ * Defined if OSAtomic*() functions are available, as provided by Darwin, and
+ * documented in the atomic(3) manual page.
+ */
+/* #undef JEMALLOC_OSATOMIC */
+
+/*
+ * Defined if __sync_add_and_fetch(uint32_t *, uint32_t) and
+ * __sync_sub_and_fetch(uint32_t *, uint32_t) are available, despite
+ * __GCC_HAVE_SYNC_COMPARE_AND_SWAP_4 not being defined (which means the
+ * functions are defined in libgcc instead of being inlines)
+ */
+#define JE_FORCE_SYNC_COMPARE_AND_SWAP_4 
+
+/*
+ * Defined if __sync_add_and_fetch(uint64_t *, uint64_t) and
+ * __sync_sub_and_fetch(uint64_t *, uint64_t) are available, despite
+ * __GCC_HAVE_SYNC_COMPARE_AND_SWAP_8 not being defined (which means the
+ * functions are defined in libgcc instead of being inlines)
+ */
+#define JE_FORCE_SYNC_COMPARE_AND_SWAP_8 
+
+/*
+ * Defined if OSSpin*() functions are available, as provided by Darwin, and
+ * documented in the spinlock(3) manual page.
+ */
+/* #undef JEMALLOC_OSSPIN */
+
+/*
+ * Defined if _malloc_thread_cleanup() exists.  At least in the case of
+ * FreeBSD, pthread_key_create() allocates, which if used during malloc
+ * bootstrapping will cause recursion into the pthreads library.  Therefore, if
+ * _malloc_thread_cleanup() exists, use it as the basis for thread cleanup in
+ * malloc_tsd.
+ */
+#define JEMALLOC_MALLOC_THREAD_CLEANUP 
+
+/*
+ * Defined if threaded initialization is known to be safe on this platform.
+ * Among other things, it must be possible to initialize a mutex without
+ * triggering allocation in order for threaded allocation to be safe.
+ */
+/* #undef JEMALLOC_THREADED_INIT */
+
+/*
+ * Defined if the pthreads implementation defines
+ * _pthread_mutex_init_calloc_cb(), in which case the function is used in order
+ * to avoid recursive allocation during mutex initialization.
+ */
+#define JEMALLOC_MUTEX_INIT_CB 1
+
+/* Defined if __attribute__((...)) syntax is supported. */
+#define JEMALLOC_HAVE_ATTR 
+#ifdef JEMALLOC_HAVE_ATTR
+#  define JEMALLOC_CATTR(s, a) __attribute__((s))
+#  define JEMALLOC_ATTR(s) JEMALLOC_CATTR(s,)
+#else
+#  define JEMALLOC_CATTR(s, a) a
+#  define JEMALLOC_ATTR(s) JEMALLOC_CATTR(s,)
+#endif
+
+/* Defined if sbrk() is supported. */
+#define JEMALLOC_HAVE_SBRK 
+
+/* Non-empty if the tls_model attribute is supported. */
+#define JEMALLOC_TLS_MODEL __attribute__((tls_model("initial-exec")))
+
+/* JEMALLOC_CC_SILENCE enables code that silences unuseful compiler warnings. */
+#define JEMALLOC_CC_SILENCE 
+
+/*
+ * JEMALLOC_DEBUG enables assertions and other sanity checks, and disables
+ * inline functions.
+ */
+/* #undef JEMALLOC_DEBUG */
+
+/* JEMALLOC_STATS enables statistics calculation. */
+#define JEMALLOC_STATS 
+
+/* JEMALLOC_PROF enables allocation profiling. */
+/* #undef JEMALLOC_PROF */
+
+/* Use libunwind for profile backtracing if defined. */
+/* #undef JEMALLOC_PROF_LIBUNWIND */
+
+/* Use libgcc for profile backtracing if defined. */
+/* #undef JEMALLOC_PROF_LIBGCC */
+
+/* Use gcc intrinsics for profile backtracing if defined. */
+/* #undef JEMALLOC_PROF_GCC */
+
+/*
+ * JEMALLOC_TCACHE enables a thread-specific caching layer for small objects.
+ * This makes it possible to allocate/deallocate objects without any locking
+ * when the cache is in the steady state.
+ */
+#define JEMALLOC_TCACHE 
+
+/*
+ * JEMALLOC_DSS enables use of sbrk(2) to allocate chunks from the data storage
+ * segment (DSS).
+ */
+#define JEMALLOC_DSS 
+
+/* Support memory filling (junk/zero/quarantine/redzone). */
+#define JEMALLOC_FILL 
+
+/* Support the experimental API. */
+#define JEMALLOC_EXPERIMENTAL 
+
+/* Support utrace(2)-based tracing. */
+#define JEMALLOC_UTRACE 
+
+/* Support Valgrind. */
+/* #undef JEMALLOC_VALGRIND */
+
+/* Support optional abort() on OOM. */
+#define JEMALLOC_XMALLOC 
+
+/* Support lazy locking (avoid locking unless a second thread is launched). */
+#define JEMALLOC_LAZY_LOCK 
+
+/* One page is 2^STATIC_PAGE_SHIFT bytes. */
+#define STATIC_PAGE_SHIFT 12
+
+/*
+ * If defined, use munmap() to unmap freed chunks, rather than storing them for
+ * later reuse.  This is automatically disabled if configuration determines
+ * that common sequences of mmap()/munmap() calls will cause virtual memory map
+ * holes.
+ */
+#define JEMALLOC_MUNMAP 
+
+/* TLS is used to map arenas and magazine caches to threads. */
+#define JEMALLOC_TLS 
+
+/*
+ * JEMALLOC_IVSALLOC enables ivsalloc(), which verifies that pointers reside
+ * within jemalloc-owned chunks before dereferencing them.
+ */
+/* #undef JEMALLOC_IVSALLOC */
+
+/*
+ * Define overrides for non-standard allocator-related functions if they
+ * are present on the system.
+ */
+/* #undef JEMALLOC_OVERRIDE_MEMALIGN */
+#define JEMALLOC_OVERRIDE_VALLOC 
+
+/*
+ * Darwin (OS X) uses zones to work around Mach-O symbol override shortcomings.
+ */
+/* #undef JEMALLOC_ZONE */
+/* #undef JEMALLOC_ZONE_VERSION */
+
+/* If defined, use mremap(...MREMAP_FIXED...) for huge realloc(). */
+/* #undef JEMALLOC_MREMAP_FIXED */
+
+/*
+ * Methods for purging unused pages differ between operating systems.
+ *
+ *   madvise(..., MADV_DONTNEED) : On Linux, this immediately discards pages,
+ *                                 such that new pages will be demand-zeroed if
+ *                                 the address region is later touched.
+ *   madvise(..., MADV_FREE) : On FreeBSD and Darwin, this marks pages as being
+ *                             unused, such that they will be discarded rather
+ *                             than swapped out.
+ */
+/* #undef JEMALLOC_PURGE_MADVISE_DONTNEED */
+#define JEMALLOC_PURGE_MADVISE_FREE 
+#ifdef JEMALLOC_PURGE_MADVISE_DONTNEED
+#  define JEMALLOC_MADV_PURGE MADV_DONTNEED
+#elif defined(JEMALLOC_PURGE_MADVISE_FREE)
+#  define JEMALLOC_MADV_PURGE MADV_FREE
+#else
+#  error "No method defined for purging unused dirty pages."
+#endif
+
+/* sizeof(void *) == 2^LG_SIZEOF_PTR. */
+#define LG_SIZEOF_PTR 3
+
+/* sizeof(int) == 2^LG_SIZEOF_INT. */
+#define LG_SIZEOF_INT 2
+
+/* sizeof(long) == 2^LG_SIZEOF_LONG. */
+#define LG_SIZEOF_LONG 3
+
+/* sizeof(intmax_t) == 2^LG_SIZEOF_INTMAX_T. */
+#define LG_SIZEOF_INTMAX_T 3

contrib/jemalloc/src/atomic.c

@@ -0,0 +1,2 @@
+#define	JEMALLOC_ATOMIC_C_
+#include "jemalloc/internal/jemalloc_internal.h"

contrib/jemalloc/src/base.c

@@ -0,0 +1,138 @@
+#define	JEMALLOC_BASE_C_
+#include "jemalloc/internal/jemalloc_internal.h"
+
+/******************************************************************************/
+/* Data. */
+
+static malloc_mutex_t	base_mtx;
+
+/*
+ * Current pages that are being used for internal memory allocations.  These
+ * pages are carved up in cacheline-size quanta, so that there is no chance of
+ * false cache line sharing.
+ */
+static void		*base_pages;
+static void		*base_next_addr;
+static void		*base_past_addr; /* Addr immediately past base_pages. */
+static extent_node_t	*base_nodes;
+
+/******************************************************************************/
+/* Function prototypes for non-inline static functions. */
+
+static bool	base_pages_alloc(size_t minsize);
+
+/******************************************************************************/
+
+static bool
+base_pages_alloc(size_t minsize)
+{
+	size_t csize;
+	bool zero;
+
+	assert(minsize != 0);
+	csize = CHUNK_CEILING(minsize);
+	zero = false;
+	base_pages = chunk_alloc(csize, chunksize, true, &zero);
+	if (base_pages == NULL)
+		return (true);
+	base_next_addr = base_pages;
+	base_past_addr = (void *)((uintptr_t)base_pages + csize);
+
+	return (false);
+}
+
+void *
+base_alloc(size_t size)
+{
+	void *ret;
+	size_t csize;
+
+	/* Round size up to nearest multiple of the cacheline size. */
+	csize = CACHELINE_CEILING(size);
+
+	malloc_mutex_lock(&base_mtx);
+	/* Make sure there's enough space for the allocation. */
+	if ((uintptr_t)base_next_addr + csize > (uintptr_t)base_past_addr) {
+		if (base_pages_alloc(csize)) {
+			malloc_mutex_unlock(&base_mtx);
+			return (NULL);
+		}
+	}
+	/* Allocate. */
+	ret = base_next_addr;
+	base_next_addr = (void *)((uintptr_t)base_next_addr + csize);
+	malloc_mutex_unlock(&base_mtx);
+
+	return (ret);
+}
+
+void *
+base_calloc(size_t number, size_t size)
+{
+	void *ret = base_alloc(number * size);
+
+	if (ret != NULL)
+		memset(ret, 0, number * size);
+
+	return (ret);
+}
+
+extent_node_t *
+base_node_alloc(void)
+{
+	extent_node_t *ret;
+
+	malloc_mutex_lock(&base_mtx);
+	if (base_nodes != NULL) {
+		ret = base_nodes;
+		base_nodes = *(extent_node_t **)ret;
+		malloc_mutex_unlock(&base_mtx);
+	} else {
+		malloc_mutex_unlock(&base_mtx);
+		ret = (extent_node_t *)base_alloc(sizeof(extent_node_t));
+	}
+
+	return (ret);
+}
+
+void
+base_node_dealloc(extent_node_t *node)
+{
+
+	malloc_mutex_lock(&base_mtx);
+	*(extent_node_t **)node = base_nodes;
+	base_nodes = node;
+	malloc_mutex_unlock(&base_mtx);
+}
+
+bool
+base_boot(void)
+{
+
+	base_nodes = NULL;
+	if (malloc_mutex_init(&base_mtx))
+		return (true);
+
+	return (false);
+}
+
+void
+base_prefork(void)
+{
+
+	malloc_mutex_prefork(&base_mtx);
+}
+
+void
+base_postfork_parent(void)
+{
+
+	malloc_mutex_postfork_parent(&base_mtx);
+}
+
+void
+base_postfork_child(void)
+{
+
+	malloc_mutex_postfork_child(&base_mtx);
+}

contrib/jemalloc/src/bitmap.c

@@ -0,0 +1,90 @@
+#define JEMALLOC_BITMAP_C_
+#include "jemalloc/internal/jemalloc_internal.h"
+
+/******************************************************************************/
+/* Function prototypes for non-inline static functions. */
+
+static size_t	bits2groups(size_t nbits);
+
+/******************************************************************************/
+
+static size_t
+bits2groups(size_t nbits)
+{
+
+	return ((nbits >> LG_BITMAP_GROUP_NBITS) +
+	    !!(nbits & BITMAP_GROUP_NBITS_MASK));
+}
+
+void
+bitmap_info_init(bitmap_info_t *binfo, size_t nbits)
+{
+	unsigned i;
+	size_t group_count;
+
+	assert(nbits > 0);
+	assert(nbits <= (ZU(1) << LG_BITMAP_MAXBITS));
+
+	/*
+	 * Compute the number of groups necessary to store nbits bits, and
+	 * progressively work upward through the levels until reaching a level
+	 * that requires only one group.
+	 */
+	binfo->levels[0].group_offset = 0;
+	group_count = bits2groups(nbits);
+	for (i = 1; group_count > 1; i++) {
+		assert(i < BITMAP_MAX_LEVELS);
+		binfo->levels[i].group_offset = binfo->levels[i-1].group_offset
+		    + group_count;
+		group_count = bits2groups(group_count);
+	}
+	binfo->levels[i].group_offset = binfo->levels[i-1].group_offset
+	    + group_count;
+	binfo->nlevels = i;
+	binfo->nbits = nbits;
+}
+
+size_t
+bitmap_info_ngroups(const bitmap_info_t *binfo)
+{
+
+	return (binfo->levels[binfo->nlevels].group_offset << LG_SIZEOF_BITMAP);
+}
+
+size_t
+bitmap_size(size_t nbits)
+{
+	bitmap_info_t binfo;
+
+	bitmap_info_init(&binfo, nbits);
+	return (bitmap_info_ngroups(&binfo));
+}
+
+void
+bitmap_init(bitmap_t *bitmap, const bitmap_info_t *binfo)
+{
+	size_t extra;
+	unsigned i;
+
+	/*
+	 * Bits are actually inverted with regard to the external bitmap
+	 * interface, so the bitmap starts out with all 1 bits, except for
+	 * trailing unused bits (if any).  Note that each group uses bit 0 to
+	 * correspond to the first logical bit in the group, so extra bits
+	 * are the most significant bits of the last group.
+	 */
+	memset(bitmap, 0xffU, binfo->levels[binfo->nlevels].group_offset <<
+	    LG_SIZEOF_BITMAP);
+	extra = (BITMAP_GROUP_NBITS - (binfo->nbits & BITMAP_GROUP_NBITS_MASK))
+	    & BITMAP_GROUP_NBITS_MASK;
+	if (extra != 0)
+		bitmap[binfo->levels[1].group_offset - 1] >>= extra;
+	for (i = 1; i < binfo->nlevels; i++) {
+		size_t group_count = binfo->levels[i].group_offset -
+		    binfo->levels[i-1].group_offset;
+		extra = (BITMAP_GROUP_NBITS - (group_count &
+		    BITMAP_GROUP_NBITS_MASK)) & BITMAP_GROUP_NBITS_MASK;
+		if (extra != 0)
+			bitmap[binfo->levels[i+1].group_offset - 1] >>= extra;
+	}
+}

contrib/jemalloc/src/chunk.c

@@ -0,0 +1,304 @@
+#define	JEMALLOC_CHUNK_C_
+#include "jemalloc/internal/jemalloc_internal.h"
+
+/******************************************************************************/
+/* Data. */
+
+size_t	opt_lg_chunk = LG_CHUNK_DEFAULT;
+
+malloc_mutex_t	chunks_mtx;
+chunk_stats_t	stats_chunks;
+
+/*
+ * Trees of chunks that were previously allocated (trees differ only in node
+ * ordering).  These are used when allocating chunks, in an attempt to re-use
+ * address space.  Depending on function, different tree orderings are needed,
+ * which is why there are two trees with the same contents.
+ */
+static extent_tree_t	chunks_szad;
+static extent_tree_t	chunks_ad;
+
+rtree_t		*chunks_rtree;
+
+/* Various chunk-related settings. */
+size_t		chunksize;
+size_t		chunksize_mask; /* (chunksize - 1). */
+size_t		chunk_npages;
+size_t		map_bias;
+size_t		arena_maxclass; /* Max size class for arenas. */
+
+/******************************************************************************/
+/* Function prototypes for non-inline static functions. */
+
+static void	*chunk_recycle(size_t size, size_t alignment, bool *zero);
+static void	chunk_record(void *chunk, size_t size);
+
+/******************************************************************************/
+
+static void *
+chunk_recycle(size_t size, size_t alignment, bool *zero)
+{
+	void *ret;
+	extent_node_t *node;
+	extent_node_t key;
+	size_t alloc_size, leadsize, trailsize;
+
+	alloc_size = size + alignment - chunksize;
+	/* Beware size_t wrap-around. */
+	if (alloc_size < size)
+		return (NULL);
+	key.addr = NULL;
+	key.size = alloc_size;
+	malloc_mutex_lock(&chunks_mtx);
+	node = extent_tree_szad_nsearch(&chunks_szad, &key);
+	if (node == NULL) {
+		malloc_mutex_unlock(&chunks_mtx);
+		return (NULL);
+	}
+	leadsize = ALIGNMENT_CEILING((uintptr_t)node->addr, alignment) -
+	    (uintptr_t)node->addr;
+	assert(alloc_size >= leadsize + size);
+	trailsize = alloc_size - leadsize - size;
+	ret = (void *)((uintptr_t)node->addr + leadsize);
+	/* Remove node from the tree. */
+	extent_tree_szad_remove(&chunks_szad, node);
+	extent_tree_ad_remove(&chunks_ad, node);
+	if (leadsize != 0) {
+		/* Insert the leading space as a smaller chunk. */
+		node->size = leadsize;
+		extent_tree_szad_insert(&chunks_szad, node);
+		extent_tree_ad_insert(&chunks_ad, node);
+		node = NULL;
+	}
+	if (trailsize != 0) {
+		/* Insert the trailing space as a smaller chunk. */
+		if (node == NULL) {
+			/*
+			 * An additional node is required, but
+			 * base_node_alloc() can cause a new base chunk to be
+			 * allocated.  Drop chunks_mtx in order to avoid
+			 * deadlock, and if node allocation fails, deallocate
+			 * the result before returning an error.
+			 */
+			malloc_mutex_unlock(&chunks_mtx);
+			node = base_node_alloc();
+			if (node == NULL) {
+				chunk_dealloc(ret, size, true);
+				return (NULL);
+			}
+			malloc_mutex_lock(&chunks_mtx);
+		}
+		node->addr = (void *)((uintptr_t)(ret) + size);
+		node->size = trailsize;
+		extent_tree_szad_insert(&chunks_szad, node);
+		extent_tree_ad_insert(&chunks_ad, node);
+		node = NULL;
+	}
+	malloc_mutex_unlock(&chunks_mtx);
+
+	if (node != NULL)
+		base_node_dealloc(node);
+#ifdef JEMALLOC_PURGE_MADVISE_FREE
+	if (*zero) {
+		VALGRIND_MAKE_MEM_UNDEFINED(ret, size);
+		memset(ret, 0, size);
+	}
+#endif
+	return (ret);
+}
+
+/*
+ * If the caller specifies (*zero == false), it is still possible to receive
+ * zeroed memory, in which case *zero is toggled to true.  arena_chunk_alloc()
+ * takes advantage of this to avoid demanding zeroed chunks, but taking
+ * advantage of them if they are returned.
+ */
+void *
+chunk_alloc(size_t size, size_t alignment, bool base, bool *zero)
+{
+	void *ret;
+
+	assert(size != 0);
+	assert((size & chunksize_mask) == 0);
+	assert((alignment & chunksize_mask) == 0);
+
+	ret = chunk_recycle(size, alignment, zero);
+	if (ret != NULL)
+		goto label_return;
+	if (config_dss) {
+		ret = chunk_alloc_dss(size, alignment, zero);
+		if (ret != NULL)
+			goto label_return;
+	}
+	ret = chunk_alloc_mmap(size, alignment);
+	if (ret != NULL) {
+		*zero = true;
+		goto label_return;
+	}
+
+	/* All strategies for allocation failed. */
+	ret = NULL;
+label_return:
+	if (config_ivsalloc && base == false && ret != NULL) {
+		if (rtree_set(chunks_rtree, (uintptr_t)ret, ret)) {
+			chunk_dealloc(ret, size, true);
+			return (NULL);
+		}
+	}
+	if ((config_stats || config_prof) && ret != NULL) {
+		bool gdump;
+		malloc_mutex_lock(&chunks_mtx);
+		if (config_stats)
+			stats_chunks.nchunks += (size / chunksize);
+		stats_chunks.curchunks += (size / chunksize);
+		if (stats_chunks.curchunks > stats_chunks.highchunks) {
+			stats_chunks.highchunks = stats_chunks.curchunks;
+			if (config_prof)
+				gdump = true;
+		} else if (config_prof)
+			gdump = false;
+		malloc_mutex_unlock(&chunks_mtx);
+		if (config_prof && opt_prof && opt_prof_gdump && gdump)
+			prof_gdump();
+	}
+
+	assert(CHUNK_ADDR2BASE(ret) == ret);
+	return (ret);
+}
+
+static void
+chunk_record(void *chunk, size_t size)
+{
+	extent_node_t *xnode, *node, *prev, key;
+
+	madvise(chunk, size, JEMALLOC_MADV_PURGE);
+
+	xnode = NULL;
+	malloc_mutex_lock(&chunks_mtx);
+	while (true) {
+		key.addr = (void *)((uintptr_t)chunk + size);
+		node = extent_tree_ad_nsearch(&chunks_ad, &key);
+		/* Try to coalesce forward. */
+		if (node != NULL && node->addr == key.addr) {
+			/*
+			 * Coalesce chunk with the following address range.
+			 * This does not change the position within chunks_ad,
+			 * so only remove/insert from/into chunks_szad.
+			 */
+			extent_tree_szad_remove(&chunks_szad, node);
+			node->addr = chunk;
+			node->size += size;
+			extent_tree_szad_insert(&chunks_szad, node);
+			break;
+		} else if (xnode == NULL) {
+			/*
+			 * It is possible that base_node_alloc() will cause a
+			 * new base chunk to be allocated, so take care not to
+			 * deadlock on chunks_mtx, and recover if another thread
+			 * deallocates an adjacent chunk while this one is busy
+			 * allocating xnode.
+			 */
+			malloc_mutex_unlock(&chunks_mtx);
+			xnode = base_node_alloc();
+			if (xnode == NULL)
+				return;
+			malloc_mutex_lock(&chunks_mtx);
+		} else {
+			/* Coalescing forward failed, so insert a new node. */
+			node = xnode;
+			xnode = NULL;
+			node->addr = chunk;
+			node->size = size;
+			extent_tree_ad_insert(&chunks_ad, node);
+			extent_tree_szad_insert(&chunks_szad, node);
+			break;
+		}
+	}
+	/* Discard xnode if it ended up unused due to a race. */
+	if (xnode != NULL)
+		base_node_dealloc(xnode);
+
+	/* Try to coalesce backward. */
+	prev = extent_tree_ad_prev(&chunks_ad, node);
+	if (prev != NULL && (void *)((uintptr_t)prev->addr + prev->size) ==
+	    chunk) {
+		/*
+		 * Coalesce chunk with the previous address range.  This does
+		 * not change the position within chunks_ad, so only
+		 * remove/insert node from/into chunks_szad.
+		 */
+		extent_tree_szad_remove(&chunks_szad, prev);
+		extent_tree_ad_remove(&chunks_ad, prev);
+
+		extent_tree_szad_remove(&chunks_szad, node);
+		node->addr = prev->addr;
+		node->size += prev->size;
+		extent_tree_szad_insert(&chunks_szad, node);
+
+		base_node_dealloc(prev);
+	}
+	malloc_mutex_unlock(&chunks_mtx);
+}
+
+void
+chunk_dealloc(void *chunk, size_t size, bool unmap)
+{
+
+	assert(chunk != NULL);
+	assert(CHUNK_ADDR2BASE(chunk) == chunk);
+	assert(size != 0);
+	assert((size & chunksize_mask) == 0);
+
+	if (config_ivsalloc)
+		rtree_set(chunks_rtree, (uintptr_t)chunk, NULL);
+	if (config_stats || config_prof) {
+		malloc_mutex_lock(&chunks_mtx);
+		stats_chunks.curchunks -= (size / chunksize);
+		malloc_mutex_unlock(&chunks_mtx);
+	}
+
+	if (unmap) {
+		if (chunk_dealloc_mmap(chunk, size) == false)
+			return;
+		chunk_record(chunk, size);
+	}
+}
+
+bool
+chunk_boot0(void)
+{
+
+	/* Set variables according to the value of opt_lg_chunk. */
+	chunksize = (ZU(1) << opt_lg_chunk);
+	assert(chunksize >= PAGE);
+	chunksize_mask = chunksize - 1;
+	chunk_npages = (chunksize >> LG_PAGE);
+
+	if (config_stats || config_prof) {
+		if (malloc_mutex_init(&chunks_mtx))
+			return (true);
+		memset(&stats_chunks, 0, sizeof(chunk_stats_t));
+	}
+	if (config_dss && chunk_dss_boot())
+		return (true);
+	extent_tree_szad_new(&chunks_szad);
+	extent_tree_ad_new(&chunks_ad);
+	if (config_ivsalloc) {
+		chunks_rtree = rtree_new((ZU(1) << (LG_SIZEOF_PTR+3)) -
+		    opt_lg_chunk);
+		if (chunks_rtree == NULL)
+			return (true);
+	}
+
+	return (false);
+}
+
+bool
+chunk_boot1(void)
+{
+
+	if (chunk_mmap_boot())
+		return (true);
+
+	return (false);
+}

contrib/jemalloc/src/chunk_dss.c

@@ -0,0 +1,159 @@
+#define	JEMALLOC_CHUNK_DSS_C_
+#include "jemalloc/internal/jemalloc_internal.h"
+/******************************************************************************/
+/* Data. */
+
+/*
+ * Protects sbrk() calls.  This avoids malloc races among threads, though it
+ * does not protect against races with threads that call sbrk() directly.
+ */
+static malloc_mutex_t	dss_mtx;
+
+/* Base address of the DSS. */
+static void		*dss_base;
+/* Current end of the DSS, or ((void *)-1) if the DSS is exhausted. */
+static void		*dss_prev;
+/* Current upper limit on DSS addresses. */
+static void		*dss_max;
+
+/******************************************************************************/
+
+#ifndef JEMALLOC_HAVE_SBRK
+static void *
+sbrk(intptr_t increment)
+{
+
+	not_implemented();
+
+	return (NULL);
+}
+#endif
+
+void *
+chunk_alloc_dss(size_t size, size_t alignment, bool *zero)
+{
+	void *ret;
+
+	cassert(config_dss);
+	assert(size > 0 && (size & chunksize_mask) == 0);
+	assert(alignment > 0 && (alignment & chunksize_mask) == 0);
+
+	/*
+	 * sbrk() uses a signed increment argument, so take care not to
+	 * interpret a huge allocation request as a negative increment.
+	 */
+	if ((intptr_t)size < 0)
+		return (NULL);
+
+	malloc_mutex_lock(&dss_mtx);
+	if (dss_prev != (void *)-1) {
+		size_t gap_size, cpad_size;
+		void *cpad, *dss_next;
+		intptr_t incr;
+
+		/*
+		 * The loop is necessary to recover from races with other
+		 * threads that are using the DSS for something other than
+		 * malloc.
+		 */
+		do {
+			/* Get the current end of the DSS. */
+			dss_max = sbrk(0);
+			/*
+			 * Calculate how much padding is necessary to
+			 * chunk-align the end of the DSS.
+			 */
+			gap_size = (chunksize - CHUNK_ADDR2OFFSET(dss_max)) &
+			    chunksize_mask;
+			/*
+			 * Compute how much chunk-aligned pad space (if any) is
+			 * necessary to satisfy alignment.  This space can be
+			 * recycled for later use.
+			 */
+			cpad = (void *)((uintptr_t)dss_max + gap_size);
+			ret = (void *)ALIGNMENT_CEILING((uintptr_t)dss_max,
+			    alignment);
+			cpad_size = (uintptr_t)ret - (uintptr_t)cpad;
+			dss_next = (void *)((uintptr_t)ret + size);
+			if ((uintptr_t)ret < (uintptr_t)dss_max ||
+			    (uintptr_t)dss_next < (uintptr_t)dss_max) {
+				/* Wrap-around. */
+				malloc_mutex_unlock(&dss_mtx);
+				return (NULL);
+			}
+			incr = gap_size + cpad_size + size;
+			dss_prev = sbrk(incr);
+			if (dss_prev == dss_max) {
+				/* Success. */
+				dss_max = dss_next;
+				malloc_mutex_unlock(&dss_mtx);
+				if (cpad_size != 0)
+					chunk_dealloc(cpad, cpad_size, true);
+				*zero = true;
+				return (ret);
+			}
+		} while (dss_prev != (void *)-1);
+	}
+	malloc_mutex_unlock(&dss_mtx);
+
+	return (NULL);
+}
+
+bool
+chunk_in_dss(void *chunk)
+{
+	bool ret;
+
+	cassert(config_dss);
+
+	malloc_mutex_lock(&dss_mtx);
+	if ((uintptr_t)chunk >= (uintptr_t)dss_base
+	    && (uintptr_t)chunk < (uintptr_t)dss_max)
+		ret = true;
+	else
+		ret = false;
+	malloc_mutex_unlock(&dss_mtx);
+
+	return (ret);
+}
+
+bool
+chunk_dss_boot(void)
+{
+
+	cassert(config_dss);
+
+	if (malloc_mutex_init(&dss_mtx))
+		return (true);
+	dss_base = sbrk(0);
+	dss_prev = dss_base;
+	dss_max = dss_base;
+
+	return (false);
+}
+
+void
+chunk_dss_prefork(void)
+{
+
+	if (config_dss)
+		malloc_mutex_prefork(&dss_mtx);
+}
+
+void
+chunk_dss_postfork_parent(void)
+{
+
+	if (config_dss)
+		malloc_mutex_postfork_parent(&dss_mtx);
+}
+
+void
+chunk_dss_postfork_child(void)
+{
+
+	if (config_dss)
+		malloc_mutex_postfork_child(&dss_mtx);
+}
+
+/******************************************************************************/

contrib/jemalloc/src/chunk_mmap.c

@@ -0,0 +1,207 @@
+#define	JEMALLOC_CHUNK_MMAP_C_
+#include "jemalloc/internal/jemalloc_internal.h"
+
+/******************************************************************************/
+/* Data. */
+
+/*
+ * Used by chunk_alloc_mmap() to decide whether to attempt the fast path and
+ * potentially avoid some system calls.
+ */
+malloc_tsd_data(static, mmap_unaligned, bool, false)
+malloc_tsd_funcs(JEMALLOC_INLINE, mmap_unaligned, bool, false,
+    malloc_tsd_no_cleanup)
+
+/******************************************************************************/
+/* Function prototypes for non-inline static functions. */
+
+static void	*pages_map(void *addr, size_t size);
+static void	pages_unmap(void *addr, size_t size);
+static void	*chunk_alloc_mmap_slow(size_t size, size_t alignment,
+    bool unaligned);
+
+/******************************************************************************/
+
+static void *
+pages_map(void *addr, size_t size)
+{
+	void *ret;
+
+	/*
+	 * We don't use MAP_FIXED here, because it can cause the *replacement*
+	 * of existing mappings, and we only want to create new mappings.
+	 */
+	ret = mmap(addr, size, PROT_READ | PROT_WRITE, MAP_PRIVATE | MAP_ANON,
+	    -1, 0);
+	assert(ret != NULL);
+
+	if (ret == MAP_FAILED)
+		ret = NULL;
+	else if (addr != NULL && ret != addr) {
+		/*
+		 * We succeeded in mapping memory, but not in the right place.
+		 */
+		if (munmap(ret, size) == -1) {
+			char buf[BUFERROR_BUF];
+
+			buferror(errno, buf, sizeof(buf));
+			malloc_printf("<jemalloc: Error in munmap(): %s\n",
+			    buf);
+			if (opt_abort)
+				abort();
+		}
+		ret = NULL;
+	}
+
+	assert(ret == NULL || (addr == NULL && ret != addr)
+	    || (addr != NULL && ret == addr));
+	return (ret);
+}
+
+static void
+pages_unmap(void *addr, size_t size)
+{
+
+	if (munmap(addr, size) == -1) {
+		char buf[BUFERROR_BUF];
+
+		buferror(errno, buf, sizeof(buf));
+		malloc_printf("<jemalloc>: Error in munmap(): %s\n", buf);
+		if (opt_abort)
+			abort();
+	}
+}
+
+static void *
+chunk_alloc_mmap_slow(size_t size, size_t alignment, bool unaligned)
+{
+	void *ret, *pages;
+	size_t alloc_size, leadsize, trailsize;
+
+	alloc_size = size + alignment - PAGE;
+	/* Beware size_t wrap-around. */
+	if (alloc_size < size)
+		return (NULL);
+	pages = pages_map(NULL, alloc_size);
+	if (pages == NULL)
+		return (NULL);
+	leadsize = ALIGNMENT_CEILING((uintptr_t)pages, alignment) -
+	    (uintptr_t)pages;
+	assert(alloc_size >= leadsize + size);
+	trailsize = alloc_size - leadsize - size;
+	ret = (void *)((uintptr_t)pages + leadsize);
+	if (leadsize != 0) {
+		/* Note that mmap() returned an unaligned mapping. */
+		unaligned = true;
+		pages_unmap(pages, leadsize);
+	}
+	if (trailsize != 0)
+		pages_unmap((void *)((uintptr_t)ret + size), trailsize);
+
+	/*
+	 * If mmap() returned an aligned mapping, reset mmap_unaligned so that
+	 * the next chunk_alloc_mmap() execution tries the fast allocation
+	 * method.
+	 */
+	if (unaligned == false && mmap_unaligned_booted) {
+		bool mu = false;
+		mmap_unaligned_tsd_set(&mu);
+	}
+
+	return (ret);
+}
+
+void *
+chunk_alloc_mmap(size_t size, size_t alignment)
+{
+	void *ret;
+
+	/*
+	 * Ideally, there would be a way to specify alignment to mmap() (like
+	 * NetBSD has), but in the absence of such a feature, we have to work
+	 * hard to efficiently create aligned mappings.  The reliable, but
+	 * slow method is to create a mapping that is over-sized, then trim the
+	 * excess.  However, that always results in at least one call to
+	 * pages_unmap().
+	 *
+	 * A more optimistic approach is to try mapping precisely the right
+	 * amount, then try to append another mapping if alignment is off.  In
+	 * practice, this works out well as long as the application is not
+	 * interleaving mappings via direct mmap() calls.  If we do run into a
+	 * situation where there is an interleaved mapping and we are unable to
+	 * extend an unaligned mapping, our best option is to switch to the
+	 * slow method until mmap() returns another aligned mapping.  This will
+	 * tend to leave a gap in the memory map that is too small to cause
+	 * later problems for the optimistic method.
+	 *
+	 * Another possible confounding factor is address space layout
+	 * randomization (ASLR), which causes mmap(2) to disregard the
+	 * requested address.  mmap_unaligned tracks whether the previous
+	 * chunk_alloc_mmap() execution received any unaligned or relocated
+	 * mappings, and if so, the current execution will immediately fall
+	 * back to the slow method.  However, we keep track of whether the fast
+	 * method would have succeeded, and if so, we make a note to try the
+	 * fast method next time.
+	 */
+
+	if (mmap_unaligned_booted && *mmap_unaligned_tsd_get() == false) {
+		size_t offset;
+
+		ret = pages_map(NULL, size);
+		if (ret == NULL)
+			return (NULL);
+
+		offset = ALIGNMENT_ADDR2OFFSET(ret, alignment);
+		if (offset != 0) {
+			bool mu = true;
+			mmap_unaligned_tsd_set(&mu);
+			/* Try to extend chunk boundary. */
+			if (pages_map((void *)((uintptr_t)ret + size),
+			    chunksize - offset) == NULL) {
+				/*
+				 * Extension failed.  Clean up, then revert to
+				 * the reliable-but-expensive method.
+				 */
+				pages_unmap(ret, size);
+				ret = chunk_alloc_mmap_slow(size, alignment,
+				    true);
+			} else {
+				/* Clean up unneeded leading space. */
+				pages_unmap(ret, chunksize - offset);
+				ret = (void *)((uintptr_t)ret + (chunksize -
+				    offset));
+			}
+		}
+	} else
+		ret = chunk_alloc_mmap_slow(size, alignment, false);
+
+	return (ret);
+}
+
+bool
+chunk_dealloc_mmap(void *chunk, size_t size)
+{
+
+	if (config_munmap)
+		pages_unmap(chunk, size);
+
+	return (config_munmap == false);
+}
+
+bool
+chunk_mmap_boot(void)
+{
+
+	/*
+	 * XXX For the non-TLS implementation of tsd, the first access from
+	 * each thread causes memory allocation.  The result is a bootstrapping
+	 * problem for this particular use case, so for now just disable it by
+	 * leaving it in an unbooted state.
+	 */
+#ifdef JEMALLOC_TLS
+	if (mmap_unaligned_tsd_boot())
+		return (true);
+#endif
+
+	return (false);
+}

contrib/jemalloc/src/ckh.c

@@ -0,0 +1,609 @@
+/*
+ *******************************************************************************
+ * Implementation of (2^1+,2) cuckoo hashing, where 2^1+ indicates that each
+ * hash bucket contains 2^n cells, for n >= 1, and 2 indicates that two hash
+ * functions are employed.  The original cuckoo hashing algorithm was described
+ * in:
+ *
+ *   Pagh, R., F.F. Rodler (2004) Cuckoo Hashing.  Journal of Algorithms
+ *     51(2):122-144.
+ *
+ * Generalization of cuckoo hashing was discussed in:
+ *
+ *   Erlingsson, U., M. Manasse, F. McSherry (2006) A cool and practical
+ *     alternative to traditional hash tables.  In Proceedings of the 7th
+ *     Workshop on Distributed Data and Structures (WDAS'06), Santa Clara, CA,
+ *     January 2006.
+ *
+ * This implementation uses precisely two hash functions because that is the
+ * fewest that can work, and supporting multiple hashes is an implementation
+ * burden.  Here is a reproduction of Figure 1 from Erlingsson et al. (2006)
+ * that shows approximate expected maximum load factors for various
+ * configurations:
+ *
+ *           |         #cells/bucket         |
+ *   #hashes |   1   |   2   |   4   |   8   |
+ *   --------+-------+-------+-------+-------+
+ *         1 | 0.006 | 0.006 | 0.03  | 0.12  |
+ *         2 | 0.49  | 0.86  |>0.93< |>0.96< |
+ *         3 | 0.91  | 0.97  | 0.98  | 0.999 |
+ *         4 | 0.97  | 0.99  | 0.999 |       |
+ *
+ * The number of cells per bucket is chosen such that a bucket fits in one cache
+ * line.  So, on 32- and 64-bit systems, we use (8,2) and (4,2) cuckoo hashing,
+ * respectively.
+ *
+ ******************************************************************************/
+#define	JEMALLOC_CKH_C_
+#include "jemalloc/internal/jemalloc_internal.h"
+
+/******************************************************************************/
+/* Function prototypes for non-inline static functions. */
+
+static bool	ckh_grow(ckh_t *ckh);
+static void	ckh_shrink(ckh_t *ckh);
+
+/******************************************************************************/
+
+/*
+ * Search bucket for key and return the cell number if found; SIZE_T_MAX
+ * otherwise.
+ */
+JEMALLOC_INLINE size_t
+ckh_bucket_search(ckh_t *ckh, size_t bucket, const void *key)
+{
+	ckhc_t *cell;
+	unsigned i;
+
+	for (i = 0; i < (ZU(1) << LG_CKH_BUCKET_CELLS); i++) {
+		cell = &ckh->tab[(bucket << LG_CKH_BUCKET_CELLS) + i];
+		if (cell->key != NULL && ckh->keycomp(key, cell->key))
+			return ((bucket << LG_CKH_BUCKET_CELLS) + i);
+	}
+
+	return (SIZE_T_MAX);
+}
+
+/*
+ * Search table for key and return cell number if found; SIZE_T_MAX otherwise.
+ */
+JEMALLOC_INLINE size_t
+ckh_isearch(ckh_t *ckh, const void *key)
+{
+	size_t hash1, hash2, bucket, cell;
+
+	assert(ckh != NULL);
+
+	ckh->hash(key, ckh->lg_curbuckets, &hash1, &hash2);
+
+	/* Search primary bucket. */
+	bucket = hash1 & ((ZU(1) << ckh->lg_curbuckets) - 1);
+	cell = ckh_bucket_search(ckh, bucket, key);
+	if (cell != SIZE_T_MAX)
+		return (cell);
+
+	/* Search secondary bucket. */
+	bucket = hash2 & ((ZU(1) << ckh->lg_curbuckets) - 1);
+	cell = ckh_bucket_search(ckh, bucket, key);
+	return (cell);
+}
+
+JEMALLOC_INLINE bool
+ckh_try_bucket_insert(ckh_t *ckh, size_t bucket, const void *key,
+    const void *data)
+{
+	ckhc_t *cell;
+	unsigned offset, i;
+
+	/*
+	 * Cycle through the cells in the bucket, starting at a random position.
+	 * The randomness avoids worst-case search overhead as buckets fill up.
+	 */
+	prng32(offset, LG_CKH_BUCKET_CELLS, ckh->prng_state, CKH_A, CKH_C);
+	for (i = 0; i < (ZU(1) << LG_CKH_BUCKET_CELLS); i++) {
+		cell = &ckh->tab[(bucket << LG_CKH_BUCKET_CELLS) +
+		    ((i + offset) & ((ZU(1) << LG_CKH_BUCKET_CELLS) - 1))];
+		if (cell->key == NULL) {
+			cell->key = key;
+			cell->data = data;
+			ckh->count++;
+			return (false);
+		}
+	}
+
+	return (true);
+}
+
+/*
+ * No space is available in bucket.  Randomly evict an item, then try to find an
+ * alternate location for that item.  Iteratively repeat this
+ * eviction/relocation procedure until either success or detection of an
+ * eviction/relocation bucket cycle.
+ */
+JEMALLOC_INLINE bool
+ckh_evict_reloc_insert(ckh_t *ckh, size_t argbucket, void const **argkey,
+    void const **argdata)
+{
+	const void *key, *data, *tkey, *tdata;
+	ckhc_t *cell;
+	size_t hash1, hash2, bucket, tbucket;
+	unsigned i;
+
+	bucket = argbucket;
+	key = *argkey;
+	data = *argdata;
+	while (true) {
+		/*
+		 * Choose a random item within the bucket to evict.  This is
+		 * critical to correct function, because without (eventually)
+		 * evicting all items within a bucket during iteration, it
+		 * would be possible to get stuck in an infinite loop if there
+		 * were an item for which both hashes indicated the same
+		 * bucket.
+		 */
+		prng32(i, LG_CKH_BUCKET_CELLS, ckh->prng_state, CKH_A, CKH_C);
+		cell = &ckh->tab[(bucket << LG_CKH_BUCKET_CELLS) + i];
+		assert(cell->key != NULL);
+
+		/* Swap cell->{key,data} and {key,data} (evict). */
+		tkey = cell->key; tdata = cell->data;
+		cell->key = key; cell->data = data;
+		key = tkey; data = tdata;
+
+#ifdef CKH_COUNT
+		ckh->nrelocs++;
+#endif
+
+		/* Find the alternate bucket for the evicted item. */
+		ckh->hash(key, ckh->lg_curbuckets, &hash1, &hash2);
+		tbucket = hash2 & ((ZU(1) << ckh->lg_curbuckets) - 1);
+		if (tbucket == bucket) {
+			tbucket = hash1 & ((ZU(1) << ckh->lg_curbuckets) - 1);
+			/*
+			 * It may be that (tbucket == bucket) still, if the
+			 * item's hashes both indicate this bucket.  However,
+			 * we are guaranteed to eventually escape this bucket
+			 * during iteration, assuming pseudo-random item
+			 * selection (true randomness would make infinite
+			 * looping a remote possibility).  The reason we can
+			 * never get trapped forever is that there are two
+			 * cases:
+			 *
+			 * 1) This bucket == argbucket, so we will quickly
+			 *    detect an eviction cycle and terminate.
+			 * 2) An item was evicted to this bucket from another,
+			 *    which means that at least one item in this bucket
+			 *    has hashes that indicate distinct buckets.
+			 */
+		}
+		/* Check for a cycle. */
+		if (tbucket == argbucket) {
+			*argkey = key;
+			*argdata = data;
+			return (true);
+		}
+
+		bucket = tbucket;
+		if (ckh_try_bucket_insert(ckh, bucket, key, data) == false)
+			return (false);
+	}
+}
+
+JEMALLOC_INLINE bool
+ckh_try_insert(ckh_t *ckh, void const**argkey, void const**argdata)
+{
+	size_t hash1, hash2, bucket;
+	const void *key = *argkey;
+	const void *data = *argdata;
+
+	ckh->hash(key, ckh->lg_curbuckets, &hash1, &hash2);
+
+	/* Try to insert in primary bucket. */
+	bucket = hash1 & ((ZU(1) << ckh->lg_curbuckets) - 1);
+	if (ckh_try_bucket_insert(ckh, bucket, key, data) == false)
+		return (false);
+
+	/* Try to insert in secondary bucket. */
+	bucket = hash2 & ((ZU(1) << ckh->lg_curbuckets) - 1);
+	if (ckh_try_bucket_insert(ckh, bucket, key, data) == false)
+		return (false);
+
+	/*
+	 * Try to find a place for this item via iterative eviction/relocation.
+	 */
+	return (ckh_evict_reloc_insert(ckh, bucket, argkey, argdata));
+}
+
+/*
+ * Try to rebuild the hash table from scratch by inserting all items from the
+ * old table into the new.
+ */
+JEMALLOC_INLINE bool
+ckh_rebuild(ckh_t *ckh, ckhc_t *aTab)
+{
+	size_t count, i, nins;
+	const void *key, *data;
+
+	count = ckh->count;
+	ckh->count = 0;
+	for (i = nins = 0; nins < count; i++) {
+		if (aTab[i].key != NULL) {
+			key = aTab[i].key;
+			data = aTab[i].data;
+			if (ckh_try_insert(ckh, &key, &data)) {
+				ckh->count = count;
+				return (true);
+			}
+			nins++;
+		}
+	}
+
+	return (false);
+}
+
+static bool
+ckh_grow(ckh_t *ckh)
+{
+	bool ret;
+	ckhc_t *tab, *ttab;
+	size_t lg_curcells;
+	unsigned lg_prevbuckets;
+
+#ifdef CKH_COUNT
+	ckh->ngrows++;
+#endif
+
+	/*
+	 * It is possible (though unlikely, given well behaved hashes) that the
+	 * table will have to be doubled more than once in order to create a
+	 * usable table.
+	 */
+	lg_prevbuckets = ckh->lg_curbuckets;
+	lg_curcells = ckh->lg_curbuckets + LG_CKH_BUCKET_CELLS;
+	while (true) {
+		size_t usize;
+
+		lg_curcells++;
+		usize = sa2u(sizeof(ckhc_t) << lg_curcells, CACHELINE);
+		if (usize == 0) {
+			ret = true;
+			goto label_return;
+		}
+		tab = (ckhc_t *)ipalloc(usize, CACHELINE, true);
+		if (tab == NULL) {
+			ret = true;
+			goto label_return;
+		}
+		/* Swap in new table. */
+		ttab = ckh->tab;
+		ckh->tab = tab;
+		tab = ttab;
+		ckh->lg_curbuckets = lg_curcells - LG_CKH_BUCKET_CELLS;
+
+		if (ckh_rebuild(ckh, tab) == false) {
+			idalloc(tab);
+			break;
+		}
+
+		/* Rebuilding failed, so back out partially rebuilt table. */
+		idalloc(ckh->tab);
+		ckh->tab = tab;
+		ckh->lg_curbuckets = lg_prevbuckets;
+	}
+
+	ret = false;
+label_return:
+	return (ret);
+}
+
+static void
+ckh_shrink(ckh_t *ckh)
+{
+	ckhc_t *tab, *ttab;
+	size_t lg_curcells, usize;
+	unsigned lg_prevbuckets;
+
+	/*
+	 * It is possible (though unlikely, given well behaved hashes) that the
+	 * table rebuild will fail.
+	 */
+	lg_prevbuckets = ckh->lg_curbuckets;
+	lg_curcells = ckh->lg_curbuckets + LG_CKH_BUCKET_CELLS - 1;
+	usize = sa2u(sizeof(ckhc_t) << lg_curcells, CACHELINE);
+	if (usize == 0)
+		return;
+	tab = (ckhc_t *)ipalloc(usize, CACHELINE, true);
+	if (tab == NULL) {
+		/*
+		 * An OOM error isn't worth propagating, since it doesn't
+		 * prevent this or future operations from proceeding.
+		 */
+		return;
+	}
+	/* Swap in new table. */
+	ttab = ckh->tab;
+	ckh->tab = tab;
+	tab = ttab;
+	ckh->lg_curbuckets = lg_curcells - LG_CKH_BUCKET_CELLS;
+
+	if (ckh_rebuild(ckh, tab) == false) {
+		idalloc(tab);
+#ifdef CKH_COUNT
+		ckh->nshrinks++;
+#endif
+		return;
+	}
+
+	/* Rebuilding failed, so back out partially rebuilt table. */
+	idalloc(ckh->tab);
+	ckh->tab = tab;
+	ckh->lg_curbuckets = lg_prevbuckets;
+#ifdef CKH_COUNT
+	ckh->nshrinkfails++;
+#endif
+}
+
+bool
+ckh_new(ckh_t *ckh, size_t minitems, ckh_hash_t *hash, ckh_keycomp_t *keycomp)
+{
+	bool ret;
+	size_t mincells, usize;
+	unsigned lg_mincells;
+
+	assert(minitems > 0);
+	assert(hash != NULL);
+	assert(keycomp != NULL);
+
+#ifdef CKH_COUNT
+	ckh->ngrows = 0;
+	ckh->nshrinks = 0;
+	ckh->nshrinkfails = 0;
+	ckh->ninserts = 0;
+	ckh->nrelocs = 0;
+#endif
+	ckh->prng_state = 42; /* Value doesn't really matter. */
+	ckh->count = 0;
+
+	/*
+	 * Find the minimum power of 2 that is large enough to fit aBaseCount
+	 * entries.  We are using (2+,2) cuckoo hashing, which has an expected
+	 * maximum load factor of at least ~0.86, so 0.75 is a conservative load
+	 * factor that will typically allow 2^aLgMinItems to fit without ever
+	 * growing the table.
+	 */
+	assert(LG_CKH_BUCKET_CELLS > 0);
+	mincells = ((minitems + (3 - (minitems % 3))) / 3) << 2;
+	for (lg_mincells = LG_CKH_BUCKET_CELLS;
+	    (ZU(1) << lg_mincells) < mincells;
+	    lg_mincells++)
+		; /* Do nothing. */
+	ckh->lg_minbuckets = lg_mincells - LG_CKH_BUCKET_CELLS;
+	ckh->lg_curbuckets = lg_mincells - LG_CKH_BUCKET_CELLS;
+	ckh->hash = hash;
+	ckh->keycomp = keycomp;
+
+	usize = sa2u(sizeof(ckhc_t) << lg_mincells, CACHELINE);
+	if (usize == 0) {
+		ret = true;
+		goto label_return;
+	}
+	ckh->tab = (ckhc_t *)ipalloc(usize, CACHELINE, true);
+	if (ckh->tab == NULL) {
+		ret = true;
+		goto label_return;
+	}
+
+	ret = false;
+label_return:
+	return (ret);
+}
+
+void
+ckh_delete(ckh_t *ckh)
+{
+
+	assert(ckh != NULL);
+
+#ifdef CKH_VERBOSE
+	malloc_printf(
+	    "%s(%p): ngrows: %"PRIu64", nshrinks: %"PRIu64","
+	    " nshrinkfails: %"PRIu64", ninserts: %"PRIu64","
+	    " nrelocs: %"PRIu64"\n", __func__, ckh,
+	    (unsigned long long)ckh->ngrows,
+	    (unsigned long long)ckh->nshrinks,
+	    (unsigned long long)ckh->nshrinkfails,
+	    (unsigned long long)ckh->ninserts,
+	    (unsigned long long)ckh->nrelocs);
+#endif
+
+	idalloc(ckh->tab);
+#ifdef JEMALLOC_DEBUG
+	memset(ckh, 0x5a, sizeof(ckh_t));
+#endif
+}
+
+size_t
+ckh_count(ckh_t *ckh)
+{
+
+	assert(ckh != NULL);
+
+	return (ckh->count);
+}
+
+bool
+ckh_iter(ckh_t *ckh, size_t *tabind, void **key, void **data)
+{
+	size_t i, ncells;
+
+	for (i = *tabind, ncells = (ZU(1) << (ckh->lg_curbuckets +
+	    LG_CKH_BUCKET_CELLS)); i < ncells; i++) {
+		if (ckh->tab[i].key != NULL) {
+			if (key != NULL)
+				*key = (void *)ckh->tab[i].key;
+			if (data != NULL)
+				*data = (void *)ckh->tab[i].data;
+			*tabind = i + 1;
+			return (false);
+		}
+	}
+
+	return (true);
+}
+
+bool
+ckh_insert(ckh_t *ckh, const void *key, const void *data)
+{
+	bool ret;
+
+	assert(ckh != NULL);
+	assert(ckh_search(ckh, key, NULL, NULL));
+
+#ifdef CKH_COUNT
+	ckh->ninserts++;
+#endif
+
+	while (ckh_try_insert(ckh, &key, &data)) {
+		if (ckh_grow(ckh)) {
+			ret = true;
+			goto label_return;
+		}
+	}
+
+	ret = false;
+label_return:
+	return (ret);
+}
+
+bool
+ckh_remove(ckh_t *ckh, const void *searchkey, void **key, void **data)
+{
+	size_t cell;
+
+	assert(ckh != NULL);
+
+	cell = ckh_isearch(ckh, searchkey);
+	if (cell != SIZE_T_MAX) {
+		if (key != NULL)
+			*key = (void *)ckh->tab[cell].key;
+		if (data != NULL)
+			*data = (void *)ckh->tab[cell].data;
+		ckh->tab[cell].key = NULL;
+		ckh->tab[cell].data = NULL; /* Not necessary. */
+
+		ckh->count--;
+		/* Try to halve the table if it is less than 1/4 full. */
+		if (ckh->count < (ZU(1) << (ckh->lg_curbuckets
+		    + LG_CKH_BUCKET_CELLS - 2)) && ckh->lg_curbuckets
+		    > ckh->lg_minbuckets) {
+			/* Ignore error due to OOM. */
+			ckh_shrink(ckh);
+		}
+
+		return (false);
+	}
+
+	return (true);
+}
+
+bool
+ckh_search(ckh_t *ckh, const void *searchkey, void **key, void **data)
+{
+	size_t cell;
+
+	assert(ckh != NULL);
+
+	cell = ckh_isearch(ckh, searchkey);
+	if (cell != SIZE_T_MAX) {
+		if (key != NULL)
+			*key = (void *)ckh->tab[cell].key;
+		if (data != NULL)
+			*data = (void *)ckh->tab[cell].data;
+		return (false);
+	}
+
+	return (true);
+}
+
+void
+ckh_string_hash(const void *key, unsigned minbits, size_t *hash1, size_t *hash2)
+{
+	size_t ret1, ret2;
+	uint64_t h;
+
+	assert(minbits <= 32 || (SIZEOF_PTR == 8 && minbits <= 64));
+	assert(hash1 != NULL);
+	assert(hash2 != NULL);
+
+	h = hash(key, strlen((const char *)key), UINT64_C(0x94122f335b332aea));
+	if (minbits <= 32) {
+		/*
+		 * Avoid doing multiple hashes, since a single hash provides
+		 * enough bits.
+		 */
+		ret1 = h & ZU(0xffffffffU);
+		ret2 = h >> 32;
+	} else {
+		ret1 = h;
+		ret2 = hash(key, strlen((const char *)key),
+		    UINT64_C(0x8432a476666bbc13));
+	}
+
+	*hash1 = ret1;
+	*hash2 = ret2;
+}
+
+bool
+ckh_string_keycomp(const void *k1, const void *k2)
+{
+
+    assert(k1 != NULL);
+    assert(k2 != NULL);
+
+    return (strcmp((char *)k1, (char *)k2) ? false : true);
+}
+
+void
+ckh_pointer_hash(const void *key, unsigned minbits, size_t *hash1,
+    size_t *hash2)
+{
+	size_t ret1, ret2;
+	uint64_t h;
+	union {
+		const void	*v;
+		uint64_t	i;
+	} u;
+
+	assert(minbits <= 32 || (SIZEOF_PTR == 8 && minbits <= 64));
+	assert(hash1 != NULL);
+	assert(hash2 != NULL);
+
+	assert(sizeof(u.v) == sizeof(u.i));
+#if (LG_SIZEOF_PTR != LG_SIZEOF_INT)
+	u.i = 0;
+#endif
+	u.v = key;
+	h = hash(&u.i, sizeof(u.i), UINT64_C(0xd983396e68886082));
+	if (minbits <= 32) {
+		/*
+		 * Avoid doing multiple hashes, since a single hash provides
+		 * enough bits.
+		 */
+		ret1 = h & ZU(0xffffffffU);
+		ret2 = h >> 32;
+	} else {
+		assert(SIZEOF_PTR == 8);
+		ret1 = h;
+		ret2 = hash(&u.i, sizeof(u.i), UINT64_C(0x5e2be9aff8709a5d));
+	}
+
+	*hash1 = ret1;
+	*hash2 = ret2;
+}
+
+bool
+ckh_pointer_keycomp(const void *k1, const void *k2)
+{
+
+	return ((k1 == k2) ? true : false);
+}

contrib/jemalloc/src/extent.c

@@ -0,0 +1,39 @@
+#define	JEMALLOC_EXTENT_C_
+#include "jemalloc/internal/jemalloc_internal.h"
+
+/******************************************************************************/
+
+static inline int
+extent_szad_comp(extent_node_t *a, extent_node_t *b)
+{
+	int ret;
+	size_t a_size = a->size;
+	size_t b_size = b->size;
+
+	ret = (a_size > b_size) - (a_size < b_size);
+	if (ret == 0) {
+		uintptr_t a_addr = (uintptr_t)a->addr;
+		uintptr_t b_addr = (uintptr_t)b->addr;
+
+		ret = (a_addr > b_addr) - (a_addr < b_addr);
+	}
+
+	return (ret);
+}
+
+/* Generate red-black tree functions. */
+rb_gen(, extent_tree_szad_, extent_tree_t, extent_node_t, link_szad,
+    extent_szad_comp)
+
+static inline int
+extent_ad_comp(extent_node_t *a, extent_node_t *b)
+{
+	uintptr_t a_addr = (uintptr_t)a->addr;
+	uintptr_t b_addr = (uintptr_t)b->addr;
+
+	return ((a_addr > b_addr) - (a_addr < b_addr));
+}
+
+/* Generate red-black tree functions. */
+rb_gen(, extent_tree_ad_, extent_tree_t, extent_node_t, link_ad,
+    extent_ad_comp)

contrib/jemalloc/src/hash.c

@@ -0,0 +1,2 @@
+#define	JEMALLOC_HASH_C_
+#include "jemalloc/internal/jemalloc_internal.h"

contrib/jemalloc/src/huge.c

@@ -0,0 +1,306 @@
+#define	JEMALLOC_HUGE_C_
+#include "jemalloc/internal/jemalloc_internal.h"
+
+/******************************************************************************/
+/* Data. */
+
+uint64_t	huge_nmalloc;
+uint64_t	huge_ndalloc;
+size_t		huge_allocated;
+
+malloc_mutex_t	huge_mtx;
+
+/******************************************************************************/
+
+/* Tree of chunks that are stand-alone huge allocations. */
+static extent_tree_t	huge;
+
+void *
+huge_malloc(size_t size, bool zero)
+{
+
+	return (huge_palloc(size, chunksize, zero));
+}
+
+void *
+huge_palloc(size_t size, size_t alignment, bool zero)
+{
+	void *ret;
+	size_t csize;
+	extent_node_t *node;
+
+	/* Allocate one or more contiguous chunks for this request. */
+
+	csize = CHUNK_CEILING(size);
+	if (csize == 0) {
+		/* size is large enough to cause size_t wrap-around. */
+		return (NULL);
+	}
+
+	/* Allocate an extent node with which to track the chunk. */
+	node = base_node_alloc();
+	if (node == NULL)
+		return (NULL);
+
+	ret = chunk_alloc(csize, alignment, false, &zero);
+	if (ret == NULL) {
+		base_node_dealloc(node);
+		return (NULL);
+	}
+
+	/* Insert node into huge. */
+	node->addr = ret;
+	node->size = csize;
+
+	malloc_mutex_lock(&huge_mtx);
+	extent_tree_ad_insert(&huge, node);
+	if (config_stats) {
+		stats_cactive_add(csize);
+		huge_nmalloc++;
+		huge_allocated += csize;
+	}
+	malloc_mutex_unlock(&huge_mtx);
+
+	if (config_fill && zero == false) {
+		if (opt_junk)
+			memset(ret, 0xa5, csize);
+		else if (opt_zero)
+			memset(ret, 0, csize);
+	}
+
+	return (ret);
+}
+
+void *
+huge_ralloc_no_move(void *ptr, size_t oldsize, size_t size, size_t extra)
+{
+
+	/*
+	 * Avoid moving the allocation if the size class can be left the same.
+	 */
+	if (oldsize > arena_maxclass
+	    && CHUNK_CEILING(oldsize) >= CHUNK_CEILING(size)
+	    && CHUNK_CEILING(oldsize) <= CHUNK_CEILING(size+extra)) {
+		assert(CHUNK_CEILING(oldsize) == oldsize);
+		if (config_fill && opt_junk && size < oldsize) {
+			memset((void *)((uintptr_t)ptr + size), 0x5a,
+			    oldsize - size);
+		}
+		return (ptr);
+	}
+
+	/* Reallocation would require a move. */
+	return (NULL);
+}
+
+void *
+huge_ralloc(void *ptr, size_t oldsize, size_t size, size_t extra,
+    size_t alignment, bool zero)
+{
+	void *ret;
+	size_t copysize;
+
+	/* Try to avoid moving the allocation. */
+	ret = huge_ralloc_no_move(ptr, oldsize, size, extra);
+	if (ret != NULL)
+		return (ret);
+
+	/*
+	 * size and oldsize are different enough that we need to use a
+	 * different size class.  In that case, fall back to allocating new
+	 * space and copying.
+	 */
+	if (alignment > chunksize)
+		ret = huge_palloc(size + extra, alignment, zero);
+	else
+		ret = huge_malloc(size + extra, zero);
+
+	if (ret == NULL) {
+		if (extra == 0)
+			return (NULL);
+		/* Try again, this time without extra. */
+		if (alignment > chunksize)
+			ret = huge_palloc(size, alignment, zero);
+		else
+			ret = huge_malloc(size, zero);
+
+		if (ret == NULL)
+			return (NULL);
+	}
+
+	/*
+	 * Copy at most size bytes (not size+extra), since the caller has no
+	 * expectation that the extra bytes will be reliably preserved.
+	 */
+	copysize = (size < oldsize) ? size : oldsize;
+
+	/*
+	 * Use mremap(2) if this is a huge-->huge reallocation, and neither the
+	 * source nor the destination are in dss.
+	 */
+#ifdef JEMALLOC_MREMAP_FIXED
+	if (oldsize >= chunksize && (config_dss == false || (chunk_in_dss(ptr)
+	    == false && chunk_in_dss(ret) == false))) {
+		size_t newsize = huge_salloc(ret);
+
+		/*
+		 * Remove ptr from the tree of huge allocations before
+		 * performing the remap operation, in order to avoid the
+		 * possibility of another thread acquiring that mapping before
+		 * this one removes it from the tree.
+		 */
+		huge_dalloc(ptr, false);
+		if (mremap(ptr, oldsize, newsize, MREMAP_MAYMOVE|MREMAP_FIXED,
+		    ret) == MAP_FAILED) {
+			/*
+			 * Assuming no chunk management bugs in the allocator,
+			 * the only documented way an error can occur here is
+			 * if the application changed the map type for a
+			 * portion of the old allocation.  This is firmly in
+			 * undefined behavior territory, so write a diagnostic
+			 * message, and optionally abort.
+			 */
+			char buf[BUFERROR_BUF];
+
+			buferror(errno, buf, sizeof(buf));
+			malloc_printf("<jemalloc>: Error in mremap(): %s\n",
+			    buf);
+			if (opt_abort)
+				abort();
+			memcpy(ret, ptr, copysize);
+			chunk_dealloc_mmap(ptr, oldsize);
+		}
+	} else
+#endif
+	{
+		memcpy(ret, ptr, copysize);
+		iqalloc(ptr);
+	}
+	return (ret);
+}
+
+void
+huge_dalloc(void *ptr, bool unmap)
+{
+	extent_node_t *node, key;
+
+	malloc_mutex_lock(&huge_mtx);
+
+	/* Extract from tree of huge allocations. */
+	key.addr = ptr;
+	node = extent_tree_ad_search(&huge, &key);
+	assert(node != NULL);
+	assert(node->addr == ptr);
+	extent_tree_ad_remove(&huge, node);
+
+	if (config_stats) {
+		stats_cactive_sub(node->size);
+		huge_ndalloc++;
+		huge_allocated -= node->size;
+	}
+
+	malloc_mutex_unlock(&huge_mtx);
+
+	if (unmap && config_fill && config_dss && opt_junk)
+		memset(node->addr, 0x5a, node->size);
+
+	chunk_dealloc(node->addr, node->size, unmap);
+
+	base_node_dealloc(node);
+}
+
+size_t
+huge_salloc(const void *ptr)
+{
+	size_t ret;
+	extent_node_t *node, key;
+
+	malloc_mutex_lock(&huge_mtx);
+
+	/* Extract from tree of huge allocations. */
+	key.addr = __DECONST(void *, ptr);
+	node = extent_tree_ad_search(&huge, &key);
+	assert(node != NULL);
+
+	ret = node->size;
+
+	malloc_mutex_unlock(&huge_mtx);
+
+	return (ret);
+}
+
+prof_ctx_t *
+huge_prof_ctx_get(const void *ptr)
+{
+	prof_ctx_t *ret;
+	extent_node_t *node, key;
+
+	malloc_mutex_lock(&huge_mtx);
+
+	/* Extract from tree of huge allocations. */
+	key.addr = __DECONST(void *, ptr);
+	node = extent_tree_ad_search(&huge, &key);
+	assert(node != NULL);
+
+	ret = node->prof_ctx;
+
+	malloc_mutex_unlock(&huge_mtx);
+
+	return (ret);
+}
+
+void
+huge_prof_ctx_set(const void *ptr, prof_ctx_t *ctx)
+{
+	extent_node_t *node, key;
+
+	malloc_mutex_lock(&huge_mtx);
+
+	/* Extract from tree of huge allocations. */
+	key.addr = __DECONST(void *, ptr);
+	node = extent_tree_ad_search(&huge, &key);
+	assert(node != NULL);
+
+	node->prof_ctx = ctx;
+
+	malloc_mutex_unlock(&huge_mtx);
+}
+
+bool
+huge_boot(void)
+{
+
+	/* Initialize chunks data. */
+	if (malloc_mutex_init(&huge_mtx))
+		return (true);
+	extent_tree_ad_new(&huge);
+
+	if (config_stats) {
+		huge_nmalloc = 0;
+		huge_ndalloc = 0;
+		huge_allocated = 0;
+	}
+
+	return (false);
+}
+
+void
+huge_prefork(void)
+{
+
+	malloc_mutex_prefork(&huge_mtx);
+}
+
+void
+huge_postfork_parent(void)
+{
+
+	malloc_mutex_postfork_parent(&huge_mtx);
+}
+
+void
+huge_postfork_child(void)
+{
+
+	malloc_mutex_postfork_child(&huge_mtx);
+}

contrib/jemalloc/src/mb.c

@@ -0,0 +1,2 @@
+#define	JEMALLOC_MB_C_
+#include "jemalloc/internal/jemalloc_internal.h"

contrib/jemalloc/src/mutex.c

@@ -0,0 +1,153 @@
+#define	JEMALLOC_MUTEX_C_
+#include "jemalloc/internal/jemalloc_internal.h"
+
+#ifdef JEMALLOC_LAZY_LOCK
+#include <dlfcn.h>
+#endif
+
+/******************************************************************************/
+/* Data. */
+
+#ifdef JEMALLOC_LAZY_LOCK
+bool isthreaded = false;
+#endif
+#ifdef JEMALLOC_MUTEX_INIT_CB
+static bool		postpone_init = true;
+static malloc_mutex_t	*postponed_mutexes = NULL;
+#endif
+
+#ifdef JEMALLOC_LAZY_LOCK
+static void	pthread_create_once(void);
+#endif
+
+/******************************************************************************/
+/*
+ * We intercept pthread_create() calls in order to toggle isthreaded if the
+ * process goes multi-threaded.
+ */
+
+#ifdef JEMALLOC_LAZY_LOCK
+static int (*pthread_create_fptr)(pthread_t *__restrict, const pthread_attr_t *,
+    void *(*)(void *), void *__restrict);
+
+static void
+pthread_create_once(void)
+{
+
+	pthread_create_fptr = dlsym(RTLD_NEXT, "pthread_create");
+	if (pthread_create_fptr == NULL) {
+		malloc_write("<jemalloc>: Error in dlsym(RTLD_NEXT, "
+		    "\"pthread_create\")\n");
+		abort();
+	}
+
+	isthreaded = true;
+}
+
+JEMALLOC_ATTR(visibility("default"))
+int
+pthread_create(pthread_t *__restrict thread,
+    const pthread_attr_t *__restrict attr, void *(*start_routine)(void *),
+    void *__restrict arg)
+{
+	static pthread_once_t once_control = PTHREAD_ONCE_INIT;
+
+	pthread_once(&once_control, pthread_create_once);
+
+	return (pthread_create_fptr(thread, attr, start_routine, arg));
+}
+#endif
+
+/******************************************************************************/
+
+#ifdef JEMALLOC_MUTEX_INIT_CB
+int	_pthread_mutex_init_calloc_cb(pthread_mutex_t *mutex,
+    void *(calloc_cb)(size_t, size_t));
+
+__weak_reference(_pthread_mutex_init_calloc_cb_stub,
+    _pthread_mutex_init_calloc_cb);
+
+int
+_pthread_mutex_init_calloc_cb_stub(pthread_mutex_t *mutex,
+    void *(calloc_cb)(size_t, size_t))
+{
+
+	return (0);
+}
+#endif
+
+bool
+malloc_mutex_init(malloc_mutex_t *mutex)
+{
+#ifdef JEMALLOC_OSSPIN
+	mutex->lock = 0;
+#elif (defined(JEMALLOC_MUTEX_INIT_CB))
+	if (postpone_init) {
+		mutex->postponed_next = postponed_mutexes;
+		postponed_mutexes = mutex;
+	} else {
+		if (_pthread_mutex_init_calloc_cb(&mutex->lock, base_calloc) !=
+		    0)
+			return (true);
+	}
+#else
+	pthread_mutexattr_t attr;
+
+	if (pthread_mutexattr_init(&attr) != 0)
+		return (true);
+	pthread_mutexattr_settype(&attr, MALLOC_MUTEX_TYPE);
+	if (pthread_mutex_init(&mutex->lock, &attr) != 0) {
+		pthread_mutexattr_destroy(&attr);
+		return (true);
+	}
+	pthread_mutexattr_destroy(&attr);
+
+#endif
+	return (false);
+}
+
+void
+malloc_mutex_prefork(malloc_mutex_t *mutex)
+{
+
+	malloc_mutex_lock(mutex);
+}
+
+void
+malloc_mutex_postfork_parent(malloc_mutex_t *mutex)
+{
+
+	malloc_mutex_unlock(mutex);
+}
+
+void
+malloc_mutex_postfork_child(malloc_mutex_t *mutex)
+{
+
+#ifdef JEMALLOC_MUTEX_INIT_CB
+	malloc_mutex_unlock(mutex);
+#else
+	if (malloc_mutex_init(mutex)) {
+		malloc_printf("<jemalloc>: Error re-initializing mutex in "
+		    "child\n");
+		if (opt_abort)
+			abort();
+	}
+#endif
+}
+
+bool
+mutex_boot(void)
+{
+
+#ifdef JEMALLOC_MUTEX_INIT_CB
+	postpone_init = false;
+	while (postponed_mutexes != NULL) {
+		if (_pthread_mutex_init_calloc_cb(&postponed_mutexes->lock,
+		    base_calloc) != 0)
+			return (true);
+		postponed_mutexes = postponed_mutexes->postponed_next;
+	}
+#endif
+	return (false);
+}

contrib/jemalloc/src/quarantine.c

@@ -0,0 +1,163 @@
+#include "jemalloc/internal/jemalloc_internal.h"
+
+/******************************************************************************/
+/* Data. */
+
+typedef struct quarantine_s quarantine_t;
+
+struct quarantine_s {
+	size_t	curbytes;
+	size_t	curobjs;
+	size_t	first;
+#define	LG_MAXOBJS_INIT 10
+	size_t	lg_maxobjs;
+	void	*objs[1]; /* Dynamically sized ring buffer. */
+};
+
+static void	quarantine_cleanup(void *arg);
+
+malloc_tsd_data(static, quarantine, quarantine_t *, NULL)
+malloc_tsd_funcs(JEMALLOC_INLINE, quarantine, quarantine_t *, NULL,
+    quarantine_cleanup)
+
+/******************************************************************************/
+/* Function prototypes for non-inline static functions. */
+
+static quarantine_t	*quarantine_init(size_t lg_maxobjs);
+static quarantine_t	*quarantine_grow(quarantine_t *quarantine);
+static void	quarantine_drain(quarantine_t *quarantine, size_t upper_bound);
+
+/******************************************************************************/
+
+static quarantine_t *
+quarantine_init(size_t lg_maxobjs)
+{
+	quarantine_t *quarantine;
+
+	quarantine = (quarantine_t *)imalloc(offsetof(quarantine_t, objs) +
+	    ((ZU(1) << lg_maxobjs) * sizeof(void *)));
+	if (quarantine == NULL)
+		return (NULL);
+	quarantine->curbytes = 0;
+	quarantine->curobjs = 0;
+	quarantine->first = 0;
+	quarantine->lg_maxobjs = lg_maxobjs;
+
+	quarantine_tsd_set(&quarantine);
+
+	return (quarantine);
+}
+
+static quarantine_t *
+quarantine_grow(quarantine_t *quarantine)
+{
+	quarantine_t *ret;
+
+	ret = quarantine_init(quarantine->lg_maxobjs + 1);
+	if (ret == NULL)
+		return (quarantine);
+
+	ret->curbytes = quarantine->curbytes;
+	if (quarantine->first + quarantine->curobjs < (ZU(1) <<
+	    quarantine->lg_maxobjs)) {
+		/* objs ring buffer data are contiguous. */
+		memcpy(ret->objs, &quarantine->objs[quarantine->first],
+		    quarantine->curobjs * sizeof(void *));
+		ret->curobjs = quarantine->curobjs;
+	} else {
+		/* objs ring buffer data wrap around. */
+		size_t ncopy = (ZU(1) << quarantine->lg_maxobjs) -
+		    quarantine->first;
+		memcpy(ret->objs, &quarantine->objs[quarantine->first], ncopy *
+		    sizeof(void *));
+		ret->curobjs = ncopy;
+		if (quarantine->curobjs != 0) {
+			memcpy(&ret->objs[ret->curobjs], quarantine->objs,
+			    quarantine->curobjs - ncopy);
+		}
+	}
+
+	return (ret);
+}
+
+static void
+quarantine_drain(quarantine_t *quarantine, size_t upper_bound)
+{
+
+	while (quarantine->curbytes > upper_bound && quarantine->curobjs > 0) {
+		void *ptr = quarantine->objs[quarantine->first];
+		size_t usize = isalloc(ptr, config_prof);
+		idalloc(ptr);
+		quarantine->curbytes -= usize;
+		quarantine->curobjs--;
+		quarantine->first = (quarantine->first + 1) & ((ZU(1) <<
+		    quarantine->lg_maxobjs) - 1);
+	}
+}
+
+void
+quarantine(void *ptr)
+{
+	quarantine_t *quarantine;
+	size_t usize = isalloc(ptr, config_prof);
+
+	assert(config_fill);
+	assert(opt_quarantine);
+
+	quarantine = *quarantine_tsd_get();
+	if (quarantine == NULL && (quarantine =
+	    quarantine_init(LG_MAXOBJS_INIT)) == NULL) {
+		idalloc(ptr);
+		return;
+	}
+	/*
+	 * Drain one or more objects if the quarantine size limit would be
+	 * exceeded by appending ptr.
+	 */
+	if (quarantine->curbytes + usize > opt_quarantine) {
+		size_t upper_bound = (opt_quarantine >= usize) ? opt_quarantine
+		    - usize : 0;
+		quarantine_drain(quarantine, upper_bound);
+	}
+	/* Grow the quarantine ring buffer if it's full. */
+	if (quarantine->curobjs == (ZU(1) << quarantine->lg_maxobjs))
+		quarantine = quarantine_grow(quarantine);
+	/* quarantine_grow() must free a slot if it fails to grow. */
+	assert(quarantine->curobjs < (ZU(1) << quarantine->lg_maxobjs));
+	/* Append ptr if its size doesn't exceed the quarantine size. */
+	if (quarantine->curbytes + usize <= opt_quarantine) {
+		size_t offset = (quarantine->first + quarantine->curobjs) &
+		    ((ZU(1) << quarantine->lg_maxobjs) - 1);
+		quarantine->objs[offset] = ptr;
+		quarantine->curbytes += usize;
+		quarantine->curobjs++;
+		if (opt_junk)
+			memset(ptr, 0x5a, usize);
+	} else {
+		assert(quarantine->curbytes == 0);
+		idalloc(ptr);
+	}
+}
+
+static void
+quarantine_cleanup(void *arg)
+{
+	quarantine_t *quarantine = *(quarantine_t **)arg;
+
+	if (quarantine != NULL) {
+		quarantine_drain(quarantine, 0);
+		idalloc(quarantine);
+	}
+}
+
+bool
+quarantine_boot(void)
+{
+
+	assert(config_fill);
+
+	if (quarantine_tsd_boot())
+		return (true);
+
+	return (false);
+}

contrib/jemalloc/src/rtree.c

@@ -0,0 +1,46 @@
+#define	JEMALLOC_RTREE_C_
+#include "jemalloc/internal/jemalloc_internal.h"
+
+rtree_t *
+rtree_new(unsigned bits)
+{
+	rtree_t *ret;
+	unsigned bits_per_level, height, i;
+
+	bits_per_level = ffs(pow2_ceil((RTREE_NODESIZE / sizeof(void *)))) - 1;
+	height = bits / bits_per_level;
+	if (height * bits_per_level != bits)
+		height++;
+	assert(height * bits_per_level >= bits);
+
+	ret = (rtree_t*)base_alloc(offsetof(rtree_t, level2bits) +
+	    (sizeof(unsigned) * height));
+	if (ret == NULL)
+		return (NULL);
+	memset(ret, 0, offsetof(rtree_t, level2bits) + (sizeof(unsigned) *
+	    height));
+
+	if (malloc_mutex_init(&ret->mutex)) {
+		/* Leak the rtree. */
+		return (NULL);
+	}
+	ret->height = height;
+	if (bits_per_level * height > bits)
+		ret->level2bits[0] = bits % bits_per_level;
+	else
+		ret->level2bits[0] = bits_per_level;
+	for (i = 1; i < height; i++)
+		ret->level2bits[i] = bits_per_level;
+
+	ret->root = (void**)base_alloc(sizeof(void *) << ret->level2bits[0]);
+	if (ret->root == NULL) {
+		/*
+		 * We leak the rtree here, since there's no generic base
+		 * deallocation.
+		 */
+		return (NULL);
+	}
+	memset(ret->root, 0, sizeof(void *) << ret->level2bits[0]);
+
+	return (ret);
+}

contrib/jemalloc/src/stats.c

@@ -0,0 +1,550 @@
+#define	JEMALLOC_STATS_C_
+#include "jemalloc/internal/jemalloc_internal.h"
+
+#define	CTL_GET(n, v, t) do {						\
+	size_t sz = sizeof(t);						\
+	xmallctl(n, v, &sz, NULL, 0);					\
+} while (0)
+
+#define	CTL_I_GET(n, v, t) do {						\
+	size_t mib[6];							\
+	size_t miblen = sizeof(mib) / sizeof(size_t);			\
+	size_t sz = sizeof(t);						\
+	xmallctlnametomib(n, mib, &miblen);				\
+	mib[2] = i;							\
+	xmallctlbymib(mib, miblen, v, &sz, NULL, 0);			\
+} while (0)
+
+#define	CTL_J_GET(n, v, t) do {						\
+	size_t mib[6];							\
+	size_t miblen = sizeof(mib) / sizeof(size_t);			\
+	size_t sz = sizeof(t);						\
+	xmallctlnametomib(n, mib, &miblen);				\
+	mib[2] = j;							\
+	xmallctlbymib(mib, miblen, v, &sz, NULL, 0);			\
+} while (0)
+
+#define	CTL_IJ_GET(n, v, t) do {					\
+	size_t mib[6];							\
+	size_t miblen = sizeof(mib) / sizeof(size_t);			\
+	size_t sz = sizeof(t);						\
+	xmallctlnametomib(n, mib, &miblen);				\
+	mib[2] = i;							\
+	mib[4] = j;							\
+	xmallctlbymib(mib, miblen, v, &sz, NULL, 0);			\
+} while (0)
+
+/******************************************************************************/
+/* Data. */
+
+bool	opt_stats_print = false;
+
+size_t	stats_cactive = 0;
+
+/******************************************************************************/
+/* Function prototypes for non-inline static functions. */
+
+static void	stats_arena_bins_print(void (*write_cb)(void *, const char *),
+    void *cbopaque, unsigned i);
+static void	stats_arena_lruns_print(void (*write_cb)(void *, const char *),
+    void *cbopaque, unsigned i);
+static void	stats_arena_print(void (*write_cb)(void *, const char *),
+    void *cbopaque, unsigned i, bool bins, bool large);
+
+/******************************************************************************/
+
+static void
+stats_arena_bins_print(void (*write_cb)(void *, const char *), void *cbopaque,
+    unsigned i)
+{
+	size_t page;
+	bool config_tcache;
+	unsigned nbins, j, gap_start;
+
+	CTL_GET("arenas.page", &page, size_t);
+
+	CTL_GET("config.tcache", &config_tcache, bool);
+	if (config_tcache) {
+		malloc_cprintf(write_cb, cbopaque,
+		    "bins:     bin  size regs pgs    allocated      nmalloc"
+		    "      ndalloc    nrequests       nfills     nflushes"
+		    "      newruns       reruns      curruns\n");
+	} else {
+		malloc_cprintf(write_cb, cbopaque,
+		    "bins:     bin  size regs pgs    allocated      nmalloc"
+		    "      ndalloc      newruns       reruns      curruns\n");
+	}
+	CTL_GET("arenas.nbins", &nbins, unsigned);
+	for (j = 0, gap_start = UINT_MAX; j < nbins; j++) {
+		uint64_t nruns;
+
+		CTL_IJ_GET("stats.arenas.0.bins.0.nruns", &nruns, uint64_t);
+		if (nruns == 0) {
+			if (gap_start == UINT_MAX)
+				gap_start = j;
+		} else {
+			size_t reg_size, run_size, allocated;
+			uint32_t nregs;
+			uint64_t nmalloc, ndalloc, nrequests, nfills, nflushes;
+			uint64_t reruns;
+			size_t curruns;
+
+			if (gap_start != UINT_MAX) {
+				if (j > gap_start + 1) {
+					/* Gap of more than one size class. */
+					malloc_cprintf(write_cb, cbopaque,
+					    "[%u..%u]\n", gap_start,
+					    j - 1);
+				} else {
+					/* Gap of one size class. */
+					malloc_cprintf(write_cb, cbopaque,
+					    "[%u]\n", gap_start);
+				}
+				gap_start = UINT_MAX;
+			}
+			CTL_J_GET("arenas.bin.0.size", &reg_size, size_t);
+			CTL_J_GET("arenas.bin.0.nregs", &nregs, uint32_t);
+			CTL_J_GET("arenas.bin.0.run_size", &run_size, size_t);
+			CTL_IJ_GET("stats.arenas.0.bins.0.allocated",
+			    &allocated, size_t);
+			CTL_IJ_GET("stats.arenas.0.bins.0.nmalloc",
+			    &nmalloc, uint64_t);
+			CTL_IJ_GET("stats.arenas.0.bins.0.ndalloc",
+			    &ndalloc, uint64_t);
+			if (config_tcache) {
+				CTL_IJ_GET("stats.arenas.0.bins.0.nrequests",
+				    &nrequests, uint64_t);
+				CTL_IJ_GET("stats.arenas.0.bins.0.nfills",
+				    &nfills, uint64_t);
+				CTL_IJ_GET("stats.arenas.0.bins.0.nflushes",
+				    &nflushes, uint64_t);
+			}
+			CTL_IJ_GET("stats.arenas.0.bins.0.nreruns", &reruns,
+			    uint64_t);
+			CTL_IJ_GET("stats.arenas.0.bins.0.curruns", &curruns,
+			    size_t);
+			if (config_tcache) {
+				malloc_cprintf(write_cb, cbopaque,
+				    "%13u %5zu %4u %3zu %12zu %12"PRIu64
+				    " %12"PRIu64" %12"PRIu64" %12"PRIu64
+				    " %12"PRIu64" %12"PRIu64" %12"PRIu64
+				    " %12zu\n",
+				    j, reg_size, nregs, run_size / page,
+				    allocated, nmalloc, ndalloc, nrequests,
+				    nfills, nflushes, nruns, reruns, curruns);
+			} else {
+				malloc_cprintf(write_cb, cbopaque,
+				    "%13u %5zu %4u %3zu %12zu %12"PRIu64
+				    " %12"PRIu64" %12"PRIu64" %12"PRIu64
+				    " %12zu\n",
+				    j, reg_size, nregs, run_size / page,
+				    allocated, nmalloc, ndalloc, nruns, reruns,
+				    curruns);
+			}
+		}
+	}
+	if (gap_start != UINT_MAX) {
+		if (j > gap_start + 1) {
+			/* Gap of more than one size class. */
+			malloc_cprintf(write_cb, cbopaque, "[%u..%u]\n",
+			    gap_start, j - 1);
+		} else {
+			/* Gap of one size class. */
+			malloc_cprintf(write_cb, cbopaque, "[%u]\n", gap_start);
+		}
+	}
+}
+
+static void
+stats_arena_lruns_print(void (*write_cb)(void *, const char *), void *cbopaque,
+    unsigned i)
+{
+	size_t page, nlruns, j;
+	ssize_t gap_start;
+
+	CTL_GET("arenas.page", &page, size_t);
+
+	malloc_cprintf(write_cb, cbopaque,
+	    "large:   size pages      nmalloc      ndalloc    nrequests"
+	    "      curruns\n");
+	CTL_GET("arenas.nlruns", &nlruns, size_t);
+	for (j = 0, gap_start = -1; j < nlruns; j++) {
+		uint64_t nmalloc, ndalloc, nrequests;
+		size_t run_size, curruns;
+
+		CTL_IJ_GET("stats.arenas.0.lruns.0.nmalloc", &nmalloc,
+		    uint64_t);
+		CTL_IJ_GET("stats.arenas.0.lruns.0.ndalloc", &ndalloc,
+		    uint64_t);
+		CTL_IJ_GET("stats.arenas.0.lruns.0.nrequests", &nrequests,
+		    uint64_t);
+		if (nrequests == 0) {
+			if (gap_start == -1)
+				gap_start = j;
+		} else {
+			CTL_J_GET("arenas.lrun.0.size", &run_size, size_t);
+			CTL_IJ_GET("stats.arenas.0.lruns.0.curruns", &curruns,
+			    size_t);
+			if (gap_start != -1) {
+				malloc_cprintf(write_cb, cbopaque, "[%zu]\n",
+				    j - gap_start);
+				gap_start = -1;
+			}
+			malloc_cprintf(write_cb, cbopaque,
+			    "%13zu %5zu %12"PRIu64" %12"PRIu64" %12"PRIu64
+			    " %12zu\n",
+			    run_size, run_size / page, nmalloc, ndalloc,
+			    nrequests, curruns);
+		}
+	}
+	if (gap_start != -1)
+		malloc_cprintf(write_cb, cbopaque, "[%zu]\n", j - gap_start);
+}
+
+static void
+stats_arena_print(void (*write_cb)(void *, const char *), void *cbopaque,
+    unsigned i, bool bins, bool large)
+{
+	unsigned nthreads;
+	size_t page, pactive, pdirty, mapped;
+	uint64_t npurge, nmadvise, purged;
+	size_t small_allocated;
+	uint64_t small_nmalloc, small_ndalloc, small_nrequests;
+	size_t large_allocated;
+	uint64_t large_nmalloc, large_ndalloc, large_nrequests;
+
+	CTL_GET("arenas.page", &page, size_t);
+
+	CTL_I_GET("stats.arenas.0.nthreads", &nthreads, unsigned);
+	malloc_cprintf(write_cb, cbopaque,
+	    "assigned threads: %u\n", nthreads);
+	CTL_I_GET("stats.arenas.0.pactive", &pactive, size_t);
+	CTL_I_GET("stats.arenas.0.pdirty", &pdirty, size_t);
+	CTL_I_GET("stats.arenas.0.npurge", &npurge, uint64_t);
+	CTL_I_GET("stats.arenas.0.nmadvise", &nmadvise, uint64_t);
+	CTL_I_GET("stats.arenas.0.purged", &purged, uint64_t);
+	malloc_cprintf(write_cb, cbopaque,
+	    "dirty pages: %zu:%zu active:dirty, %"PRIu64" sweep%s,"
+	    " %"PRIu64" madvise%s, %"PRIu64" purged\n",
+	    pactive, pdirty, npurge, npurge == 1 ? "" : "s",
+	    nmadvise, nmadvise == 1 ? "" : "s", purged);
+
+	malloc_cprintf(write_cb, cbopaque,
+	    "            allocated      nmalloc      ndalloc    nrequests\n");
+	CTL_I_GET("stats.arenas.0.small.allocated", &small_allocated, size_t);
+	CTL_I_GET("stats.arenas.0.small.nmalloc", &small_nmalloc, uint64_t);
+	CTL_I_GET("stats.arenas.0.small.ndalloc", &small_ndalloc, uint64_t);
+	CTL_I_GET("stats.arenas.0.small.nrequests", &small_nrequests, uint64_t);
+	malloc_cprintf(write_cb, cbopaque,
+	    "small:   %12zu %12"PRIu64" %12"PRIu64" %12"PRIu64"\n",
+	    small_allocated, small_nmalloc, small_ndalloc, small_nrequests);
+	CTL_I_GET("stats.arenas.0.large.allocated", &large_allocated, size_t);
+	CTL_I_GET("stats.arenas.0.large.nmalloc", &large_nmalloc, uint64_t);
+	CTL_I_GET("stats.arenas.0.large.ndalloc", &large_ndalloc, uint64_t);
+	CTL_I_GET("stats.arenas.0.large.nrequests", &large_nrequests, uint64_t);
+	malloc_cprintf(write_cb, cbopaque,
+	    "large:   %12zu %12"PRIu64" %12"PRIu64" %12"PRIu64"\n",
+	    large_allocated, large_nmalloc, large_ndalloc, large_nrequests);
+	malloc_cprintf(write_cb, cbopaque,
+	    "total:   %12zu %12"PRIu64" %12"PRIu64" %12"PRIu64"\n",
+	    small_allocated + large_allocated,
+	    small_nmalloc + large_nmalloc,
+	    small_ndalloc + large_ndalloc,
+	    small_nrequests + large_nrequests);
+	malloc_cprintf(write_cb, cbopaque, "active:  %12zu\n", pactive * page);
+	CTL_I_GET("stats.arenas.0.mapped", &mapped, size_t);
+	malloc_cprintf(write_cb, cbopaque, "mapped:  %12zu\n", mapped);
+
+	if (bins)
+		stats_arena_bins_print(write_cb, cbopaque, i);
+	if (large)
+		stats_arena_lruns_print(write_cb, cbopaque, i);
+}
+
+void
+stats_print(void (*write_cb)(void *, const char *), void *cbopaque,
+    const char *opts)
+{
+	int err;
+	uint64_t epoch;
+	size_t u64sz;
+	bool general = true;
+	bool merged = true;
+	bool unmerged = true;
+	bool bins = true;
+	bool large = true;
+
+	/*
+	 * Refresh stats, in case mallctl() was called by the application.
+	 *
+	 * Check for OOM here, since refreshing the ctl cache can trigger
+	 * allocation.  In practice, none of the subsequent mallctl()-related
+	 * calls in this function will cause OOM if this one succeeds.
+	 * */
+	epoch = 1;
+	u64sz = sizeof(uint64_t);
+	err = je_mallctl("epoch", &epoch, &u64sz, &epoch, sizeof(uint64_t));
+	if (err != 0) {
+		if (err == EAGAIN) {
+			malloc_write("<jemalloc>: Memory allocation failure in "
+			    "mallctl(\"epoch\", ...)\n");
+			return;
+		}
+		malloc_write("<jemalloc>: Failure in mallctl(\"epoch\", "
+		    "...)\n");
+		abort();
+	}
+
+	if (write_cb == NULL) {
+		/*
+		 * The caller did not provide an alternate write_cb callback
+		 * function, so use the default one.  malloc_write() is an
+		 * inline function, so use malloc_message() directly here.
+		 */
+		write_cb = je_malloc_message;
+		cbopaque = NULL;
+	}
+
+	if (opts != NULL) {
+		unsigned i;
+
+		for (i = 0; opts[i] != '\0'; i++) {
+			switch (opts[i]) {
+			case 'g':
+				general = false;
+				break;
+			case 'm':
+				merged = false;
+				break;
+			case 'a':
+				unmerged = false;
+				break;
+			case 'b':
+				bins = false;
+				break;
+			case 'l':
+				large = false;
+				break;
+			default:;
+			}
+		}
+	}
+
+	write_cb(cbopaque, "___ Begin jemalloc statistics ___\n");
+	if (general) {
+		int err;
+		const char *cpv;
+		bool bv;
+		unsigned uv;
+		ssize_t ssv;
+		size_t sv, bsz, ssz, sssz, cpsz;
+
+		bsz = sizeof(bool);
+		ssz = sizeof(size_t);
+		sssz = sizeof(ssize_t);
+		cpsz = sizeof(const char *);
+
+		CTL_GET("version", &cpv, const char *);
+		malloc_cprintf(write_cb, cbopaque, "Version: %s\n", cpv);
+		CTL_GET("config.debug", &bv, bool);
+		malloc_cprintf(write_cb, cbopaque, "Assertions %s\n",
+		    bv ? "enabled" : "disabled");
+
+#define OPT_WRITE_BOOL(n)						\
+		if ((err = je_mallctl("opt."#n, &bv, &bsz, NULL, 0))	\
+		    == 0) {						\
+			malloc_cprintf(write_cb, cbopaque,		\
+			    "  opt."#n": %s\n", bv ? "true" : "false");	\
+		}
+#define OPT_WRITE_SIZE_T(n)						\
+		if ((err = je_mallctl("opt."#n, &sv, &ssz, NULL, 0))	\
+		    == 0) {						\
+			malloc_cprintf(write_cb, cbopaque,		\
+			"  opt."#n": %zu\n", sv);			\
+		}
+#define OPT_WRITE_SSIZE_T(n)						\
+		if ((err = je_mallctl("opt."#n, &ssv, &sssz, NULL, 0))	\
+		    == 0) {						\
+			malloc_cprintf(write_cb, cbopaque,		\
+			    "  opt."#n": %zd\n", ssv);			\
+		}
+#define OPT_WRITE_CHAR_P(n)						\
+		if ((err = je_mallctl("opt."#n, &cpv, &cpsz, NULL, 0))	\
+		    == 0) {						\
+			malloc_cprintf(write_cb, cbopaque,		\
+			    "  opt."#n": \"%s\"\n", cpv);		\
+		}
+
+		write_cb(cbopaque, "Run-time option settings:\n");
+		OPT_WRITE_BOOL(abort)
+		OPT_WRITE_SIZE_T(lg_chunk)
+		OPT_WRITE_SIZE_T(narenas)
+		OPT_WRITE_SSIZE_T(lg_dirty_mult)
+		OPT_WRITE_BOOL(stats_print)
+		OPT_WRITE_BOOL(junk)
+		OPT_WRITE_SIZE_T(quarantine)
+		OPT_WRITE_BOOL(redzone)
+		OPT_WRITE_BOOL(zero)
+		OPT_WRITE_BOOL(utrace)
+		OPT_WRITE_BOOL(valgrind)
+		OPT_WRITE_BOOL(xmalloc)
+		OPT_WRITE_BOOL(tcache)
+		OPT_WRITE_SSIZE_T(lg_tcache_max)
+		OPT_WRITE_BOOL(prof)
+		OPT_WRITE_CHAR_P(prof_prefix)
+		OPT_WRITE_BOOL(prof_active)
+		OPT_WRITE_SSIZE_T(lg_prof_sample)
+		OPT_WRITE_BOOL(prof_accum)
+		OPT_WRITE_SSIZE_T(lg_prof_interval)
+		OPT_WRITE_BOOL(prof_gdump)
+		OPT_WRITE_BOOL(prof_leak)
+
+#undef OPT_WRITE_BOOL
+#undef OPT_WRITE_SIZE_T
+#undef OPT_WRITE_SSIZE_T
+#undef OPT_WRITE_CHAR_P
+
+		malloc_cprintf(write_cb, cbopaque, "CPUs: %u\n", ncpus);
+
+		CTL_GET("arenas.narenas", &uv, unsigned);
+		malloc_cprintf(write_cb, cbopaque, "Max arenas: %u\n", uv);
+
+		malloc_cprintf(write_cb, cbopaque, "Pointer size: %zu\n",
+		    sizeof(void *));
+
+		CTL_GET("arenas.quantum", &sv, size_t);
+		malloc_cprintf(write_cb, cbopaque, "Quantum size: %zu\n", sv);
+
+		CTL_GET("arenas.page", &sv, size_t);
+		malloc_cprintf(write_cb, cbopaque, "Page size: %zu\n", sv);
+
+		CTL_GET("opt.lg_dirty_mult", &ssv, ssize_t);
+		if (ssv >= 0) {
+			malloc_cprintf(write_cb, cbopaque,
+			    "Min active:dirty page ratio per arena: %u:1\n",
+			    (1U << ssv));
+		} else {
+			write_cb(cbopaque,
+			    "Min active:dirty page ratio per arena: N/A\n");
+		}
+		if ((err = je_mallctl("arenas.tcache_max", &sv, &ssz, NULL, 0))
+		    == 0) {
+			malloc_cprintf(write_cb, cbopaque,
+			    "Maximum thread-cached size class: %zu\n", sv);
+		}
+		if ((err = je_mallctl("opt.prof", &bv, &bsz, NULL, 0)) == 0 &&
+		    bv) {
+			CTL_GET("opt.lg_prof_sample", &sv, size_t);
+			malloc_cprintf(write_cb, cbopaque,
+			    "Average profile sample interval: %"PRIu64
+			    " (2^%zu)\n", (((uint64_t)1U) << sv), sv);
+
+			CTL_GET("opt.lg_prof_interval", &ssv, ssize_t);
+			if (ssv >= 0) {
+				malloc_cprintf(write_cb, cbopaque,
+				    "Average profile dump interval: %"PRIu64
+				    " (2^%zd)\n",
+				    (((uint64_t)1U) << ssv), ssv);
+			} else {
+				write_cb(cbopaque,
+				    "Average profile dump interval: N/A\n");
+			}
+		}
+		CTL_GET("opt.lg_chunk", &sv, size_t);
+		malloc_cprintf(write_cb, cbopaque, "Chunk size: %zu (2^%zu)\n",
+		    (ZU(1) << sv), sv);
+	}
+
+	if (config_stats) {
+		size_t *cactive;
+		size_t allocated, active, mapped;
+		size_t chunks_current, chunks_high;
+		uint64_t chunks_total;
+		size_t huge_allocated;
+		uint64_t huge_nmalloc, huge_ndalloc;
+
+		CTL_GET("stats.cactive", &cactive, size_t *);
+		CTL_GET("stats.allocated", &allocated, size_t);
+		CTL_GET("stats.active", &active, size_t);
+		CTL_GET("stats.mapped", &mapped, size_t);
+		malloc_cprintf(write_cb, cbopaque,
+		    "Allocated: %zu, active: %zu, mapped: %zu\n",
+		    allocated, active, mapped);
+		malloc_cprintf(write_cb, cbopaque,
+		    "Current active ceiling: %zu\n", atomic_read_z(cactive));
+
+		/* Print chunk stats. */
+		CTL_GET("stats.chunks.total", &chunks_total, uint64_t);
+		CTL_GET("stats.chunks.high", &chunks_high, size_t);
+		CTL_GET("stats.chunks.current", &chunks_current, size_t);
+		malloc_cprintf(write_cb, cbopaque, "chunks: nchunks   "
+		    "highchunks    curchunks\n");
+		malloc_cprintf(write_cb, cbopaque, "  %13"PRIu64"%13zu%13zu\n",
+		    chunks_total, chunks_high, chunks_current);
+
+		/* Print huge stats. */
+		CTL_GET("stats.huge.nmalloc", &huge_nmalloc, uint64_t);
+		CTL_GET("stats.huge.ndalloc", &huge_ndalloc, uint64_t);
+		CTL_GET("stats.huge.allocated", &huge_allocated, size_t);
+		malloc_cprintf(write_cb, cbopaque,
+		    "huge: nmalloc      ndalloc    allocated\n");
+		malloc_cprintf(write_cb, cbopaque,
+		    " %12"PRIu64" %12"PRIu64" %12zu\n",
+		    huge_nmalloc, huge_ndalloc, huge_allocated);
+
+		if (merged) {
+			unsigned narenas;
+
+			CTL_GET("arenas.narenas", &narenas, unsigned);
+			{
+				bool initialized[narenas];
+				size_t isz;
+				unsigned i, ninitialized;
+
+				isz = sizeof(initialized);
+				xmallctl("arenas.initialized", initialized,
+				    &isz, NULL, 0);
+				for (i = ninitialized = 0; i < narenas; i++) {
+					if (initialized[i])
+						ninitialized++;
+				}
+
+				if (ninitialized > 1 || unmerged == false) {
+					/* Print merged arena stats. */
+					malloc_cprintf(write_cb, cbopaque,
+					    "\nMerged arenas stats:\n");
+					stats_arena_print(write_cb, cbopaque,
+					    narenas, bins, large);
+				}
+			}
+		}
+
+		if (unmerged) {
+			unsigned narenas;
+
+			/* Print stats for each arena. */
+
+			CTL_GET("arenas.narenas", &narenas, unsigned);
+			{
+				bool initialized[narenas];
+				size_t isz;
+				unsigned i;
+
+				isz = sizeof(initialized);
+				xmallctl("arenas.initialized", initialized,
+				    &isz, NULL, 0);
+
+				for (i = 0; i < narenas; i++) {
+					if (initialized[i]) {
+						malloc_cprintf(write_cb,
+						    cbopaque,
+						    "\narenas[%u]:\n", i);
+						stats_arena_print(write_cb,
+						    cbopaque, i, bins, large);
+					}
+				}
+			}
+		}
+	}
+	write_cb(cbopaque, "--- End jemalloc statistics ---\n");
+}

contrib/jemalloc/src/tcache.c

@@ -0,0 +1,435 @@
+#define	JEMALLOC_TCACHE_C_
+#include "jemalloc/internal/jemalloc_internal.h"
+
+/******************************************************************************/
+/* Data. */
+
+malloc_tsd_data(, tcache, tcache_t *, NULL)
+malloc_tsd_data(, tcache_enabled, tcache_enabled_t, tcache_enabled_default)
+
+bool	opt_tcache = true;
+ssize_t	opt_lg_tcache_max = LG_TCACHE_MAXCLASS_DEFAULT;
+
+tcache_bin_info_t	*tcache_bin_info;
+static unsigned		stack_nelms; /* Total stack elms per tcache. */
+
+size_t			nhbins;
+size_t			tcache_maxclass;
+
+/******************************************************************************/
+
+void *
+tcache_alloc_small_hard(tcache_t *tcache, tcache_bin_t *tbin, size_t binind)
+{
+	void *ret;
+
+	arena_tcache_fill_small(tcache->arena, tbin, binind,
+	    config_prof ? tcache->prof_accumbytes : 0);
+	if (config_prof)
+		tcache->prof_accumbytes = 0;
+	ret = tcache_alloc_easy(tbin);
+
+	return (ret);
+}
+
+void
+tcache_bin_flush_small(tcache_bin_t *tbin, size_t binind, unsigned rem,
+    tcache_t *tcache)
+{
+	void *ptr;
+	unsigned i, nflush, ndeferred;
+	bool merged_stats = false;
+
+	assert(binind < NBINS);
+	assert(rem <= tbin->ncached);
+
+	for (nflush = tbin->ncached - rem; nflush > 0; nflush = ndeferred) {
+		/* Lock the arena bin associated with the first object. */
+		arena_chunk_t *chunk = (arena_chunk_t *)CHUNK_ADDR2BASE(
+		    tbin->avail[0]);
+		arena_t *arena = chunk->arena;
+		arena_bin_t *bin = &arena->bins[binind];
+
+		if (config_prof && arena == tcache->arena) {
+			malloc_mutex_lock(&arena->lock);
+			arena_prof_accum(arena, tcache->prof_accumbytes);
+			malloc_mutex_unlock(&arena->lock);
+			tcache->prof_accumbytes = 0;
+		}
+
+		malloc_mutex_lock(&bin->lock);
+		if (config_stats && arena == tcache->arena) {
+			assert(merged_stats == false);
+			merged_stats = true;
+			bin->stats.nflushes++;
+			bin->stats.nrequests += tbin->tstats.nrequests;
+			tbin->tstats.nrequests = 0;
+		}
+		ndeferred = 0;
+		for (i = 0; i < nflush; i++) {
+			ptr = tbin->avail[i];
+			assert(ptr != NULL);
+			chunk = (arena_chunk_t *)CHUNK_ADDR2BASE(ptr);
+			if (chunk->arena == arena) {
+				size_t pageind = ((uintptr_t)ptr -
+				    (uintptr_t)chunk) >> LG_PAGE;
+				arena_chunk_map_t *mapelm =
+				    &chunk->map[pageind-map_bias];
+				if (config_fill && opt_junk) {
+					arena_alloc_junk_small(ptr,
+					    &arena_bin_info[binind], true);
+				}
+				arena_dalloc_bin(arena, chunk, ptr, mapelm);
+			} else {
+				/*
+				 * This object was allocated via a different
+				 * arena bin than the one that is currently
+				 * locked.  Stash the object, so that it can be
+				 * handled in a future pass.
+				 */
+				tbin->avail[ndeferred] = ptr;
+				ndeferred++;
+			}
+		}
+		malloc_mutex_unlock(&bin->lock);
+	}
+	if (config_stats && merged_stats == false) {
+		/*
+		 * The flush loop didn't happen to flush to this thread's
+		 * arena, so the stats didn't get merged.  Manually do so now.
+		 */
+		arena_bin_t *bin = &tcache->arena->bins[binind];
+		malloc_mutex_lock(&bin->lock);
+		bin->stats.nflushes++;
+		bin->stats.nrequests += tbin->tstats.nrequests;
+		tbin->tstats.nrequests = 0;
+		malloc_mutex_unlock(&bin->lock);
+	}
+
+	memmove(tbin->avail, &tbin->avail[tbin->ncached - rem],
+	    rem * sizeof(void *));
+	tbin->ncached = rem;
+	if ((int)tbin->ncached < tbin->low_water)
+		tbin->low_water = tbin->ncached;
+}
+
+void
+tcache_bin_flush_large(tcache_bin_t *tbin, size_t binind, unsigned rem,
+    tcache_t *tcache)
+{
+	void *ptr;
+	unsigned i, nflush, ndeferred;
+	bool merged_stats = false;
+
+	assert(binind < nhbins);
+	assert(rem <= tbin->ncached);
+
+	for (nflush = tbin->ncached - rem; nflush > 0; nflush = ndeferred) {
+		/* Lock the arena associated with the first object. */
+		arena_chunk_t *chunk = (arena_chunk_t *)CHUNK_ADDR2BASE(
+		    tbin->avail[0]);
+		arena_t *arena = chunk->arena;
+
+		malloc_mutex_lock(&arena->lock);
+		if ((config_prof || config_stats) && arena == tcache->arena) {
+			if (config_prof) {
+				arena_prof_accum(arena,
+				    tcache->prof_accumbytes);
+				tcache->prof_accumbytes = 0;
+			}
+			if (config_stats) {
+				merged_stats = true;
+				arena->stats.nrequests_large +=
+				    tbin->tstats.nrequests;
+				arena->stats.lstats[binind - NBINS].nrequests +=
+				    tbin->tstats.nrequests;
+				tbin->tstats.nrequests = 0;
+			}
+		}
+		ndeferred = 0;
+		for (i = 0; i < nflush; i++) {
+			ptr = tbin->avail[i];
+			assert(ptr != NULL);
+			chunk = (arena_chunk_t *)CHUNK_ADDR2BASE(ptr);
+			if (chunk->arena == arena)
+				arena_dalloc_large(arena, chunk, ptr);
+			else {
+				/*
+				 * This object was allocated via a different
+				 * arena than the one that is currently locked.
+				 * Stash the object, so that it can be handled
+				 * in a future pass.
+				 */
+				tbin->avail[ndeferred] = ptr;
+				ndeferred++;
+			}
+		}
+		malloc_mutex_unlock(&arena->lock);
+	}
+	if (config_stats && merged_stats == false) {
+		/*
+		 * The flush loop didn't happen to flush to this thread's
+		 * arena, so the stats didn't get merged.  Manually do so now.
+		 */
+		arena_t *arena = tcache->arena;
+		malloc_mutex_lock(&arena->lock);
+		arena->stats.nrequests_large += tbin->tstats.nrequests;
+		arena->stats.lstats[binind - NBINS].nrequests +=
+		    tbin->tstats.nrequests;
+		tbin->tstats.nrequests = 0;
+		malloc_mutex_unlock(&arena->lock);
+	}
+
+	memmove(tbin->avail, &tbin->avail[tbin->ncached - rem],
+	    rem * sizeof(void *));
+	tbin->ncached = rem;
+	if ((int)tbin->ncached < tbin->low_water)
+		tbin->low_water = tbin->ncached;
+}
+
+void
+tcache_arena_associate(tcache_t *tcache, arena_t *arena)
+{
+
+	if (config_stats) {
+		/* Link into list of extant tcaches. */
+		malloc_mutex_lock(&arena->lock);
+		ql_elm_new(tcache, link);
+		ql_tail_insert(&arena->tcache_ql, tcache, link);
+		malloc_mutex_unlock(&arena->lock);
+	}
+	tcache->arena = arena;
+}
+
+void
+tcache_arena_dissociate(tcache_t *tcache)
+{
+
+	if (config_stats) {
+		/* Unlink from list of extant tcaches. */
+		malloc_mutex_lock(&tcache->arena->lock);
+		ql_remove(&tcache->arena->tcache_ql, tcache, link);
+		malloc_mutex_unlock(&tcache->arena->lock);
+		tcache_stats_merge(tcache, tcache->arena);
+	}
+}
+
+tcache_t *
+tcache_create(arena_t *arena)
+{
+	tcache_t *tcache;
+	size_t size, stack_offset;
+	unsigned i;
+
+	size = offsetof(tcache_t, tbins) + (sizeof(tcache_bin_t) * nhbins);
+	/* Naturally align the pointer stacks. */
+	size = PTR_CEILING(size);
+	stack_offset = size;
+	size += stack_nelms * sizeof(void *);
+	/*
+	 * Round up to the nearest multiple of the cacheline size, in order to
+	 * avoid the possibility of false cacheline sharing.
+	 *
+	 * That this works relies on the same logic as in ipalloc(), but we
+	 * cannot directly call ipalloc() here due to tcache bootstrapping
+	 * issues.
+	 */
+	size = (size + CACHELINE_MASK) & (-CACHELINE);
+
+	if (size <= SMALL_MAXCLASS)
+		tcache = (tcache_t *)arena_malloc_small(arena, size, true);
+	else if (size <= tcache_maxclass)
+		tcache = (tcache_t *)arena_malloc_large(arena, size, true);
+	else
+		tcache = (tcache_t *)icalloc(size);
+
+	if (tcache == NULL)
+		return (NULL);
+
+	tcache_arena_associate(tcache, arena);
+
+	assert((TCACHE_NSLOTS_SMALL_MAX & 1U) == 0);
+	for (i = 0; i < nhbins; i++) {
+		tcache->tbins[i].lg_fill_div = 1;
+		tcache->tbins[i].avail = (void **)((uintptr_t)tcache +
+		    (uintptr_t)stack_offset);
+		stack_offset += tcache_bin_info[i].ncached_max * sizeof(void *);
+	}
+
+	tcache_tsd_set(&tcache);
+
+	return (tcache);
+}
+
+void
+tcache_destroy(tcache_t *tcache)
+{
+	unsigned i;
+	size_t tcache_size;
+
+	tcache_arena_dissociate(tcache);
+
+	for (i = 0; i < NBINS; i++) {
+		tcache_bin_t *tbin = &tcache->tbins[i];
+		tcache_bin_flush_small(tbin, i, 0, tcache);
+
+		if (config_stats && tbin->tstats.nrequests != 0) {
+			arena_t *arena = tcache->arena;
+			arena_bin_t *bin = &arena->bins[i];
+			malloc_mutex_lock(&bin->lock);
+			bin->stats.nrequests += tbin->tstats.nrequests;
+			malloc_mutex_unlock(&bin->lock);
+		}
+	}
+
+	for (; i < nhbins; i++) {
+		tcache_bin_t *tbin = &tcache->tbins[i];
+		tcache_bin_flush_large(tbin, i, 0, tcache);
+
+		if (config_stats && tbin->tstats.nrequests != 0) {
+			arena_t *arena = tcache->arena;
+			malloc_mutex_lock(&arena->lock);
+			arena->stats.nrequests_large += tbin->tstats.nrequests;
+			arena->stats.lstats[i - NBINS].nrequests +=
+			    tbin->tstats.nrequests;
+			malloc_mutex_unlock(&arena->lock);
+		}
+	}
+
+	if (config_prof && tcache->prof_accumbytes > 0) {
+		malloc_mutex_lock(&tcache->arena->lock);
+		arena_prof_accum(tcache->arena, tcache->prof_accumbytes);
+		malloc_mutex_unlock(&tcache->arena->lock);
+	}
+
+	tcache_size = arena_salloc(tcache, false);
+	if (tcache_size <= SMALL_MAXCLASS) {
+		arena_chunk_t *chunk = CHUNK_ADDR2BASE(tcache);
+		arena_t *arena = chunk->arena;
+		size_t pageind = ((uintptr_t)tcache - (uintptr_t)chunk) >>
+		    LG_PAGE;
+		arena_chunk_map_t *mapelm = &chunk->map[pageind-map_bias];
+		arena_run_t *run = (arena_run_t *)((uintptr_t)chunk +
+		    (uintptr_t)((pageind - (mapelm->bits >> LG_PAGE)) <<
+		    LG_PAGE));
+		arena_bin_t *bin = run->bin;
+
+		malloc_mutex_lock(&bin->lock);
+		arena_dalloc_bin(arena, chunk, tcache, mapelm);
+		malloc_mutex_unlock(&bin->lock);
+	} else if (tcache_size <= tcache_maxclass) {
+		arena_chunk_t *chunk = CHUNK_ADDR2BASE(tcache);
+		arena_t *arena = chunk->arena;
+
+		malloc_mutex_lock(&arena->lock);
+		arena_dalloc_large(arena, chunk, tcache);
+		malloc_mutex_unlock(&arena->lock);
+	} else
+		idalloc(tcache);
+}
+
+void
+tcache_thread_cleanup(void *arg)
+{
+	tcache_t *tcache = *(tcache_t **)arg;
+
+	if (tcache == TCACHE_STATE_DISABLED) {
+		/* Do nothing. */
+	} else if (tcache == TCACHE_STATE_REINCARNATED) {
+		/*
+		 * Another destructor called an allocator function after this
+		 * destructor was called.  Reset tcache to
+		 * TCACHE_STATE_PURGATORY in order to receive another callback.
+		 */
+		tcache = TCACHE_STATE_PURGATORY;
+		tcache_tsd_set(&tcache);
+	} else if (tcache == TCACHE_STATE_PURGATORY) {
+		/*
+		 * The previous time this destructor was called, we set the key
+		 * to TCACHE_STATE_PURGATORY so that other destructors wouldn't
+		 * cause re-creation of the tcache.  This time, do nothing, so
+		 * that the destructor will not be called again.
+		 */
+	} else if (tcache != NULL) {
+		assert(tcache != TCACHE_STATE_PURGATORY);
+		tcache_destroy(tcache);
+		tcache = TCACHE_STATE_PURGATORY;
+		tcache_tsd_set(&tcache);
+	}
+}
+
+void
+tcache_stats_merge(tcache_t *tcache, arena_t *arena)
+{
+	unsigned i;
+
+	/* Merge and reset tcache stats. */
+	for (i = 0; i < NBINS; i++) {
+		arena_bin_t *bin = &arena->bins[i];
+		tcache_bin_t *tbin = &tcache->tbins[i];
+		malloc_mutex_lock(&bin->lock);
+		bin->stats.nrequests += tbin->tstats.nrequests;
+		malloc_mutex_unlock(&bin->lock);
+		tbin->tstats.nrequests = 0;
+	}
+
+	for (; i < nhbins; i++) {
+		malloc_large_stats_t *lstats = &arena->stats.lstats[i - NBINS];
+		tcache_bin_t *tbin = &tcache->tbins[i];
+		arena->stats.nrequests_large += tbin->tstats.nrequests;
+		lstats->nrequests += tbin->tstats.nrequests;
+		tbin->tstats.nrequests = 0;
+	}
+}
+
+bool
+tcache_boot0(void)
+{
+	unsigned i;
+
+	/*
+	 * If necessary, clamp opt_lg_tcache_max, now that arena_maxclass is
+	 * known.
+	 */
+	if (opt_lg_tcache_max < 0 || (1U << opt_lg_tcache_max) < SMALL_MAXCLASS)
+		tcache_maxclass = SMALL_MAXCLASS;
+	else if ((1U << opt_lg_tcache_max) > arena_maxclass)
+		tcache_maxclass = arena_maxclass;
+	else
+		tcache_maxclass = (1U << opt_lg_tcache_max);
+
+	nhbins = NBINS + (tcache_maxclass >> LG_PAGE);
+
+	/* Initialize tcache_bin_info. */
+	tcache_bin_info = (tcache_bin_info_t *)base_alloc(nhbins *
+	    sizeof(tcache_bin_info_t));
+	if (tcache_bin_info == NULL)
+		return (true);
+	stack_nelms = 0;
+	for (i = 0; i < NBINS; i++) {
+		if ((arena_bin_info[i].nregs << 1) <= TCACHE_NSLOTS_SMALL_MAX) {
+			tcache_bin_info[i].ncached_max =
+			    (arena_bin_info[i].nregs << 1);
+		} else {
+			tcache_bin_info[i].ncached_max =
+			    TCACHE_NSLOTS_SMALL_MAX;
+		}
+		stack_nelms += tcache_bin_info[i].ncached_max;
+	}
+	for (; i < nhbins; i++) {
+		tcache_bin_info[i].ncached_max = TCACHE_NSLOTS_LARGE;
+		stack_nelms += tcache_bin_info[i].ncached_max;
+	}
+
+	return (false);
+}
+
+bool
+tcache_boot1(void)
+{
+
+	if (tcache_tsd_boot() || tcache_enabled_tsd_boot())
+		return (true);
+
+	return (false);
+}

contrib/jemalloc/src/tsd.c

@@ -0,0 +1,72 @@
+#define	JEMALLOC_TSD_C_
+#include "jemalloc/internal/jemalloc_internal.h"
+
+/******************************************************************************/
+/* Data. */
+
+static unsigned ncleanups;
+static malloc_tsd_cleanup_t cleanups[MALLOC_TSD_CLEANUPS_MAX];
+
+/******************************************************************************/
+
+void *
+malloc_tsd_malloc(size_t size)
+{
+
+	/* Avoid choose_arena() in order to dodge bootstrapping issues. */
+	return arena_malloc(arenas[0], size, false, false);
+}
+
+void
+malloc_tsd_dalloc(void *wrapper)
+{
+
+	idalloc(wrapper);
+}
+
+void
+malloc_tsd_no_cleanup(void *arg)
+{
+
+	not_reached();
+}
+
+#ifdef JEMALLOC_MALLOC_THREAD_CLEANUP
+void
+_malloc_thread_cleanup(void)
+{
+	bool pending[ncleanups], again;
+	unsigned i;
+
+	for (i = 0; i < ncleanups; i++)
+		pending[i] = true;
+
+	do {
+		again = false;
+		for (i = 0; i < ncleanups; i++) {
+			if (pending[i]) {
+				pending[i] = cleanups[i].f(cleanups[i].arg);
+				if (pending[i])
+					again = true;
+			}
+		}
+	} while (again);
+}
+#endif
+
+void
+malloc_tsd_cleanup_register(bool (*f)(void *), void *arg)
+{
+
+	assert(ncleanups < MALLOC_TSD_CLEANUPS_MAX);
+	cleanups[ncleanups].f = f;
+	cleanups[ncleanups].arg = arg;
+	ncleanups++;
+}
+
+void
+malloc_tsd_boot(void)
+{
+
+	ncleanups = 0;
+}

contrib/jemalloc/src/util.c

@@ -0,0 +1,635 @@
+#define	assert(e) do {							\
+	if (config_debug && !(e)) {					\
+		malloc_write("<jemalloc>: Failed assertion\n");		\
+		abort();						\
+	}								\
+} while (0)
+
+#define	not_reached() do {						\
+	if (config_debug) {						\
+		malloc_write("<jemalloc>: Unreachable code reached\n");	\
+		abort();						\
+	}								\
+} while (0)
+
+#define	not_implemented() do {						\
+	if (config_debug) {						\
+		malloc_write("<jemalloc>: Not implemented\n");		\
+		abort();						\
+	}								\
+} while (0)
+
+#define	JEMALLOC_UTIL_C_
+#include "jemalloc/internal/jemalloc_internal.h"
+
+/******************************************************************************/
+/* Function prototypes for non-inline static functions. */
+
+static void	wrtmessage(void *cbopaque, const char *s);
+#define	U2S_BUFSIZE	((1U << (LG_SIZEOF_INTMAX_T + 3)) + 1)
+static char	*u2s(uintmax_t x, unsigned base, bool uppercase, char *s,
+    size_t *slen_p);
+#define	D2S_BUFSIZE	(1 + U2S_BUFSIZE)
+static char	*d2s(intmax_t x, char sign, char *s, size_t *slen_p);
+#define	O2S_BUFSIZE	(1 + U2S_BUFSIZE)
+static char	*o2s(uintmax_t x, bool alt_form, char *s, size_t *slen_p);
+#define	X2S_BUFSIZE	(2 + U2S_BUFSIZE)
+static char	*x2s(uintmax_t x, bool alt_form, bool uppercase, char *s,
+    size_t *slen_p);
+
+/******************************************************************************/
+
+/* malloc_message() setup. */
+JEMALLOC_CATTR(visibility("hidden"), static)
+void
+wrtmessage(void *cbopaque, const char *s)
+{
+
+#ifdef SYS_write
+	/*
+	 * Use syscall(2) rather than write(2) when possible in order to avoid
+	 * the possibility of memory allocation within libc.  This is necessary
+	 * on FreeBSD; most operating systems do not have this problem though.
+	 */
+	UNUSED int result = syscall(SYS_write, STDERR_FILENO, s, strlen(s));
+#else
+	UNUSED int result = write(STDERR_FILENO, s, strlen(s));
+#endif
+}
+
+void	(*je_malloc_message)(void *, const char *s)
+    JEMALLOC_ATTR(visibility("default")) = wrtmessage;
+
+JEMALLOC_CATTR(visibility("hidden"), static)
+void
+wrtmessage_1_0(const char *s1, const char *s2, const char *s3,
+    const char *s4)
+{
+
+	wrtmessage(NULL, s1);
+	wrtmessage(NULL, s2);
+	wrtmessage(NULL, s3);
+	wrtmessage(NULL, s4);
+}
+
+void	(*__malloc_message_1_0)(const char *s1, const char *s2, const char *s3,
+    const char *s4) = wrtmessage_1_0;
+__sym_compat(_malloc_message, __malloc_message_1_0, FBSD_1.0);
+
+/*
+ * glibc provides a non-standard strerror_r() when _GNU_SOURCE is defined, so
+ * provide a wrapper.
+ */
+int
+buferror(int errnum, char *buf, size_t buflen)
+{
+#ifdef _GNU_SOURCE
+	char *b = strerror_r(errno, buf, buflen);
+	if (b != buf) {
+		strncpy(buf, b, buflen);
+		buf[buflen-1] = '\0';
+	}
+	return (0);
+#else
+	return (strerror_r(errno, buf, buflen));
+#endif
+}
+
+uintmax_t
+malloc_strtoumax(const char *nptr, char **endptr, int base)
+{
+	uintmax_t ret, digit;
+	int b;
+	bool neg;
+	const char *p, *ns;
+
+	if (base < 0 || base == 1 || base > 36) {
+		errno = EINVAL;
+		return (UINTMAX_MAX);
+	}
+	b = base;
+
+	/* Swallow leading whitespace and get sign, if any. */
+	neg = false;
+	p = nptr;
+	while (true) {
+		switch (*p) {
+		case '\t': case '\n': case '\v': case '\f': case '\r': case ' ':
+			p++;
+			break;
+		case '-':
+			neg = true;
+			/* Fall through. */
+		case '+':
+			p++;
+			/* Fall through. */
+		default:
+			goto label_prefix;
+		}
+	}
+
+	/* Get prefix, if any. */
+	label_prefix:
+	/*
+	 * Note where the first non-whitespace/sign character is so that it is
+	 * possible to tell whether any digits are consumed (e.g., "  0" vs.
+	 * "  -x").
+	 */
+	ns = p;
+	if (*p == '0') {
+		switch (p[1]) {
+		case '0': case '1': case '2': case '3': case '4': case '5':
+		case '6': case '7':
+			if (b == 0)
+				b = 8;
+			if (b == 8)
+				p++;
+			break;
+		case 'x':
+			switch (p[2]) {
+			case '0': case '1': case '2': case '3': case '4':
+			case '5': case '6': case '7': case '8': case '9':
+			case 'A': case 'B': case 'C': case 'D': case 'E':
+			case 'F':
+			case 'a': case 'b': case 'c': case 'd': case 'e':
+			case 'f':
+				if (b == 0)
+					b = 16;
+				if (b == 16)
+					p += 2;
+				break;
+			default:
+				break;
+			}
+			break;
+		default:
+			break;
+		}
+	}
+	if (b == 0)
+		b = 10;
+
+	/* Convert. */
+	ret = 0;
+	while ((*p >= '0' && *p <= '9' && (digit = *p - '0') < b)
+	    || (*p >= 'A' && *p <= 'Z' && (digit = 10 + *p - 'A') < b)
+	    || (*p >= 'a' && *p <= 'z' && (digit = 10 + *p - 'a') < b)) {
+		uintmax_t pret = ret;
+		ret *= b;
+		ret += digit;
+		if (ret < pret) {
+			/* Overflow. */
+			errno = ERANGE;
+			return (UINTMAX_MAX);
+		}
+		p++;
+	}
+	if (neg)
+		ret = -ret;
+
+	if (endptr != NULL) {
+		if (p == ns) {
+			/* No characters were converted. */
+			*endptr = (char *)nptr;
+		} else
+			*endptr = (char *)p;
+	}
+
+	return (ret);
+}
+
+static char *
+u2s(uintmax_t x, unsigned base, bool uppercase, char *s, size_t *slen_p)
+{
+	unsigned i;
+
+	i = U2S_BUFSIZE - 1;
+	s[i] = '\0';
+	switch (base) {
+	case 10:
+		do {
+			i--;
+			s[i] = "0123456789"[x % (uint64_t)10];
+			x /= (uint64_t)10;
+		} while (x > 0);
+		break;
+	case 16: {
+		const char *digits = (uppercase)
+		    ? "0123456789ABCDEF"
+		    : "0123456789abcdef";
+
+		do {
+			i--;
+			s[i] = digits[x & 0xf];
+			x >>= 4;
+		} while (x > 0);
+		break;
+	} default: {
+		const char *digits = (uppercase)
+		    ? "0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZ"
+		    : "0123456789abcdefghijklmnopqrstuvwxyz";
+
+		assert(base >= 2 && base <= 36);
+		do {
+			i--;
+			s[i] = digits[x % (uint64_t)base];
+			x /= (uint64_t)base;
+		} while (x > 0);
+	}}
+
+	*slen_p = U2S_BUFSIZE - 1 - i;
+	return (&s[i]);
+}
+
+static char *
+d2s(intmax_t x, char sign, char *s, size_t *slen_p)
+{
+	bool neg;
+
+	if ((neg = (x < 0)))
+		x = -x;
+	s = u2s(x, 10, false, s, slen_p);
+	if (neg)
+		sign = '-';
+	switch (sign) {
+	case '-':
+		if (neg == false)
+			break;
+		/* Fall through. */
+	case ' ':
+	case '+':
+		s--;
+		(*slen_p)++;
+		*s = sign;
+		break;
+	default: not_reached();
+	}
+	return (s);
+}
+
+static char *
+o2s(uintmax_t x, bool alt_form, char *s, size_t *slen_p)
+{
+
+	s = u2s(x, 8, false, s, slen_p);
+	if (alt_form && *s != '0') {
+		s--;
+		(*slen_p)++;
+		*s = '0';
+	}
+	return (s);
+}
+
+static char *
+x2s(uintmax_t x, bool alt_form, bool uppercase, char *s, size_t *slen_p)
+{
+
+	s = u2s(x, 16, uppercase, s, slen_p);
+	if (alt_form) {
+		s -= 2;
+		(*slen_p) += 2;
+		memcpy(s, uppercase ? "0X" : "0x", 2);
+	}
+	return (s);
+}
+
+int
+malloc_vsnprintf(char *str, size_t size, const char *format, va_list ap)
+{
+	int ret;
+	size_t i;
+	const char *f;
+	va_list tap;
+
+#define	APPEND_C(c) do {						\
+	if (i < size)							\
+		str[i] = (c);						\
+	i++;								\
+} while (0)
+#define	APPEND_S(s, slen) do {						\
+	if (i < size) {							\
+		size_t cpylen = (slen <= size - i) ? slen : size - i;	\
+		memcpy(&str[i], s, cpylen);				\
+	}								\
+	i += slen;							\
+} while (0)
+#define	APPEND_PADDED_S(s, slen, width, left_justify) do {		\
+	/* Left padding. */						\
+	size_t pad_len = (width == -1) ? 0 : ((slen < (size_t)width) ?	\
+	    (size_t)width - slen : 0);					\
+	if (left_justify == false && pad_len != 0) {			\
+		size_t j;						\
+		for (j = 0; j < pad_len; j++)				\
+			APPEND_C(' ');					\
+	}								\
+	/* Value. */							\
+	APPEND_S(s, slen);						\
+	/* Right padding. */						\
+	if (left_justify && pad_len != 0) {				\
+		size_t j;						\
+		for (j = 0; j < pad_len; j++)				\
+			APPEND_C(' ');					\
+	}								\
+} while (0)
+#define GET_ARG_NUMERIC(val, len) do {					\
+	switch (len) {							\
+	case '?':							\
+		val = va_arg(ap, int);					\
+		break;							\
+	case 'l':							\
+		val = va_arg(ap, long);					\
+		break;							\
+	case 'q':							\
+		val = va_arg(ap, long long);				\
+		break;							\
+	case 'j':							\
+		val = va_arg(ap, intmax_t);				\
+		break;							\
+	case 't':							\
+		val = va_arg(ap, ptrdiff_t);				\
+		break;							\
+	case 'z':							\
+		val = va_arg(ap, ssize_t);				\
+		break;							\
+	case 'p': /* Synthetic; used for %p. */				\
+		val = va_arg(ap, uintptr_t);				\
+		break;							\
+	default: not_reached();						\
+	}								\
+} while (0)
+
+	if (config_debug)
+		va_copy(tap, ap);
+
+	i = 0;
+	f = format;
+	while (true) {
+		switch (*f) {
+		case '\0': goto label_out;
+		case '%': {
+			bool alt_form = false;
+			bool zero_pad = false;
+			bool left_justify = false;
+			bool plus_space = false;
+			bool plus_plus = false;
+			int prec = -1;
+			int width = -1;
+			char len = '?';
+
+			f++;
+			if (*f == '%') {
+				/* %% */
+				APPEND_C(*f);
+				break;
+			}
+			/* Flags. */
+			while (true) {
+				switch (*f) {
+				case '#':
+					assert(alt_form == false);
+					alt_form = true;
+					break;
+				case '0':
+					assert(zero_pad == false);
+					zero_pad = true;
+					break;
+				case '-':
+					assert(left_justify == false);
+					left_justify = true;
+					break;
+				case ' ':
+					assert(plus_space == false);
+					plus_space = true;
+					break;
+				case '+':
+					assert(plus_plus == false);
+					plus_plus = true;
+					break;
+				default: goto label_width;
+				}
+				f++;
+			}
+			/* Width. */
+			label_width:
+			switch (*f) {
+			case '*':
+				width = va_arg(ap, int);
+				f++;
+				break;
+			case '0': case '1': case '2': case '3': case '4':
+			case '5': case '6': case '7': case '8': case '9': {
+				uintmax_t uwidth;
+				errno = 0;
+				uwidth = malloc_strtoumax(f, (char **)&f, 10);
+				assert(uwidth != UINTMAX_MAX || errno !=
+				    ERANGE);
+				width = (int)uwidth;
+				if (*f == '.') {
+					f++;
+					goto label_precision;
+				} else
+					goto label_length;
+				break;
+			} case '.':
+				f++;
+				goto label_precision;
+			default: goto label_length;
+			}
+			/* Precision. */
+			label_precision:
+			switch (*f) {
+			case '*':
+				prec = va_arg(ap, int);
+				f++;
+				break;
+			case '0': case '1': case '2': case '3': case '4':
+			case '5': case '6': case '7': case '8': case '9': {
+				uintmax_t uprec;
+				errno = 0;
+				uprec = malloc_strtoumax(f, (char **)&f, 10);
+				assert(uprec != UINTMAX_MAX || errno != ERANGE);
+				prec = (int)uprec;
+				break;
+			}
+			default: break;
+			}
+			/* Length. */
+			label_length:
+			switch (*f) {
+			case 'l':
+				f++;
+				if (*f == 'l') {
+					len = 'q';
+					f++;
+				} else
+					len = 'l';
+				break;
+			case 'j':
+				len = 'j';
+				f++;
+				break;
+			case 't':
+				len = 't';
+				f++;
+				break;
+			case 'z':
+				len = 'z';
+				f++;
+				break;
+			default: break;
+			}
+			/* Conversion specifier. */
+			switch (*f) {
+				char *s;
+				size_t slen;
+			case 'd': case 'i': {
+				intmax_t val JEMALLOC_CC_SILENCE_INIT(0);
+				char buf[D2S_BUFSIZE];
+
+				GET_ARG_NUMERIC(val, len);
+				s = d2s(val, (plus_plus ? '+' : (plus_space ?
+				    ' ' : '-')), buf, &slen);
+				APPEND_PADDED_S(s, slen, width, left_justify);
+				f++;
+				break;
+			} case 'o': {
+				uintmax_t val JEMALLOC_CC_SILENCE_INIT(0);
+				char buf[O2S_BUFSIZE];
+
+				GET_ARG_NUMERIC(val, len);
+				s = o2s(val, alt_form, buf, &slen);
+				APPEND_PADDED_S(s, slen, width, left_justify);
+				f++;
+				break;
+			} case 'u': {
+				uintmax_t val JEMALLOC_CC_SILENCE_INIT(0);
+				char buf[U2S_BUFSIZE];
+
+				GET_ARG_NUMERIC(val, len);
+				s = u2s(val, 10, false, buf, &slen);
+				APPEND_PADDED_S(s, slen, width, left_justify);
+				f++;
+				break;
+			} case 'x': case 'X': {
+				uintmax_t val JEMALLOC_CC_SILENCE_INIT(0);
+				char buf[X2S_BUFSIZE];
+
+				GET_ARG_NUMERIC(val, len);
+				s = x2s(val, alt_form, *f == 'X', buf, &slen);
+				APPEND_PADDED_S(s, slen, width, left_justify);
+				f++;
+				break;
+			} case 'c': {
+				unsigned char val;
+				char buf[2];
+
+				assert(len == '?' || len == 'l');
+				assert_not_implemented(len != 'l');
+				val = va_arg(ap, int);
+				buf[0] = val;
+				buf[1] = '\0';
+				APPEND_PADDED_S(buf, 1, width, left_justify);
+				f++;
+				break;
+			} case 's':
+				assert(len == '?' || len == 'l');
+				assert_not_implemented(len != 'l');
+				s = va_arg(ap, char *);
+				slen = (prec == -1) ? strlen(s) : prec;
+				APPEND_PADDED_S(s, slen, width, left_justify);
+				f++;
+				break;
+			case 'p': {
+				uintmax_t val;
+				char buf[X2S_BUFSIZE];
+
+				GET_ARG_NUMERIC(val, 'p');
+				s = x2s(val, true, false, buf, &slen);
+				APPEND_PADDED_S(s, slen, width, left_justify);
+				f++;
+				break;
+			}
+			default: not_implemented();
+			}
+			break;
+		} default: {
+			APPEND_C(*f);
+			f++;
+			break;
+		}}
+	}
+	label_out:
+	if (i < size)
+		str[i] = '\0';
+	else
+		str[size - 1] = '\0';
+	ret = i;
+
+#undef APPEND_C
+#undef APPEND_S
+#undef APPEND_PADDED_S
+#undef GET_ARG_NUMERIC
+	return (ret);
+}
+
+JEMALLOC_ATTR(format(printf, 3, 4))
+int
+malloc_snprintf(char *str, size_t size, const char *format, ...)
+{
+	int ret;
+	va_list ap;
+
+	va_start(ap, format);
+	ret = malloc_vsnprintf(str, size, format, ap);
+	va_end(ap);
+
+	return (ret);
+}
+
+void
+malloc_vcprintf(void (*write_cb)(void *, const char *), void *cbopaque,
+    const char *format, va_list ap)
+{
+	char buf[MALLOC_PRINTF_BUFSIZE];
+
+	if (write_cb == NULL) {
+		/*
+		 * The caller did not provide an alternate write_cb callback
+		 * function, so use the default one.  malloc_write() is an
+		 * inline function, so use malloc_message() directly here.
+		 */
+		write_cb = je_malloc_message;
+		cbopaque = NULL;
+	}
+
+	malloc_vsnprintf(buf, sizeof(buf), format, ap);
+	write_cb(cbopaque, buf);
+}
+
+/*
+ * Print to a callback function in such a way as to (hopefully) avoid memory
+ * allocation.
+ */
+JEMALLOC_ATTR(format(printf, 3, 4))
+void
+malloc_cprintf(void (*write_cb)(void *, const char *), void *cbopaque,
+    const char *format, ...)
+{
+	va_list ap;
+
+	va_start(ap, format);
+	malloc_vcprintf(write_cb, cbopaque, format, ap);
+	va_end(ap);
+}
+
+/* Print to stderr in such a way as to avoid memory allocation. */
+JEMALLOC_ATTR(format(printf, 1, 2))
+void
+malloc_printf(const char *format, ...)
+{
+	va_list ap;
+
+	va_start(ap, format);
+	malloc_vcprintf(NULL, NULL, format, ap);
+	va_end(ap);
+}

include/malloc_np.h

@@ -33,9 +33,36 @@
 #define	_MALLOC_NP_H_
 #include <sys/cdefs.h>
 #include <sys/types.h>
+#include <strings.h>
 
 __BEGIN_DECLS
 size_t	malloc_usable_size(const void *ptr);
+
+void	malloc_stats_print(void (*write_cb)(void *, const char *),
+    void *cbopaque, const char *opts);
+int	mallctl(const char *name, void *oldp, size_t *oldlenp, void *newp,
+    size_t newlen);
+int	mallctlnametomib(const char *name, size_t *mibp, size_t *miblenp);
+int	mallctlbymib(const size_t *mib, size_t miblen, void *oldp,
+    size_t *oldlenp, void *newp, size_t newlen);
+
+#define	ALLOCM_LG_ALIGN(la)	(la)
+#define	ALLOCM_ALIGN(a)	(ffsl(a)-1)
+#define	ALLOCM_ZERO	((int)0x40)
+#define	ALLOCM_NO_MOVE	((int)0x80)
+
+#define	ALLOCM_SUCCESS		0
+#define	ALLOCM_ERR_OOM		1
+#define	ALLOCM_ERR_NOT_MOVED	2
+
+int	allocm(void **ptr, size_t *rsize, size_t size, int flags)
+    __attribute__(nonnull(1));
+int	rallocm(void **ptr, size_t *rsize, size_t size, size_t extra,
+    int flags) __attribute__(nonnull(1));
+int	sallocm(const void *ptr, size_t *rsize, int flags)
+    __attribute__(nonnull(1));
+int	dallocm(void *ptr, int flags) __attribute__(nonnull(1));
+int	nallocm(size_t *rsize, size_t size, int flags);
 __END_DECLS
 
 #endif /* _MALLOC_NP_H_ */

include/stdlib.h

@@ -155,7 +155,7 @@ _Noreturn void	 _Exit(int);
  * If we're in a mode greater than C99, expose C11 functions.
  */
 #if __ISO_C_VISIBLE >= 2011 || __cplusplus >= 201103L
-void *	aligned_alloc(size_t, size_t);
+void *	aligned_alloc(size_t, size_t) __malloc_like;
 int	at_quick_exit(void (*)(void));
 _Noreturn void
 	quick_exit(int);
@@ -228,9 +228,8 @@ int	 unlockpt(int);
 #endif /* __XSI_VISIBLE */
 
 #if __BSD_VISIBLE
-extern const char *_malloc_options;
-extern void (*_malloc_message)(const char *, const char *, const char *,
-	    const char *);
+extern const char *malloc_conf;
+extern void (*malloc_message)(void *, const char *);
 
 /*
  * The alloca() function can't be implemented in C, and on some

lib/libc/Makefile

@@ -79,6 +79,7 @@ NOASM=
 .include "${.CURDIR}/resolv/Makefile.inc"
 .include "${.CURDIR}/stdio/Makefile.inc"
 .include "${.CURDIR}/stdlib/Makefile.inc"
+.include "${.CURDIR}/stdlib/jemalloc/Makefile.inc"
 .include "${.CURDIR}/stdtime/Makefile.inc"
 .include "${.CURDIR}/string/Makefile.inc"
 .include "${.CURDIR}/sys/Makefile.inc"

lib/libc/gen/tls.c

@@ -39,6 +39,11 @@
 
 #include "libc_private.h"
 
+/* Provided by jemalloc to avoid bootstrapping issues. */
+void	*a0malloc(size_t size);
+void	*a0calloc(size_t num, size_t size);
+void	a0free(void *ptr);
+
 __weak_reference(__libc_allocate_tls, _rtld_allocate_tls);
 __weak_reference(__libc_free_tls, _rtld_free_tls);
 
@@ -120,8 +125,8 @@ __libc_free_tls(void *tcb, size_t tcbsize, size_t tcbalign __unused)
 
 	tls = (Elf_Addr **)((Elf_Addr)tcb + tcbsize - TLS_TCB_SIZE);
 	dtv = tls[0];
-	free(dtv);
-	free(tcb);
+	a0free(dtv);
+	a0free(tcb);
 }
 
 /*
@@ -137,18 +142,18 @@ __libc_allocate_tls(void *oldtcb, size_t tcbsize, size_t tcbalign __unused)
 	if (oldtcb != NULL && tcbsize == TLS_TCB_SIZE)
 		return (oldtcb);
 
-	tcb = calloc(1, tls_static_space + tcbsize - TLS_TCB_SIZE);
+	tcb = a0calloc(1, tls_static_space + tcbsize - TLS_TCB_SIZE);
 	tls = (Elf_Addr **)(tcb + tcbsize - TLS_TCB_SIZE);
 
 	if (oldtcb != NULL) {
 		memcpy(tls, oldtcb, tls_static_space);
-		free(oldtcb);
+		a0free(oldtcb);
 
 		/* Adjust the DTV. */
 		dtv = tls[0];
 		dtv[2] = (Elf_Addr)tls + TLS_TCB_SIZE;
 	} else {
-		dtv = malloc(3 * sizeof(Elf_Addr));
+		dtv = a0malloc(3 * sizeof(Elf_Addr));
 		tls[0] = dtv;
 		dtv[0] = 1;
 		dtv[1] = 1;
@@ -189,8 +194,8 @@ __libc_free_tls(void *tcb, size_t tcbsize __unused, size_t tcbalign)
 	dtv = ((Elf_Addr**)tcb)[1];
 	tlsend = (Elf_Addr) tcb;
 	tlsstart = tlsend - size;
-	free((void*) tlsstart);
-	free(dtv);
+	a0free((void*) tlsstart);
+	a0free(dtv);
 }
 
 /*
@@ -208,8 +213,8 @@ __libc_allocate_tls(void *oldtls, size_t tcbsize, size_t tcbalign)
 
 	if (tcbsize < 2 * sizeof(Elf_Addr))
 		tcbsize = 2 * sizeof(Elf_Addr);
-	tls = calloc(1, size + tcbsize);
-	dtv = malloc(3 * sizeof(Elf_Addr));
+	tls = a0calloc(1, size + tcbsize);
+	dtv = a0malloc(3 * sizeof(Elf_Addr));
 
 	segbase = (Elf_Addr)(tls + size);
 	((Elf_Addr*)segbase)[0] = segbase;

lib/libc/stdlib/Makefile.inc

@@ -7,7 +7,7 @@
 MISRCS+=_Exit.c a64l.c abort.c abs.c atexit.c atof.c atoi.c atol.c atoll.c \
 	bsearch.c div.c exit.c getenv.c getopt.c getopt_long.c \
 	getsubopt.c hcreate.c heapsort.c imaxabs.c imaxdiv.c \
-	insque.c l64a.c labs.c ldiv.c llabs.c lldiv.c lsearch.c malloc.c \
+	insque.c l64a.c labs.c ldiv.c llabs.c lldiv.c lsearch.c \
 	merge.c ptsname.c qsort.c qsort_r.c quick_exit.c radixsort.c rand.c \
 	random.c reallocf.c realpath.c remque.c strfmon.c strtoimax.c \
 	strtol.c strtoll.c strtoq.c strtoul.c strtonum.c strtoull.c \
@@ -18,18 +18,17 @@ SYM_MAPS+= ${.CURDIR}/stdlib/Symbol.map
 # machine-dependent stdlib sources
 .sinclude "${.CURDIR}/${LIBC_ARCH}/stdlib/Makefile.inc"
 
-MAN+=	a64l.3 abort.3 abs.3 aligned_alloc.3 alloca.3 atexit.3 atof.3 \
+MAN+=	a64l.3 abort.3 abs.3 alloca.3 atexit.3 atof.3 \
 	atoi.3 atol.3 at_quick_exit.3 bsearch.3 \
 	div.3 exit.3 getenv.3 getopt.3 getopt_long.3 getsubopt.3 \
 	hcreate.3 imaxabs.3 imaxdiv.3 insque.3 labs.3 ldiv.3 llabs.3 lldiv.3 \
-	lsearch.3 malloc.3 memory.3 ptsname.3 qsort.3 \
+	lsearch.3 memory.3 ptsname.3 qsort.3 \
 	quick_exit.3 \
-	radixsort.3 rand.3 random.3 \
+	radixsort.3 rand.3 random.3 reallocf.3 \
 	realpath.3 strfmon.3 strtod.3 strtol.3 strtonum.3 strtoul.3 system.3 \
 	tsearch.3
 
 MLINKS+=a64l.3 l64a.3 a64l.3 l64a_r.3
-MLINKS+=aligned_alloc.3 posix_memalign.3
 MLINKS+=atol.3 atoll.3
 MLINKS+=exit.3 _Exit.3
 MLINKS+=getenv.3 putenv.3 getenv.3 setenv.3 getenv.3 unsetenv.3
@@ -46,11 +45,4 @@ MLINKS+=radixsort.3 sradixsort.3
 MLINKS+=strtod.3 strtof.3 strtod.3 strtold.3
 MLINKS+=strtol.3 strtoll.3 strtol.3 strtoq.3 strtol.3 strtoimax.3
 MLINKS+=strtoul.3 strtoull.3 strtoul.3 strtouq.3 strtoul.3 strtoumax.3
-MLINKS+=malloc.3 calloc.3 malloc.3 free.3 malloc.3 malloc.conf.5 \
-	malloc.3 realloc.3 malloc.3 reallocf.3 malloc.3 malloc_usable_size.3
 MLINKS+=tsearch.3 tdelete.3 tsearch.3 tfind.3 tsearch.3 twalk.3
-
-.if defined(MALLOC_PRODUCTION)
-CFLAGS+=	-DMALLOC_PRODUCTION
-.endif
-

lib/libc/stdlib/Symbol.map

@@ -47,14 +47,6 @@ FBSD_1.0 {
 	lldiv;
 	lsearch;
 	lfind;
-	_malloc_options;
-	_malloc_message;
-	malloc;
-	posix_memalign;
-	calloc;
-	realloc;
-	free;
-	malloc_usable_size;
 	mergesort;
 	putenv;
 	qsort_r;
@@ -93,7 +85,6 @@ FBSD_1.0 {
 };
 
 FBSD_1.3 {
-	aligned_alloc;
 	at_quick_exit;
 	atof_l;
 	atoi_l;
@@ -114,9 +105,6 @@ FBSD_1.3 {
 };
 
 FBSDprivate_1.0 {
-	_malloc_thread_cleanup;
-	_malloc_prefork;
-	_malloc_postfork;
 	__system;
 	_system;
 };

lib/libc/stdlib/aligned_alloc.3

@@ -1,126 +0,0 @@
-.\" Copyright (C) 2006 Jason Evans <jasone@FreeBSD.org>.
-.\" All rights reserved.
-.\"
-.\" Redistribution and use in source and binary forms, with or without
-.\" modification, are permitted provided that the following conditions
-.\" are met:
-.\" 1. Redistributions of source code must retain the above copyright
-.\"    notice(s), this list of conditions and the following disclaimer as
-.\"    the first lines of this file unmodified other than the possible
-.\"    addition of one or more copyright notices.
-.\" 2. Redistributions in binary form must reproduce the above copyright
-.\"    notice(s), this list of conditions and the following disclaimer in
-.\"    the documentation and/or other materials provided with the
-.\"    distribution.
-.\"
-.\" THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDER(S) ``AS IS'' AND ANY
-.\" EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
-.\" IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
-.\" PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE COPYRIGHT HOLDER(S) BE
-.\" LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
-.\" CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
-.\" SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR
-.\" BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
-.\" WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE
-.\" OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE,
-.\" EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
-.\"
-.\" $FreeBSD$
-.\"
-.Dd January 7, 2011
-.Dt ALIGNED_ALLOC 3
-.Os
-.Sh NAME
-.Nm aligned_alloc ,
-.Nm posix_memalign
-.Nd aligned memory allocation
-.Sh LIBRARY
-.Lb libc
-.Sh SYNOPSIS
-.In stdlib.h
-.Ft void *
-.Fn aligned_alloc "size_t alignment" "size_t size"
-.Ft int
-.Fn posix_memalign "void **ptr" "size_t alignment" "size_t size"
-.Sh DESCRIPTION
-The
-.Fn aligned_alloc
-and
-.Fn posix_memalign
-functions allocate
-.Fa size
-bytes of memory such that the allocation's base address is an even multiple of
-.Fa alignment .
-The
-.Fn aligned_alloc
-function returns the allocation, while the
-.Fn posix_memalign
-function stores the allocation in the value pointed to by
-.Fa ptr .
-.Pp
-The requested
-.Fa alignment
-must be a power of 2 at least as large as
-.Fn sizeof "void *" .
-.Pp
-Memory that is allocated via
-.Fn aligned_alloc
-and
-.Fn posix_memalign
-can be used as an argument in subsequent calls to
-.Xr realloc 3 ,
-.Xr reallocf 3 ,
-and
-.Xr free 3 .
-.Sh RETURN VALUES
-The
-.Fn aligned_alloc
-function returns a pointer to the allocation if successful; otherwise a
-NULL pointer is returned and
-.Va errno
-is set to an error value.
-.Pp
-The
-.Fn posix_memalign
-function returns the value 0 if successful; otherwise it returns an error value.
-.Sh ERRORS
-The
-.Fn aligned_alloc
-and
-.Fn posix_memalign
-functions will fail if:
-.Bl -tag -width Er
-.It Bq Er EINVAL
-The
-.Fa alignment
-parameter is not a power of 2 at least as large as
-.Fn sizeof "void *" .
-.It Bq Er ENOMEM
-Memory allocation error.
-.El
-.Sh SEE ALSO
-.Xr free 3 ,
-.Xr malloc 3 ,
-.Xr realloc 3 ,
-.Xr reallocf 3 ,
-.Xr valloc 3
-.Sh STANDARDS
-The
-.Fn aligned_alloc
-function conforms to
-.St -isoC-2011 .
-.Pp
-The
-.Fn posix_memalign
-function conforms to
-.St -p1003.1-2001 .
-.Sh HISTORY
-The
-.Fn posix_memalign
-function first appeared in
-.Fx 7.0 .
-.Pp
-The
-.Fn aligned_alloc
-function first appeared in
-.Fx 10.0 .

lib/libc/stdlib/jemalloc/Makefile.inc

@@ -0,0 +1,46 @@
+# $FreeBSD$
+
+.PATH: ${.CURDIR}/stdlib/jemalloc
+
+JEMALLOCSRCS:= jemalloc.c arena.c atomic.c base.c bitmap.c chunk.c \
+	chunk_dss.c chunk_mmap.c ckh.c ctl.c extent.c hash.c huge.c mb.c \
+	mutex.c prof.c quarantine.c rtree.c stats.c tcache.c util.c tsd.c
+
+SYM_MAPS+=${.CURDIR}/stdlib/jemalloc/Symbol.map
+
+CFLAGS+=-I${.CURDIR}/../../contrib/jemalloc/include
+
+.for src in ${JEMALLOCSRCS}
+MISRCS+=jemalloc_${src}
+CLEANFILES+=jemalloc_${src}
+jemalloc_${src}:
+	ln -sf ${.CURDIR}/../../contrib/jemalloc/src/${src} ${.TARGET}
+.endfor
+
+MAN+=jemalloc.3
+CLEANFILES+=jemalloc.3
+jemalloc.3:
+	ln -sf ${.CURDIR}/../../contrib/jemalloc/doc/jemalloc.3 ${.TARGET}
+
+MLINKS+= \
+	jemalloc.3 malloc.3 \
+	jemalloc.3 calloc.3 \
+	jemalloc.3 posix_memalign.3 \
+	jemalloc.3 aligned_alloc.3 \
+	jemalloc.3 realloc.3 \
+	jemalloc.3 free.3 \
+	jemalloc.3 malloc_usable_size.3 \
+	jemalloc.3 malloc_stats_print.3 \
+	jemalloc.3 mallctl.3 \
+	jemalloc.3 mallctlnametomib.3 \
+	jemalloc.3 mallctlbymib.3 \
+	jemalloc.3 allocm.3 \
+	jemalloc.3 rallocm.3 \
+	jemalloc.3 sallocm.3 \
+	jemalloc.3 dallocm.3 \
+	jemalloc.3 nallocm.3 \
+	jemalloc.3 malloc.conf.5
+
+.if defined(MALLOC_PRODUCTION)
+CFLAGS+=	-DMALLOC_PRODUCTION
+.endif

lib/libc/stdlib/jemalloc/Symbol.map

@@ -0,0 +1,35 @@
+/*
+ * $FreeBSD$
+ */
+
+FBSD_1.0 {
+	_malloc_options;
+	_malloc_message;
+	malloc;
+	posix_memalign;
+	calloc;
+	realloc;
+	free;
+	malloc_usable_size;
+};
+
+FBSD_1.3 {
+	malloc_conf;
+	malloc_message;
+	aligned_alloc;
+	malloc_stats_print;
+	mallctl;
+	mallctlnametomib;
+	mallctlbymib;
+	allocm;
+	rallocm;
+	sallocm;
+	dallocm;
+	nallocm;
+};
+
+FBSDprivate_1.0 {
+	_malloc_thread_cleanup;
+	_malloc_prefork;
+	_malloc_postfork;
+};

lib/libc/stdlib/malloc.3

@@ -1,591 +0,0 @@
-.\" Copyright (c) 1980, 1991, 1993
-.\"	The Regents of the University of California.  All rights reserved.
-.\"
-.\" This code is derived from software contributed to Berkeley by
-.\" the American National Standards Committee X3, on Information
-.\" Processing Systems.
-.\"
-.\" Redistribution and use in source and binary forms, with or without
-.\" modification, are permitted provided that the following conditions
-.\" are met:
-.\" 1. Redistributions of source code must retain the above copyright
-.\"    notice, this list of conditions and the following disclaimer.
-.\" 2. Redistributions in binary form must reproduce the above copyright
-.\"    notice, this list of conditions and the following disclaimer in the
-.\"    documentation and/or other materials provided with the distribution.
-.\" 3. Neither the name of the University nor the names of its contributors
-.\"    may be used to endorse or promote products derived from this software
-.\"    without specific prior written permission.
-.\"
-.\" THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
-.\" ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
-.\" IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
-.\" ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
-.\" FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
-.\" DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
-.\" OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
-.\" HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
-.\" LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
-.\" OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
-.\" SUCH DAMAGE.
-.\"
-.\"     @(#)malloc.3	8.1 (Berkeley) 6/4/93
-.\" $FreeBSD$
-.\"
-.Dd January 31, 2010
-.Dt MALLOC 3
-.Os
-.Sh NAME
-.Nm malloc , calloc , realloc , free , reallocf , malloc_usable_size
-.Nd general purpose memory allocation functions
-.Sh LIBRARY
-.Lb libc
-.Sh SYNOPSIS
-.In stdlib.h
-.Ft void *
-.Fn malloc "size_t size"
-.Ft void *
-.Fn calloc "size_t number" "size_t size"
-.Ft void *
-.Fn realloc "void *ptr" "size_t size"
-.Ft void *
-.Fn reallocf "void *ptr" "size_t size"
-.Ft void
-.Fn free "void *ptr"
-.Ft const char *
-.Va _malloc_options ;
-.Ft void
-.Fn \*(lp*_malloc_message\*(rp "const char *p1" "const char *p2" "const char *p3" "const char *p4"
-.In malloc_np.h
-.Ft size_t
-.Fn malloc_usable_size "const void *ptr"
-.Sh DESCRIPTION
-The
-.Fn malloc
-function allocates
-.Fa size
-bytes of uninitialized memory.
-The allocated space is suitably aligned (after possible pointer coercion)
-for storage of any type of object.
-.Pp
-The
-.Fn calloc
-function allocates space for
-.Fa number
-objects,
-each
-.Fa size
-bytes in length.
-The result is identical to calling
-.Fn malloc
-with an argument of
-.Dq "number * size" ,
-with the exception that the allocated memory is explicitly initialized
-to zero bytes.
-.Pp
-The
-.Fn realloc
-function changes the size of the previously allocated memory referenced by
-.Fa ptr
-to
-.Fa size
-bytes.
-The contents of the memory are unchanged up to the lesser of the new and
-old sizes.
-If the new size is larger,
-the contents of the newly allocated portion of the memory are undefined.
-Upon success, the memory referenced by
-.Fa ptr
-is freed and a pointer to the newly allocated memory is returned.
-Note that
-.Fn realloc
-and
-.Fn reallocf
-may move the memory allocation, resulting in a different return value than
-.Fa ptr .
-If
-.Fa ptr
-is
-.Dv NULL ,
-the
-.Fn realloc
-function behaves identically to
-.Fn malloc
-for the specified size.
-.Pp
-The
-.Fn reallocf
-function is identical to the
-.Fn realloc
-function, except that it
-will free the passed pointer when the requested memory cannot be allocated.
-This is a
-.Fx
-specific API designed to ease the problems with traditional coding styles
-for
-.Fn realloc
-causing memory leaks in libraries.
-.Pp
-The
-.Fn free
-function causes the allocated memory referenced by
-.Fa ptr
-to be made available for future allocations.
-If
-.Fa ptr
-is
-.Dv NULL ,
-no action occurs.
-.Pp
-The
-.Fn malloc_usable_size
-function returns the usable size of the allocation pointed to by
-.Fa ptr .
-The return value may be larger than the size that was requested during
-allocation.
-The
-.Fn malloc_usable_size
-function is not a mechanism for in-place
-.Fn realloc ;
-rather it is provided solely as a tool for introspection purposes.
-Any discrepancy between the requested allocation size and the size reported by
-.Fn malloc_usable_size
-should not be depended on, since such behavior is entirely
-implementation-dependent.
-.Sh TUNING
-Once, when the first call is made to one of these memory allocation
-routines, various flags will be set or reset, which affects the
-workings of this allocator implementation.
-.Pp
-The
-.Dq name
-of the file referenced by the symbolic link named
-.Pa /etc/malloc.conf ,
-the value of the environment variable
-.Ev MALLOC_OPTIONS ,
-and the string pointed to by the global variable
-.Va _malloc_options
-will be interpreted, in that order, from left to right as flags.
-.Pp
-Each flag is a single letter, optionally prefixed by a non-negative base 10
-integer repetition count.
-For example,
-.Dq 3N
-is equivalent to
-.Dq NNN .
-Some flags control parameter magnitudes, where uppercase increases the
-magnitude, and lowercase decreases the magnitude.
-Other flags control boolean parameters, where uppercase indicates that a
-behavior is set, or on, and lowercase means that a behavior is not set, or off.
-.Bl -tag -width indent
-.It A
-All warnings (except for the warning about unknown
-flags being set) become fatal.
-The process will call
-.Xr abort 3
-in these cases.
-.It C
-Double/halve the size of the maximum size class that is a multiple of the
-cacheline size (64).
-Above this size, subpage spacing (256 bytes) is used for size classes.
-The default value is 512 bytes.
-.It D
-Use
-.Xr sbrk 2
-to acquire memory in the data storage segment (DSS).
-This option is enabled by default.
-See the
-.Dq M
-option for related information and interactions.
-.It E
-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.
-.It F
-Halve/double the per-arena minimum ratio of active to dirty pages.
-Some dirty unused pages may be allowed to accumulate, within the limit set by
-the ratio, before informing the kernel about at least half of those pages via
-.Xr madvise 2 .
-This provides the kernel with sufficient information to recycle dirty pages if
-physical memory becomes scarce and the pages remain unused.
-The default minimum ratio is 32:1;
-.Ev MALLOC_OPTIONS=6F
-will disable dirty page purging.
-.It G
-Double/halve the approximate interval (counted in terms of
-thread-specific cache allocation/deallocation events) between full
-thread-specific cache garbage collection sweeps.
-Garbage collection is actually performed incrementally, one size
-class at a time, in order to avoid large collection pauses.
-The default sweep interval is 8192;
-.Ev MALLOC_OPTIONS=14g
-will disable garbage collection.
-.It H
-Double/halve the number of thread-specific cache slots per size
-class.
-When there are multiple threads, each thread uses a
-thread-specific cache for small and medium objects.
-Thread-specific caching allows many allocations to be satisfied
-without performing any thread synchronization, at the cost of
-increased memory use.
-See the
-.Dq G
-option for related tuning information.
-The default number of cache slots is 128;
-.Ev MALLOC_OPTIONS=7h
-will disable thread-specific caching.
-Note that one cache slot per size class is not a valid
-configuration due to implementation details.
-.It J
-Each byte of new memory allocated by
-.Fn malloc ,
-.Fn realloc ,
-or
-.Fn reallocf
-will be initialized to 0xa5.
-All memory returned by
-.Fn free ,
-.Fn realloc ,
-or
-.Fn reallocf
-will be initialized to 0x5a.
-This is intended for debugging and will impact performance negatively.
-.It K
-Double/halve the virtual memory chunk size.
-The default chunk size is 4 MiB.
-.It M
-Use
-.Xr mmap 2
-to acquire anonymously mapped memory.
-This option is enabled by default.
-If both the
-.Dq D
-and
-.Dq M
-options are enabled, the allocator prefers anonymous mappings over the DSS,
-but allocation only fails if memory cannot be acquired via either method.
-If neither option is enabled, then the
-.Dq M
-option is implicitly enabled in order to assure that there is a method for
-acquiring memory.
-.It N
-Double/halve the number of arenas.
-The default number of arenas is two times the number of CPUs, or one if there
-is a single CPU.
-.It P
-Various statistics are printed at program exit via an
-.Xr atexit 3
-function.
-This has the potential to cause deadlock for a multi-threaded process that exits
-while one or more threads are executing in the memory allocation functions.
-Therefore, this option should only be used with care; it is primarily intended
-as a performance tuning aid during application development.
-.It Q
-Double/halve the size of the maximum size class that is a multiple of the
-quantum (8 or 16 bytes, depending on architecture).
-Above this size, cacheline spacing is used for size classes.
-The default value is 128 bytes.
-.It U
-Generate
-.Dq utrace
-entries for
-.Xr ktrace 1 ,
-for all operations.
-Consult the source for details on this option.
-.It V
-Attempting to allocate zero bytes will return a
-.Dv NULL
-pointer instead of a valid pointer.
-(The default behavior is to make a minimal allocation and return a
-pointer to it.)
-This option is provided for System V compatibility.
-This option is incompatible with the
-.Dq X
-option.
-.It X
-Rather than return failure for any allocation function, display a diagnostic
-message on
-.Dv STDERR_FILENO
-and cause the program to drop core (using
-.Xr abort 3 ) .
-This option should be set at compile time by including the following in the
-source code:
-.Bd -literal -offset indent
-_malloc_options = "X";
-.Ed
-.It Z
-Each byte of new memory allocated by
-.Fn malloc ,
-.Fn realloc ,
-or
-.Fn reallocf
-will be initialized to 0.
-Note that this initialization only happens once for each byte, so
-.Fn realloc
-and
-.Fn reallocf
-calls do not zero memory that was previously allocated.
-This is intended for debugging and will impact performance negatively.
-.El
-.Pp
-The
-.Dq J
-and
-.Dq Z
-options are intended for testing and debugging.
-An application which changes its behavior when these options are used
-is flawed.
-.Sh IMPLEMENTATION NOTES
-Traditionally, allocators have used
-.Xr sbrk 2
-to obtain memory, which is suboptimal for several reasons, including race
-conditions, increased fragmentation, and artificial limitations on maximum
-usable memory.
-This allocator uses both
-.Xr sbrk 2
-and
-.Xr mmap 2
-by default, but it can be configured at run time to use only one or the other.
-If resource limits are not a primary concern, the preferred configuration is
-.Ev MALLOC_OPTIONS=dM
-or
-.Ev MALLOC_OPTIONS=DM .
-When so configured, the
-.Ar datasize
-resource limit has little practical effect for typical applications; use
-.Ev MALLOC_OPTIONS=Dm
-if that is a concern.
-Regardless of allocator configuration, the
-.Ar vmemoryuse
-resource limit can be used to bound the total virtual memory used by a
-process, as described in
-.Xr limits 1 .
-.Pp
-This allocator uses multiple arenas in order to reduce lock contention for
-threaded programs on multi-processor systems.
-This works well with regard to threading scalability, but incurs some costs.
-There is a small fixed per-arena overhead, and additionally, arenas manage
-memory completely independently of each other, which means a small fixed
-increase in overall memory fragmentation.
-These overheads are not generally an issue, given the number of arenas normally
-used.
-Note that using substantially more arenas than the default is not likely to
-improve performance, mainly due to reduced cache performance.
-However, it may make sense to reduce the number of arenas if an application
-does not make much use of the allocation functions.
-.Pp
-In addition to multiple arenas, this allocator supports thread-specific caching
-for small and medium objects, in order to make it possible to completely avoid
-synchronization for most small and medium allocation requests.
-Such caching allows very fast allocation in the common case, but it increases
-memory usage and fragmentation, since a bounded number of objects can remain
-allocated in each thread cache.
-.Pp
-Memory is conceptually broken into equal-sized chunks, where the chunk size is
-a power of two that is greater than the page size.
-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.
-Small objects are smaller than one page.
-Medium objects range from one page to an upper limit determined at run time (see
-the
-.Dq E
-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
-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).
-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.
-Each run maintains a bitmap that tracks which regions are in use.
-Allocation requests that are no more than half the quantum (8 or 16, depending
-on architecture) are rounded up to the nearest power of two.
-Allocation requests that are more than half the quantum, but no more than the
-minimum cacheline-multiple size class (see the
-.Dq Q
-option) are rounded up to the nearest multiple of the quantum.
-Allocation requests that are more than the minimum cacheline-multiple size
-class, but no more than the minimum subpage-multiple size class (see the
-.Dq C
-option) are rounded up to the nearest multiple of the cacheline size (64).
-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
-.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
-rounded up to the nearest multiple of the chunk size.
-.Pp
-Allocations are packed tightly together, which can be an issue for
-multi-threaded applications.
-If you need to assure that allocations do not suffer from cacheline sharing,
-round your allocation requests up to the nearest multiple of the cacheline
-size.
-.Sh DEBUGGING MALLOC PROBLEMS
-The first thing to do is to set the
-.Dq A
-option.
-This option forces a coredump (if possible) at the first sign of trouble,
-rather than the normal policy of trying to continue if at all possible.
-.Pp
-It is probably also a good idea to recompile the program with suitable
-options and symbols for debugger support.
-.Pp
-If the program starts to give unusual results, coredump or generally behave
-differently without emitting any of the messages mentioned in the next
-section, it is likely because it depends on the storage being filled with
-zero bytes.
-Try running it with the
-.Dq Z
-option set;
-if that improves the situation, this diagnosis has been confirmed.
-If the program still misbehaves,
-the likely problem is accessing memory outside the allocated area.
-.Pp
-Alternatively, if the symptoms are not easy to reproduce, setting the
-.Dq J
-option may help provoke the problem.
-.Pp
-In truly difficult cases, the
-.Dq U
-option, if supported by the kernel, can provide a detailed trace of
-all calls made to these functions.
-.Pp
-Unfortunately this implementation does not provide much detail about
-the problems it detects; the performance impact for storing such information
-would be prohibitive.
-There are a number of allocator implementations available on the Internet
-which focus on detecting and pinpointing problems by trading performance for
-extra sanity checks and detailed diagnostics.
-.Sh DIAGNOSTIC MESSAGES
-If any of the memory allocation/deallocation functions detect an error or
-warning condition, a message will be printed to file descriptor
-.Dv STDERR_FILENO .
-Errors will result in the process dumping core.
-If the
-.Dq A
-option is set, all warnings are treated as errors.
-.Pp
-The
-.Va _malloc_message
-variable allows the programmer to override the function which emits the text
-strings forming the errors and warnings if for some reason the
-.Dv STDERR_FILENO
-file descriptor is not suitable for this.
-Please note that doing anything which tries to allocate memory in this function
-is likely to result in a crash or deadlock.
-.Pp
-All messages are prefixed by
-.Dq Ao Ar progname Ac Ns Li : (malloc) .
-.Sh RETURN VALUES
-The
-.Fn malloc
-and
-.Fn calloc
-functions return a pointer to the allocated memory if successful; otherwise
-a
-.Dv NULL
-pointer is returned and
-.Va errno
-is set to
-.Er ENOMEM .
-.Pp
-The
-.Fn realloc
-and
-.Fn reallocf
-functions return a pointer, possibly identical to
-.Fa ptr ,
-to the allocated memory
-if successful; otherwise a
-.Dv NULL
-pointer is returned, and
-.Va errno
-is set to
-.Er ENOMEM
-if the error was the result of an allocation failure.
-The
-.Fn realloc
-function always leaves the original buffer intact
-when an error occurs, whereas
-.Fn reallocf
-deallocates it in this case.
-.Pp
-The
-.Fn free
-function returns no value.
-.Pp
-The
-.Fn malloc_usable_size
-function returns the usable size of the allocation pointed to by
-.Fa ptr .
-.Sh ENVIRONMENT
-The following environment variables affect the execution of the allocation
-functions:
-.Bl -tag -width ".Ev MALLOC_OPTIONS"
-.It Ev MALLOC_OPTIONS
-If the environment variable
-.Ev MALLOC_OPTIONS
-is set, the characters it contains will be interpreted as flags to the
-allocation functions.
-.El
-.Sh EXAMPLES
-To dump core whenever a problem occurs:
-.Bd -literal -offset indent
-ln -s 'A' /etc/malloc.conf
-.Ed
-.Pp
-To specify in the source that a program does no return value checking
-on calls to these functions:
-.Bd -literal -offset indent
-_malloc_options = "X";
-.Ed
-.Sh SEE ALSO
-.Xr limits 1 ,
-.Xr madvise 2 ,
-.Xr mmap 2 ,
-.Xr sbrk 2 ,
-.Xr alloca 3 ,
-.Xr atexit 3 ,
-.Xr getpagesize 3 ,
-.Xr getpagesizes 3 ,
-.Xr memory 3 ,
-.Xr posix_memalign 3
-.Sh STANDARDS
-The
-.Fn malloc ,
-.Fn calloc ,
-.Fn realloc
-and
-.Fn free
-functions conform to
-.St -isoC .
-.Sh HISTORY
-The
-.Fn reallocf
-function first appeared in
-.Fx 3.0 .
-.Pp
-The
-.Fn malloc_usable_size
-function first appeared in
-.Fx 7.0 .

lib/libc/stdlib/reallocf.3

@@ -0,0 +1,82 @@
+.\" Copyright (c) 1980, 1991, 1993
+.\"	The Regents of the University of California.  All rights reserved.
+.\"
+.\" This code is derived from software contributed to Berkeley by
+.\" the American National Standards Committee X3, on Information
+.\" Processing Systems.
+.\"
+.\" Redistribution and use in source and binary forms, with or without
+.\" modification, are permitted provided that the following conditions
+.\" are met:
+.\" 1. Redistributions of source code must retain the above copyright
+.\"    notice, this list of conditions and the following disclaimer.
+.\" 2. Redistributions in binary form must reproduce the above copyright
+.\"    notice, this list of conditions and the following disclaimer in the
+.\"    documentation and/or other materials provided with the distribution.
+.\" 3. Neither the name of the University nor the names of its contributors
+.\"    may be used to endorse or promote products derived from this software
+.\"    without specific prior written permission.
+.\"
+.\" THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
+.\" ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+.\" IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
+.\" ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
+.\" FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
+.\" DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
+.\" OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
+.\" HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
+.\" LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
+.\" OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
+.\" SUCH DAMAGE.
+.\"
+.\"     @(#)malloc.3	8.1 (Berkeley) 6/4/93
+.\" $FreeBSD$
+.\"
+.Dd January 31, 2010
+.Dt MALLOC 3
+.Os
+.Sh NAME
+.Nm reallocf
+.Nd memory reallocation function
+.Sh LIBRARY
+.Lb libc
+.Sh SYNOPSIS
+.In stdlib.h
+.Ft void *
+.Fn reallocf "void *ptr" "size_t size"
+.Sh DESCRIPTION
+The
+.Fn reallocf
+function is identical to the
+.Fn realloc
+function, except that it
+will free the passed pointer when the requested memory cannot be allocated.
+This is a
+.Fx
+specific API designed to ease the problems with traditional coding styles
+for
+.Fn realloc
+causing memory leaks in libraries.
+.Sh RETURN VALUES
+The
+.Fn reallocf
+function returns a pointer, possibly identical to
+.Fa ptr ,
+to the allocated memory
+if successful; otherwise a
+.Dv NULL
+pointer is returned, and
+.Va errno
+is set to
+.Er ENOMEM
+if the error was the result of an allocation failure.
+The
+.Fn reallocf
+function deletes the original buffer when an error occurs.
+.Sh SEE ALSO
+.Xr realloc 3
+.Sh HISTORY
+The
+.Fn reallocf
+function first appeared in
+.Fx 3.0 .