Merge branch 'dev'

When configuring out-of-tree (source directory is not build directory),
the generated include files from the build directory should have higher
priority than those in the source dir.

This is especially helpful when cross-compiling.

Signed-off-by: Christoph Muellner <christoph.muellner@theobroma-systems.com>

.appveyor.yml

@@ -0,0 +1,42 @@
+version: '{build}'
+
+environment:
+  matrix:
+  - MSYSTEM: MINGW64
+    CPU: x86_64
+    MSVC: amd64
+  - MSYSTEM: MINGW32
+    CPU: i686
+    MSVC: x86
+  - MSYSTEM: MINGW64
+    CPU: x86_64
+  - MSYSTEM: MINGW32
+    CPU: i686
+  - MSYSTEM: MINGW64
+    CPU: x86_64
+    MSVC: amd64
+    CONFIG_FLAGS: --enable-debug
+  - MSYSTEM: MINGW32
+    CPU: i686
+    MSVC: x86
+    CONFIG_FLAGS: --enable-debug
+  - MSYSTEM: MINGW64
+    CPU: x86_64
+    CONFIG_FLAGS: --enable-debug
+  - MSYSTEM: MINGW32
+    CPU: i686
+    CONFIG_FLAGS: --enable-debug
+
+install:
+  - set PATH=c:\msys64\%MSYSTEM%\bin;c:\msys64\usr\bin;%PATH%
+  - if defined MSVC call "c:\Program Files (x86)\Microsoft Visual Studio 14.0\VC\vcvarsall.bat" %MSVC%
+  - if defined MSVC pacman --noconfirm -Rsc mingw-w64-%CPU%-gcc gcc
+  - pacman --noconfirm -Suy mingw-w64-%CPU%-make
+
+build_script:
+  - bash -c "autoconf"
+  - bash -c "./configure $CONFIG_FLAGS"
+  - mingw32-make
+  - file lib/jemalloc.dll
+  - mingw32-make tests
+  - mingw32-make -k check

.autom4te.cfg

@@ -0,0 +1,3 @@
+begin-language: "Autoconf-without-aclocal-m4"
+args: --no-cache
+end-language: "Autoconf-without-aclocal-m4"

.gitattributes

@@ -0,0 +1 @@
+* text=auto eol=lf

.gitignore

@@ -1,8 +1,6 @@
-/*.gcov.*
-
-/autom4te.cache/
-
+/bin/jemalloc-config
 /bin/jemalloc.sh
+/bin/jeprof
 
 /config.stamp
 /config.log
@@ -15,14 +13,20 @@
 /doc/jemalloc.html
 /doc/jemalloc.3
 
+/jemalloc.pc
+
 /lib/
 
 /Makefile
 
-/include/jemalloc/internal/jemalloc_internal.h
+/include/jemalloc/internal/jemalloc_preamble.h
 /include/jemalloc/internal/jemalloc_internal_defs.h
+/include/jemalloc/internal/private_namespace.gen.h
 /include/jemalloc/internal/private_namespace.h
-/include/jemalloc/internal/private_unnamespace.h
+/include/jemalloc/internal/private_namespace_jet.gen.h
+/include/jemalloc/internal/private_namespace_jet.h
+/include/jemalloc/internal/private_symbols.awk
+/include/jemalloc/internal/private_symbols_jet.awk
 /include/jemalloc/internal/public_namespace.h
 /include/jemalloc/internal/public_symbols.txt
 /include/jemalloc/internal/public_unnamespace.h
@@ -35,10 +39,12 @@
 /include/jemalloc/jemalloc_protos.h
 /include/jemalloc/jemalloc_protos_jet.h
 /include/jemalloc/jemalloc_rename.h
+/include/jemalloc/jemalloc_typedefs.h
 
 /src/*.[od]
-/src/*.gcda
-/src/*.gcno
+/src/*.sym
+
+/run_tests.out/
 
 /test/test.sh
 test/include/test/jemalloc_test.h
@@ -47,26 +53,41 @@ test/include/test/jemalloc_test_defs.h
 /test/integration/[A-Za-z]*
 !/test/integration/[A-Za-z]*.*
 /test/integration/*.[od]
-/test/integration/*.gcda
-/test/integration/*.gcno
 /test/integration/*.out
 
+/test/integration/cpp/[A-Za-z]*
+!/test/integration/cpp/[A-Za-z]*.*
+/test/integration/cpp/*.[od]
+/test/integration/cpp/*.out
+
 /test/src/*.[od]
-/test/src/*.gcda
-/test/src/*.gcno
 
 /test/stress/[A-Za-z]*
 !/test/stress/[A-Za-z]*.*
 /test/stress/*.[od]
-/test/stress/*.gcda
-/test/stress/*.gcno
 /test/stress/*.out
 
 /test/unit/[A-Za-z]*
 !/test/unit/[A-Za-z]*.*
 /test/unit/*.[od]
-/test/unit/*.gcda
-/test/unit/*.gcno
 /test/unit/*.out
 
 /VERSION
+
+*.pdb
+*.sdf
+*.opendb
+*.VC.db
+*.opensdf
+*.cachefile
+*.suo
+*.user
+*.sln.docstates
+*.tmp
+.vs/
+/msvc/Win32/
+/msvc/x64/
+/msvc/projects/*/*/Debug*/
+/msvc/projects/*/*/Release*/
+/msvc/projects/*/*/Win32/
+/msvc/projects/*/*/x64/

.travis.yml

@@ -0,0 +1,156 @@
+language: generic
+dist: precise
+
+matrix:
+  include:
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: osx
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=clang CXX=clang++ COMPILER_FLAGS="" CONFIGURE_FLAGS="" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="-m32" CONFIGURE_FLAGS="" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+      addons:
+        apt:
+          packages:
+            - gcc-multilib
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--enable-debug" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--enable-prof" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--disable-stats" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--with-malloc-conf=tcache:false" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--with-malloc-conf=dss:primary" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--with-malloc-conf=percpu_arena:percpu" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--with-malloc-conf=background_thread:true" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: osx
+      env: CC=clang CXX=clang++ COMPILER_FLAGS="" CONFIGURE_FLAGS="" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: osx
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="-m32" CONFIGURE_FLAGS="" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: osx
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--enable-debug" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: osx
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--disable-stats" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: osx
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--with-malloc-conf=tcache:false" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=clang CXX=clang++ COMPILER_FLAGS="-m32" CONFIGURE_FLAGS="" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+      addons:
+        apt:
+          packages:
+            - gcc-multilib
+    - os: linux
+      env: CC=clang CXX=clang++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--enable-debug" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=clang CXX=clang++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--enable-prof" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=clang CXX=clang++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--disable-stats" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=clang CXX=clang++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--with-malloc-conf=tcache:false" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=clang CXX=clang++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--with-malloc-conf=dss:primary" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=clang CXX=clang++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--with-malloc-conf=percpu_arena:percpu" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=clang CXX=clang++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--with-malloc-conf=background_thread:true" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="-m32" CONFIGURE_FLAGS="--enable-debug" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+      addons:
+        apt:
+          packages:
+            - gcc-multilib
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="-m32" CONFIGURE_FLAGS="--enable-prof" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+      addons:
+        apt:
+          packages:
+            - gcc-multilib
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="-m32" CONFIGURE_FLAGS="--disable-stats" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+      addons:
+        apt:
+          packages:
+            - gcc-multilib
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="-m32" CONFIGURE_FLAGS="--with-malloc-conf=tcache:false" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+      addons:
+        apt:
+          packages:
+            - gcc-multilib
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="-m32" CONFIGURE_FLAGS="--with-malloc-conf=dss:primary" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+      addons:
+        apt:
+          packages:
+            - gcc-multilib
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="-m32" CONFIGURE_FLAGS="--with-malloc-conf=percpu_arena:percpu" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+      addons:
+        apt:
+          packages:
+            - gcc-multilib
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="-m32" CONFIGURE_FLAGS="--with-malloc-conf=background_thread:true" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+      addons:
+        apt:
+          packages:
+            - gcc-multilib
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--enable-debug --enable-prof" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--enable-debug --disable-stats" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--enable-debug --with-malloc-conf=tcache:false" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--enable-debug --with-malloc-conf=dss:primary" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--enable-debug --with-malloc-conf=percpu_arena:percpu" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--enable-debug --with-malloc-conf=background_thread:true" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--enable-prof --disable-stats" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--enable-prof --with-malloc-conf=tcache:false" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--enable-prof --with-malloc-conf=dss:primary" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--enable-prof --with-malloc-conf=percpu_arena:percpu" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--enable-prof --with-malloc-conf=background_thread:true" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--disable-stats --with-malloc-conf=tcache:false" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--disable-stats --with-malloc-conf=dss:primary" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--disable-stats --with-malloc-conf=percpu_arena:percpu" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--disable-stats --with-malloc-conf=background_thread:true" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--with-malloc-conf=tcache:false,dss:primary" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--with-malloc-conf=tcache:false,percpu_arena:percpu" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--with-malloc-conf=tcache:false,background_thread:true" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--with-malloc-conf=dss:primary,percpu_arena:percpu" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--with-malloc-conf=dss:primary,background_thread:true" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+    - os: linux
+      env: CC=gcc CXX=g++ COMPILER_FLAGS="" CONFIGURE_FLAGS="--with-malloc-conf=percpu_arena:percpu,background_thread:true" EXTRA_CFLAGS="-Werror -Wno-array-bounds"
+
+
+before_script:
+  - autoconf
+  - ./configure ${COMPILER_FLAGS:+       CC="$CC $COMPILER_FLAGS"       CXX="$CXX $COMPILER_FLAGS" }       $CONFIGURE_FLAGS
+  - make -j3
+  - make -j3 tests
+
+script:
+  - make check
+

COPYING

@@ -1,10 +1,10 @@
 Unless otherwise specified, files in the jemalloc source distribution are
 subject to the following license:
 --------------------------------------------------------------------------------
-Copyright (C) 2002-2014 Jason Evans <jasone@canonware.com>.
+Copyright (C) 2002-2018 Jason Evans <jasone@canonware.com>.
 All rights reserved.
 Copyright (C) 2007-2012 Mozilla Foundation.  All rights reserved.
-Copyright (C) 2009-2014 Facebook, Inc.  All rights reserved.
+Copyright (C) 2009-2018 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:

ChangeLog

@@ -1,10 +1,844 @@
 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:
+bug fixes are all mentioned, but some internal enhancements are omitted here for
+brevity.  Much more detail can be found in the git revision history:
 
     https://github.com/jemalloc/jemalloc
 
+* 5.1.0 (May 4th, 2018)
+
+  This release is primarily about fine-tuning, ranging from several new features
+  to numerous notable performance and portability enhancements.  The release and
+  prior dev versions have been running in multiple large scale applications for
+  months, and the cumulative improvements are substantial in many cases.
+
+  Given the long and successful production runs, this release is likely a good
+  candidate for applications to upgrade, from both jemalloc 5.0 and before.  For
+  performance-critical applications, the newly added TUNING.md provides
+  guidelines on jemalloc tuning.
+
+  New features:
+  - Implement transparent huge page support for internal metadata.  (@interwq)
+  - Add opt.thp to allow enabling / disabling transparent huge pages for all
+    mappings.  (@interwq)
+  - Add maximum background thread count option.  (@djwatson)
+  - Allow prof_active to control opt.lg_prof_interval and prof.gdump.
+    (@interwq)
+  - Allow arena index lookup based on allocation addresses via mallctl.
+    (@lionkov)
+  - Allow disabling initial-exec TLS model.  (@davidtgoldblatt, @KenMacD)
+  - Add opt.lg_extent_max_active_fit to set the max ratio between the size of
+    the active extent selected (to split off from) and the size of the requested
+    allocation.  (@interwq, @davidtgoldblatt)
+  - Add retain_grow_limit to set the max size when growing virtual address
+    space.  (@interwq)
+  - Add mallctl interfaces:
+    + arena.<i>.retain_grow_limit  (@interwq)
+    + arenas.lookup  (@lionkov)
+    + max_background_threads  (@djwatson)
+    + opt.lg_extent_max_active_fit  (@interwq)
+    + opt.max_background_threads  (@djwatson)
+    + opt.metadata_thp  (@interwq)
+    + opt.thp  (@interwq)
+    + stats.metadata_thp  (@interwq)
+
+  Portability improvements:
+  - Support GNU/kFreeBSD configuration.  (@paravoid)
+  - Support m68k, nios2 and SH3 architectures.  (@paravoid)
+  - Fall back to FD_CLOEXEC when O_CLOEXEC is unavailable.  (@zonyitoo)
+  - Fix symbol listing for cross-compiling.  (@tamird)
+  - Fix high bits computation on ARM.  (@davidtgoldblatt, @paravoid)
+  - Disable the CPU_SPINWAIT macro for Power.  (@davidtgoldblatt, @marxin)
+  - Fix MSVC 2015 & 2017 builds.  (@rustyx)
+  - Improve RISC-V support.  (@EdSchouten)
+  - Set name mangling script in strict mode.  (@nicolov)
+  - Avoid MADV_HUGEPAGE on ARM.  (@marxin)
+  - Modify configure to determine return value of strerror_r.
+    (@davidtgoldblatt, @cferris1000)
+  - Make sure CXXFLAGS is tested with CPP compiler.  (@nehaljwani)
+  - Fix 32-bit build on MSVC.  (@rustyx)
+  - Fix external symbol on MSVC.  (@maksqwe)
+  - Avoid a printf format specifier warning.  (@jasone)
+  - Add configure option --disable-initial-exec-tls which can allow jemalloc to
+    be dynamically loaded after program startup.  (@davidtgoldblatt, @KenMacD)
+  - AArch64: Add ILP32 support.  (@cmuellner)
+  - Add --with-lg-vaddr configure option to support cross compiling.
+    (@cmuellner, @davidtgoldblatt)
+
+  Optimizations and refactors:
+  - Improve active extent fit with extent_max_active_fit.  This considerably
+    reduces fragmentation over time and improves virtual memory and metadata
+    usage.  (@davidtgoldblatt, @interwq)
+  - Eagerly coalesce large extents to reduce fragmentation.  (@interwq)
+  - sdallocx: only read size info when page aligned (i.e. possibly sampled),
+    which speeds up the sized deallocation path significantly.  (@interwq)
+  - Avoid attempting new mappings for in place expansion with retain, since
+    it rarely succeeds in practice and causes high overhead.  (@interwq)
+  - Refactor OOM handling in newImpl.  (@wqfish)
+  - Add internal fine-grained logging functionality for debugging use.
+    (@davidtgoldblatt)
+  - Refactor arena / tcache interactions.  (@davidtgoldblatt)
+  - Refactor extent management with dumpable flag.  (@davidtgoldblatt)
+  - Add runtime detection of lazy purging.  (@interwq)
+  - Use pairing heap instead of red-black tree for extents_avail.  (@djwatson)
+  - Use sysctl on startup in FreeBSD.  (@trasz)
+  - Use thread local prng state instead of atomic.  (@djwatson)
+  - Make decay to always purge one more extent than before, because in
+    practice large extents are usually the ones that cross the decay threshold.
+    Purging the additional extent helps save memory as well as reduce VM
+    fragmentation.  (@interwq)
+  - Fast division by dynamic values.  (@davidtgoldblatt)
+  - Improve the fit for aligned allocation.  (@interwq, @edwinsmith)
+  - Refactor extent_t bitpacking.  (@rkmisra)
+  - Optimize the generated assembly for ticker operations.  (@davidtgoldblatt)
+  - Convert stats printing to use a structured text emitter.  (@davidtgoldblatt)
+  - Remove preserve_lru feature for extents management.  (@djwatson)
+  - Consolidate two memory loads into one on the fast deallocation path.
+    (@davidtgoldblatt, @interwq)
+
+  Bug fixes (most of the issues are only relevant to jemalloc 5.0):
+  - Fix deadlock with multithreaded fork in OS X.  (@davidtgoldblatt)
+  - Validate returned file descriptor before use.  (@zonyitoo)
+  - Fix a few background thread initialization and shutdown issues.  (@interwq)
+  - Fix an extent coalesce + decay race by taking both coalescing extents off
+    the LRU list.  (@interwq)
+  - Fix potentially unbound increase during decay, caused by one thread keep
+    stashing memory to purge while other threads generating new pages.  The
+    number of pages to purge is checked to prevent this.  (@interwq)
+  - Fix a FreeBSD bootstrap assertion.  (@strejda, @interwq)
+  - Handle 32 bit mutex counters.  (@rkmisra)
+  - Fix a indexing bug when creating background threads.  (@davidtgoldblatt,
+    @binliu19)
+  - Fix arguments passed to extent_init.  (@yuleniwo, @interwq)
+  - Fix addresses used for ordering mutexes.  (@rkmisra)
+  - Fix abort_conf processing during bootstrap.  (@interwq)
+  - Fix include path order for out-of-tree builds.  (@cmuellner)
+
+  Incompatible changes:
+  - Remove --disable-thp.  (@interwq)
+  - Remove mallctl interfaces:
+    + config.thp  (@interwq)
+
+  Documentation:
+  - Add TUNING.md.  (@interwq, @davidtgoldblatt, @djwatson)
+
+* 5.0.1 (July 1, 2017)
+
+  This bugfix release fixes several issues, most of which are obscure enough
+  that typical applications are not impacted.
+
+  Bug fixes:
+  - Update decay->nunpurged before purging, in order to avoid potential update
+    races and subsequent incorrect purging volume.  (@interwq)
+  - Only abort on dlsym(3) error if the failure impacts an enabled feature (lazy
+    locking and/or background threads).  This mitigates an initialization
+    failure bug for which we still do not have a clear reproduction test case.
+    (@interwq)
+  - Modify tsd management so that it neither crashes nor leaks if a thread's
+    only allocation activity is to call free() after TLS destructors have been
+    executed.  This behavior was observed when operating with GNU libc, and is
+    unlikely to be an issue with other libc implementations.  (@interwq)
+  - Mask signals during background thread creation.  This prevents signals from
+    being inadvertently delivered to background threads.  (@jasone,
+    @davidtgoldblatt, @interwq)
+  - Avoid inactivity checks within background threads, in order to prevent
+    recursive mutex acquisition.  (@interwq)
+  - Fix extent_grow_retained() to use the specified hooks when the
+    arena.<i>.extent_hooks mallctl is used to override the default hooks.
+    (@interwq)
+  - Add missing reentrancy support for custom extent hooks which allocate.
+    (@interwq)
+  - Post-fork(2), re-initialize the list of tcaches associated with each arena
+    to contain no tcaches except the forking thread's.  (@interwq)
+  - Add missing post-fork(2) mutex reinitialization for extent_grow_mtx.  This
+    fixes potential deadlocks after fork(2).  (@interwq)
+  - Enforce minimum autoconf version (currently 2.68), since 2.63 is known to
+    generate corrupt configure scripts.  (@jasone)
+  - Ensure that the configured page size (--with-lg-page) is no larger than the
+    configured huge page size (--with-lg-hugepage).  (@jasone)
+
+* 5.0.0 (June 13, 2017)
+
+  Unlike all previous jemalloc releases, this release does not use naturally
+  aligned "chunks" for virtual memory management, and instead uses page-aligned
+  "extents".  This change has few externally visible effects, but the internal
+  impacts are... extensive.  Many other internal changes combine to make this
+  the most cohesively designed version of jemalloc so far, with ample
+  opportunity for further enhancements.
+
+  Continuous integration is now an integral aspect of development thanks to the
+  efforts of @davidtgoldblatt, and the dev branch tends to remain reasonably
+  stable on the tested platforms (Linux, FreeBSD, macOS, and Windows).  As a
+  side effect the official release frequency may decrease over time.
+
+  New features:
+  - Implement optional per-CPU arena support; threads choose which arena to use
+    based on current CPU rather than on fixed thread-->arena associations.
+    (@interwq)
+  - Implement two-phase decay of unused dirty pages.  Pages transition from
+    dirty-->muzzy-->clean, where the first phase transition relies on
+    madvise(... MADV_FREE) semantics, and the second phase transition discards
+    pages such that they are replaced with demand-zeroed pages on next access.
+    (@jasone)
+  - Increase decay time resolution from seconds to milliseconds.  (@jasone)
+  - Implement opt-in per CPU background threads, and use them for asynchronous
+    decay-driven unused dirty page purging.  (@interwq)
+  - Add mutex profiling, which collects a variety of statistics useful for
+    diagnosing overhead/contention issues.  (@interwq)
+  - Add C++ new/delete operator bindings.  (@djwatson)
+  - Support manually created arena destruction, such that all data and metadata
+    are discarded.  Add MALLCTL_ARENAS_DESTROYED for accessing merged stats
+    associated with destroyed arenas.  (@jasone)
+  - Add MALLCTL_ARENAS_ALL as a fixed index for use in accessing
+    merged/destroyed arena statistics via mallctl.  (@jasone)
+  - Add opt.abort_conf to optionally abort if invalid configuration options are
+    detected during initialization.  (@interwq)
+  - Add opt.stats_print_opts, so that e.g. JSON output can be selected for the
+    stats dumped during exit if opt.stats_print is true.  (@jasone)
+  - Add --with-version=VERSION for use when embedding jemalloc into another
+    project's git repository.  (@jasone)
+  - Add --disable-thp to support cross compiling.  (@jasone)
+  - Add --with-lg-hugepage to support cross compiling.  (@jasone)
+  - Add mallctl interfaces (various authors):
+    + background_thread
+    + opt.abort_conf
+    + opt.retain
+    + opt.percpu_arena
+    + opt.background_thread
+    + opt.{dirty,muzzy}_decay_ms
+    + opt.stats_print_opts
+    + arena.<i>.initialized
+    + arena.<i>.destroy
+    + arena.<i>.{dirty,muzzy}_decay_ms
+    + arena.<i>.extent_hooks
+    + arenas.{dirty,muzzy}_decay_ms
+    + arenas.bin.<i>.slab_size
+    + arenas.nlextents
+    + arenas.lextent.<i>.size
+    + arenas.create
+    + stats.background_thread.{num_threads,num_runs,run_interval}
+    + stats.mutexes.{ctl,background_thread,prof,reset}.
+      {num_ops,num_spin_acq,num_wait,max_wait_time,total_wait_time,max_num_thds,
+      num_owner_switch}
+    + stats.arenas.<i>.{dirty,muzzy}_decay_ms
+    + stats.arenas.<i>.uptime
+    + stats.arenas.<i>.{pmuzzy,base,internal,resident}
+    + stats.arenas.<i>.{dirty,muzzy}_{npurge,nmadvise,purged}
+    + stats.arenas.<i>.bins.<j>.{nslabs,reslabs,curslabs}
+    + stats.arenas.<i>.bins.<j>.mutex.
+      {num_ops,num_spin_acq,num_wait,max_wait_time,total_wait_time,max_num_thds,
+      num_owner_switch}
+    + stats.arenas.<i>.lextents.<j>.{nmalloc,ndalloc,nrequests,curlextents}
+    + stats.arenas.i.mutexes.{large,extent_avail,extents_dirty,extents_muzzy,
+      extents_retained,decay_dirty,decay_muzzy,base,tcache_list}.
+      {num_ops,num_spin_acq,num_wait,max_wait_time,total_wait_time,max_num_thds,
+      num_owner_switch}
+
+  Portability improvements:
+  - Improve reentrant allocation support, such that deadlock is less likely if
+    e.g. a system library call in turn allocates memory.  (@davidtgoldblatt,
+    @interwq)
+  - Support static linking of jemalloc with glibc.  (@djwatson)
+
+  Optimizations and refactors:
+  - Organize virtual memory as "extents" of virtual memory pages, rather than as
+    naturally aligned "chunks", and store all metadata in arbitrarily distant
+    locations.  This reduces virtual memory external fragmentation, and will
+    interact better with huge pages (not yet explicitly supported).  (@jasone)
+  - Fold large and huge size classes together; only small and large size classes
+    remain.  (@jasone)
+  - Unify the allocation paths, and merge most fast-path branching decisions.
+    (@davidtgoldblatt, @interwq)
+  - Embed per thread automatic tcache into thread-specific data, which reduces
+    conditional branches and dereferences.  Also reorganize tcache to increase
+    fast-path data locality.  (@interwq)
+  - Rewrite atomics to closely model the C11 API, convert various
+    synchronization from mutex-based to atomic, and use the explicit memory
+    ordering control to resolve various hypothetical races without increasing
+    synchronization overhead.  (@davidtgoldblatt)
+  - Extensively optimize rtree via various methods:
+    + Add multiple layers of rtree lookup caching, since rtree lookups are now
+      part of fast-path deallocation.  (@interwq)
+    + Determine rtree layout at compile time.  (@jasone)
+    + Make the tree shallower for common configurations.  (@jasone)
+    + Embed the root node in the top-level rtree data structure, thus avoiding
+      one level of indirection.  (@jasone)
+    + Further specialize leaf elements as compared to internal node elements,
+      and directly embed extent metadata needed for fast-path deallocation.
+      (@jasone)
+    + Ignore leading always-zero address bits (architecture-specific).
+      (@jasone)
+  - Reorganize headers (ongoing work) to make them hermetic, and disentangle
+    various module dependencies.  (@davidtgoldblatt)
+  - Convert various internal data structures such as size class metadata from
+    boot-time-initialized to compile-time-initialized.  Propagate resulting data
+    structure simplifications, such as making arena metadata fixed-size.
+    (@jasone)
+  - Simplify size class lookups when constrained to size classes that are
+    multiples of the page size.  This speeds lookups, but the primary benefit is
+    complexity reduction in code that was the source of numerous regressions.
+    (@jasone)
+  - Lock individual extents when possible for localized extent operations,
+    rather than relying on a top-level arena lock.  (@davidtgoldblatt, @jasone)
+  - Use first fit layout policy instead of best fit, in order to improve
+    packing.  (@jasone)
+  - If munmap(2) is not in use, use an exponential series to grow each arena's
+    virtual memory, so that the number of disjoint virtual memory mappings
+    remains low.  (@jasone)
+  - Implement per arena base allocators, so that arenas never share any virtual
+    memory pages.  (@jasone)
+  - Automatically generate private symbol name mangling macros.  (@jasone)
+
+  Incompatible changes:
+  - Replace chunk hooks with an expanded/normalized set of extent hooks.
+    (@jasone)
+  - Remove ratio-based purging.  (@jasone)
+  - Remove --disable-tcache.  (@jasone)
+  - Remove --disable-tls.  (@jasone)
+  - Remove --enable-ivsalloc.  (@jasone)
+  - Remove --with-lg-size-class-group.  (@jasone)
+  - Remove --with-lg-tiny-min.  (@jasone)
+  - Remove --disable-cc-silence.  (@jasone)
+  - Remove --enable-code-coverage.  (@jasone)
+  - Remove --disable-munmap (replaced by opt.retain).  (@jasone)
+  - Remove Valgrind support.  (@jasone)
+  - Remove quarantine support.  (@jasone)
+  - Remove redzone support.  (@jasone)
+  - Remove mallctl interfaces (various authors):
+    + config.munmap
+    + config.tcache
+    + config.tls
+    + config.valgrind
+    + opt.lg_chunk
+    + opt.purge
+    + opt.lg_dirty_mult
+    + opt.decay_time
+    + opt.quarantine
+    + opt.redzone
+    + opt.thp
+    + arena.<i>.lg_dirty_mult
+    + arena.<i>.decay_time
+    + arena.<i>.chunk_hooks
+    + arenas.initialized
+    + arenas.lg_dirty_mult
+    + arenas.decay_time
+    + arenas.bin.<i>.run_size
+    + arenas.nlruns
+    + arenas.lrun.<i>.size
+    + arenas.nhchunks
+    + arenas.hchunk.<i>.size
+    + arenas.extend
+    + stats.cactive
+    + stats.arenas.<i>.lg_dirty_mult
+    + stats.arenas.<i>.decay_time
+    + stats.arenas.<i>.metadata.{mapped,allocated}
+    + stats.arenas.<i>.{npurge,nmadvise,purged}
+    + stats.arenas.<i>.huge.{allocated,nmalloc,ndalloc,nrequests}
+    + stats.arenas.<i>.bins.<j>.{nruns,reruns,curruns}
+    + stats.arenas.<i>.lruns.<j>.{nmalloc,ndalloc,nrequests,curruns}
+    + stats.arenas.<i>.hchunks.<j>.{nmalloc,ndalloc,nrequests,curhchunks}
+
+  Bug fixes:
+  - Improve interval-based profile dump triggering to dump only one profile when
+    a single allocation's size exceeds the interval.  (@jasone)
+  - Use prefixed function names (as controlled by --with-jemalloc-prefix) when
+    pruning backtrace frames in jeprof.  (@jasone)
+
+* 4.5.0 (February 28, 2017)
+
+  This is the first release to benefit from much broader continuous integration
+  testing, thanks to @davidtgoldblatt.  Had we had this testing infrastructure
+  in place for prior releases, it would have caught all of the most serious
+  regressions fixed by this release.
+
+  New features:
+  - Add --disable-thp and the opt.thp mallctl to provide opt-out mechanisms for
+    transparent huge page integration.  (@jasone)
+  - Update zone allocator integration to work with macOS 10.12.  (@glandium)
+  - Restructure *CFLAGS configuration, so that CFLAGS behaves typically, and
+    EXTRA_CFLAGS provides a way to specify e.g. -Werror during building, but not
+    during configuration.  (@jasone, @ronawho)
+
+  Bug fixes:
+  - Fix DSS (sbrk(2)-based) allocation.  This regression was first released in
+    4.3.0.  (@jasone)
+  - Handle race in per size class utilization computation.  This functionality
+    was first released in 4.0.0.  (@interwq)
+  - Fix lock order reversal during gdump.  (@jasone)
+  - Fix/refactor tcache synchronization.  This regression was first released in
+    4.0.0.  (@jasone)
+  - Fix various JSON-formatted malloc_stats_print() bugs.  This functionality
+    was first released in 4.3.0.  (@jasone)
+  - Fix huge-aligned allocation.  This regression was first released in 4.4.0.
+    (@jasone)
+  - When transparent huge page integration is enabled, detect what state pages
+    start in according to the kernel's current operating mode, and only convert
+    arena chunks to non-huge during purging if that is not their initial state.
+    This functionality was first released in 4.4.0.  (@jasone)
+  - Fix lg_chunk clamping for the --enable-cache-oblivious --disable-fill case.
+    This regression was first released in 4.0.0.  (@jasone, @428desmo)
+  - Properly detect sparc64 when building for Linux.  (@glaubitz)
+
+* 4.4.0 (December 3, 2016)
+
+  New features:
+  - Add configure support for *-*-linux-android.  (@cferris1000, @jasone)
+  - Add the --disable-syscall configure option, for use on systems that place
+    security-motivated limitations on syscall(2).  (@jasone)
+  - Add support for Debian GNU/kFreeBSD.  (@thesam)
+
+  Optimizations:
+  - Add extent serial numbers and use them where appropriate as a sort key that
+    is higher priority than address, so that the allocation policy prefers older
+    extents.  This tends to improve locality (decrease fragmentation) when
+    memory grows downward.  (@jasone)
+  - Refactor madvise(2) configuration so that MADV_FREE is detected and utilized
+    on Linux 4.5 and newer.  (@jasone)
+  - Mark partially purged arena chunks as non-huge-page.  This improves
+    interaction with Linux's transparent huge page functionality.  (@jasone)
+
+  Bug fixes:
+  - Fix size class computations for edge conditions involving extremely large
+    allocations.  This regression was first released in 4.0.0.  (@jasone,
+    @ingvarha)
+  - Remove overly restrictive assertions related to the cactive statistic.  This
+    regression was first released in 4.1.0.  (@jasone)
+  - Implement a more reliable detection scheme for os_unfair_lock on macOS.
+    (@jszakmeister)
+
+* 4.3.1 (November 7, 2016)
+
+  Bug fixes:
+  - Fix a severe virtual memory leak.  This regression was first released in
+    4.3.0.  (@interwq, @jasone)
+  - Refactor atomic and prng APIs to restore support for 32-bit platforms that
+    use pre-C11 toolchains, e.g. FreeBSD's mips.  (@jasone)
+
+* 4.3.0 (November 4, 2016)
+
+  This is the first release that passes the test suite for multiple Windows
+  configurations, thanks in large part to @glandium setting up continuous
+  integration via AppVeyor (and Travis CI for Linux and OS X).
+
+  New features:
+  - Add "J" (JSON) support to malloc_stats_print().  (@jasone)
+  - Add Cray compiler support.  (@ronawho)
+
+  Optimizations:
+  - Add/use adaptive spinning for bootstrapping and radix tree node
+    initialization.  (@jasone)
+
+  Bug fixes:
+  - Fix large allocation to search starting in the optimal size class heap,
+    which can substantially reduce virtual memory churn and fragmentation.  This
+    regression was first released in 4.0.0.  (@mjp41, @jasone)
+  - Fix stats.arenas.<i>.nthreads accounting.  (@interwq)
+  - Fix and simplify decay-based purging.  (@jasone)
+  - Make DSS (sbrk(2)-related) operations lockless, which resolves potential
+    deadlocks during thread exit.  (@jasone)
+  - Fix over-sized allocation of radix tree leaf nodes.  (@mjp41, @ogaun,
+    @jasone)
+  - Fix over-sized allocation of arena_t (plus associated stats) data
+    structures.  (@jasone, @interwq)
+  - Fix EXTRA_CFLAGS to not affect configuration.  (@jasone)
+  - Fix a Valgrind integration bug.  (@ronawho)
+  - Disallow 0x5a junk filling when running in Valgrind.  (@jasone)
+  - Fix a file descriptor leak on Linux.  This regression was first released in
+    4.2.0.  (@vsarunas, @jasone)
+  - Fix static linking of jemalloc with glibc.  (@djwatson)
+  - Use syscall(2) rather than {open,read,close}(2) during boot on Linux.  This
+    works around other libraries' system call wrappers performing reentrant
+    allocation.  (@kspinka, @Whissi, @jasone)
+  - Fix OS X default zone replacement to work with OS X 10.12.  (@glandium,
+    @jasone)
+  - Fix cached memory management to avoid needless commit/decommit operations
+    during purging, which resolves permanent virtual memory map fragmentation
+    issues on Windows.  (@mjp41, @jasone)
+  - Fix TSD fetches to avoid (recursive) allocation.  This is relevant to
+    non-TLS and Windows configurations.  (@jasone)
+  - Fix malloc_conf overriding to work on Windows.  (@jasone)
+  - Forcibly disable lazy-lock on Windows (was forcibly *enabled*).  (@jasone)
+
+* 4.2.1 (June 8, 2016)
+
+  Bug fixes:
+  - Fix bootstrapping issues for configurations that require allocation during
+    tsd initialization (e.g. --disable-tls).  (@cferris1000, @jasone)
+  - Fix gettimeofday() version of nstime_update().  (@ronawho)
+  - Fix Valgrind regressions in calloc() and chunk_alloc_wrapper().  (@ronawho)
+  - Fix potential VM map fragmentation regression.  (@jasone)
+  - Fix opt_zero-triggered in-place huge reallocation zeroing.  (@jasone)
+  - Fix heap profiling context leaks in reallocation edge cases.  (@jasone)
+
+* 4.2.0 (May 12, 2016)
+
+  New features:
+  - Add the arena.<i>.reset mallctl, which makes it possible to discard all of
+    an arena's allocations in a single operation.  (@jasone)
+  - Add the stats.retained and stats.arenas.<i>.retained statistics.  (@jasone)
+  - Add the --with-version configure option.  (@jasone)
+  - Support --with-lg-page values larger than actual page size.  (@jasone)
+
+  Optimizations:
+  - Use pairing heaps rather than red-black trees for various hot data
+    structures.  (@djwatson, @jasone)
+  - Streamline fast paths of rtree operations.  (@jasone)
+  - Optimize the fast paths of calloc() and [m,d,sd]allocx().  (@jasone)
+  - Decommit unused virtual memory if the OS does not overcommit.  (@jasone)
+  - Specify MAP_NORESERVE on Linux if [heuristic] overcommit is active, in order
+    to avoid unfortunate interactions during fork(2).  (@jasone)
+
+  Bug fixes:
+  - Fix chunk accounting related to triggering gdump profiles.  (@jasone)
+  - Link against librt for clock_gettime(2) if glibc < 2.17.  (@jasone)
+  - Scale leak report summary according to sampling probability.  (@jasone)
+
+* 4.1.1 (May 3, 2016)
+
+  This bugfix release resolves a variety of mostly minor issues, though the
+  bitmap fix is critical for 64-bit Windows.
+
+  Bug fixes:
+  - Fix the linear scan version of bitmap_sfu() to shift by the proper amount
+    even when sizeof(long) is not the same as sizeof(void *), as on 64-bit
+    Windows.  (@jasone)
+  - Fix hashing functions to avoid unaligned memory accesses (and resulting
+    crashes).  This is relevant at least to some ARM-based platforms.
+    (@rkmisra)
+  - Fix fork()-related lock rank ordering reversals.  These reversals were
+    unlikely to cause deadlocks in practice except when heap profiling was
+    enabled and active.  (@jasone)
+  - Fix various chunk leaks in OOM code paths.  (@jasone)
+  - Fix malloc_stats_print() to print opt.narenas correctly.  (@jasone)
+  - Fix MSVC-specific build/test issues.  (@rustyx, @yuslepukhin)
+  - Fix a variety of test failures that were due to test fragility rather than
+    core bugs.  (@jasone)
+
+* 4.1.0 (February 28, 2016)
+
+  This release is primarily about optimizations, but it also incorporates a lot
+  of portability-motivated refactoring and enhancements.  Many people worked on
+  this release, to an extent that even with the omission here of minor changes
+  (see git revision history), and of the people who reported and diagnosed
+  issues, so much of the work was contributed that starting with this release,
+  changes are annotated with author credits to help reflect the collaborative
+  effort involved.
+
+  New features:
+  - Implement decay-based unused dirty page purging, a major optimization with
+    mallctl API impact.  This is an alternative to the existing ratio-based
+    unused dirty page purging, and is intended to eventually become the sole
+    purging mechanism.  New mallctls:
+    + opt.purge
+    + opt.decay_time
+    + arena.<i>.decay
+    + arena.<i>.decay_time
+    + arenas.decay_time
+    + stats.arenas.<i>.decay_time
+    (@jasone, @cevans87)
+  - Add --with-malloc-conf, which makes it possible to embed a default
+    options string during configuration.  This was motivated by the desire to
+    specify --with-malloc-conf=purge:decay , since the default must remain
+    purge:ratio until the 5.0.0 release.  (@jasone)
+  - Add MS Visual Studio 2015 support.  (@rustyx, @yuslepukhin)
+  - Make *allocx() size class overflow behavior defined.  The maximum
+    size class is now less than PTRDIFF_MAX to protect applications against
+    numerical overflow, and all allocation functions are guaranteed to indicate
+    errors rather than potentially crashing if the request size exceeds the
+    maximum size class.  (@jasone)
+  - jeprof:
+    + Add raw heap profile support.  (@jasone)
+    + Add --retain and --exclude for backtrace symbol filtering.  (@jasone)
+
+  Optimizations:
+  - Optimize the fast path to combine various bootstrapping and configuration
+    checks and execute more streamlined code in the common case.  (@interwq)
+  - Use linear scan for small bitmaps (used for small object tracking).  In
+    addition to speeding up bitmap operations on 64-bit systems, this reduces
+    allocator metadata overhead by approximately 0.2%.  (@djwatson)
+  - Separate arena_avail trees, which substantially speeds up run tree
+    operations.  (@djwatson)
+  - Use memoization (boot-time-computed table) for run quantization.  Separate
+    arena_avail trees reduced the importance of this optimization.  (@jasone)
+  - Attempt mmap-based in-place huge reallocation.  This can dramatically speed
+    up incremental huge reallocation.  (@jasone)
+
+  Incompatible changes:
+  - Make opt.narenas unsigned rather than size_t.  (@jasone)
+
+  Bug fixes:
+  - Fix stats.cactive accounting regression.  (@rustyx, @jasone)
+  - Handle unaligned keys in hash().  This caused problems for some ARM systems.
+    (@jasone, @cferris1000)
+  - Refactor arenas array.  In addition to fixing a fork-related deadlock, this
+    makes arena lookups faster and simpler.  (@jasone)
+  - Move retained memory allocation out of the default chunk allocation
+    function, to a location that gets executed even if the application installs
+    a custom chunk allocation function.  This resolves a virtual memory leak.
+    (@buchgr)
+  - Fix a potential tsd cleanup leak.  (@cferris1000, @jasone)
+  - Fix run quantization.  In practice this bug had no impact unless
+    applications requested memory with alignment exceeding one page.
+    (@jasone, @djwatson)
+  - Fix LinuxThreads-specific bootstrapping deadlock.  (Cosmin Paraschiv)
+  - jeprof:
+    + Don't discard curl options if timeout is not defined.  (@djwatson)
+    + Detect failed profile fetches.  (@djwatson)
+  - Fix stats.arenas.<i>.{dss,lg_dirty_mult,decay_time,pactive,pdirty} for
+    --disable-stats case.  (@jasone)
+
+* 4.0.4 (October 24, 2015)
+
+  This bugfix release fixes another xallocx() regression.  No other regressions
+  have come to light in over a month, so this is likely a good starting point
+  for people who prefer to wait for "dot one" releases with all the major issues
+  shaken out.
+
+  Bug fixes:
+  - Fix xallocx(..., MALLOCX_ZERO to zero the last full trailing page of large
+    allocations that have been randomly assigned an offset of 0 when
+    --enable-cache-oblivious configure option is enabled.
+
+* 4.0.3 (September 24, 2015)
+
+  This bugfix release continues the trend of xallocx() and heap profiling fixes.
+
+  Bug fixes:
+  - Fix xallocx(..., MALLOCX_ZERO) to zero all trailing bytes of large
+    allocations when --enable-cache-oblivious configure option is enabled.
+  - Fix xallocx(..., MALLOCX_ZERO) to zero trailing bytes of huge allocations
+    when resizing from/to a size class that is not a multiple of the chunk size.
+  - Fix prof_tctx_dump_iter() to filter out nodes that were created after heap
+    profile dumping started.
+  - Work around a potentially bad thread-specific data initialization
+    interaction with NPTL (glibc's pthreads implementation).
+
+* 4.0.2 (September 21, 2015)
+
+  This bugfix release addresses a few bugs specific to heap profiling.
+
+  Bug fixes:
+  - Fix ixallocx_prof_sample() to never modify nor create sampled small
+    allocations.  xallocx() is in general incapable of moving small allocations,
+    so this fix removes buggy code without loss of generality.
+  - Fix irallocx_prof_sample() to always allocate large regions, even when
+    alignment is non-zero.
+  - Fix prof_alloc_rollback() to read tdata from thread-specific data rather
+    than dereferencing a potentially invalid tctx.
+
+* 4.0.1 (September 15, 2015)
+
+  This is a bugfix release that is somewhat high risk due to the amount of
+  refactoring required to address deep xallocx() problems.  As a side effect of
+  these fixes, xallocx() now tries harder to partially fulfill requests for
+  optional extra space.  Note that a couple of minor heap profiling
+  optimizations are included, but these are better thought of as performance
+  fixes that were integral to discovering most of the other bugs.
+
+  Optimizations:
+  - Avoid a chunk metadata read in arena_prof_tctx_set(), since it is in the
+    fast path when heap profiling is enabled.  Additionally, split a special
+    case out into arena_prof_tctx_reset(), which also avoids chunk metadata
+    reads.
+  - Optimize irallocx_prof() to optimistically update the sampler state.  The
+    prior implementation appears to have been a holdover from when
+    rallocx()/xallocx() functionality was combined as rallocm().
+
+  Bug fixes:
+  - Fix TLS configuration such that it is enabled by default for platforms on
+    which it works correctly.
+  - Fix arenas_cache_cleanup() and arena_get_hard() to handle
+    allocation/deallocation within the application's thread-specific data
+    cleanup functions even after arenas_cache is torn down.
+  - Fix xallocx() bugs related to size+extra exceeding HUGE_MAXCLASS.
+  - Fix chunk purge hook calls for in-place huge shrinking reallocation to
+    specify the old chunk size rather than the new chunk size.  This bug caused
+    no correctness issues for the default chunk purge function, but was
+    visible to custom functions set via the "arena.<i>.chunk_hooks" mallctl.
+  - Fix heap profiling bugs:
+    + Fix heap profiling to distinguish among otherwise identical sample sites
+      with interposed resets (triggered via the "prof.reset" mallctl).  This bug
+      could cause data structure corruption that would most likely result in a
+      segfault.
+    + Fix irealloc_prof() to prof_alloc_rollback() on OOM.
+    + Make one call to prof_active_get_unlocked() per allocation event, and use
+      the result throughout the relevant functions that handle an allocation
+      event.  Also add a missing check in prof_realloc().  These fixes protect
+      allocation events against concurrent prof_active changes.
+    + Fix ixallocx_prof() to pass usize_max and zero to ixallocx_prof_sample()
+      in the correct order.
+    + Fix prof_realloc() to call prof_free_sampled_object() after calling
+      prof_malloc_sample_object().  Prior to this fix, if tctx and old_tctx were
+      the same, the tctx could have been prematurely destroyed.
+  - Fix portability bugs:
+    + Don't bitshift by negative amounts when encoding/decoding run sizes in
+      chunk header maps.  This affected systems with page sizes greater than 8
+      KiB.
+    + Rename index_t to szind_t to avoid an existing type on Solaris.
+    + Add JEMALLOC_CXX_THROW to the memalign() function prototype, in order to
+      match glibc and avoid compilation errors when including both
+      jemalloc/jemalloc.h and malloc.h in C++ code.
+    + Don't assume that /bin/sh is appropriate when running size_classes.sh
+      during configuration.
+    + Consider __sparcv9 a synonym for __sparc64__ when defining LG_QUANTUM.
+    + Link tests to librt if it contains clock_gettime(2).
+
+* 4.0.0 (August 17, 2015)
+
+  This version contains many speed and space optimizations, both minor and
+  major.  The major themes are generalization, unification, and simplification.
+  Although many of these optimizations cause no visible behavior change, their
+  cumulative effect is substantial.
+
+  New features:
+  - Normalize size class spacing to be consistent across the complete size
+    range.  By default there are four size classes per size doubling, but this
+    is now configurable via the --with-lg-size-class-group option.  Also add the
+    --with-lg-page, --with-lg-page-sizes, --with-lg-quantum, and
+    --with-lg-tiny-min options, which can be used to tweak page and size class
+    settings.  Impacts:
+    + Worst case performance for incrementally growing/shrinking reallocation
+      is improved because there are far fewer size classes, and therefore
+      copying happens less often.
+    + Internal fragmentation is limited to 20% for all but the smallest size
+      classes (those less than four times the quantum).  (1B + 4 KiB)
+      and (1B + 4 MiB) previously suffered nearly 50% internal fragmentation.
+    + Chunk fragmentation tends to be lower because there are fewer distinct run
+      sizes to pack.
+  - Add support for explicit tcaches.  The "tcache.create", "tcache.flush", and
+    "tcache.destroy" mallctls control tcache lifetime and flushing, and the
+    MALLOCX_TCACHE(tc) and MALLOCX_TCACHE_NONE flags to the *allocx() API
+    control which tcache is used for each operation.
+  - Implement per thread heap profiling, as well as the ability to
+    enable/disable heap profiling on a per thread basis.  Add the "prof.reset",
+    "prof.lg_sample", "thread.prof.name", "thread.prof.active",
+    "opt.prof_thread_active_init", "prof.thread_active_init", and
+    "thread.prof.active" mallctls.
+  - Add support for per arena application-specified chunk allocators, configured
+    via the "arena.<i>.chunk_hooks" mallctl.
+  - Refactor huge allocation to be managed by arenas, so that arenas now
+    function as general purpose independent allocators.  This is important in
+    the context of user-specified chunk allocators, aside from the scalability
+    benefits.  Related new statistics:
+    + The "stats.arenas.<i>.huge.allocated", "stats.arenas.<i>.huge.nmalloc",
+      "stats.arenas.<i>.huge.ndalloc", and "stats.arenas.<i>.huge.nrequests"
+      mallctls provide high level per arena huge allocation statistics.
+    + The "arenas.nhchunks", "arenas.hchunk.<i>.size",
+      "stats.arenas.<i>.hchunks.<j>.nmalloc",
+      "stats.arenas.<i>.hchunks.<j>.ndalloc",
+      "stats.arenas.<i>.hchunks.<j>.nrequests", and
+      "stats.arenas.<i>.hchunks.<j>.curhchunks" mallctls provide per size class
+      statistics.
+  - Add the 'util' column to malloc_stats_print() output, which reports the
+    proportion of available regions that are currently in use for each small
+    size class.
+  - Add "alloc" and "free" modes for for junk filling (see the "opt.junk"
+    mallctl), so that it is possible to separately enable junk filling for
+    allocation versus deallocation.
+  - Add the jemalloc-config script, which provides information about how
+    jemalloc was configured, and how to integrate it into application builds.
+  - Add metadata statistics, which are accessible via the "stats.metadata",
+    "stats.arenas.<i>.metadata.mapped", and
+    "stats.arenas.<i>.metadata.allocated" mallctls.
+  - Add the "stats.resident" mallctl, which reports the upper limit of
+    physically resident memory mapped by the allocator.
+  - Add per arena control over unused dirty page purging, via the
+    "arenas.lg_dirty_mult", "arena.<i>.lg_dirty_mult", and
+    "stats.arenas.<i>.lg_dirty_mult" mallctls.
+  - Add the "prof.gdump" mallctl, which makes it possible to toggle the gdump
+    feature on/off during program execution.
+  - Add sdallocx(), which implements sized deallocation.  The primary
+    optimization over dallocx() is the removal of a metadata read, which often
+    suffers an L1 cache miss.
+  - Add missing header includes in jemalloc/jemalloc.h, so that applications
+    only have to #include <jemalloc/jemalloc.h>.
+  - Add support for additional platforms:
+    + Bitrig
+    + Cygwin
+    + DragonFlyBSD
+    + iOS
+    + OpenBSD
+    + OpenRISC/or1k
+
+  Optimizations:
+  - Maintain dirty runs in per arena LRUs rather than in per arena trees of
+    dirty-run-containing chunks.  In practice this change significantly reduces
+    dirty page purging volume.
+  - Integrate whole chunks into the unused dirty page purging machinery.  This
+    reduces the cost of repeated huge allocation/deallocation, because it
+    effectively introduces a cache of chunks.
+  - Split the arena chunk map into two separate arrays, in order to increase
+    cache locality for the frequently accessed bits.
+  - Move small run metadata out of runs, into arena chunk headers.  This reduces
+    run fragmentation, smaller runs reduce external fragmentation for small size
+    classes, and packed (less uniformly aligned) metadata layout improves CPU
+    cache set distribution.
+  - Randomly distribute large allocation base pointer alignment relative to page
+    boundaries in order to more uniformly utilize CPU cache sets.  This can be
+    disabled via the --disable-cache-oblivious configure option, and queried via
+    the "config.cache_oblivious" mallctl.
+  - Micro-optimize the fast paths for the public API functions.
+  - Refactor thread-specific data to reside in a single structure.  This assures
+    that only a single TLS read is necessary per call into the public API.
+  - Implement in-place huge allocation growing and shrinking.
+  - Refactor rtree (radix tree for chunk lookups) to be lock-free, and make
+    additional optimizations that reduce maximum lookup depth to one or two
+    levels.  This resolves what was a concurrency bottleneck for per arena huge
+    allocation, because a global data structure is critical for determining
+    which arenas own which huge allocations.
+
+  Incompatible changes:
+  - Replace --enable-cc-silence with --disable-cc-silence to suppress spurious
+    warnings by default.
+  - Assure that the constness of malloc_usable_size()'s return type matches that
+    of the system implementation.
+  - Change the heap profile dump format to support per thread heap profiling,
+    rename pprof to jeprof, and enhance it with the --thread=<n> option.  As a
+    result, the bundled jeprof must now be used rather than the upstream
+    (gperftools) pprof.
+  - Disable "opt.prof_final" by default, in order to avoid atexit(3), which can
+    internally deadlock on some platforms.
+  - Change the "arenas.nlruns" mallctl type from size_t to unsigned.
+  - Replace the "stats.arenas.<i>.bins.<j>.allocated" mallctl with
+    "stats.arenas.<i>.bins.<j>.curregs".
+  - Ignore MALLOC_CONF in set{uid,gid,cap} binaries.
+  - Ignore MALLOCX_ARENA(a) in dallocx(), in favor of using the
+    MALLOCX_TCACHE(tc) and MALLOCX_TCACHE_NONE flags to control tcache usage.
+
+  Removed features:
+  - Remove the *allocm() API, which is superseded by the *allocx() API.
+  - Remove the --enable-dss options, and make dss non-optional on all platforms
+    which support sbrk(2).
+  - Remove the "arenas.purge" mallctl, which was obsoleted by the
+    "arena.<i>.purge" mallctl in 3.1.0.
+  - Remove the unnecessary "opt.valgrind" mallctl; jemalloc automatically
+    detects whether it is running inside Valgrind.
+  - Remove the "stats.huge.allocated", "stats.huge.nmalloc", and
+    "stats.huge.ndalloc" mallctls.
+  - Remove the --enable-mremap option.
+  - Remove the "stats.chunks.current", "stats.chunks.total", and
+    "stats.chunks.high" mallctls.
+
+  Bug fixes:
+  - Fix the cactive statistic to decrease (rather than increase) when active
+    memory decreases.  This regression was first released in 3.5.0.
+  - Fix OOM handling in memalign() and valloc().  A variant of this bug existed
+    in all releases since 2.0.0, which introduced these functions.
+  - Fix an OOM-related regression in arena_tcache_fill_small(), which could
+    cause cache corruption on OOM.  This regression was present in all releases
+    from 2.2.0 through 3.6.0.
+  - Fix size class overflow handling for malloc(), posix_memalign(), memalign(),
+    calloc(), and realloc() when profiling is enabled.
+  - Fix the "arena.<i>.dss" mallctl to return an error if "primary" or
+    "secondary" precedence is specified, but sbrk(2) is not supported.
+  - Fix fallback lg_floor() implementations to handle extremely large inputs.
+  - Ensure the default purgeable zone is after the default zone on OS X.
+  - Fix latent bugs in atomic_*().
+  - Fix the "arena.<i>.dss" mallctl to handle read-only calls.
+  - Fix tls_model configuration to enable the initial-exec model when possible.
+  - Mark malloc_conf as a weak symbol so that the application can override it.
+  - Correctly detect glibc's adaptive pthread mutexes.
+  - Fix the --without-export configure option.
+
 * 3.6.0 (March 31, 2014)
 
   This version contains a critical bug fix for a regression present in 3.5.0 and
@@ -21,7 +855,7 @@ found in the git revision history:
     backtracing to be reliable.
   - Use dss allocation precedence for huge allocations as well as small/large
     allocations.
-  - Fix test assertion failure message formatting.  This bug did not manifect on
+  - Fix test assertion failure message formatting.  This bug did not manifest on
     x86_64 systems because of implementation subtleties in va_list.
   - Fix inconsequential test failures for hash and SFMT code.
 
@@ -516,7 +1350,7 @@ found in the git revision history:
   - 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
+  - Compute various additional run-time statistics, including per size class
     statistics for large objects.
   - Expose malloc_stats_print(), which can be called repeatedly by the
     application.

INSTALL

@@ -1,306 +0,0 @@
-Building and installing jemalloc can be as simple as typing the following while
-in the root directory of the source tree:
-
-    ./configure
-    make
-    make install
-
-=== Advanced configuration =====================================================
-
-The 'configure' script supports numerous options that allow control of which
-functionality is enabled, where jemalloc is installed, etc.  Optionally, pass
-any of the following arguments (not a definitive list) to 'configure':
-
---help
-    Print a definitive list of options.
-
---prefix=<install-root-dir>
-    Set the base directory in which to install.  For example:
-
-        ./configure --prefix=/usr/local
-
-    will cause files to be installed into /usr/local/include, /usr/local/lib,
-    and /usr/local/man.
-
---with-rpath=<colon-separated-rpath>
-    Embed one or more library paths, so that libjemalloc can find the libraries
-    it is linked to.  This works only on ELF-based systems.
-
---with-mangling=<map>
-    Mangle public symbols specified in <map> which is a comma-separated list of
-    name:mangled pairs.
-
-    For example, to use ld's --wrap option as an alternative method for
-    overriding libc's malloc implementation, specify something like:
-
-      --with-mangling=malloc:__wrap_malloc,free:__wrap_free[...]
-
-    Note that mangling happens prior to application of the prefix specified by
-    --with-jemalloc-prefix, and mangled symbols are then ignored when applying
-    the prefix.
-
---with-jemalloc-prefix=<prefix>
-    Prefix all public APIs with <prefix>.  For example, if <prefix> is
-    "prefix_", API changes like the following occur:
-
-      malloc()         --> prefix_malloc()
-      malloc_conf      --> prefix_malloc_conf
-      /etc/malloc.conf --> /etc/prefix_malloc.conf
-      MALLOC_CONF      --> PREFIX_MALLOC_CONF
-
-    This makes it possible to use jemalloc at the same time as the system
-    allocator, or even to use multiple copies of jemalloc simultaneously.
-
-    By default, the prefix is "", except on OS X, where it is "je_".  On OS X,
-    jemalloc overlays the default malloc zone, but makes no attempt to actually
-    replace the "malloc", "calloc", etc. symbols.
-
---without-export
-    Don't export public APIs. This can be useful when building jemalloc as a
-    static library, or to avoid exporting public APIs when using the zone
-    allocator on OSX.
-
---with-private-namespace=<prefix>
-    Prefix all library-private APIs with <prefix>je_.  For shared libraries,
-    symbol visibility mechanisms prevent these symbols from being exported, but
-    for static libraries, naming collisions are a real possibility.  By
-    default, <prefix> is empty, which results in a symbol prefix of je_ .
-
---with-install-suffix=<suffix>
-    Append <suffix> to the base name of all installed files, such that multiple
-    versions of jemalloc can coexist in the same installation directory.  For
-    example, libjemalloc.so.0 becomes libjemalloc<suffix>.so.0.
-
---enable-cc-silence
-    Enable code that silences non-useful compiler warnings.  This is helpful
-    when trying to tell serious warnings from those due to compiler
-    limitations, but it potentially incurs a performance penalty.
-
---enable-debug
-    Enable assertions and validation code.  This incurs a substantial
-    performance hit, but is very useful during application development.
-    Implies --enable-ivsalloc.
-
---enable-code-coverage
-    Enable code coverage support, for use during jemalloc test development.
-    Additional testing targets are available if this option is enabled:
-
-      coverage
-      coverage_unit
-      coverage_integration
-      coverage_stress
-
-    These targets do not clear code coverage results from previous runs, and
-    there are interactions between the various coverage targets, so it is
-    usually advisable to run 'make clean' between repeated code coverage runs.
-
---enable-ivsalloc
-    Enable validation code, which verifies that pointers reside within
-    jemalloc-owned chunks before dereferencing them. This incurs a substantial
-    performance hit.
-
---disable-stats
-    Disable statistics gathering functionality.  See the "opt.stats_print"
-    option documentation for usage details.
-
---enable-prof
-    Enable heap profiling and leak detection functionality.  See the "opt.prof"
-    option documentation for usage details.  When enabled, there are several
-    approaches to backtracing, and the configure script chooses the first one
-    in the following list that appears to function correctly:
-
-    + libunwind      (requires --enable-prof-libunwind)
-    + libgcc         (unless --disable-prof-libgcc)
-    + gcc intrinsics (unless --disable-prof-gcc)
-
---enable-prof-libunwind
-    Use the libunwind library (http://www.nongnu.org/libunwind/) for stack
-    backtracing.
-
---disable-prof-libgcc
-    Disable the use of libgcc's backtracing functionality.
-
---disable-prof-gcc
-    Disable the use of gcc intrinsics for backtracing.
-
---with-static-libunwind=<libunwind.a>
-    Statically link against the specified libunwind.a rather than dynamically
-    linking with -lunwind.
-
---disable-tcache
-    Disable thread-specific caches for small objects.  Objects are cached and
-    released in bulk, thus reducing the total number of mutex operations.  See
-    the "opt.tcache" option for usage details.
-
---enable-mremap
-    Enable huge realloc() via mremap(2).  mremap() is disabled by default
-    because the flavor used is specific to Linux, which has a quirk in its
-    virtual memory allocation algorithm that causes semi-permanent VM map holes
-    under normal jemalloc operation.
-
---disable-munmap
-    Disable virtual memory deallocation via munmap(2); instead keep track of
-    the virtual memory for later use.  munmap() is disabled by default (i.e.
-    --disable-munmap is implied) on Linux, which has a quirk in its virtual
-    memory allocation algorithm that causes semi-permanent VM map holes under
-    normal jemalloc operation.
-
---enable-dss
-    Enable support for page allocation/deallocation via sbrk(2), in addition to
-    mmap(2).
-
---disable-fill
-    Disable support for junk/zero filling of memory, quarantine, and redzones.
-    See the "opt.junk", "opt.zero", "opt.quarantine", and "opt.redzone" option
-    documentation for usage details.
-
---disable-valgrind
-    Disable support for Valgrind.
-
---disable-experimental
-    Disable support for the experimental API (*allocm()).
-
---disable-zone-allocator
-    Disable zone allocator for Darwin. This means jemalloc won't be hooked as
-    the default allocator on OSX/iOS.
-
---enable-utrace
-    Enable utrace(2)-based allocation tracing.  This feature is not broadly
-    portable (FreeBSD has it, but Linux and OS X do not).
-
---enable-xmalloc
-    Enable support for optional immediate termination due to out-of-memory
-    errors, as is commonly implemented by "xmalloc" wrapper function for malloc.
-    See the "opt.xmalloc" option documentation for usage details.
-
---enable-lazy-lock
-    Enable code that wraps pthread_create() to detect when an application
-    switches from single-threaded to multi-threaded mode, so that it can avoid
-    mutex locking/unlocking operations while in single-threaded mode.  In
-    practice, this feature usually has little impact on performance unless
-    thread-specific caching is disabled.
-
---disable-tls
-    Disable thread-local storage (TLS), which allows for fast access to
-    thread-local variables via the __thread keyword.  If TLS is available,
-    jemalloc uses it for several purposes.
-
---with-xslroot=<path>
-    Specify where to find DocBook XSL stylesheets when building the
-    documentation.
-
-The following environment variables (not a definitive list) impact configure's
-behavior:
-
-CFLAGS="?"
-    Pass these flags to the compiler.  You probably shouldn't define this unless
-    you know what you are doing.  (Use EXTRA_CFLAGS instead.)
-
-EXTRA_CFLAGS="?"
-    Append these flags to CFLAGS.  This makes it possible to add flags such as
-    -Werror, while allowing the configure script to determine what other flags
-    are appropriate for the specified configuration.
-
-    The configure script specifically checks whether an optimization flag (-O*)
-    is specified in EXTRA_CFLAGS, and refrains from specifying an optimization
-    level if it finds that one has already been specified.
-
-CPPFLAGS="?"
-    Pass these flags to the C preprocessor.  Note that CFLAGS is not passed to
-    'cpp' when 'configure' is looking for include files, so you must use
-    CPPFLAGS instead if you need to help 'configure' find header files.
-
-LD_LIBRARY_PATH="?"
-    'ld' uses this colon-separated list to find libraries.
-
-LDFLAGS="?"
-    Pass these flags when linking.
-
-PATH="?"
-    'configure' uses this to find programs.
-
-=== Advanced compilation =======================================================
-
-To build only parts of jemalloc, use the following targets:
-
-    build_lib_shared
-    build_lib_static
-    build_lib
-    build_doc_html
-    build_doc_man
-    build_doc
-
-To install only parts of jemalloc, use the following targets:
-
-    install_bin
-    install_include
-    install_lib_shared
-    install_lib_static
-    install_lib
-    install_doc_html
-    install_doc_man
-    install_doc
-
-To clean up build results to varying degrees, use the following make targets:
-
-    clean
-    distclean
-    relclean
-
-=== Advanced installation ======================================================
-
-Optionally, define make variables when invoking make, including (not
-exclusively):
-
-INCLUDEDIR="?"
-    Use this as the installation prefix for header files.
-
-LIBDIR="?"
-    Use this as the installation prefix for libraries.
-
-MANDIR="?"
-    Use this as the installation prefix for man pages.
-
-DESTDIR="?"
-    Prepend DESTDIR to INCLUDEDIR, LIBDIR, DATADIR, and MANDIR.  This is useful
-    when installing to a different path than was specified via --prefix.
-
-CC="?"
-    Use this to invoke the C compiler.
-
-CFLAGS="?"
-    Pass these flags to the compiler.
-
-CPPFLAGS="?"
-    Pass these flags to the C preprocessor.
-
-LDFLAGS="?"
-    Pass these flags when linking.
-
-PATH="?"
-    Use this to search for programs used during configuration and building.
-
-=== Development ================================================================
-
-If you intend to make non-trivial changes to jemalloc, use the 'autogen.sh'
-script rather than 'configure'.  This re-generates 'configure', enables
-configuration dependency rules, and enables re-generation of automatically
-generated source files.
-
-The build system supports using an object directory separate from the source
-tree.  For example, you can create an 'obj' directory, and from within that
-directory, issue configuration and build commands:
-
-    autoconf
-    mkdir obj
-    cd obj
-    ../configure --enable-autogen
-    make
-
-=== Documentation ==============================================================
-
-The manual page is generated in both html and roff formats.  Any web browser
-can be used to view the html manual.  The roff manual page can be formatted
-prior to installation via the following command:
-
-    nroff -man -t doc/jemalloc.3

INSTALL.md

@@ -0,0 +1,423 @@
+Building and installing a packaged release of jemalloc can be as simple as
+typing the following while in the root directory of the source tree:
+
+    ./configure
+    make
+    make install
+
+If building from unpackaged developer sources, the simplest command sequence
+that might work is:
+
+    ./autogen.sh
+    make dist
+    make
+    make install
+
+Note that documentation is not built by the default target because doing so
+would create a dependency on xsltproc in packaged releases, hence the
+requirement to either run 'make dist' or avoid installing docs via the various
+install_* targets documented below.
+
+
+## Advanced configuration
+
+The 'configure' script supports numerous options that allow control of which
+functionality is enabled, where jemalloc is installed, etc.  Optionally, pass
+any of the following arguments (not a definitive list) to 'configure':
+
+* `--help`
+
+    Print a definitive list of options.
+
+* `--prefix=<install-root-dir>`
+
+    Set the base directory in which to install.  For example:
+
+        ./configure --prefix=/usr/local
+
+    will cause files to be installed into /usr/local/include, /usr/local/lib,
+    and /usr/local/man.
+
+* `--with-version=(<major>.<minor>.<bugfix>-<nrev>-g<gid>|VERSION)`
+
+    The VERSION file is mandatory for successful configuration, and the
+    following steps are taken to assure its presence:
+    1) If --with-version=<major>.<minor>.<bugfix>-<nrev>-g<gid> is specified,
+       generate VERSION using the specified value.
+    2) If --with-version is not specified in either form and the source
+       directory is inside a git repository, try to generate VERSION via 'git
+       describe' invocations that pattern-match release tags.
+    3) If VERSION is missing, generate it with a bogus version:
+       0.0.0-0-g0000000000000000000000000000000000000000
+
+    Note that --with-version=VERSION bypasses (1) and (2), which simplifies
+    VERSION configuration when embedding a jemalloc release into another
+    project's git repository.
+
+* `--with-rpath=<colon-separated-rpath>`
+
+    Embed one or more library paths, so that libjemalloc can find the libraries
+    it is linked to.  This works only on ELF-based systems.
+
+* `--with-mangling=<map>`
+
+    Mangle public symbols specified in <map> which is a comma-separated list of
+    name:mangled pairs.
+
+    For example, to use ld's --wrap option as an alternative method for
+    overriding libc's malloc implementation, specify something like:
+
+      --with-mangling=malloc:__wrap_malloc,free:__wrap_free[...]
+
+    Note that mangling happens prior to application of the prefix specified by
+    --with-jemalloc-prefix, and mangled symbols are then ignored when applying
+    the prefix.
+
+* `--with-jemalloc-prefix=<prefix>`
+
+    Prefix all public APIs with <prefix>.  For example, if <prefix> is
+    "prefix_", API changes like the following occur:
+
+      malloc()         --> prefix_malloc()
+      malloc_conf      --> prefix_malloc_conf
+      /etc/malloc.conf --> /etc/prefix_malloc.conf
+      MALLOC_CONF      --> PREFIX_MALLOC_CONF
+
+    This makes it possible to use jemalloc at the same time as the system
+    allocator, or even to use multiple copies of jemalloc simultaneously.
+
+    By default, the prefix is "", except on OS X, where it is "je_".  On OS X,
+    jemalloc overlays the default malloc zone, but makes no attempt to actually
+    replace the "malloc", "calloc", etc. symbols.
+
+* `--without-export`
+
+    Don't export public APIs.  This can be useful when building jemalloc as a
+    static library, or to avoid exporting public APIs when using the zone
+    allocator on OSX.
+
+* `--with-private-namespace=<prefix>`
+
+    Prefix all library-private APIs with <prefix>je_.  For shared libraries,
+    symbol visibility mechanisms prevent these symbols from being exported, but
+    for static libraries, naming collisions are a real possibility.  By
+    default, <prefix> is empty, which results in a symbol prefix of je_ .
+
+* `--with-install-suffix=<suffix>`
+
+    Append <suffix> to the base name of all installed files, such that multiple
+    versions of jemalloc can coexist in the same installation directory.  For
+    example, libjemalloc.so.0 becomes libjemalloc<suffix>.so.0.
+
+* `--with-malloc-conf=<malloc_conf>`
+
+    Embed `<malloc_conf>` as a run-time options string that is processed prior to
+    the malloc_conf global variable, the /etc/malloc.conf symlink, and the
+    MALLOC_CONF environment variable.  For example, to change the default decay
+    time to 30 seconds:
+
+      --with-malloc-conf=decay_ms:30000
+
+* `--enable-debug`
+
+    Enable assertions and validation code.  This incurs a substantial
+    performance hit, but is very useful during application development.
+
+* `--disable-stats`
+
+    Disable statistics gathering functionality.  See the "opt.stats_print"
+    option documentation for usage details.
+
+* `--enable-prof`
+
+    Enable heap profiling and leak detection functionality.  See the "opt.prof"
+    option documentation for usage details.  When enabled, there are several
+    approaches to backtracing, and the configure script chooses the first one
+    in the following list that appears to function correctly:
+
+    + libunwind      (requires --enable-prof-libunwind)
+    + libgcc         (unless --disable-prof-libgcc)
+    + gcc intrinsics (unless --disable-prof-gcc)
+
+* `--enable-prof-libunwind`
+
+    Use the libunwind library (http://www.nongnu.org/libunwind/) for stack
+    backtracing.
+
+* `--disable-prof-libgcc`
+
+    Disable the use of libgcc's backtracing functionality.
+
+* `--disable-prof-gcc`
+
+    Disable the use of gcc intrinsics for backtracing.
+
+* `--with-static-libunwind=<libunwind.a>`
+
+    Statically link against the specified libunwind.a rather than dynamically
+    linking with -lunwind.
+
+* `--disable-fill`
+
+    Disable support for junk/zero filling of memory.  See the "opt.junk" and
+    "opt.zero" option documentation for usage details.
+
+* `--disable-zone-allocator`
+
+    Disable zone allocator for Darwin.  This means jemalloc won't be hooked as
+    the default allocator on OSX/iOS.
+
+* `--enable-utrace`
+
+    Enable utrace(2)-based allocation tracing.  This feature is not broadly
+    portable (FreeBSD has it, but Linux and OS X do not).
+
+* `--enable-xmalloc`
+
+    Enable support for optional immediate termination due to out-of-memory
+    errors, as is commonly implemented by "xmalloc" wrapper function for malloc.
+    See the "opt.xmalloc" option documentation for usage details.
+
+* `--enable-lazy-lock`
+
+    Enable code that wraps pthread_create() to detect when an application
+    switches from single-threaded to multi-threaded mode, so that it can avoid
+    mutex locking/unlocking operations while in single-threaded mode.  In
+    practice, this feature usually has little impact on performance unless
+    thread-specific caching is disabled.
+
+* `--disable-cache-oblivious`
+
+    Disable cache-oblivious large allocation alignment for large allocation
+    requests with no alignment constraints.  If this feature is disabled, all
+    large allocations are page-aligned as an implementation artifact, which can
+    severely harm CPU cache utilization.  However, the cache-oblivious layout
+    comes at the cost of one extra page per large allocation, which in the
+    most extreme case increases physical memory usage for the 16 KiB size class
+    to 20 KiB.
+
+* `--disable-syscall`
+
+    Disable use of syscall(2) rather than {open,read,write,close}(2).  This is
+    intended as a workaround for systems that place security limitations on
+    syscall(2).
+
+* `--disable-cxx`
+
+    Disable C++ integration.  This will cause new and delete operator
+    implementations to be omitted.
+
+* `--with-xslroot=<path>`
+
+    Specify where to find DocBook XSL stylesheets when building the
+    documentation.
+
+* `--with-lg-page=<lg-page>`
+
+    Specify the base 2 log of the allocator page size, which must in turn be at
+    least as large as the system page size.  By default the configure script
+    determines the host's page size and sets the allocator page size equal to
+    the system page size, so this option need not be specified unless the
+    system page size may change between configuration and execution, e.g. when
+    cross compiling.
+
+* `--with-lg-page-sizes=<lg-page-sizes>`
+
+    Specify the comma-separated base 2 logs of the page sizes to support.  This
+    option may be useful when cross compiling in combination with
+    `--with-lg-page`, but its primary use case is for integration with FreeBSD's
+    libc, wherein jemalloc is embedded.
+
+* `--with-lg-hugepage=<lg-hugepage>`
+
+    Specify the base 2 log of the system huge page size.  This option is useful
+    when cross compiling, or when overriding the default for systems that do
+    not explicitly support huge pages.
+
+* `--with-lg-quantum=<lg-quantum>`
+
+    Specify the base 2 log of the minimum allocation alignment.  jemalloc needs
+    to know the minimum alignment that meets the following C standard
+    requirement (quoted from the April 12, 2011 draft of the C11 standard):
+
+    >  The pointer returned if the allocation succeeds is suitably aligned so
+      that it may be assigned to a pointer to any type of object with a
+      fundamental alignment requirement and then used to access such an object
+      or an array of such objects in the space allocated [...]
+
+    This setting is architecture-specific, and although jemalloc includes known
+    safe values for the most commonly used modern architectures, there is a
+    wrinkle related to GNU libc (glibc) that may impact your choice of
+    <lg-quantum>.  On most modern architectures, this mandates 16-byte
+    alignment (<lg-quantum>=4), but the glibc developers chose not to meet this
+    requirement for performance reasons.  An old discussion can be found at
+    <https://sourceware.org/bugzilla/show_bug.cgi?id=206> .  Unlike glibc,
+    jemalloc does follow the C standard by default (caveat: jemalloc
+    technically cheats for size classes smaller than the quantum), but the fact
+    that Linux systems already work around this allocator noncompliance means
+    that it is generally safe in practice to let jemalloc's minimum alignment
+    follow glibc's lead.  If you specify `--with-lg-quantum=3` during
+    configuration, jemalloc will provide additional size classes that are not
+    16-byte-aligned (24, 40, and 56).
+
+* `--with-lg-vaddr=<lg-vaddr>`
+
+    Specify the number of significant virtual address bits.  By default, the
+    configure script attempts to detect virtual address size on those platforms
+    where it knows how, and picks a default otherwise.  This option may be
+    useful when cross-compiling.
+
+* `--disable-initial-exec-tls`
+
+    Disable the initial-exec TLS model for jemalloc's internal thread-local
+    storage (on those platforms that support explicit settings).  This can allow
+    jemalloc to be dynamically loaded after program startup (e.g. using dlopen).
+    Note that in this case, there will be two malloc implementations operating
+    in the same process, which will almost certainly result in confusing runtime
+    crashes if pointers leak from one implementation to the other.
+
+The following environment variables (not a definitive list) impact configure's
+behavior:
+
+* `CFLAGS="?"`
+* `CXXFLAGS="?"`
+
+    Pass these flags to the C/C++ compiler.  Any flags set by the configure
+    script are prepended, which means explicitly set flags generally take
+    precedence.  Take care when specifying flags such as -Werror, because
+    configure tests may be affected in undesirable ways.
+
+* `EXTRA_CFLAGS="?"`
+* `EXTRA_CXXFLAGS="?"`
+
+    Append these flags to CFLAGS/CXXFLAGS, without passing them to the
+    compiler(s) during configuration.  This makes it possible to add flags such
+    as -Werror, while allowing the configure script to determine what other
+    flags are appropriate for the specified configuration.
+
+* `CPPFLAGS="?"`
+
+    Pass these flags to the C preprocessor.  Note that CFLAGS is not passed to
+    'cpp' when 'configure' is looking for include files, so you must use
+    CPPFLAGS instead if you need to help 'configure' find header files.
+
+* `LD_LIBRARY_PATH="?"`
+
+    'ld' uses this colon-separated list to find libraries.
+
+* `LDFLAGS="?"`
+
+    Pass these flags when linking.
+
+* `PATH="?"`
+
+    'configure' uses this to find programs.
+
+In some cases it may be necessary to work around configuration results that do
+not match reality.  For example, Linux 4.5 added support for the MADV_FREE flag
+to madvise(2), which can cause problems if building on a host with MADV_FREE
+support and deploying to a target without.  To work around this, use a cache
+file to override the relevant configuration variable defined in configure.ac,
+e.g.:
+
+    echo "je_cv_madv_free=no" > config.cache && ./configure -C
+
+
+## Advanced compilation
+
+To build only parts of jemalloc, use the following targets:
+
+    build_lib_shared
+    build_lib_static
+    build_lib
+    build_doc_html
+    build_doc_man
+    build_doc
+
+To install only parts of jemalloc, use the following targets:
+
+    install_bin
+    install_include
+    install_lib_shared
+    install_lib_static
+    install_lib_pc
+    install_lib
+    install_doc_html
+    install_doc_man
+    install_doc
+
+To clean up build results to varying degrees, use the following make targets:
+
+    clean
+    distclean
+    relclean
+
+
+## Advanced installation
+
+Optionally, define make variables when invoking make, including (not
+exclusively):
+
+* `INCLUDEDIR="?"`
+
+    Use this as the installation prefix for header files.
+
+* `LIBDIR="?"`
+
+    Use this as the installation prefix for libraries.
+
+* `MANDIR="?"`
+
+    Use this as the installation prefix for man pages.
+
+* `DESTDIR="?"`
+
+    Prepend DESTDIR to INCLUDEDIR, LIBDIR, DATADIR, and MANDIR.  This is useful
+    when installing to a different path than was specified via --prefix.
+
+* `CC="?"`
+
+    Use this to invoke the C compiler.
+
+* `CFLAGS="?"`
+
+    Pass these flags to the compiler.
+
+* `CPPFLAGS="?"`
+
+    Pass these flags to the C preprocessor.
+
+* `LDFLAGS="?"`
+
+    Pass these flags when linking.
+
+* `PATH="?"`
+
+    Use this to search for programs used during configuration and building.
+
+
+## Development
+
+If you intend to make non-trivial changes to jemalloc, use the 'autogen.sh'
+script rather than 'configure'.  This re-generates 'configure', enables
+configuration dependency rules, and enables re-generation of automatically
+generated source files.
+
+The build system supports using an object directory separate from the source
+tree.  For example, you can create an 'obj' directory, and from within that
+directory, issue configuration and build commands:
+
+    autoconf
+    mkdir obj
+    cd obj
+    ../configure --enable-autogen
+    make
+
+
+## Documentation
+
+The manual page is generated in both html and roff formats.  Any web browser
+can be used to view the html manual.  The roff manual page can be formatted
+prior to installation via the following command:
+
+    nroff -man -t doc/jemalloc.3

Makefile.in

@@ -9,6 +9,7 @@ vpath % .
 SHELL := /bin/sh
 
 CC := @CC@
+CXX := @CXX@
 
 # Configuration parameters.
 DESTDIR =
@@ -23,8 +24,15 @@ abs_srcroot := @abs_srcroot@
 abs_objroot := @abs_objroot@
 
 # Build parameters.
-CPPFLAGS := @CPPFLAGS@ -I$(srcroot)include -I$(objroot)include
-CFLAGS := @CFLAGS@
+CPPFLAGS := @CPPFLAGS@ -I$(objroot)include -I$(srcroot)include
+CONFIGURE_CFLAGS := @CONFIGURE_CFLAGS@
+SPECIFIED_CFLAGS := @SPECIFIED_CFLAGS@
+EXTRA_CFLAGS := @EXTRA_CFLAGS@
+CFLAGS := $(strip $(CONFIGURE_CFLAGS) $(SPECIFIED_CFLAGS) $(EXTRA_CFLAGS))
+CONFIGURE_CXXFLAGS := @CONFIGURE_CXXFLAGS@
+SPECIFIED_CXXFLAGS := @SPECIFIED_CXXFLAGS@
+EXTRA_CXXFLAGS := @EXTRA_CXXFLAGS@
+CXXFLAGS := $(strip $(CONFIGURE_CXXFLAGS) $(SPECIFIED_CXXFLAGS) $(EXTRA_CXXFLAGS))
 LDFLAGS := @LDFLAGS@
 EXTRA_LDFLAGS := @EXTRA_LDFLAGS@
 LIBS := @LIBS@
@@ -42,23 +50,29 @@ XSLTPROC := @XSLTPROC@
 AUTOCONF := @AUTOCONF@
 _RPATH = @RPATH@
 RPATH = $(if $(1),$(call _RPATH,$(1)))
-cfghdrs_in := @cfghdrs_in@
+cfghdrs_in := $(addprefix $(srcroot),@cfghdrs_in@)
 cfghdrs_out := @cfghdrs_out@
-cfgoutputs_in := @cfgoutputs_in@
+cfgoutputs_in := $(addprefix $(srcroot),@cfgoutputs_in@)
 cfgoutputs_out := @cfgoutputs_out@
 enable_autogen := @enable_autogen@
-enable_code_coverage := @enable_code_coverage@
-enable_experimental := @enable_experimental@
+enable_prof := @enable_prof@
 enable_zone_allocator := @enable_zone_allocator@
+MALLOC_CONF := @JEMALLOC_CPREFIX@MALLOC_CONF
+link_whole_archive := @link_whole_archive@
 DSO_LDFLAGS = @DSO_LDFLAGS@
 SOREV = @SOREV@
 PIC_CFLAGS = @PIC_CFLAGS@
 CTARGET = @CTARGET@
 LDTARGET = @LDTARGET@
+TEST_LD_MODE = @TEST_LD_MODE@
 MKLIB = @MKLIB@
 AR = @AR@
 ARFLAGS = @ARFLAGS@
+DUMP_SYMS = @DUMP_SYMS@
+AWK := @AWK@
 CC_MM = @CC_MM@
+LM := @LM@
+INSTALL = @INSTALL@
 
 ifeq (macho, $(ABI))
 TEST_LIBRARY_PATH := DYLD_FALLBACK_LIBRARY_PATH="$(objroot)lib"
@@ -73,16 +87,38 @@ endif
 LIBJEMALLOC := $(LIBPREFIX)jemalloc$(install_suffix)
 
 # Lists of files.
-BINS := $(srcroot)bin/pprof $(objroot)bin/jemalloc.sh
+BINS := $(objroot)bin/jemalloc-config $(objroot)bin/jemalloc.sh $(objroot)bin/jeprof
 C_HDRS := $(objroot)include/jemalloc/jemalloc$(install_suffix).h
-C_SRCS := $(srcroot)src/jemalloc.c $(srcroot)src/arena.c \
-	$(srcroot)src/atomic.c $(srcroot)src/base.c $(srcroot)src/bitmap.c \
-	$(srcroot)src/chunk.c $(srcroot)src/chunk_dss.c \
-	$(srcroot)src/chunk_mmap.c $(srcroot)src/ckh.c $(srcroot)src/ctl.c \
-	$(srcroot)src/extent.c $(srcroot)src/hash.c $(srcroot)src/huge.c \
-	$(srcroot)src/mb.c $(srcroot)src/mutex.c $(srcroot)src/prof.c \
-	$(srcroot)src/quarantine.c $(srcroot)src/rtree.c $(srcroot)src/stats.c \
-	$(srcroot)src/tcache.c $(srcroot)src/util.c $(srcroot)src/tsd.c
+C_SRCS := $(srcroot)src/jemalloc.c \
+	$(srcroot)src/arena.c \
+	$(srcroot)src/background_thread.c \
+	$(srcroot)src/base.c \
+	$(srcroot)src/bin.c \
+	$(srcroot)src/bitmap.c \
+	$(srcroot)src/ckh.c \
+	$(srcroot)src/ctl.c \
+	$(srcroot)src/div.c \
+	$(srcroot)src/extent.c \
+	$(srcroot)src/extent_dss.c \
+	$(srcroot)src/extent_mmap.c \
+	$(srcroot)src/hash.c \
+	$(srcroot)src/hooks.c \
+	$(srcroot)src/large.c \
+	$(srcroot)src/log.c \
+	$(srcroot)src/malloc_io.c \
+	$(srcroot)src/mutex.c \
+	$(srcroot)src/mutex_pool.c \
+	$(srcroot)src/nstime.c \
+	$(srcroot)src/pages.c \
+	$(srcroot)src/prng.c \
+	$(srcroot)src/prof.c \
+	$(srcroot)src/rtree.c \
+	$(srcroot)src/stats.c \
+	$(srcroot)src/sz.c \
+	$(srcroot)src/tcache.c \
+	$(srcroot)src/ticker.c \
+	$(srcroot)src/tsd.c \
+	$(srcroot)src/witness.c
 ifeq ($(enable_zone_allocator), 1)
 C_SRCS += $(srcroot)src/zone.c
 endif
@@ -98,57 +134,119 @@ DSOS := $(objroot)lib/$(LIBJEMALLOC).$(SOREV)
 ifneq ($(SOREV),$(SO))
 DSOS += $(objroot)lib/$(LIBJEMALLOC).$(SO)
 endif
+ifeq (1, $(link_whole_archive))
+LJEMALLOC := -Wl,--whole-archive -L$(objroot)lib -l$(LIBJEMALLOC) -Wl,--no-whole-archive
+else
+LJEMALLOC := $(objroot)lib/$(LIBJEMALLOC).$(IMPORTLIB)
+endif
+PC := $(objroot)jemalloc.pc
 MAN3 := $(objroot)doc/jemalloc$(install_suffix).3
 DOCS_XML := $(objroot)doc/jemalloc$(install_suffix).xml
-DOCS_HTML := $(DOCS_XML:$(objroot)%.xml=$(srcroot)%.html)
-DOCS_MAN3 := $(DOCS_XML:$(objroot)%.xml=$(srcroot)%.3)
+DOCS_HTML := $(DOCS_XML:$(objroot)%.xml=$(objroot)%.html)
+DOCS_MAN3 := $(DOCS_XML:$(objroot)%.xml=$(objroot)%.3)
 DOCS := $(DOCS_HTML) $(DOCS_MAN3)
-C_TESTLIB_SRCS := $(srcroot)test/src/math.c $(srcroot)test/src/mtx.c \
+C_TESTLIB_SRCS := $(srcroot)test/src/btalloc.c $(srcroot)test/src/btalloc_0.c \
+	$(srcroot)test/src/btalloc_1.c $(srcroot)test/src/math.c \
+	$(srcroot)test/src/mtx.c $(srcroot)test/src/mq.c \
 	$(srcroot)test/src/SFMT.c $(srcroot)test/src/test.c \
-	$(srcroot)test/src/thd.c
-C_UTIL_INTEGRATION_SRCS := $(srcroot)src/util.c
-TESTS_UNIT := $(srcroot)test/unit/bitmap.c \
+	$(srcroot)test/src/thd.c $(srcroot)test/src/timer.c
+ifeq (1, $(link_whole_archive))
+C_UTIL_INTEGRATION_SRCS :=
+C_UTIL_CPP_SRCS :=
+else
+C_UTIL_INTEGRATION_SRCS := $(srcroot)src/nstime.c $(srcroot)src/malloc_io.c
+C_UTIL_CPP_SRCS := $(srcroot)src/nstime.c $(srcroot)src/malloc_io.c
+endif
+TESTS_UNIT := \
+	$(srcroot)test/unit/a0.c \
+	$(srcroot)test/unit/arena_reset.c \
+	$(srcroot)test/unit/atomic.c \
+	$(srcroot)test/unit/background_thread.c \
+	$(srcroot)test/unit/background_thread_enable.c \
+	$(srcroot)test/unit/base.c \
+	$(srcroot)test/unit/bitmap.c \
 	$(srcroot)test/unit/ckh.c \
+	$(srcroot)test/unit/decay.c \
+	$(srcroot)test/unit/div.c \
+	$(srcroot)test/unit/emitter.c \
+	$(srcroot)test/unit/extent_quantize.c \
+	$(srcroot)test/unit/fork.c \
 	$(srcroot)test/unit/hash.c \
+	$(srcroot)test/unit/hooks.c \
 	$(srcroot)test/unit/junk.c \
+	$(srcroot)test/unit/junk_alloc.c \
+	$(srcroot)test/unit/junk_free.c \
+	$(srcroot)test/unit/log.c \
 	$(srcroot)test/unit/mallctl.c \
+	$(srcroot)test/unit/malloc_io.c \
 	$(srcroot)test/unit/math.c \
 	$(srcroot)test/unit/mq.c \
 	$(srcroot)test/unit/mtx.c \
+	$(srcroot)test/unit/pack.c \
+	$(srcroot)test/unit/pages.c \
+	$(srcroot)test/unit/ph.c \
+	$(srcroot)test/unit/prng.c \
 	$(srcroot)test/unit/prof_accum.c \
+	$(srcroot)test/unit/prof_active.c \
 	$(srcroot)test/unit/prof_gdump.c \
 	$(srcroot)test/unit/prof_idump.c \
+	$(srcroot)test/unit/prof_reset.c \
+	$(srcroot)test/unit/prof_tctx.c \
+	$(srcroot)test/unit/prof_thread_name.c \
 	$(srcroot)test/unit/ql.c \
 	$(srcroot)test/unit/qr.c \
-	$(srcroot)test/unit/quarantine.c \
 	$(srcroot)test/unit/rb.c \
+	$(srcroot)test/unit/retained.c \
 	$(srcroot)test/unit/rtree.c \
 	$(srcroot)test/unit/SFMT.c \
+	$(srcroot)test/unit/size_classes.c \
+	$(srcroot)test/unit/slab.c \
+	$(srcroot)test/unit/smoothstep.c \
+	$(srcroot)test/unit/spin.c \
 	$(srcroot)test/unit/stats.c \
+	$(srcroot)test/unit/stats_print.c \
+	$(srcroot)test/unit/ticker.c \
+	$(srcroot)test/unit/nstime.c \
 	$(srcroot)test/unit/tsd.c \
-	$(srcroot)test/unit/util.c \
+	$(srcroot)test/unit/witness.c \
 	$(srcroot)test/unit/zero.c
-TESTS_UNIT_AUX := $(srcroot)test/unit/prof_accum_a.c \
-	$(srcroot)test/unit/prof_accum_b.c
+ifeq (@enable_prof@, 1)
+TESTS_UNIT += \
+	$(srcroot)test/unit/arena_reset_prof.c
+endif
 TESTS_INTEGRATION := $(srcroot)test/integration/aligned_alloc.c \
 	$(srcroot)test/integration/allocated.c \
+	$(srcroot)test/integration/extent.c \
 	$(srcroot)test/integration/mallocx.c \
-	$(srcroot)test/integration/mremap.c \
+	$(srcroot)test/integration/MALLOCX_ARENA.c \
+	$(srcroot)test/integration/overflow.c \
 	$(srcroot)test/integration/posix_memalign.c \
 	$(srcroot)test/integration/rallocx.c \
+	$(srcroot)test/integration/sdallocx.c \
 	$(srcroot)test/integration/thread_arena.c \
 	$(srcroot)test/integration/thread_tcache_enabled.c \
 	$(srcroot)test/integration/xallocx.c
-ifeq ($(enable_experimental), 1)
-TESTS_INTEGRATION += $(srcroot)test/integration/allocm.c \
-	$(srcroot)test/integration/MALLOCX_ARENA.c \
-	$(srcroot)test/integration/rallocm.c
+ifeq (@enable_cxx@, 1)
+CPP_SRCS := $(srcroot)src/jemalloc_cpp.cpp
+TESTS_INTEGRATION_CPP := $(srcroot)test/integration/cpp/basic.cpp
+else
+CPP_SRCS :=
+TESTS_INTEGRATION_CPP :=
 endif
-TESTS_STRESS :=
-TESTS := $(TESTS_UNIT) $(TESTS_INTEGRATION) $(TESTS_STRESS)
+TESTS_STRESS := $(srcroot)test/stress/microbench.c
 
+TESTS := $(TESTS_UNIT) $(TESTS_INTEGRATION) $(TESTS_INTEGRATION_CPP) $(TESTS_STRESS)
+
+PRIVATE_NAMESPACE_HDRS := $(objroot)include/jemalloc/internal/private_namespace.h $(objroot)include/jemalloc/internal/private_namespace_jet.h
+PRIVATE_NAMESPACE_GEN_HDRS := $(PRIVATE_NAMESPACE_HDRS:%.h=%.gen.h)
+C_SYM_OBJS := $(C_SRCS:$(srcroot)%.c=$(objroot)%.sym.$(O))
+C_SYMS := $(C_SRCS:$(srcroot)%.c=$(objroot)%.sym)
 C_OBJS := $(C_SRCS:$(srcroot)%.c=$(objroot)%.$(O))
+CPP_OBJS := $(CPP_SRCS:$(srcroot)%.cpp=$(objroot)%.$(O))
 C_PIC_OBJS := $(C_SRCS:$(srcroot)%.c=$(objroot)%.pic.$(O))
+CPP_PIC_OBJS := $(CPP_SRCS:$(srcroot)%.cpp=$(objroot)%.pic.$(O))
+C_JET_SYM_OBJS := $(C_SRCS:$(srcroot)%.c=$(objroot)%.jet.sym.$(O))
+C_JET_SYMS := $(C_SRCS:$(srcroot)%.c=$(objroot)%.jet.sym)
 C_JET_OBJS := $(C_SRCS:$(srcroot)%.c=$(objroot)%.jet.$(O))
 C_TESTLIB_UNIT_OBJS := $(C_TESTLIB_SRCS:$(srcroot)%.c=$(objroot)%.unit.$(O))
 C_TESTLIB_INTEGRATION_OBJS := $(C_TESTLIB_SRCS:$(srcroot)%.c=$(objroot)%.integration.$(O))
@@ -157,27 +255,28 @@ C_TESTLIB_STRESS_OBJS := $(C_TESTLIB_SRCS:$(srcroot)%.c=$(objroot)%.stress.$(O))
 C_TESTLIB_OBJS := $(C_TESTLIB_UNIT_OBJS) $(C_TESTLIB_INTEGRATION_OBJS) $(C_UTIL_INTEGRATION_OBJS) $(C_TESTLIB_STRESS_OBJS)
 
 TESTS_UNIT_OBJS := $(TESTS_UNIT:$(srcroot)%.c=$(objroot)%.$(O))
-TESTS_UNIT_AUX_OBJS := $(TESTS_UNIT_AUX:$(srcroot)%.c=$(objroot)%.$(O))
 TESTS_INTEGRATION_OBJS := $(TESTS_INTEGRATION:$(srcroot)%.c=$(objroot)%.$(O))
+TESTS_INTEGRATION_CPP_OBJS := $(TESTS_INTEGRATION_CPP:$(srcroot)%.cpp=$(objroot)%.$(O))
 TESTS_STRESS_OBJS := $(TESTS_STRESS:$(srcroot)%.c=$(objroot)%.$(O))
-TESTS_OBJS := $(TESTS_UNIT_OBJS) $(TESTS_UNIT_AUX_OBJS) $(TESTS_INTEGRATION_OBJS) $(TESTS_STRESS_OBJS)
+TESTS_OBJS := $(TESTS_UNIT_OBJS) $(TESTS_INTEGRATION_OBJS) $(TESTS_STRESS_OBJS)
+TESTS_CPP_OBJS := $(TESTS_INTEGRATION_CPP_OBJS)
 
 .PHONY: all dist build_doc_html build_doc_man build_doc
 .PHONY: install_bin install_include install_lib
 .PHONY: install_doc_html install_doc_man install_doc install
 .PHONY: tests check clean distclean relclean
 
-.SECONDARY : $(TESTS_OBJS)
+.SECONDARY : $(PRIVATE_NAMESPACE_GEN_HDRS) $(TESTS_OBJS) $(TESTS_CPP_OBJS)
 
 # Default target.
 all: build_lib
 
 dist: build_doc
 
-$(srcroot)doc/%.html : $(objroot)doc/%.xml $(srcroot)doc/stylesheet.xsl $(objroot)doc/html.xsl
+$(objroot)doc/%.html : $(objroot)doc/%.xml $(srcroot)doc/stylesheet.xsl $(objroot)doc/html.xsl
 	$(XSLTPROC) -o $@ $(objroot)doc/html.xsl $<
 
-$(srcroot)doc/%.3 : $(objroot)doc/%.xml $(srcroot)doc/stylesheet.xsl $(objroot)doc/manpages.xsl
+$(objroot)doc/%.3 : $(objroot)doc/%.xml $(srcroot)doc/stylesheet.xsl $(objroot)doc/manpages.xsl
 	$(XSLTPROC) -o $@ $(objroot)doc/manpages.xsl $<
 
 build_doc_html: $(DOCS_HTML)
@@ -188,18 +287,32 @@ build_doc: $(DOCS)
 # Include generated dependency files.
 #
 ifdef CC_MM
+-include $(C_SYM_OBJS:%.$(O)=%.d)
 -include $(C_OBJS:%.$(O)=%.d)
+-include $(CPP_OBJS:%.$(O)=%.d)
 -include $(C_PIC_OBJS:%.$(O)=%.d)
+-include $(CPP_PIC_OBJS:%.$(O)=%.d)
+-include $(C_JET_SYM_OBJS:%.$(O)=%.d)
 -include $(C_JET_OBJS:%.$(O)=%.d)
 -include $(C_TESTLIB_OBJS:%.$(O)=%.d)
 -include $(TESTS_OBJS:%.$(O)=%.d)
+-include $(TESTS_CPP_OBJS:%.$(O)=%.d)
 endif
 
+$(C_SYM_OBJS): $(objroot)src/%.sym.$(O): $(srcroot)src/%.c
+$(C_SYM_OBJS): CPPFLAGS += -DJEMALLOC_NO_PRIVATE_NAMESPACE
+$(C_SYMS): $(objroot)src/%.sym: $(objroot)src/%.sym.$(O)
 $(C_OBJS): $(objroot)src/%.$(O): $(srcroot)src/%.c
+$(CPP_OBJS): $(objroot)src/%.$(O): $(srcroot)src/%.cpp
 $(C_PIC_OBJS): $(objroot)src/%.pic.$(O): $(srcroot)src/%.c
 $(C_PIC_OBJS): CFLAGS += $(PIC_CFLAGS)
+$(CPP_PIC_OBJS): $(objroot)src/%.pic.$(O): $(srcroot)src/%.cpp
+$(CPP_PIC_OBJS): CXXFLAGS += $(PIC_CFLAGS)
+$(C_JET_SYM_OBJS): $(objroot)src/%.jet.sym.$(O): $(srcroot)src/%.c
+$(C_JET_SYM_OBJS): CPPFLAGS += -DJEMALLOC_JET -DJEMALLOC_NO_PRIVATE_NAMESPACE
+$(C_JET_SYMS): $(objroot)src/%.jet.sym: $(objroot)src/%.jet.sym.$(O)
 $(C_JET_OBJS): $(objroot)src/%.jet.$(O): $(srcroot)src/%.c
-$(C_JET_OBJS): CFLAGS += -DJEMALLOC_JET
+$(C_JET_OBJS): CPPFLAGS += -DJEMALLOC_JET
 $(C_TESTLIB_UNIT_OBJS): $(objroot)test/src/%.unit.$(O): $(srcroot)test/src/%.c
 $(C_TESTLIB_UNIT_OBJS): CPPFLAGS += -DJEMALLOC_UNIT_TEST
 $(C_TESTLIB_INTEGRATION_OBJS): $(objroot)test/src/%.integration.$(O): $(srcroot)test/src/%.c
@@ -209,112 +322,147 @@ $(C_TESTLIB_STRESS_OBJS): $(objroot)test/src/%.stress.$(O): $(srcroot)test/src/%
 $(C_TESTLIB_STRESS_OBJS): CPPFLAGS += -DJEMALLOC_STRESS_TEST -DJEMALLOC_STRESS_TESTLIB
 $(C_TESTLIB_OBJS): CPPFLAGS += -I$(srcroot)test/include -I$(objroot)test/include
 $(TESTS_UNIT_OBJS): CPPFLAGS += -DJEMALLOC_UNIT_TEST
-$(TESTS_UNIT_AUX_OBJS): CPPFLAGS += -DJEMALLOC_UNIT_TEST
-define make-unit-link-dep
-$(1): TESTS_UNIT_LINK_OBJS += $(2)
-$(1): $(2)
-endef
-$(foreach test, $(TESTS_UNIT:$(srcroot)test/unit/%.c=$(objroot)test/unit/%$(EXE)), $(eval $(call make-unit-link-dep,$(test),$(filter $(test:%=%_a.$(O)) $(test:%=%_b.$(O)),$(TESTS_UNIT_AUX_OBJS)))))
 $(TESTS_INTEGRATION_OBJS): CPPFLAGS += -DJEMALLOC_INTEGRATION_TEST
+$(TESTS_INTEGRATION_CPP_OBJS): CPPFLAGS += -DJEMALLOC_INTEGRATION_CPP_TEST
 $(TESTS_STRESS_OBJS): CPPFLAGS += -DJEMALLOC_STRESS_TEST
 $(TESTS_OBJS): $(objroot)test/%.$(O): $(srcroot)test/%.c
+$(TESTS_CPP_OBJS): $(objroot)test/%.$(O): $(srcroot)test/%.cpp
 $(TESTS_OBJS): CPPFLAGS += -I$(srcroot)test/include -I$(objroot)test/include
+$(TESTS_CPP_OBJS): CPPFLAGS += -I$(srcroot)test/include -I$(objroot)test/include
 ifneq ($(IMPORTLIB),$(SO))
-$(C_OBJS): CPPFLAGS += -DDLLEXPORT
+$(CPP_OBJS) $(C_SYM_OBJS) $(C_OBJS) $(C_JET_SYM_OBJS) $(C_JET_OBJS): CPPFLAGS += -DDLLEXPORT
 endif
 
-ifndef CC_MM
 # Dependencies.
+ifndef CC_MM
 HEADER_DIRS = $(srcroot)include/jemalloc/internal \
 	$(objroot)include/jemalloc $(objroot)include/jemalloc/internal
-HEADERS = $(wildcard $(foreach dir,$(HEADER_DIRS),$(dir)/*.h))
-$(C_OBJS) $(C_PIC_OBJS) $(C_JET_OBJS) $(C_TESTLIB_OBJS) $(TESTS_OBJS): $(HEADERS)
-$(TESTS_OBJS): $(objroot)test/unit/jemalloc_test.h
+HEADERS = $(filter-out $(PRIVATE_NAMESPACE_HDRS),$(wildcard $(foreach dir,$(HEADER_DIRS),$(dir)/*.h)))
+$(C_SYM_OBJS) $(C_OBJS) $(CPP_OBJS) $(C_PIC_OBJS) $(CPP_PIC_OBJS) $(C_JET_SYM_OBJS) $(C_JET_OBJS) $(C_TESTLIB_OBJS) $(TESTS_OBJS) $(TESTS_CPP_OBJS): $(HEADERS)
+$(TESTS_OBJS) $(TESTS_CPP_OBJS): $(objroot)test/include/test/jemalloc_test.h
 endif
 
-$(C_OBJS) $(C_PIC_OBJS) $(C_JET_OBJS) $(C_TESTLIB_OBJS) $(TESTS_OBJS): %.$(O):
+$(C_OBJS) $(CPP_OBJS) $(C_PIC_OBJS) $(CPP_PIC_OBJS) $(C_TESTLIB_INTEGRATION_OBJS) $(C_UTIL_INTEGRATION_OBJS) $(TESTS_INTEGRATION_OBJS) $(TESTS_INTEGRATION_CPP_OBJS): $(objroot)include/jemalloc/internal/private_namespace.h
+$(C_JET_OBJS) $(C_TESTLIB_UNIT_OBJS) $(C_TESTLIB_STRESS_OBJS) $(TESTS_UNIT_OBJS) $(TESTS_STRESS_OBJS): $(objroot)include/jemalloc/internal/private_namespace_jet.h
+
+$(C_SYM_OBJS) $(C_OBJS) $(C_PIC_OBJS) $(C_JET_SYM_OBJS) $(C_JET_OBJS) $(C_TESTLIB_OBJS) $(TESTS_OBJS): %.$(O):
 	@mkdir -p $(@D)
 	$(CC) $(CFLAGS) -c $(CPPFLAGS) $(CTARGET) $<
 ifdef CC_MM
 	@$(CC) -MM $(CPPFLAGS) -MT $@ -o $(@:%.$(O)=%.d) $<
 endif
 
+$(C_SYMS): %.sym:
+	@mkdir -p $(@D)
+	$(DUMP_SYMS) $< | $(AWK) -f $(objroot)include/jemalloc/internal/private_symbols.awk > $@
+
+$(C_JET_SYMS): %.sym:
+	@mkdir -p $(@D)
+	$(DUMP_SYMS) $< | $(AWK) -f $(objroot)include/jemalloc/internal/private_symbols_jet.awk > $@
+
+$(objroot)include/jemalloc/internal/private_namespace.gen.h: $(C_SYMS)
+	$(SHELL) $(srcroot)include/jemalloc/internal/private_namespace.sh $^ > $@
+
+$(objroot)include/jemalloc/internal/private_namespace_jet.gen.h: $(C_JET_SYMS)
+	$(SHELL) $(srcroot)include/jemalloc/internal/private_namespace.sh $^ > $@
+
+%.h: %.gen.h
+	@if ! `cmp -s $< $@` ; then echo "cp $< $<"; cp $< $@ ; fi
+
+$(CPP_OBJS) $(CPP_PIC_OBJS) $(TESTS_CPP_OBJS): %.$(O):
+	@mkdir -p $(@D)
+	$(CXX) $(CXXFLAGS) -c $(CPPFLAGS) $(CTARGET) $<
+ifdef CC_MM
+	@$(CXX) -MM $(CPPFLAGS) -MT $@ -o $(@:%.$(O)=%.d) $<
+endif
+
 ifneq ($(SOREV),$(SO))
 %.$(SO) : %.$(SOREV)
 	@mkdir -p $(@D)
 	ln -sf $(<F) $@
 endif
 
-$(objroot)lib/$(LIBJEMALLOC).$(SOREV) : $(if $(PIC_CFLAGS),$(C_PIC_OBJS),$(C_OBJS))
+$(objroot)lib/$(LIBJEMALLOC).$(SOREV) : $(if $(PIC_CFLAGS),$(C_PIC_OBJS),$(C_OBJS)) $(if $(PIC_CFLAGS),$(CPP_PIC_OBJS),$(CPP_OBJS))
 	@mkdir -p $(@D)
 	$(CC) $(DSO_LDFLAGS) $(call RPATH,$(RPATH_EXTRA)) $(LDTARGET) $+ $(LDFLAGS) $(LIBS) $(EXTRA_LDFLAGS)
 
-$(objroot)lib/$(LIBJEMALLOC)_pic.$(A) : $(C_PIC_OBJS)
-$(objroot)lib/$(LIBJEMALLOC).$(A) : $(C_OBJS)
-$(objroot)lib/$(LIBJEMALLOC)_s.$(A) : $(C_OBJS)
+$(objroot)lib/$(LIBJEMALLOC)_pic.$(A) : $(C_PIC_OBJS) $(CPP_PIC_OBJS)
+$(objroot)lib/$(LIBJEMALLOC).$(A) : $(C_OBJS) $(CPP_OBJS)
+$(objroot)lib/$(LIBJEMALLOC)_s.$(A) : $(C_OBJS) $(CPP_OBJS)
 
 $(STATIC_LIBS):
 	@mkdir -p $(@D)
 	$(AR) $(ARFLAGS)@AROUT@ $+
 
-$(objroot)test/unit/%$(EXE): $(objroot)test/unit/%.$(O) $(TESTS_UNIT_LINK_OBJS) $(C_JET_OBJS) $(C_TESTLIB_UNIT_OBJS)
+$(objroot)test/unit/%$(EXE): $(objroot)test/unit/%.$(O) $(C_JET_OBJS) $(C_TESTLIB_UNIT_OBJS)
 	@mkdir -p $(@D)
-	$(CC) $(LDTARGET) $(filter %.$(O),$^) $(call RPATH,$(objroot)lib) $(LDFLAGS) $(filter-out -lm,$(LIBS)) -lm $(EXTRA_LDFLAGS)
+	$(CC) $(LDTARGET) $(filter %.$(O),$^) $(call RPATH,$(objroot)lib) $(LDFLAGS) $(filter-out -lm,$(LIBS)) $(LM) $(EXTRA_LDFLAGS)
 
 $(objroot)test/integration/%$(EXE): $(objroot)test/integration/%.$(O) $(C_TESTLIB_INTEGRATION_OBJS) $(C_UTIL_INTEGRATION_OBJS) $(objroot)lib/$(LIBJEMALLOC).$(IMPORTLIB)
 	@mkdir -p $(@D)
-	$(CC) $(LDTARGET) $(filter %.$(O),$^) $(call RPATH,$(objroot)lib) $(objroot)lib/$(LIBJEMALLOC).$(IMPORTLIB) $(LDFLAGS) $(filter-out -lm,$(filter -lpthread,$(LIBS))) -lm $(EXTRA_LDFLAGS)
+	$(CC) $(TEST_LD_MODE) $(LDTARGET) $(filter %.$(O),$^) $(call RPATH,$(objroot)lib) $(LJEMALLOC) $(LDFLAGS) $(filter-out -lm,$(filter -lrt -lpthread -lstdc++,$(LIBS))) $(LM) $(EXTRA_LDFLAGS)
+
+$(objroot)test/integration/cpp/%$(EXE): $(objroot)test/integration/cpp/%.$(O) $(C_TESTLIB_INTEGRATION_OBJS) $(C_UTIL_INTEGRATION_OBJS) $(objroot)lib/$(LIBJEMALLOC).$(IMPORTLIB)
+	@mkdir -p $(@D)
+	$(CXX) $(LDTARGET) $(filter %.$(O),$^) $(call RPATH,$(objroot)lib) $(objroot)lib/$(LIBJEMALLOC).$(IMPORTLIB) $(LDFLAGS) $(filter-out -lm,$(LIBS)) -lm $(EXTRA_LDFLAGS)
 
 $(objroot)test/stress/%$(EXE): $(objroot)test/stress/%.$(O) $(C_JET_OBJS) $(C_TESTLIB_STRESS_OBJS) $(objroot)lib/$(LIBJEMALLOC).$(IMPORTLIB)
 	@mkdir -p $(@D)
-	$(CC) $(LDTARGET) $(filter %.$(O),$^) $(call RPATH,$(objroot)lib) $(objroot)lib/$(LIBJEMALLOC).$(IMPORTLIB) $(LDFLAGS) $(filter-out -lm,$(LIBS)) -lm $(EXTRA_LDFLAGS)
+	$(CC) $(TEST_LD_MODE) $(LDTARGET) $(filter %.$(O),$^) $(call RPATH,$(objroot)lib) $(objroot)lib/$(LIBJEMALLOC).$(IMPORTLIB) $(LDFLAGS) $(filter-out -lm,$(LIBS)) $(LM) $(EXTRA_LDFLAGS)
 
 build_lib_shared: $(DSOS)
 build_lib_static: $(STATIC_LIBS)
 build_lib: build_lib_shared build_lib_static
 
 install_bin:
-	install -d $(BINDIR)
+	$(INSTALL) -d $(BINDIR)
 	@for b in $(BINS); do \
-	echo "install -m 755 $$b $(BINDIR)"; \
-	install -m 755 $$b $(BINDIR); \
+	echo "$(INSTALL) -m 755 $$b $(BINDIR)"; \
+	$(INSTALL) -m 755 $$b $(BINDIR); \
 done
 
 install_include:
-	install -d $(INCLUDEDIR)/jemalloc
+	$(INSTALL) -d $(INCLUDEDIR)/jemalloc
 	@for h in $(C_HDRS); do \
-	echo "install -m 644 $$h $(INCLUDEDIR)/jemalloc"; \
-	install -m 644 $$h $(INCLUDEDIR)/jemalloc; \
+	echo "$(INSTALL) -m 644 $$h $(INCLUDEDIR)/jemalloc"; \
+	$(INSTALL) -m 644 $$h $(INCLUDEDIR)/jemalloc; \
 done
 
 install_lib_shared: $(DSOS)
-	install -d $(LIBDIR)
-	install -m 755 $(objroot)lib/$(LIBJEMALLOC).$(SOREV) $(LIBDIR)
+	$(INSTALL) -d $(LIBDIR)
+	$(INSTALL) -m 755 $(objroot)lib/$(LIBJEMALLOC).$(SOREV) $(LIBDIR)
 ifneq ($(SOREV),$(SO))
 	ln -sf $(LIBJEMALLOC).$(SOREV) $(LIBDIR)/$(LIBJEMALLOC).$(SO)
 endif
 
 install_lib_static: $(STATIC_LIBS)
-	install -d $(LIBDIR)
+	$(INSTALL) -d $(LIBDIR)
 	@for l in $(STATIC_LIBS); do \
-	echo "install -m 755 $$l $(LIBDIR)"; \
-	install -m 755 $$l $(LIBDIR); \
+	echo "$(INSTALL) -m 755 $$l $(LIBDIR)"; \
+	$(INSTALL) -m 755 $$l $(LIBDIR); \
 done
 
-install_lib: install_lib_shared install_lib_static
+install_lib_pc: $(PC)
+	$(INSTALL) -d $(LIBDIR)/pkgconfig
+	@for l in $(PC); do \
+	echo "$(INSTALL) -m 644 $$l $(LIBDIR)/pkgconfig"; \
+	$(INSTALL) -m 644 $$l $(LIBDIR)/pkgconfig; \
+done
+
+install_lib: install_lib_shared install_lib_static install_lib_pc
 
 install_doc_html:
-	install -d $(DATADIR)/doc/jemalloc$(install_suffix)
+	$(INSTALL) -d $(DATADIR)/doc/jemalloc$(install_suffix)
 	@for d in $(DOCS_HTML); do \
-	echo "install -m 644 $$d $(DATADIR)/doc/jemalloc$(install_suffix)"; \
-	install -m 644 $$d $(DATADIR)/doc/jemalloc$(install_suffix); \
+	echo "$(INSTALL) -m 644 $$d $(DATADIR)/doc/jemalloc$(install_suffix)"; \
+	$(INSTALL) -m 644 $$d $(DATADIR)/doc/jemalloc$(install_suffix); \
 done
 
 install_doc_man:
-	install -d $(MANDIR)/man3
+	$(INSTALL) -d $(MANDIR)/man3
 	@for d in $(DOCS_MAN3); do \
-	echo "install -m 644 $$d $(MANDIR)/man3"; \
-	install -m 644 $$d $(MANDIR)/man3; \
+	echo "$(INSTALL) -m 644 $$d $(MANDIR)/man3"; \
+	$(INSTALL) -m 644 $$d $(MANDIR)/man3; \
 done
 
 install_doc: install_doc_html install_doc_man
@@ -322,7 +470,7 @@ install_doc: install_doc_html install_doc_man
 install: install_bin install_include install_lib install_doc
 
 tests_unit: $(TESTS_UNIT:$(srcroot)%.c=$(objroot)%$(EXE))
-tests_integration: $(TESTS_INTEGRATION:$(srcroot)%.c=$(objroot)%$(EXE))
+tests_integration: $(TESTS_INTEGRATION:$(srcroot)%.c=$(objroot)%$(EXE)) $(TESTS_INTEGRATION_CPP:$(srcroot)%.cpp=$(objroot)%$(EXE))
 tests_stress: $(TESTS_STRESS:$(srcroot)%.c=$(objroot)%$(EXE))
 tests: tests_unit tests_integration tests_stress
 
@@ -330,78 +478,61 @@ check_unit_dir:
 	@mkdir -p $(objroot)test/unit
 check_integration_dir:
 	@mkdir -p $(objroot)test/integration
-check_stress_dir:
+stress_dir:
 	@mkdir -p $(objroot)test/stress
-check_dir: check_unit_dir check_integration_dir check_stress_dir
+check_dir: check_unit_dir check_integration_dir
 
 check_unit: tests_unit check_unit_dir
 	$(SHELL) $(objroot)test/test.sh $(TESTS_UNIT:$(srcroot)%.c=$(objroot)%)
-check_integration: tests_integration check_integration_dir
-	$(SHELL) $(objroot)test/test.sh $(TESTS_INTEGRATION:$(srcroot)%.c=$(objroot)%)
-check_stress: tests_stress check_stress_dir
-	$(SHELL) $(objroot)test/test.sh $(TESTS_STRESS:$(srcroot)%.c=$(objroot)%)
-check: tests check_dir
-	$(SHELL) $(objroot)test/test.sh $(TESTS:$(srcroot)%.c=$(objroot)%)
-
-ifeq ($(enable_code_coverage), 1)
-coverage_unit: check_unit
-	$(SHELL) $(srcroot)coverage.sh $(srcroot)src jet $(C_JET_OBJS)
-	$(SHELL) $(srcroot)coverage.sh $(srcroot)test/src unit $(C_TESTLIB_UNIT_OBJS)
-	$(SHELL) $(srcroot)coverage.sh $(srcroot)test/unit unit $(TESTS_UNIT_OBJS)
-
-coverage_integration: check_integration
-	$(SHELL) $(srcroot)coverage.sh $(srcroot)src pic $(C_PIC_OBJS)
-	$(SHELL) $(srcroot)coverage.sh $(srcroot)src integration $(C_UTIL_INTEGRATION_OBJS)
-	$(SHELL) $(srcroot)coverage.sh $(srcroot)test/src integration $(C_TESTLIB_INTEGRATION_OBJS)
-	$(SHELL) $(srcroot)coverage.sh $(srcroot)test/integration integration $(TESTS_INTEGRATION_OBJS)
-
-coverage_stress: check_stress
-	$(SHELL) $(srcroot)coverage.sh $(srcroot)src pic $(C_PIC_OBJS)
-	$(SHELL) $(srcroot)coverage.sh $(srcroot)src jet $(C_JET_OBJS)
-	$(SHELL) $(srcroot)coverage.sh $(srcroot)test/src stress $(C_TESTLIB_STRESS_OBJS)
-	$(SHELL) $(srcroot)coverage.sh $(srcroot)test/stress stress $(TESTS_STRESS_OBJS)
-
-coverage: check
-	$(SHELL) $(srcroot)coverage.sh $(srcroot)src pic $(C_PIC_OBJS)
-	$(SHELL) $(srcroot)coverage.sh $(srcroot)src jet $(C_JET_OBJS)
-	$(SHELL) $(srcroot)coverage.sh $(srcroot)src integration $(C_UTIL_INTEGRATION_OBJS)
-	$(SHELL) $(srcroot)coverage.sh $(srcroot)test/src unit $(C_TESTLIB_UNIT_OBJS)
-	$(SHELL) $(srcroot)coverage.sh $(srcroot)test/src integration $(C_TESTLIB_INTEGRATION_OBJS)
-	$(SHELL) $(srcroot)coverage.sh $(srcroot)test/src stress $(C_TESTLIB_STRESS_OBJS)
-	$(SHELL) $(srcroot)coverage.sh $(srcroot)test/unit unit $(TESTS_UNIT_OBJS) $(TESTS_UNIT_AUX_OBJS)
-	$(SHELL) $(srcroot)coverage.sh $(srcroot)test/integration integration $(TESTS_INTEGRATION_OBJS)
-	$(SHELL) $(srcroot)coverage.sh $(srcroot)test/stress integration $(TESTS_STRESS_OBJS)
+check_integration_prof: tests_integration check_integration_dir
+ifeq ($(enable_prof), 1)
+	$(MALLOC_CONF)="prof:true" $(SHELL) $(objroot)test/test.sh $(TESTS_INTEGRATION:$(srcroot)%.c=$(objroot)%) $(TESTS_INTEGRATION_CPP:$(srcroot)%.cpp=$(objroot)%)
+	$(MALLOC_CONF)="prof:true,prof_active:false" $(SHELL) $(objroot)test/test.sh $(TESTS_INTEGRATION:$(srcroot)%.c=$(objroot)%) $(TESTS_INTEGRATION_CPP:$(srcroot)%.cpp=$(objroot)%)
 endif
+check_integration_decay: tests_integration check_integration_dir
+	$(MALLOC_CONF)="dirty_decay_ms:-1,muzzy_decay_ms:-1" $(SHELL) $(objroot)test/test.sh $(TESTS_INTEGRATION:$(srcroot)%.c=$(objroot)%) $(TESTS_INTEGRATION_CPP:$(srcroot)%.cpp=$(objroot)%)
+	$(MALLOC_CONF)="dirty_decay_ms:0,muzzy_decay_ms:0" $(SHELL) $(objroot)test/test.sh $(TESTS_INTEGRATION:$(srcroot)%.c=$(objroot)%) $(TESTS_INTEGRATION_CPP:$(srcroot)%.cpp=$(objroot)%)
+check_integration: tests_integration check_integration_dir
+	$(SHELL) $(objroot)test/test.sh $(TESTS_INTEGRATION:$(srcroot)%.c=$(objroot)%) $(TESTS_INTEGRATION_CPP:$(srcroot)%.cpp=$(objroot)%)
+stress: tests_stress stress_dir
+	$(SHELL) $(objroot)test/test.sh $(TESTS_STRESS:$(srcroot)%.c=$(objroot)%)
+check: check_unit check_integration check_integration_decay check_integration_prof
 
 clean:
+	rm -f $(PRIVATE_NAMESPACE_HDRS)
+	rm -f $(PRIVATE_NAMESPACE_GEN_HDRS)
+	rm -f $(C_SYM_OBJS)
+	rm -f $(C_SYMS)
 	rm -f $(C_OBJS)
+	rm -f $(CPP_OBJS)
 	rm -f $(C_PIC_OBJS)
+	rm -f $(CPP_PIC_OBJS)
+	rm -f $(C_JET_SYM_OBJS)
+	rm -f $(C_JET_SYMS)
 	rm -f $(C_JET_OBJS)
 	rm -f $(C_TESTLIB_OBJS)
+	rm -f $(C_SYM_OBJS:%.$(O)=%.d)
 	rm -f $(C_OBJS:%.$(O)=%.d)
-	rm -f $(C_OBJS:%.$(O)=%.gcda)
-	rm -f $(C_OBJS:%.$(O)=%.gcno)
+	rm -f $(CPP_OBJS:%.$(O)=%.d)
 	rm -f $(C_PIC_OBJS:%.$(O)=%.d)
-	rm -f $(C_PIC_OBJS:%.$(O)=%.gcda)
-	rm -f $(C_PIC_OBJS:%.$(O)=%.gcno)
+	rm -f $(CPP_PIC_OBJS:%.$(O)=%.d)
+	rm -f $(C_JET_SYM_OBJS:%.$(O)=%.d)
 	rm -f $(C_JET_OBJS:%.$(O)=%.d)
-	rm -f $(C_JET_OBJS:%.$(O)=%.gcda)
-	rm -f $(C_JET_OBJS:%.$(O)=%.gcno)
 	rm -f $(C_TESTLIB_OBJS:%.$(O)=%.d)
-	rm -f $(C_TESTLIB_OBJS:%.$(O)=%.gcda)
-	rm -f $(C_TESTLIB_OBJS:%.$(O)=%.gcno)
 	rm -f $(TESTS_OBJS:%.$(O)=%$(EXE))
 	rm -f $(TESTS_OBJS)
 	rm -f $(TESTS_OBJS:%.$(O)=%.d)
-	rm -f $(TESTS_OBJS:%.$(O)=%.gcda)
-	rm -f $(TESTS_OBJS:%.$(O)=%.gcno)
 	rm -f $(TESTS_OBJS:%.$(O)=%.out)
+	rm -f $(TESTS_CPP_OBJS:%.$(O)=%$(EXE))
+	rm -f $(TESTS_CPP_OBJS)
+	rm -f $(TESTS_CPP_OBJS:%.$(O)=%.d)
+	rm -f $(TESTS_CPP_OBJS:%.$(O)=%.out)
 	rm -f $(DSOS) $(STATIC_LIBS)
-	rm -f $(objroot)*.gcov.*
 
 distclean: clean
-	rm -rf $(objroot)autom4te.cache
+	rm -f $(objroot)bin/jemalloc-config
 	rm -f $(objroot)bin/jemalloc.sh
+	rm -f $(objroot)bin/jeprof
 	rm -f $(objroot)config.log
 	rm -f $(objroot)config.status
 	rm -f $(objroot)config.stamp
@@ -410,7 +541,7 @@ distclean: clean
 
 relclean: distclean
 	rm -f $(objroot)configure
-	rm -f $(srcroot)VERSION
+	rm -f $(objroot)VERSION
 	rm -f $(DOCS_HTML)
 	rm -f $(DOCS_MAN3)
 

README

@@ -3,12 +3,12 @@ fragmentation avoidance and scalable concurrency support.  jemalloc first came
 into use as the FreeBSD libc allocator in 2005, and since then it has found its
 way into numerous applications that rely on its predictable behavior.  In 2010
 jemalloc development efforts broadened to include developer support features
-such as heap profiling, Valgrind integration, and extensive monitoring/tuning
-hooks.  Modern jemalloc releases continue to be integrated back into FreeBSD,
-and therefore versatility remains critical.  Ongoing development efforts trend
-toward making jemalloc among the best allocators for a broad range of demanding
-applications, and eliminating/mitigating weaknesses that have practical
-repercussions for real world applications.
+such as heap profiling and extensive monitoring/tuning hooks.  Modern jemalloc
+releases continue to be integrated back into FreeBSD, and therefore versatility
+remains critical.  Ongoing development efforts trend toward making jemalloc
+among the best allocators for a broad range of demanding applications, and
+eliminating/mitigating weaknesses that have practical repercussions for real
+world applications.
 
 The COPYING file contains copyright and licensing information.
 
@@ -17,4 +17,4 @@ jemalloc.
 
 The ChangeLog file contains a brief summary of changes for each release.
 
-URL: http://www.canonware.com/jemalloc/
+URL: http://jemalloc.net/

TUNING.md

@@ -0,0 +1,129 @@
+This document summarizes the common approaches for performance fine tuning with
+jemalloc (as of 5.1.0).  The default configuration of jemalloc tends to work
+reasonably well in practice, and most applications should not have to tune any
+options. However, in order to cover a wide range of applications and avoid
+pathological cases, the default setting is sometimes kept conservative and
+suboptimal, even for many common workloads.  When jemalloc is properly tuned for
+a specific application / workload, it is common to improve system level metrics
+by a few percent, or make favorable trade-offs.
+
+
+## Notable runtime options for performance tuning
+
+Runtime options can be set via
+[malloc_conf](http://jemalloc.net/jemalloc.3.html#tuning).
+
+* [background_thread](http://jemalloc.net/jemalloc.3.html#background_thread)
+
+    Enabling jemalloc background threads generally improves the tail latency for
+    application threads, since unused memory purging is shifted to the dedicated
+    background threads.  In addition, unintended purging delay caused by
+    application inactivity is avoided with background threads.
+
+    Suggested: `background_thread:true` when jemalloc managed threads can be
+    allowed.
+
+* [metadata_thp](http://jemalloc.net/jemalloc.3.html#opt.metadata_thp)
+
+    Allowing jemalloc to utilize transparent huge pages for its internal
+    metadata usually reduces TLB misses significantly, especially for programs
+    with large memory footprint and frequent allocation / deallocation
+    activities.  Metadata memory usage may increase due to the use of huge
+    pages.
+
+    Suggested for allocation intensive programs: `metadata_thp:auto` or
+    `metadata_thp:always`, which is expected to improve CPU utilization at a
+    small memory cost.
+
+* [dirty_decay_ms](http://jemalloc.net/jemalloc.3.html#opt.dirty_decay_ms) and
+  [muzzy_decay_ms](http://jemalloc.net/jemalloc.3.html#opt.muzzy_decay_ms)
+
+    Decay time determines how fast jemalloc returns unused pages back to the
+    operating system, and therefore provides a fairly straightforward trade-off
+    between CPU and memory usage.  Shorter decay time purges unused pages faster
+    to reduces memory usage (usually at the cost of more CPU cycles spent on
+    purging), and vice versa.
+
+    Suggested: tune the values based on the desired trade-offs.
+
+* [narenas](http://jemalloc.net/jemalloc.3.html#opt.narenas)
+
+    By default jemalloc uses multiple arenas to reduce internal lock contention.
+    However high arena count may also increase overall memory fragmentation,
+    since arenas manage memory independently.  When high degree of parallelism
+    is not expected at the allocator level, lower number of arenas often
+    improves memory usage.
+
+    Suggested: if low parallelism is expected, try lower arena count while
+    monitoring CPU and memory usage.
+
+* [percpu_arena](http://jemalloc.net/jemalloc.3.html#opt.percpu_arena)
+
+    Enable dynamic thread to arena association based on running CPU.  This has
+    the potential to improve locality, e.g. when thread to CPU affinity is
+    present.
+    
+    Suggested: try `percpu_arena:percpu` or `percpu_arena:phycpu` if
+    thread migration between processors is expected to be infrequent.
+
+Examples:
+
+* High resource consumption application, prioritizing CPU utilization:
+
+    `background_thread:true,metadata_thp:auto` combined with relaxed decay time
+    (increased `dirty_decay_ms` and / or `muzzy_decay_ms`,
+    e.g. `dirty_decay_ms:30000,muzzy_decay_ms:30000`).
+
+* High resource consumption application, prioritizing memory usage:
+
+    `background_thread:true` combined with shorter decay time (decreased
+    `dirty_decay_ms` and / or `muzzy_decay_ms`,
+    e.g. `dirty_decay_ms:5000,muzzy_decay_ms:5000`), and lower arena count
+    (e.g. number of CPUs).
+
+* Low resource consumption application:
+
+    `narenas:1,lg_tcache_max:13` combined with shorter decay time (decreased
+    `dirty_decay_ms` and / or `muzzy_decay_ms`,e.g.
+    `dirty_decay_ms:1000,muzzy_decay_ms:0`).
+
+* Extremely conservative -- minimize memory usage at all costs, only suitable when
+allocation activity is very rare:
+
+    `narenas:1,tcache:false,dirty_decay_ms:0,muzzy_decay_ms:0`
+
+Note that it is recommended to combine the options with `abort_conf:true` which
+aborts immediately on illegal options.
+
+## Beyond runtime options
+
+In addition to the runtime options, there are a number of programmatic ways to
+improve application performance with jemalloc.
+
+* [Explicit arenas](http://jemalloc.net/jemalloc.3.html#arenas.create)
+
+    Manually created arenas can help performance in various ways, e.g. by
+    managing locality and contention for specific usages.  For example,
+    applications can explicitly allocate frequently accessed objects from a
+    dedicated arena with
+    [mallocx()](http://jemalloc.net/jemalloc.3.html#MALLOCX_ARENA) to improve
+    locality.  In addition, explicit arenas often benefit from individually
+    tuned options, e.g. relaxed [decay
+    time](http://jemalloc.net/jemalloc.3.html#arena.i.dirty_decay_ms) if
+    frequent reuse is expected.
+
+* [Extent hooks](http://jemalloc.net/jemalloc.3.html#arena.i.extent_hooks)
+
+    Extent hooks allow customization for managing underlying memory.  One use
+    case for performance purpose is to utilize huge pages -- for example,
+    [HHVM](https://github.com/facebook/hhvm/blob/master/hphp/util/alloc.cpp)
+    uses explicit arenas with customized extent hooks to manage 1GB huge pages
+    for frequently accessed data, which reduces TLB misses significantly.
+
+* [Explicit thread-to-arena
+  binding](http://jemalloc.net/jemalloc.3.html#thread.arena)
+
+    It is common for some threads in an application to have different memory
+    access / allocation patterns.  Threads with heavy workloads often benefit
+    from explicit binding, e.g. binding very active threads to dedicated arenas
+    may reduce contention at the allocator level.

bin/jemalloc-config.in

@@ -0,0 +1,83 @@
+#!/bin/sh
+
+usage() {
+	cat <<EOF
+Usage:
+  @BINDIR@/jemalloc-config <option>
+Options:
+  --help | -h  : Print usage.
+  --version    : Print jemalloc version.
+  --revision   : Print shared library revision number.
+  --config     : Print configure options used to build jemalloc.
+  --prefix     : Print installation directory prefix.
+  --bindir     : Print binary installation directory.
+  --datadir    : Print data installation directory.
+  --includedir : Print include installation directory.
+  --libdir     : Print library installation directory.
+  --mandir     : Print manual page installation directory.
+  --cc         : Print compiler used to build jemalloc.
+  --cflags     : Print compiler flags used to build jemalloc.
+  --cppflags   : Print preprocessor flags used to build jemalloc.
+  --cxxflags   : Print C++ compiler flags used to build jemalloc.
+  --ldflags    : Print library flags used to build jemalloc.
+  --libs       : Print libraries jemalloc was linked against.
+EOF
+}
+
+prefix="@prefix@"
+exec_prefix="@exec_prefix@"
+
+case "$1" in
+--help | -h)
+	usage
+	exit 0
+	;;
+--version)
+	echo "@jemalloc_version@"
+	;;
+--revision)
+	echo "@rev@"
+	;;
+--config)
+	echo "@CONFIG@"
+	;;
+--prefix)
+	echo "@PREFIX@"
+	;;
+--bindir)
+	echo "@BINDIR@"
+	;;
+--datadir)
+	echo "@DATADIR@"
+	;;
+--includedir)
+	echo "@INCLUDEDIR@"
+	;;
+--libdir)
+	echo "@LIBDIR@"
+	;;
+--mandir)
+	echo "@MANDIR@"
+	;;
+--cc)
+	echo "@CC@"
+	;;
+--cflags)
+	echo "@CFLAGS@"
+	;;
+--cppflags)
+	echo "@CPPFLAGS@"
+	;;
+--cxxflags)
+	echo "@CXXFLAGS@"
+	;;
+--ldflags)
+	echo "@LDFLAGS@ @EXTRA_LDFLAGS@"
+	;;
+--libs)
+	echo "@LIBS@"
+	;;
+*)
+	usage
+	exit 1
+esac

build-aux/config.guess

@@ -1,8 +1,8 @@
 #! /bin/sh
 # Attempt to guess a canonical system name.
-#   Copyright 1992-2013 Free Software Foundation, Inc.
+#   Copyright 1992-2016 Free Software Foundation, Inc.
 
-timestamp='2013-06-10'
+timestamp='2016-10-02'
 
 # This file is free software; you can redistribute it and/or modify it
 # under the terms of the GNU General Public License as published by
@@ -24,12 +24,12 @@ timestamp='2013-06-10'
 # program.  This Exception is an additional permission under section 7
 # of the GNU General Public License, version 3 ("GPLv3").
 #
-# Originally written by Per Bothner.
+# Originally written by Per Bothner; maintained since 2000 by Ben Elliston.
 #
 # You can get the latest version of this script from:
-# http://git.savannah.gnu.org/gitweb/?p=config.git;a=blob_plain;f=config.guess;hb=HEAD
+# http://git.savannah.gnu.org/gitweb/?p=config.git;a=blob_plain;f=config.guess
 #
-# Please send patches with a ChangeLog entry to config-patches@gnu.org.
+# Please send patches to <config-patches@gnu.org>.
 
 
 me=`echo "$0" | sed -e 's,.*/,,'`
@@ -50,7 +50,7 @@ version="\
 GNU config.guess ($timestamp)
 
 Originally written by Per Bothner.
-Copyright 1992-2013 Free Software Foundation, Inc.
+Copyright 1992-2016 Free Software Foundation, Inc.
 
 This is free software; see the source for copying conditions.  There is NO
 warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE."
@@ -149,7 +149,7 @@ Linux|GNU|GNU/*)
 	LIBC=gnu
 	#endif
 	EOF
-	eval `$CC_FOR_BUILD -E $dummy.c 2>/dev/null | grep '^LIBC'`
+	eval `$CC_FOR_BUILD -E $dummy.c 2>/dev/null | grep '^LIBC' | sed 's, ,,g'`
 	;;
 esac
 
@@ -168,19 +168,29 @@ case "${UNAME_MACHINE}:${UNAME_SYSTEM}:${UNAME_RELEASE}:${UNAME_VERSION}" in
 	# Note: NetBSD doesn't particularly care about the vendor
 	# portion of the name.  We always set it to "unknown".
 	sysctl="sysctl -n hw.machine_arch"
-	UNAME_MACHINE_ARCH=`(/sbin/$sysctl 2>/dev/null || \
-	    /usr/sbin/$sysctl 2>/dev/null || echo unknown)`
+	UNAME_MACHINE_ARCH=`(uname -p 2>/dev/null || \
+	    /sbin/$sysctl 2>/dev/null || \
+	    /usr/sbin/$sysctl 2>/dev/null || \
+	    echo unknown)`
 	case "${UNAME_MACHINE_ARCH}" in
 	    armeb) machine=armeb-unknown ;;
 	    arm*) machine=arm-unknown ;;
 	    sh3el) machine=shl-unknown ;;
 	    sh3eb) machine=sh-unknown ;;
 	    sh5el) machine=sh5le-unknown ;;
+	    earmv*)
+		arch=`echo ${UNAME_MACHINE_ARCH} | sed -e 's,^e\(armv[0-9]\).*$,\1,'`
+		endian=`echo ${UNAME_MACHINE_ARCH} | sed -ne 's,^.*\(eb\)$,\1,p'`
+		machine=${arch}${endian}-unknown
+		;;
 	    *) machine=${UNAME_MACHINE_ARCH}-unknown ;;
 	esac
 	# The Operating System including object format, if it has switched
-	# to ELF recently, or will in the future.
+	# to ELF recently (or will in the future) and ABI.
 	case "${UNAME_MACHINE_ARCH}" in
+	    earm*)
+		os=netbsdelf
+		;;
 	    arm*|i386|m68k|ns32k|sh3*|sparc|vax)
 		eval $set_cc_for_build
 		if echo __ELF__ | $CC_FOR_BUILD -E - 2>/dev/null \
@@ -197,6 +207,13 @@ case "${UNAME_MACHINE}:${UNAME_SYSTEM}:${UNAME_RELEASE}:${UNAME_VERSION}" in
 		os=netbsd
 		;;
 	esac
+	# Determine ABI tags.
+	case "${UNAME_MACHINE_ARCH}" in
+	    earm*)
+		expr='s/^earmv[0-9]/-eabi/;s/eb$//'
+		abi=`echo ${UNAME_MACHINE_ARCH} | sed -e "$expr"`
+		;;
+	esac
 	# The OS release
 	# Debian GNU/NetBSD machines have a different userland, and
 	# thus, need a distinct triplet. However, they do not need
@@ -207,13 +224,13 @@ case "${UNAME_MACHINE}:${UNAME_SYSTEM}:${UNAME_RELEASE}:${UNAME_VERSION}" in
 		release='-gnu'
 		;;
 	    *)
-		release=`echo ${UNAME_RELEASE}|sed -e 's/[-_].*/\./'`
+		release=`echo ${UNAME_RELEASE} | sed -e 's/[-_].*//' | cut -d. -f1,2`
 		;;
 	esac
 	# Since CPU_TYPE-MANUFACTURER-KERNEL-OPERATING_SYSTEM:
 	# contains redundant information, the shorter form:
 	# CPU_TYPE-MANUFACTURER-OPERATING_SYSTEM is used.
-	echo "${machine}-${os}${release}"
+	echo "${machine}-${os}${release}${abi}"
 	exit ;;
     *:Bitrig:*:*)
 	UNAME_MACHINE_ARCH=`arch | sed 's/Bitrig.//'`
@@ -223,6 +240,10 @@ case "${UNAME_MACHINE}:${UNAME_SYSTEM}:${UNAME_RELEASE}:${UNAME_VERSION}" in
 	UNAME_MACHINE_ARCH=`arch | sed 's/OpenBSD.//'`
 	echo ${UNAME_MACHINE_ARCH}-unknown-openbsd${UNAME_RELEASE}
 	exit ;;
+    *:LibertyBSD:*:*)
+	UNAME_MACHINE_ARCH=`arch | sed 's/^.*BSD\.//'`
+	echo ${UNAME_MACHINE_ARCH}-unknown-libertybsd${UNAME_RELEASE}
+	exit ;;
     *:ekkoBSD:*:*)
 	echo ${UNAME_MACHINE}-unknown-ekkobsd${UNAME_RELEASE}
 	exit ;;
@@ -235,6 +256,9 @@ case "${UNAME_MACHINE}:${UNAME_SYSTEM}:${UNAME_RELEASE}:${UNAME_VERSION}" in
     *:MirBSD:*:*)
 	echo ${UNAME_MACHINE}-unknown-mirbsd${UNAME_RELEASE}
 	exit ;;
+    *:Sortix:*:*)
+	echo ${UNAME_MACHINE}-unknown-sortix
+	exit ;;
     alpha:OSF1:*:*)
 	case $UNAME_RELEASE in
 	*4.0)
@@ -251,42 +275,42 @@ case "${UNAME_MACHINE}:${UNAME_SYSTEM}:${UNAME_RELEASE}:${UNAME_VERSION}" in
 	ALPHA_CPU_TYPE=`/usr/sbin/psrinfo -v | sed -n -e 's/^  The alpha \(.*\) processor.*$/\1/p' | head -n 1`
 	case "$ALPHA_CPU_TYPE" in
 	    "EV4 (21064)")
-		UNAME_MACHINE="alpha" ;;
+		UNAME_MACHINE=alpha ;;
 	    "EV4.5 (21064)")
-		UNAME_MACHINE="alpha" ;;
+		UNAME_MACHINE=alpha ;;
 	    "LCA4 (21066/21068)")
-		UNAME_MACHINE="alpha" ;;
+		UNAME_MACHINE=alpha ;;
 	    "EV5 (21164)")
-		UNAME_MACHINE="alphaev5" ;;
+		UNAME_MACHINE=alphaev5 ;;
 	    "EV5.6 (21164A)")
-		UNAME_MACHINE="alphaev56" ;;
+		UNAME_MACHINE=alphaev56 ;;
 	    "EV5.6 (21164PC)")
-		UNAME_MACHINE="alphapca56" ;;
+		UNAME_MACHINE=alphapca56 ;;
 	    "EV5.7 (21164PC)")
-		UNAME_MACHINE="alphapca57" ;;
+		UNAME_MACHINE=alphapca57 ;;
 	    "EV6 (21264)")
-		UNAME_MACHINE="alphaev6" ;;
+		UNAME_MACHINE=alphaev6 ;;
 	    "EV6.7 (21264A)")
-		UNAME_MACHINE="alphaev67" ;;
+		UNAME_MACHINE=alphaev67 ;;
 	    "EV6.8CB (21264C)")
-		UNAME_MACHINE="alphaev68" ;;
+		UNAME_MACHINE=alphaev68 ;;
 	    "EV6.8AL (21264B)")
-		UNAME_MACHINE="alphaev68" ;;
+		UNAME_MACHINE=alphaev68 ;;
 	    "EV6.8CX (21264D)")
-		UNAME_MACHINE="alphaev68" ;;
+		UNAME_MACHINE=alphaev68 ;;
 	    "EV6.9A (21264/EV69A)")
-		UNAME_MACHINE="alphaev69" ;;
+		UNAME_MACHINE=alphaev69 ;;
 	    "EV7 (21364)")
-		UNAME_MACHINE="alphaev7" ;;
+		UNAME_MACHINE=alphaev7 ;;
 	    "EV7.9 (21364A)")
-		UNAME_MACHINE="alphaev79" ;;
+		UNAME_MACHINE=alphaev79 ;;
 	esac
 	# A Pn.n version is a patched version.
 	# A Vn.n version is a released version.
 	# A Tn.n version is a released field test version.
 	# A Xn.n version is an unreleased experimental baselevel.
 	# 1.2 uses "1.2" for uname -r.
-	echo ${UNAME_MACHINE}-dec-osf`echo ${UNAME_RELEASE} | sed -e 's/^[PVTX]//' | tr 'ABCDEFGHIJKLMNOPQRSTUVWXYZ' 'abcdefghijklmnopqrstuvwxyz'`
+	echo ${UNAME_MACHINE}-dec-osf`echo ${UNAME_RELEASE} | sed -e 's/^[PVTX]//' | tr ABCDEFGHIJKLMNOPQRSTUVWXYZ abcdefghijklmnopqrstuvwxyz`
 	# Reset EXIT trap before exiting to avoid spurious non-zero exit code.
 	exitcode=$?
 	trap '' 0
@@ -359,16 +383,16 @@ case "${UNAME_MACHINE}:${UNAME_SYSTEM}:${UNAME_RELEASE}:${UNAME_VERSION}" in
 	exit ;;
     i86pc:SunOS:5.*:* | i86xen:SunOS:5.*:*)
 	eval $set_cc_for_build
-	SUN_ARCH="i386"
+	SUN_ARCH=i386
 	# If there is a compiler, see if it is configured for 64-bit objects.
 	# Note that the Sun cc does not turn __LP64__ into 1 like gcc does.
 	# This test works for both compilers.
-	if [ "$CC_FOR_BUILD" != 'no_compiler_found' ]; then
+	if [ "$CC_FOR_BUILD" != no_compiler_found ]; then
 	    if (echo '#ifdef __amd64'; echo IS_64BIT_ARCH; echo '#endif') | \
-		(CCOPTS= $CC_FOR_BUILD -E - 2>/dev/null) | \
+		(CCOPTS="" $CC_FOR_BUILD -E - 2>/dev/null) | \
 		grep IS_64BIT_ARCH >/dev/null
 	    then
-		SUN_ARCH="x86_64"
+		SUN_ARCH=x86_64
 	    fi
 	fi
 	echo ${SUN_ARCH}-pc-solaris2`echo ${UNAME_RELEASE}|sed -e 's/[^.]*//'`
@@ -393,7 +417,7 @@ case "${UNAME_MACHINE}:${UNAME_SYSTEM}:${UNAME_RELEASE}:${UNAME_VERSION}" in
 	exit ;;
     sun*:*:4.2BSD:*)
 	UNAME_RELEASE=`(sed 1q /etc/motd | awk '{print substr($5,1,3)}') 2>/dev/null`
-	test "x${UNAME_RELEASE}" = "x" && UNAME_RELEASE=3
+	test "x${UNAME_RELEASE}" = x && UNAME_RELEASE=3
 	case "`/bin/arch`" in
 	    sun3)
 		echo m68k-sun-sunos${UNAME_RELEASE}
@@ -579,8 +603,9 @@ EOF
 	else
 		IBM_ARCH=powerpc
 	fi
-	if [ -x /usr/bin/oslevel ] ; then
-		IBM_REV=`/usr/bin/oslevel`
+	if [ -x /usr/bin/lslpp ] ; then
+		IBM_REV=`/usr/bin/lslpp -Lqc bos.rte.libc |
+			   awk -F: '{ print $3 }' | sed s/[0-9]*$/0/`
 	else
 		IBM_REV=${UNAME_VERSION}.${UNAME_RELEASE}
 	fi
@@ -617,13 +642,13 @@ EOF
 		    sc_cpu_version=`/usr/bin/getconf SC_CPU_VERSION 2>/dev/null`
 		    sc_kernel_bits=`/usr/bin/getconf SC_KERNEL_BITS 2>/dev/null`
 		    case "${sc_cpu_version}" in
-		      523) HP_ARCH="hppa1.0" ;; # CPU_PA_RISC1_0
-		      528) HP_ARCH="hppa1.1" ;; # CPU_PA_RISC1_1
+		      523) HP_ARCH=hppa1.0 ;; # CPU_PA_RISC1_0
+		      528) HP_ARCH=hppa1.1 ;; # CPU_PA_RISC1_1
 		      532)                      # CPU_PA_RISC2_0
 			case "${sc_kernel_bits}" in
-			  32) HP_ARCH="hppa2.0n" ;;
-			  64) HP_ARCH="hppa2.0w" ;;
-			  '') HP_ARCH="hppa2.0" ;;   # HP-UX 10.20
+			  32) HP_ARCH=hppa2.0n ;;
+			  64) HP_ARCH=hppa2.0w ;;
+			  '') HP_ARCH=hppa2.0 ;;   # HP-UX 10.20
 			esac ;;
 		    esac
 		fi
@@ -662,11 +687,11 @@ EOF
 		    exit (0);
 		}
 EOF
-		    (CCOPTS= $CC_FOR_BUILD -o $dummy $dummy.c 2>/dev/null) && HP_ARCH=`$dummy`
+		    (CCOPTS="" $CC_FOR_BUILD -o $dummy $dummy.c 2>/dev/null) && HP_ARCH=`$dummy`
 		    test -z "$HP_ARCH" && HP_ARCH=hppa
 		fi ;;
 	esac
-	if [ ${HP_ARCH} = "hppa2.0w" ]
+	if [ ${HP_ARCH} = hppa2.0w ]
 	then
 	    eval $set_cc_for_build
 
@@ -679,12 +704,12 @@ EOF
 	    # $ CC_FOR_BUILD="cc +DA2.0w" ./config.guess
 	    # => hppa64-hp-hpux11.23
 
-	    if echo __LP64__ | (CCOPTS= $CC_FOR_BUILD -E - 2>/dev/null) |
+	    if echo __LP64__ | (CCOPTS="" $CC_FOR_BUILD -E - 2>/dev/null) |
 		grep -q __LP64__
 	    then
-		HP_ARCH="hppa2.0w"
+		HP_ARCH=hppa2.0w
 	    else
-		HP_ARCH="hppa64"
+		HP_ARCH=hppa64
 	    fi
 	fi
 	echo ${HP_ARCH}-hp-hpux${HPUX_REV}
@@ -789,14 +814,14 @@ EOF
 	echo craynv-cray-unicosmp${UNAME_RELEASE} | sed -e 's/\.[^.]*$/.X/'
 	exit ;;
     F30[01]:UNIX_System_V:*:* | F700:UNIX_System_V:*:*)
-	FUJITSU_PROC=`uname -m | tr 'ABCDEFGHIJKLMNOPQRSTUVWXYZ' 'abcdefghijklmnopqrstuvwxyz'`
-	FUJITSU_SYS=`uname -p | tr 'ABCDEFGHIJKLMNOPQRSTUVWXYZ' 'abcdefghijklmnopqrstuvwxyz' | sed -e 's/\///'`
+	FUJITSU_PROC=`uname -m | tr ABCDEFGHIJKLMNOPQRSTUVWXYZ abcdefghijklmnopqrstuvwxyz`
+	FUJITSU_SYS=`uname -p | tr ABCDEFGHIJKLMNOPQRSTUVWXYZ abcdefghijklmnopqrstuvwxyz | sed -e 's/\///'`
 	FUJITSU_REL=`echo ${UNAME_RELEASE} | sed -e 's/ /_/'`
 	echo "${FUJITSU_PROC}-fujitsu-${FUJITSU_SYS}${FUJITSU_REL}"
 	exit ;;
     5000:UNIX_System_V:4.*:*)
-	FUJITSU_SYS=`uname -p | tr 'ABCDEFGHIJKLMNOPQRSTUVWXYZ' 'abcdefghijklmnopqrstuvwxyz' | sed -e 's/\///'`
-	FUJITSU_REL=`echo ${UNAME_RELEASE} | tr 'ABCDEFGHIJKLMNOPQRSTUVWXYZ' 'abcdefghijklmnopqrstuvwxyz' | sed -e 's/ /_/'`
+	FUJITSU_SYS=`uname -p | tr ABCDEFGHIJKLMNOPQRSTUVWXYZ abcdefghijklmnopqrstuvwxyz | sed -e 's/\///'`
+	FUJITSU_REL=`echo ${UNAME_RELEASE} | tr ABCDEFGHIJKLMNOPQRSTUVWXYZ abcdefghijklmnopqrstuvwxyz | sed -e 's/ /_/'`
 	echo "sparc-fujitsu-${FUJITSU_SYS}${FUJITSU_REL}"
 	exit ;;
     i*86:BSD/386:*:* | i*86:BSD/OS:*:* | *:Ascend\ Embedded/OS:*:*)
@@ -826,7 +851,7 @@ EOF
     *:MINGW*:*)
 	echo ${UNAME_MACHINE}-pc-mingw32
 	exit ;;
-    i*:MSYS*:*)
+    *:MSYS*:*)
 	echo ${UNAME_MACHINE}-pc-msys
 	exit ;;
     i*:windows32*:*)
@@ -878,7 +903,7 @@ EOF
 	exit ;;
     *:GNU/*:*:*)
 	# other systems with GNU libc and userland
-	echo ${UNAME_MACHINE}-unknown-`echo ${UNAME_SYSTEM} | sed 's,^[^/]*/,,' | tr '[A-Z]' '[a-z]'``echo ${UNAME_RELEASE}|sed -e 's/[-(].*//'`-${LIBC}
+	echo ${UNAME_MACHINE}-unknown-`echo ${UNAME_SYSTEM} | sed 's,^[^/]*/,,' | tr "[:upper:]" "[:lower:]"``echo ${UNAME_RELEASE}|sed -e 's/[-(].*//'`-${LIBC}
 	exit ;;
     i*86:Minix:*:*)
 	echo ${UNAME_MACHINE}-pc-minix
@@ -901,7 +926,7 @@ EOF
 	  EV68*) UNAME_MACHINE=alphaev68 ;;
 	esac
 	objdump --private-headers /bin/sh | grep -q ld.so.1
-	if test "$?" = 0 ; then LIBC="gnulibc1" ; fi
+	if test "$?" = 0 ; then LIBC=gnulibc1 ; fi
 	echo ${UNAME_MACHINE}-unknown-linux-${LIBC}
 	exit ;;
     arc:Linux:*:* | arceb:Linux:*:*)
@@ -932,6 +957,9 @@ EOF
     crisv32:Linux:*:*)
 	echo ${UNAME_MACHINE}-axis-linux-${LIBC}
 	exit ;;
+    e2k:Linux:*:*)
+	echo ${UNAME_MACHINE}-unknown-linux-${LIBC}
+	exit ;;
     frv:Linux:*:*)
 	echo ${UNAME_MACHINE}-unknown-linux-${LIBC}
 	exit ;;
@@ -944,6 +972,9 @@ EOF
     ia64:Linux:*:*)
 	echo ${UNAME_MACHINE}-unknown-linux-${LIBC}
 	exit ;;
+    k1om:Linux:*:*)
+	echo ${UNAME_MACHINE}-unknown-linux-${LIBC}
+	exit ;;
     m32r*:Linux:*:*)
 	echo ${UNAME_MACHINE}-unknown-linux-${LIBC}
 	exit ;;
@@ -969,10 +1000,13 @@ EOF
 	eval `$CC_FOR_BUILD -E $dummy.c 2>/dev/null | grep '^CPU'`
 	test x"${CPU}" != x && { echo "${CPU}-unknown-linux-${LIBC}"; exit; }
 	;;
-    or1k:Linux:*:*)
+    mips64el:Linux:*:*)
 	echo ${UNAME_MACHINE}-unknown-linux-${LIBC}
 	exit ;;
-    or32:Linux:*:*)
+    openrisc*:Linux:*:*)
+	echo or1k-unknown-linux-${LIBC}
+	exit ;;
+    or32:Linux:*:* | or1k*:Linux:*:*)
 	echo ${UNAME_MACHINE}-unknown-linux-${LIBC}
 	exit ;;
     padre:Linux:*:*)
@@ -1001,6 +1035,9 @@ EOF
     ppcle:Linux:*:*)
 	echo powerpcle-unknown-linux-${LIBC}
 	exit ;;
+    riscv32:Linux:*:* | riscv64:Linux:*:*)
+	echo ${UNAME_MACHINE}-unknown-linux-${LIBC}
+	exit ;;
     s390:Linux:*:* | s390x:Linux:*:*)
 	echo ${UNAME_MACHINE}-ibm-linux-${LIBC}
 	exit ;;
@@ -1020,7 +1057,7 @@ EOF
 	echo ${UNAME_MACHINE}-dec-linux-${LIBC}
 	exit ;;
     x86_64:Linux:*:*)
-	echo ${UNAME_MACHINE}-unknown-linux-${LIBC}
+	echo ${UNAME_MACHINE}-pc-linux-${LIBC}
 	exit ;;
     xtensa*:Linux:*:*)
 	echo ${UNAME_MACHINE}-unknown-linux-${LIBC}
@@ -1099,7 +1136,7 @@ EOF
 	# uname -m prints for DJGPP always 'pc', but it prints nothing about
 	# the processor, so we play safe by assuming i586.
 	# Note: whatever this is, it MUST be the same as what config.sub
-	# prints for the "djgpp" host, or else GDB configury will decide that
+	# prints for the "djgpp" host, or else GDB configure will decide that
 	# this is a cross-build.
 	echo i586-pc-msdosdjgpp
 	exit ;;
@@ -1248,6 +1285,9 @@ EOF
     SX-8R:SUPER-UX:*:*)
 	echo sx8r-nec-superux${UNAME_RELEASE}
 	exit ;;
+    SX-ACE:SUPER-UX:*:*)
+	echo sxace-nec-superux${UNAME_RELEASE}
+	exit ;;
     Power*:Rhapsody:*:*)
 	echo powerpc-apple-rhapsody${UNAME_RELEASE}
 	exit ;;
@@ -1260,22 +1300,32 @@ EOF
 	if test "$UNAME_PROCESSOR" = unknown ; then
 	    UNAME_PROCESSOR=powerpc
 	fi
-	if [ "$CC_FOR_BUILD" != 'no_compiler_found' ]; then
-	    if (echo '#ifdef __LP64__'; echo IS_64BIT_ARCH; echo '#endif') | \
-		(CCOPTS= $CC_FOR_BUILD -E - 2>/dev/null) | \
-		grep IS_64BIT_ARCH >/dev/null
-	    then
-		case $UNAME_PROCESSOR in
-		    i386) UNAME_PROCESSOR=x86_64 ;;
-		    powerpc) UNAME_PROCESSOR=powerpc64 ;;
-		esac
+	if test `echo "$UNAME_RELEASE" | sed -e 's/\..*//'` -le 10 ; then
+	    if [ "$CC_FOR_BUILD" != no_compiler_found ]; then
+		if (echo '#ifdef __LP64__'; echo IS_64BIT_ARCH; echo '#endif') | \
+		    (CCOPTS="" $CC_FOR_BUILD -E - 2>/dev/null) | \
+		    grep IS_64BIT_ARCH >/dev/null
+		then
+		    case $UNAME_PROCESSOR in
+			i386) UNAME_PROCESSOR=x86_64 ;;
+			powerpc) UNAME_PROCESSOR=powerpc64 ;;
+		    esac
+		fi
 	    fi
+	elif test "$UNAME_PROCESSOR" = i386 ; then
+	    # Avoid executing cc on OS X 10.9, as it ships with a stub
+	    # that puts up a graphical alert prompting to install
+	    # developer tools.  Any system running Mac OS X 10.7 or
+	    # later (Darwin 11 and later) is required to have a 64-bit
+	    # processor. This is not true of the ARM version of Darwin
+	    # that Apple uses in portable devices.
+	    UNAME_PROCESSOR=x86_64
 	fi
 	echo ${UNAME_PROCESSOR}-apple-darwin${UNAME_RELEASE}
 	exit ;;
     *:procnto*:*:* | *:QNX:[0123456789]*:*)
 	UNAME_PROCESSOR=`uname -p`
-	if test "$UNAME_PROCESSOR" = "x86"; then
+	if test "$UNAME_PROCESSOR" = x86; then
 		UNAME_PROCESSOR=i386
 		UNAME_MACHINE=pc
 	fi
@@ -1306,7 +1356,7 @@ EOF
 	# "uname -m" is not consistent, so use $cputype instead. 386
 	# is converted to i386 for consistency with other x86
 	# operating systems.
-	if test "$cputype" = "386"; then
+	if test "$cputype" = 386; then
 	    UNAME_MACHINE=i386
 	else
 	    UNAME_MACHINE="$cputype"
@@ -1348,7 +1398,7 @@ EOF
 	echo i386-pc-xenix
 	exit ;;
     i*86:skyos:*:*)
-	echo ${UNAME_MACHINE}-pc-skyos`echo ${UNAME_RELEASE}` | sed -e 's/ .*$//'
+	echo ${UNAME_MACHINE}-pc-skyos`echo ${UNAME_RELEASE} | sed -e 's/ .*$//'`
 	exit ;;
     i*86:rdos:*:*)
 	echo ${UNAME_MACHINE}-pc-rdos
@@ -1359,171 +1409,25 @@ EOF
     x86_64:VMkernel:*:*)
 	echo ${UNAME_MACHINE}-unknown-esx
 	exit ;;
+    amd64:Isilon\ OneFS:*:*)
+	echo x86_64-unknown-onefs
+	exit ;;
 esac
 
-eval $set_cc_for_build
-cat >$dummy.c <<EOF
-#ifdef _SEQUENT_
-# include <sys/types.h>
-# include <sys/utsname.h>
-#endif
-main ()
-{
-#if defined (sony)
-#if defined (MIPSEB)
-  /* BFD wants "bsd" instead of "newsos".  Perhaps BFD should be changed,
-     I don't know....  */
-  printf ("mips-sony-bsd\n"); exit (0);
-#else
-#include <sys/param.h>
-  printf ("m68k-sony-newsos%s\n",
-#ifdef NEWSOS4
-	"4"
-#else
-	""
-#endif
-	); exit (0);
-#endif
-#endif
-
-#if defined (__arm) && defined (__acorn) && defined (__unix)
-  printf ("arm-acorn-riscix\n"); exit (0);
-#endif
-
-#if defined (hp300) && !defined (hpux)
-  printf ("m68k-hp-bsd\n"); exit (0);
-#endif
-
-#if defined (NeXT)
-#if !defined (__ARCHITECTURE__)
-#define __ARCHITECTURE__ "m68k"
-#endif
-  int version;
-  version=`(hostinfo | sed -n 's/.*NeXT Mach \([0-9]*\).*/\1/p') 2>/dev/null`;
-  if (version < 4)
-    printf ("%s-next-nextstep%d\n", __ARCHITECTURE__, version);
-  else
-    printf ("%s-next-openstep%d\n", __ARCHITECTURE__, version);
-  exit (0);
-#endif
-
-#if defined (MULTIMAX) || defined (n16)
-#if defined (UMAXV)
-  printf ("ns32k-encore-sysv\n"); exit (0);
-#else
-#if defined (CMU)
-  printf ("ns32k-encore-mach\n"); exit (0);
-#else
-  printf ("ns32k-encore-bsd\n"); exit (0);
-#endif
-#endif
-#endif
-
-#if defined (__386BSD__)
-  printf ("i386-pc-bsd\n"); exit (0);
-#endif
-
-#if defined (sequent)
-#if defined (i386)
-  printf ("i386-sequent-dynix\n"); exit (0);
-#endif
-#if defined (ns32000)
-  printf ("ns32k-sequent-dynix\n"); exit (0);
-#endif
-#endif
-
-#if defined (_SEQUENT_)
-    struct utsname un;
-
-    uname(&un);
-
-    if (strncmp(un.version, "V2", 2) == 0) {
-	printf ("i386-sequent-ptx2\n"); exit (0);
-    }
-    if (strncmp(un.version, "V1", 2) == 0) { /* XXX is V1 correct? */
-	printf ("i386-sequent-ptx1\n"); exit (0);
-    }
-    printf ("i386-sequent-ptx\n"); exit (0);
-
-#endif
-
-#if defined (vax)
-# if !defined (ultrix)
-#  include <sys/param.h>
-#  if defined (BSD)
-#   if BSD == 43
-      printf ("vax-dec-bsd4.3\n"); exit (0);
-#   else
-#    if BSD == 199006
-      printf ("vax-dec-bsd4.3reno\n"); exit (0);
-#    else
-      printf ("vax-dec-bsd\n"); exit (0);
-#    endif
-#   endif
-#  else
-    printf ("vax-dec-bsd\n"); exit (0);
-#  endif
-# else
-    printf ("vax-dec-ultrix\n"); exit (0);
-# endif
-#endif
-
-#if defined (alliant) && defined (i860)
-  printf ("i860-alliant-bsd\n"); exit (0);
-#endif
-
-  exit (1);
-}
-EOF
-
-$CC_FOR_BUILD -o $dummy $dummy.c 2>/dev/null && SYSTEM_NAME=`$dummy` &&
-	{ echo "$SYSTEM_NAME"; exit; }
-
-# Apollos put the system type in the environment.
-
-test -d /usr/apollo && { echo ${ISP}-apollo-${SYSTYPE}; exit; }
-
-# Convex versions that predate uname can use getsysinfo(1)
-
-if [ -x /usr/convex/getsysinfo ]
-then
-    case `getsysinfo -f cpu_type` in
-    c1*)
-	echo c1-convex-bsd
-	exit ;;
-    c2*)
-	if getsysinfo -f scalar_acc
-	then echo c32-convex-bsd
-	else echo c2-convex-bsd
-	fi
-	exit ;;
-    c34*)
-	echo c34-convex-bsd
-	exit ;;
-    c38*)
-	echo c38-convex-bsd
-	exit ;;
-    c4*)
-	echo c4-convex-bsd
-	exit ;;
-    esac
-fi
-
 cat >&2 <<EOF
 $0: unable to guess system type
 
-This script, last modified $timestamp, has failed to recognize
-the operating system you are using. It is advised that you
-download the most up to date version of the config scripts from
+This script (version $timestamp), has failed to recognize the
+operating system you are using. If your script is old, overwrite
+config.guess and config.sub with the latest versions from:
 
-  http://git.savannah.gnu.org/gitweb/?p=config.git;a=blob_plain;f=config.guess;hb=HEAD
+  http://git.savannah.gnu.org/gitweb/?p=config.git;a=blob_plain;f=config.guess
 and
-  http://git.savannah.gnu.org/gitweb/?p=config.git;a=blob_plain;f=config.sub;hb=HEAD
+  http://git.savannah.gnu.org/gitweb/?p=config.git;a=blob_plain;f=config.sub
 
-If the version you run ($0) is already up to date, please
-send the following data and any information you think might be
-pertinent to <config-patches@gnu.org> in order to provide the needed
-information to handle your system.
+If $0 has already been updated, send the following data and any
+information you think might be pertinent to config-patches@gnu.org to
+provide the necessary information to handle your system.
 
 config.guess timestamp = $timestamp
 

build-aux/config.sub

@@ -1,8 +1,8 @@
 #! /bin/sh
 # Configuration validation subroutine script.
-#   Copyright 1992-2013 Free Software Foundation, Inc.
+#   Copyright 1992-2016 Free Software Foundation, Inc.
 
-timestamp='2013-10-01'
+timestamp='2016-11-04'
 
 # This file is free software; you can redistribute it and/or modify it
 # under the terms of the GNU General Public License as published by
@@ -25,7 +25,7 @@ timestamp='2013-10-01'
 # of the GNU General Public License, version 3 ("GPLv3").
 
 
-# Please send patches with a ChangeLog entry to config-patches@gnu.org.
+# Please send patches to <config-patches@gnu.org>.
 #
 # Configuration subroutine to validate and canonicalize a configuration type.
 # Supply the specified configuration type as an argument.
@@ -33,7 +33,7 @@ timestamp='2013-10-01'
 # Otherwise, we print the canonical config type on stdout and succeed.
 
 # You can get the latest version of this script from:
-# http://git.savannah.gnu.org/gitweb/?p=config.git;a=blob_plain;f=config.sub;hb=HEAD
+# http://git.savannah.gnu.org/gitweb/?p=config.git;a=blob_plain;f=config.sub
 
 # This file is supposed to be the same for all GNU packages
 # and recognize all the CPU types, system types and aliases
@@ -53,8 +53,7 @@ timestamp='2013-10-01'
 me=`echo "$0" | sed -e 's,.*/,,'`
 
 usage="\
-Usage: $0 [OPTION] CPU-MFR-OPSYS
-       $0 [OPTION] ALIAS
+Usage: $0 [OPTION] CPU-MFR-OPSYS or ALIAS
 
 Canonicalize a configuration name.
 
@@ -68,7 +67,7 @@ Report bugs and patches to <config-patches@gnu.org>."
 version="\
 GNU config.sub ($timestamp)
 
-Copyright 1992-2013 Free Software Foundation, Inc.
+Copyright 1992-2016 Free Software Foundation, Inc.
 
 This is free software; see the source for copying conditions.  There is NO
 warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE."
@@ -117,8 +116,8 @@ maybe_os=`echo $1 | sed 's/^\(.*\)-\([^-]*-[^-]*\)$/\2/'`
 case $maybe_os in
   nto-qnx* | linux-gnu* | linux-android* | linux-dietlibc | linux-newlib* | \
   linux-musl* | linux-uclibc* | uclinux-uclibc* | uclinux-gnu* | kfreebsd*-gnu* | \
-  knetbsd*-gnu* | netbsd*-gnu* | \
-  kopensolaris*-gnu* | \
+  knetbsd*-gnu* | netbsd*-gnu* | netbsd*-eabi* | \
+  kopensolaris*-gnu* | cloudabi*-eabi* | \
   storm-chaos* | os2-emx* | rtmk-nova*)
     os=-$maybe_os
     basic_machine=`echo $1 | sed 's/^\(.*\)-\([^-]*-[^-]*\)$/\1/'`
@@ -255,12 +254,13 @@ case $basic_machine in
 	| arc | arceb \
 	| arm | arm[bl]e | arme[lb] | armv[2-8] | armv[3-8][lb] | armv7[arm] \
 	| avr | avr32 \
+	| ba \
 	| be32 | be64 \
 	| bfin \
 	| c4x | c8051 | clipper \
 	| d10v | d30v | dlx | dsp16xx \
-	| epiphany \
-	| fido | fr30 | frv \
+	| e2k | epiphany \
+	| fido | fr30 | frv | ft32 \
 	| h8300 | h8500 | hppa | hppa1.[01] | hppa2.0 | hppa2.0[nw] | hppa64 \
 	| hexagon \
 	| i370 | i860 | i960 | ia64 \
@@ -283,8 +283,10 @@ case $basic_machine in
 	| mips64vr5900 | mips64vr5900el \
 	| mipsisa32 | mipsisa32el \
 	| mipsisa32r2 | mipsisa32r2el \
+	| mipsisa32r6 | mipsisa32r6el \
 	| mipsisa64 | mipsisa64el \
 	| mipsisa64r2 | mipsisa64r2el \
+	| mipsisa64r6 | mipsisa64r6el \
 	| mipsisa64sb1 | mipsisa64sb1el \
 	| mipsisa64sr71k | mipsisa64sr71kel \
 	| mipsr5900 | mipsr5900el \
@@ -296,14 +298,15 @@ case $basic_machine in
 	| nds32 | nds32le | nds32be \
 	| nios | nios2 | nios2eb | nios2el \
 	| ns16k | ns32k \
-	| open8 \
-	| or1k | or32 \
+	| open8 | or1k | or1knd | or32 \
 	| pdp10 | pdp11 | pj | pjl \
 	| powerpc | powerpc64 | powerpc64le | powerpcle \
+	| pru \
 	| pyramid \
+	| riscv32 | riscv64 \
 	| rl78 | rx \
 	| score \
-	| sh | sh[1234] | sh[24]a | sh[24]aeb | sh[23]e | sh[34]eb | sheb | shbe | shle | sh[1234]le | sh3ele \
+	| sh | sh[1234] | sh[24]a | sh[24]aeb | sh[23]e | sh[234]eb | sheb | shbe | shle | sh[1234]le | sh3ele \
 	| sh64 | sh64le \
 	| sparc | sparc64 | sparc64b | sparc64v | sparc86x | sparclet | sparclite \
 	| sparcv8 | sparcv9 | sparcv9b | sparcv9v \
@@ -311,6 +314,7 @@ case $basic_machine in
 	| tahoe | tic4x | tic54x | tic55x | tic6x | tic80 | tron \
 	| ubicom32 \
 	| v850 | v850e | v850e1 | v850e2 | v850es | v850e2v3 \
+	| visium \
 	| we32k \
 	| x86 | xc16x | xstormy16 | xtensa \
 	| z8k | z80)
@@ -325,6 +329,9 @@ case $basic_machine in
 	c6x)
 		basic_machine=tic6x-unknown
 		;;
+	leon|leon[3-9])
+		basic_machine=sparc-$basic_machine
+		;;
 	m6811 | m68hc11 | m6812 | m68hc12 | m68hcs12x | nvptx | picochip)
 		basic_machine=$basic_machine-unknown
 		os=-none
@@ -370,12 +377,13 @@ case $basic_machine in
 	| alphapca5[67]-* | alpha64pca5[67]-* | arc-* | arceb-* \
 	| arm-*  | armbe-* | armle-* | armeb-* | armv*-* \
 	| avr-* | avr32-* \
+	| ba-* \
 	| be32-* | be64-* \
 	| bfin-* | bs2000-* \
 	| c[123]* | c30-* | [cjt]90-* | c4x-* \
 	| c8051-* | clipper-* | craynv-* | cydra-* \
 	| d10v-* | d30v-* | dlx-* \
-	| elxsi-* \
+	| e2k-* | elxsi-* \
 	| f30[01]-* | f700-* | fido-* | fr30-* | frv-* | fx80-* \
 	| h8300-* | h8500-* \
 	| hppa-* | hppa1.[01]-* | hppa2.0-* | hppa2.0[nw]-* | hppa64-* \
@@ -402,8 +410,10 @@ case $basic_machine in
 	| mips64vr5900-* | mips64vr5900el-* \
 	| mipsisa32-* | mipsisa32el-* \
 	| mipsisa32r2-* | mipsisa32r2el-* \
+	| mipsisa32r6-* | mipsisa32r6el-* \
 	| mipsisa64-* | mipsisa64el-* \
 	| mipsisa64r2-* | mipsisa64r2el-* \
+	| mipsisa64r6-* | mipsisa64r6el-* \
 	| mipsisa64sb1-* | mipsisa64sb1el-* \
 	| mipsisa64sr71k-* | mipsisa64sr71kel-* \
 	| mipsr5900-* | mipsr5900el-* \
@@ -415,16 +425,19 @@ case $basic_machine in
 	| nios-* | nios2-* | nios2eb-* | nios2el-* \
 	| none-* | np1-* | ns16k-* | ns32k-* \
 	| open8-* \
+	| or1k*-* \
 	| orion-* \
 	| pdp10-* | pdp11-* | pj-* | pjl-* | pn-* | power-* \
 	| powerpc-* | powerpc64-* | powerpc64le-* | powerpcle-* \
+	| pru-* \
 	| pyramid-* \
+	| riscv32-* | riscv64-* \
 	| rl78-* | romp-* | rs6000-* | rx-* \
 	| sh-* | sh[1234]-* | sh[24]a-* | sh[24]aeb-* | sh[23]e-* | sh[34]eb-* | sheb-* | shbe-* \
 	| shle-* | sh[1234]le-* | sh3ele-* | sh64-* | sh64le-* \
 	| sparc-* | sparc64-* | sparc64b-* | sparc64v-* | sparc86x-* | sparclet-* \
 	| sparclite-* \
-	| sparcv8-* | sparcv9-* | sparcv9b-* | sparcv9v-* | sv1-* | sx?-* \
+	| sparcv8-* | sparcv9-* | sparcv9b-* | sparcv9v-* | sv1-* | sx*-* \
 	| tahoe-* \
 	| tic30-* | tic4x-* | tic54x-* | tic55x-* | tic6x-* | tic80-* \
 	| tile*-* \
@@ -432,6 +445,7 @@ case $basic_machine in
 	| ubicom32-* \
 	| v850-* | v850e-* | v850e1-* | v850es-* | v850e2-* | v850e2v3-* \
 	| vax-* \
+	| visium-* \
 	| we32k-* \
 	| x86-* | x86_64-* | xc16x-* | xps100-* \
 	| xstormy16-* | xtensa*-* \
@@ -508,6 +522,9 @@ case $basic_machine in
 		basic_machine=i386-pc
 		os=-aros
 		;;
+	asmjs)
+		basic_machine=asmjs-unknown
+		;;
 	aux)
 		basic_machine=m68k-apple
 		os=-aux
@@ -628,6 +645,14 @@ case $basic_machine in
 		basic_machine=m68k-bull
 		os=-sysv3
 		;;
+	e500v[12])
+		basic_machine=powerpc-unknown
+		os=$os"spe"
+		;;
+	e500v[12]-*)
+		basic_machine=powerpc-`echo $basic_machine | sed 's/^[^-]*-//'`
+		os=$os"spe"
+		;;
 	ebmon29k)
 		basic_machine=a29k-amd
 		os=-ebmon
@@ -769,6 +794,9 @@ case $basic_machine in
 		basic_machine=m68k-isi
 		os=-sysv
 		;;
+	leon-*|leon[3-9]-*)
+		basic_machine=sparc-`echo $basic_machine | sed 's/-.*//'`
+		;;
 	m68knommu)
 		basic_machine=m68k-unknown
 		os=-linux
@@ -824,6 +852,10 @@ case $basic_machine in
 		basic_machine=powerpc-unknown
 		os=-morphos
 		;;
+	moxiebox)
+		basic_machine=moxie-unknown
+		os=-moxiebox
+		;;
 	msdos)
 		basic_machine=i386-pc
 		os=-msdos
@@ -1000,7 +1032,7 @@ case $basic_machine in
 	ppc-* | ppcbe-*)
 		basic_machine=powerpc-`echo $basic_machine | sed 's/^[^-]*-//'`
 		;;
-	ppcle | powerpclittle | ppc-le | powerpc-little)
+	ppcle | powerpclittle)
 		basic_machine=powerpcle-unknown
 		;;
 	ppcle-* | powerpclittle-*)
@@ -1010,7 +1042,7 @@ case $basic_machine in
 		;;
 	ppc64-*) basic_machine=powerpc64-`echo $basic_machine | sed 's/^[^-]*-//'`
 		;;
-	ppc64le | powerpc64little | ppc64-le | powerpc64-little)
+	ppc64le | powerpc64little)
 		basic_machine=powerpc64le-unknown
 		;;
 	ppc64le-* | powerpc64little-*)
@@ -1356,27 +1388,28 @@ case $os in
 	      | -hpux* | -unos* | -osf* | -luna* | -dgux* | -auroraux* | -solaris* \
 	      | -sym* | -kopensolaris* | -plan9* \
 	      | -amigaos* | -amigados* | -msdos* | -newsos* | -unicos* | -aof* \
-	      | -aos* | -aros* \
+	      | -aos* | -aros* | -cloudabi* | -sortix* \
 	      | -nindy* | -vxsim* | -vxworks* | -ebmon* | -hms* | -mvs* \
 	      | -clix* | -riscos* | -uniplus* | -iris* | -rtu* | -xenix* \
 	      | -hiux* | -386bsd* | -knetbsd* | -mirbsd* | -netbsd* \
-	      | -bitrig* | -openbsd* | -solidbsd* \
+	      | -bitrig* | -openbsd* | -solidbsd* | -libertybsd* \
 	      | -ekkobsd* | -kfreebsd* | -freebsd* | -riscix* | -lynxos* \
 	      | -bosx* | -nextstep* | -cxux* | -aout* | -elf* | -oabi* \
 	      | -ptx* | -coff* | -ecoff* | -winnt* | -domain* | -vsta* \
 	      | -udi* | -eabi* | -lites* | -ieee* | -go32* | -aux* \
 	      | -chorusos* | -chorusrdb* | -cegcc* \
 	      | -cygwin* | -msys* | -pe* | -psos* | -moss* | -proelf* | -rtems* \
-	      | -mingw32* | -mingw64* | -linux-gnu* | -linux-android* \
+	      | -midipix* | -mingw32* | -mingw64* | -linux-gnu* | -linux-android* \
 	      | -linux-newlib* | -linux-musl* | -linux-uclibc* \
-	      | -uxpv* | -beos* | -mpeix* | -udk* \
+	      | -uxpv* | -beos* | -mpeix* | -udk* | -moxiebox* \
 	      | -interix* | -uwin* | -mks* | -rhapsody* | -darwin* | -opened* \
 	      | -openstep* | -oskit* | -conix* | -pw32* | -nonstopux* \
 	      | -storm-chaos* | -tops10* | -tenex* | -tops20* | -its* \
 	      | -os2* | -vos* | -palmos* | -uclinux* | -nucleus* \
 	      | -morphos* | -superux* | -rtmk* | -rtmk-nova* | -windiss* \
 	      | -powermax* | -dnix* | -nx6 | -nx7 | -sei* | -dragonfly* \
-	      | -skyos* | -haiku* | -rdos* | -toppers* | -drops* | -es*)
+	      | -skyos* | -haiku* | -rdos* | -toppers* | -drops* | -es* \
+	      | -onefs* | -tirtos* | -phoenix* | -fuchsia*)
 	# Remember, each alternative MUST END IN *, to match a version number.
 		;;
 	-qnx*)
@@ -1508,6 +1541,8 @@ case $os in
 		;;
 	-nacl*)
 		;;
+	-ios)
+		;;
 	-none)
 		;;
 	*)
@@ -1594,9 +1629,6 @@ case $basic_machine in
 	mips*-*)
 		os=-elf
 		;;
-	or1k-*)
-		os=-elf
-		;;
 	or32-*)
 		os=-coff
 		;;

coverage.sh

@@ -1,16 +0,0 @@
-#!/bin/sh
-
-set -e
-
-objdir=$1
-suffix=$2
-shift 2
-objs=$@
-
-gcov -b -p -f -o "${objdir}" ${objs}
-
-# Move gcov outputs so that subsequent gcov invocations won't clobber results
-# for the same sources with different compilation flags.
-for f in `find . -maxdepth 1 -type f -name '*.gcov'` ; do
-  mv "${f}" "${f}.${suffix}"
-done

doc/html.xsl.in

@@ -1,4 +1,5 @@
 <xsl:stylesheet xmlns:xsl="http://www.w3.org/1999/XSL/Transform" version="1.0">
   <xsl:import href="@XSLROOT@/html/docbook.xsl"/>
   <xsl:import href="@abs_srcroot@doc/stylesheet.xsl"/>
+  <xsl:output method="xml" encoding="utf-8"/>
 </xsl:stylesheet>

doc/stylesheet.xsl

@@ -1,7 +1,10 @@
 <xsl:stylesheet xmlns:xsl="http://www.w3.org/1999/XSL/Transform" version="1.0">
   <xsl:param name="funcsynopsis.style">ansi</xsl:param>
-  <xsl:param name="function.parens" select="1"/>
+  <xsl:param name="function.parens" select="0"/>
+  <xsl:template match="function">
+    <xsl:call-template name="inline.monoseq"/>
+  </xsl:template>
   <xsl:template match="mallctl">
-    "<xsl:call-template name="inline.monoseq"/>"
+    <quote><xsl:call-template name="inline.monoseq"/></quote>
   </xsl:template>
 </xsl:stylesheet>

include/jemalloc/internal/arena_externs.h

@@ -0,0 +1,94 @@
+#ifndef JEMALLOC_INTERNAL_ARENA_EXTERNS_H
+#define JEMALLOC_INTERNAL_ARENA_EXTERNS_H
+
+#include "jemalloc/internal/bin.h"
+#include "jemalloc/internal/extent_dss.h"
+#include "jemalloc/internal/pages.h"
+#include "jemalloc/internal/size_classes.h"
+#include "jemalloc/internal/stats.h"
+
+extern ssize_t opt_dirty_decay_ms;
+extern ssize_t opt_muzzy_decay_ms;
+
+extern percpu_arena_mode_t opt_percpu_arena;
+extern const char *percpu_arena_mode_names[];
+
+extern const uint64_t h_steps[SMOOTHSTEP_NSTEPS];
+extern malloc_mutex_t arenas_lock;
+
+void arena_basic_stats_merge(tsdn_t *tsdn, arena_t *arena,
+    unsigned *nthreads, const char **dss, ssize_t *dirty_decay_ms,
+    ssize_t *muzzy_decay_ms, size_t *nactive, size_t *ndirty, size_t *nmuzzy);
+void arena_stats_merge(tsdn_t *tsdn, arena_t *arena, unsigned *nthreads,
+    const char **dss, ssize_t *dirty_decay_ms, ssize_t *muzzy_decay_ms,
+    size_t *nactive, size_t *ndirty, size_t *nmuzzy, arena_stats_t *astats,
+    bin_stats_t *bstats, arena_stats_large_t *lstats);
+void arena_extents_dirty_dalloc(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, extent_t *extent);
+#ifdef JEMALLOC_JET
+size_t arena_slab_regind(extent_t *slab, szind_t binind, const void *ptr);
+#endif
+extent_t *arena_extent_alloc_large(tsdn_t *tsdn, arena_t *arena,
+    size_t usize, size_t alignment, bool *zero);
+void arena_extent_dalloc_large_prep(tsdn_t *tsdn, arena_t *arena,
+    extent_t *extent);
+void arena_extent_ralloc_large_shrink(tsdn_t *tsdn, arena_t *arena,
+    extent_t *extent, size_t oldsize);
+void arena_extent_ralloc_large_expand(tsdn_t *tsdn, arena_t *arena,
+    extent_t *extent, size_t oldsize);
+ssize_t arena_dirty_decay_ms_get(arena_t *arena);
+bool arena_dirty_decay_ms_set(tsdn_t *tsdn, arena_t *arena, ssize_t decay_ms);
+ssize_t arena_muzzy_decay_ms_get(arena_t *arena);
+bool arena_muzzy_decay_ms_set(tsdn_t *tsdn, arena_t *arena, ssize_t decay_ms);
+void arena_decay(tsdn_t *tsdn, arena_t *arena, bool is_background_thread,
+    bool all);
+void arena_reset(tsd_t *tsd, arena_t *arena);
+void arena_destroy(tsd_t *tsd, arena_t *arena);
+void arena_tcache_fill_small(tsdn_t *tsdn, arena_t *arena, tcache_t *tcache,
+    cache_bin_t *tbin, szind_t binind, uint64_t prof_accumbytes);
+void arena_alloc_junk_small(void *ptr, const bin_info_t *bin_info,
+    bool zero);
+
+typedef void (arena_dalloc_junk_small_t)(void *, const bin_info_t *);
+extern arena_dalloc_junk_small_t *JET_MUTABLE arena_dalloc_junk_small;
+
+void *arena_malloc_hard(tsdn_t *tsdn, arena_t *arena, size_t size,
+    szind_t ind, bool zero);
+void *arena_palloc(tsdn_t *tsdn, arena_t *arena, size_t usize,
+    size_t alignment, bool zero, tcache_t *tcache);
+void arena_prof_promote(tsdn_t *tsdn, const void *ptr, size_t usize);
+void arena_dalloc_promoted(tsdn_t *tsdn, void *ptr, tcache_t *tcache,
+    bool slow_path);
+void arena_dalloc_bin_junked_locked(tsdn_t *tsdn, arena_t *arena,
+    extent_t *extent, void *ptr);
+void arena_dalloc_small(tsdn_t *tsdn, void *ptr);
+bool arena_ralloc_no_move(tsdn_t *tsdn, void *ptr, size_t oldsize, size_t size,
+    size_t extra, bool zero);
+void *arena_ralloc(tsdn_t *tsdn, arena_t *arena, void *ptr, size_t oldsize,
+    size_t size, size_t alignment, bool zero, tcache_t *tcache);
+dss_prec_t arena_dss_prec_get(arena_t *arena);
+bool arena_dss_prec_set(arena_t *arena, dss_prec_t dss_prec);
+ssize_t arena_dirty_decay_ms_default_get(void);
+bool arena_dirty_decay_ms_default_set(ssize_t decay_ms);
+ssize_t arena_muzzy_decay_ms_default_get(void);
+bool arena_muzzy_decay_ms_default_set(ssize_t decay_ms);
+bool arena_retain_grow_limit_get_set(tsd_t *tsd, arena_t *arena,
+    size_t *old_limit, size_t *new_limit);
+unsigned arena_nthreads_get(arena_t *arena, bool internal);
+void arena_nthreads_inc(arena_t *arena, bool internal);
+void arena_nthreads_dec(arena_t *arena, bool internal);
+size_t arena_extent_sn_next(arena_t *arena);
+arena_t *arena_new(tsdn_t *tsdn, unsigned ind, extent_hooks_t *extent_hooks);
+void arena_boot(void);
+void arena_prefork0(tsdn_t *tsdn, arena_t *arena);
+void arena_prefork1(tsdn_t *tsdn, arena_t *arena);
+void arena_prefork2(tsdn_t *tsdn, arena_t *arena);
+void arena_prefork3(tsdn_t *tsdn, arena_t *arena);
+void arena_prefork4(tsdn_t *tsdn, arena_t *arena);
+void arena_prefork5(tsdn_t *tsdn, arena_t *arena);
+void arena_prefork6(tsdn_t *tsdn, arena_t *arena);
+void arena_prefork7(tsdn_t *tsdn, arena_t *arena);
+void arena_postfork_parent(tsdn_t *tsdn, arena_t *arena);
+void arena_postfork_child(tsdn_t *tsdn, arena_t *arena);
+
+#endif /* JEMALLOC_INTERNAL_ARENA_EXTERNS_H */

include/jemalloc/internal/arena_inlines_a.h

@@ -0,0 +1,57 @@
+#ifndef JEMALLOC_INTERNAL_ARENA_INLINES_A_H
+#define JEMALLOC_INTERNAL_ARENA_INLINES_A_H
+
+static inline unsigned
+arena_ind_get(const arena_t *arena) {
+	return base_ind_get(arena->base);
+}
+
+static inline void
+arena_internal_add(arena_t *arena, size_t size) {
+	atomic_fetch_add_zu(&arena->stats.internal, size, ATOMIC_RELAXED);
+}
+
+static inline void
+arena_internal_sub(arena_t *arena, size_t size) {
+	atomic_fetch_sub_zu(&arena->stats.internal, size, ATOMIC_RELAXED);
+}
+
+static inline size_t
+arena_internal_get(arena_t *arena) {
+	return atomic_load_zu(&arena->stats.internal, ATOMIC_RELAXED);
+}
+
+static inline bool
+arena_prof_accum(tsdn_t *tsdn, arena_t *arena, uint64_t accumbytes) {
+	cassert(config_prof);
+
+	if (likely(prof_interval == 0 || !prof_active_get_unlocked())) {
+		return false;
+	}
+
+	return prof_accum_add(tsdn, &arena->prof_accum, accumbytes);
+}
+
+static inline void
+percpu_arena_update(tsd_t *tsd, unsigned cpu) {
+	assert(have_percpu_arena);
+	arena_t *oldarena = tsd_arena_get(tsd);
+	assert(oldarena != NULL);
+	unsigned oldind = arena_ind_get(oldarena);
+
+	if (oldind != cpu) {
+		unsigned newind = cpu;
+		arena_t *newarena = arena_get(tsd_tsdn(tsd), newind, true);
+		assert(newarena != NULL);
+
+		/* Set new arena/tcache associations. */
+		arena_migrate(tsd, oldind, newind);
+		tcache_t *tcache = tcache_get(tsd);
+		if (tcache != NULL) {
+			tcache_arena_reassociate(tsd_tsdn(tsd), tcache,
+			    newarena);
+		}
+	}
+}
+
+#endif /* JEMALLOC_INTERNAL_ARENA_INLINES_A_H */

include/jemalloc/internal/arena_inlines_b.h

@@ -0,0 +1,354 @@
+#ifndef JEMALLOC_INTERNAL_ARENA_INLINES_B_H
+#define JEMALLOC_INTERNAL_ARENA_INLINES_B_H
+
+#include "jemalloc/internal/jemalloc_internal_types.h"
+#include "jemalloc/internal/mutex.h"
+#include "jemalloc/internal/rtree.h"
+#include "jemalloc/internal/size_classes.h"
+#include "jemalloc/internal/sz.h"
+#include "jemalloc/internal/ticker.h"
+
+JEMALLOC_ALWAYS_INLINE prof_tctx_t *
+arena_prof_tctx_get(tsdn_t *tsdn, const void *ptr, alloc_ctx_t *alloc_ctx) {
+	cassert(config_prof);
+	assert(ptr != NULL);
+
+	/* Static check. */
+	if (alloc_ctx == NULL) {
+		const extent_t *extent = iealloc(tsdn, ptr);
+		if (unlikely(!extent_slab_get(extent))) {
+			return large_prof_tctx_get(tsdn, extent);
+		}
+	} else {
+		if (unlikely(!alloc_ctx->slab)) {
+			return large_prof_tctx_get(tsdn, iealloc(tsdn, ptr));
+		}
+	}
+	return (prof_tctx_t *)(uintptr_t)1U;
+}
+
+JEMALLOC_ALWAYS_INLINE void
+arena_prof_tctx_set(tsdn_t *tsdn, const void *ptr, UNUSED size_t usize,
+    alloc_ctx_t *alloc_ctx, prof_tctx_t *tctx) {
+	cassert(config_prof);
+	assert(ptr != NULL);
+
+	/* Static check. */
+	if (alloc_ctx == NULL) {
+		extent_t *extent = iealloc(tsdn, ptr);
+		if (unlikely(!extent_slab_get(extent))) {
+			large_prof_tctx_set(tsdn, extent, tctx);
+		}
+	} else {
+		if (unlikely(!alloc_ctx->slab)) {
+			large_prof_tctx_set(tsdn, iealloc(tsdn, ptr), tctx);
+		}
+	}
+}
+
+static inline void
+arena_prof_tctx_reset(tsdn_t *tsdn, const void *ptr, UNUSED prof_tctx_t *tctx) {
+	cassert(config_prof);
+	assert(ptr != NULL);
+
+	extent_t *extent = iealloc(tsdn, ptr);
+	assert(!extent_slab_get(extent));
+
+	large_prof_tctx_reset(tsdn, extent);
+}
+
+JEMALLOC_ALWAYS_INLINE void
+arena_decay_ticks(tsdn_t *tsdn, arena_t *arena, unsigned nticks) {
+	tsd_t *tsd;
+	ticker_t *decay_ticker;
+
+	if (unlikely(tsdn_null(tsdn))) {
+		return;
+	}
+	tsd = tsdn_tsd(tsdn);
+	decay_ticker = decay_ticker_get(tsd, arena_ind_get(arena));
+	if (unlikely(decay_ticker == NULL)) {
+		return;
+	}
+	if (unlikely(ticker_ticks(decay_ticker, nticks))) {
+		arena_decay(tsdn, arena, false, false);
+	}
+}
+
+JEMALLOC_ALWAYS_INLINE void
+arena_decay_tick(tsdn_t *tsdn, arena_t *arena) {
+	malloc_mutex_assert_not_owner(tsdn, &arena->decay_dirty.mtx);
+	malloc_mutex_assert_not_owner(tsdn, &arena->decay_muzzy.mtx);
+
+	arena_decay_ticks(tsdn, arena, 1);
+}
+
+JEMALLOC_ALWAYS_INLINE void *
+arena_malloc(tsdn_t *tsdn, arena_t *arena, size_t size, szind_t ind, bool zero,
+    tcache_t *tcache, bool slow_path) {
+	assert(!tsdn_null(tsdn) || tcache == NULL);
+	assert(size != 0);
+
+	if (likely(tcache != NULL)) {
+		if (likely(size <= SMALL_MAXCLASS)) {
+			return tcache_alloc_small(tsdn_tsd(tsdn), arena,
+			    tcache, size, ind, zero, slow_path);
+		}
+		if (likely(size <= tcache_maxclass)) {
+			return tcache_alloc_large(tsdn_tsd(tsdn), arena,
+			    tcache, size, ind, zero, slow_path);
+		}
+		/* (size > tcache_maxclass) case falls through. */
+		assert(size > tcache_maxclass);
+	}
+
+	return arena_malloc_hard(tsdn, arena, size, ind, zero);
+}
+
+JEMALLOC_ALWAYS_INLINE arena_t *
+arena_aalloc(tsdn_t *tsdn, const void *ptr) {
+	return extent_arena_get(iealloc(tsdn, ptr));
+}
+
+JEMALLOC_ALWAYS_INLINE size_t
+arena_salloc(tsdn_t *tsdn, const void *ptr) {
+	assert(ptr != NULL);
+
+	rtree_ctx_t rtree_ctx_fallback;
+	rtree_ctx_t *rtree_ctx = tsdn_rtree_ctx(tsdn, &rtree_ctx_fallback);
+
+	szind_t szind = rtree_szind_read(tsdn, &extents_rtree, rtree_ctx,
+	    (uintptr_t)ptr, true);
+	assert(szind != NSIZES);
+
+	return sz_index2size(szind);
+}
+
+JEMALLOC_ALWAYS_INLINE size_t
+arena_vsalloc(tsdn_t *tsdn, const void *ptr) {
+	/*
+	 * Return 0 if ptr is not within an extent managed by jemalloc.  This
+	 * function has two extra costs relative to isalloc():
+	 * - The rtree calls cannot claim to be dependent lookups, which induces
+	 *   rtree lookup load dependencies.
+	 * - The lookup may fail, so there is an extra branch to check for
+	 *   failure.
+	 */
+
+	rtree_ctx_t rtree_ctx_fallback;
+	rtree_ctx_t *rtree_ctx = tsdn_rtree_ctx(tsdn, &rtree_ctx_fallback);
+
+	extent_t *extent;
+	szind_t szind;
+	if (rtree_extent_szind_read(tsdn, &extents_rtree, rtree_ctx,
+	    (uintptr_t)ptr, false, &extent, &szind)) {
+		return 0;
+	}
+
+	if (extent == NULL) {
+		return 0;
+	}
+	assert(extent_state_get(extent) == extent_state_active);
+	/* Only slab members should be looked up via interior pointers. */
+	assert(extent_addr_get(extent) == ptr || extent_slab_get(extent));
+
+	assert(szind != NSIZES);
+
+	return sz_index2size(szind);
+}
+
+static inline void
+arena_dalloc_no_tcache(tsdn_t *tsdn, void *ptr) {
+	assert(ptr != NULL);
+
+	rtree_ctx_t rtree_ctx_fallback;
+	rtree_ctx_t *rtree_ctx = tsdn_rtree_ctx(tsdn, &rtree_ctx_fallback);
+
+	szind_t szind;
+	bool slab;
+	rtree_szind_slab_read(tsdn, &extents_rtree, rtree_ctx, (uintptr_t)ptr,
+	    true, &szind, &slab);
+
+	if (config_debug) {
+		extent_t *extent = rtree_extent_read(tsdn, &extents_rtree,
+		    rtree_ctx, (uintptr_t)ptr, true);
+		assert(szind == extent_szind_get(extent));
+		assert(szind < NSIZES);
+		assert(slab == extent_slab_get(extent));
+	}
+
+	if (likely(slab)) {
+		/* Small allocation. */
+		arena_dalloc_small(tsdn, ptr);
+	} else {
+		extent_t *extent = iealloc(tsdn, ptr);
+		large_dalloc(tsdn, extent);
+	}
+}
+
+JEMALLOC_ALWAYS_INLINE void
+arena_dalloc(tsdn_t *tsdn, void *ptr, tcache_t *tcache,
+    alloc_ctx_t *alloc_ctx, bool slow_path) {
+	assert(!tsdn_null(tsdn) || tcache == NULL);
+	assert(ptr != NULL);
+
+	if (unlikely(tcache == NULL)) {
+		arena_dalloc_no_tcache(tsdn, ptr);
+		return;
+	}
+
+	szind_t szind;
+	bool slab;
+	rtree_ctx_t *rtree_ctx;
+	if (alloc_ctx != NULL) {
+		szind = alloc_ctx->szind;
+		slab = alloc_ctx->slab;
+		assert(szind != NSIZES);
+	} else {
+		rtree_ctx = tsd_rtree_ctx(tsdn_tsd(tsdn));
+		rtree_szind_slab_read(tsdn, &extents_rtree, rtree_ctx,
+		    (uintptr_t)ptr, true, &szind, &slab);
+	}
+
+	if (config_debug) {
+		rtree_ctx = tsd_rtree_ctx(tsdn_tsd(tsdn));
+		extent_t *extent = rtree_extent_read(tsdn, &extents_rtree,
+		    rtree_ctx, (uintptr_t)ptr, true);
+		assert(szind == extent_szind_get(extent));
+		assert(szind < NSIZES);
+		assert(slab == extent_slab_get(extent));
+	}
+
+	if (likely(slab)) {
+		/* Small allocation. */
+		tcache_dalloc_small(tsdn_tsd(tsdn), tcache, ptr, szind,
+		    slow_path);
+	} else {
+		if (szind < nhbins) {
+			if (config_prof && unlikely(szind < NBINS)) {
+				arena_dalloc_promoted(tsdn, ptr, tcache,
+				    slow_path);
+			} else {
+				tcache_dalloc_large(tsdn_tsd(tsdn), tcache, ptr,
+				    szind, slow_path);
+			}
+		} else {
+			extent_t *extent = iealloc(tsdn, ptr);
+			large_dalloc(tsdn, extent);
+		}
+	}
+}
+
+static inline void
+arena_sdalloc_no_tcache(tsdn_t *tsdn, void *ptr, size_t size) {
+	assert(ptr != NULL);
+	assert(size <= LARGE_MAXCLASS);
+
+	szind_t szind;
+	bool slab;
+	if (!config_prof || !opt_prof) {
+		/*
+		 * There is no risk of being confused by a promoted sampled
+		 * object, so base szind and slab on the given size.
+		 */
+		szind = sz_size2index(size);
+		slab = (szind < NBINS);
+	}
+
+	if ((config_prof && opt_prof) || config_debug) {
+		rtree_ctx_t rtree_ctx_fallback;
+		rtree_ctx_t *rtree_ctx = tsdn_rtree_ctx(tsdn,
+		    &rtree_ctx_fallback);
+
+		rtree_szind_slab_read(tsdn, &extents_rtree, rtree_ctx,
+		    (uintptr_t)ptr, true, &szind, &slab);
+
+		assert(szind == sz_size2index(size));
+		assert((config_prof && opt_prof) || slab == (szind < NBINS));
+
+		if (config_debug) {
+			extent_t *extent = rtree_extent_read(tsdn,
+			    &extents_rtree, rtree_ctx, (uintptr_t)ptr, true);
+			assert(szind == extent_szind_get(extent));
+			assert(slab == extent_slab_get(extent));
+		}
+	}
+
+	if (likely(slab)) {
+		/* Small allocation. */
+		arena_dalloc_small(tsdn, ptr);
+	} else {
+		extent_t *extent = iealloc(tsdn, ptr);
+		large_dalloc(tsdn, extent);
+	}
+}
+
+JEMALLOC_ALWAYS_INLINE void
+arena_sdalloc(tsdn_t *tsdn, void *ptr, size_t size, tcache_t *tcache,
+    alloc_ctx_t *alloc_ctx, bool slow_path) {
+	assert(!tsdn_null(tsdn) || tcache == NULL);
+	assert(ptr != NULL);
+	assert(size <= LARGE_MAXCLASS);
+
+	if (unlikely(tcache == NULL)) {
+		arena_sdalloc_no_tcache(tsdn, ptr, size);
+		return;
+	}
+
+	szind_t szind;
+	bool slab;
+	UNUSED alloc_ctx_t local_ctx;
+	if (config_prof && opt_prof) {
+		if (alloc_ctx == NULL) {
+			/* Uncommon case and should be a static check. */
+			rtree_ctx_t rtree_ctx_fallback;
+			rtree_ctx_t *rtree_ctx = tsdn_rtree_ctx(tsdn,
+			    &rtree_ctx_fallback);
+			rtree_szind_slab_read(tsdn, &extents_rtree, rtree_ctx,
+			    (uintptr_t)ptr, true, &local_ctx.szind,
+			    &local_ctx.slab);
+			assert(local_ctx.szind == sz_size2index(size));
+			alloc_ctx = &local_ctx;
+		}
+		slab = alloc_ctx->slab;
+		szind = alloc_ctx->szind;
+	} else {
+		/*
+		 * There is no risk of being confused by a promoted sampled
+		 * object, so base szind and slab on the given size.
+		 */
+		szind = sz_size2index(size);
+		slab = (szind < NBINS);
+	}
+
+	if (config_debug) {
+		rtree_ctx_t *rtree_ctx = tsd_rtree_ctx(tsdn_tsd(tsdn));
+		rtree_szind_slab_read(tsdn, &extents_rtree, rtree_ctx,
+		    (uintptr_t)ptr, true, &szind, &slab);
+		extent_t *extent = rtree_extent_read(tsdn,
+		    &extents_rtree, rtree_ctx, (uintptr_t)ptr, true);
+		assert(szind == extent_szind_get(extent));
+		assert(slab == extent_slab_get(extent));
+	}
+
+	if (likely(slab)) {
+		/* Small allocation. */
+		tcache_dalloc_small(tsdn_tsd(tsdn), tcache, ptr, szind,
+		    slow_path);
+	} else {
+		if (szind < nhbins) {
+			if (config_prof && unlikely(szind < NBINS)) {
+				arena_dalloc_promoted(tsdn, ptr, tcache,
+				    slow_path);
+			} else {
+				tcache_dalloc_large(tsdn_tsd(tsdn),
+				    tcache, ptr, szind, slow_path);
+			}
+		} else {
+			extent_t *extent = iealloc(tsdn, ptr);
+			large_dalloc(tsdn, extent);
+		}
+	}
+}
+
+#endif /* JEMALLOC_INTERNAL_ARENA_INLINES_B_H */

include/jemalloc/internal/arena_stats.h

@@ -0,0 +1,237 @@
+#ifndef JEMALLOC_INTERNAL_ARENA_STATS_H
+#define JEMALLOC_INTERNAL_ARENA_STATS_H
+
+#include "jemalloc/internal/atomic.h"
+#include "jemalloc/internal/mutex.h"
+#include "jemalloc/internal/mutex_prof.h"
+#include "jemalloc/internal/size_classes.h"
+
+/*
+ * In those architectures that support 64-bit atomics, we use atomic updates for
+ * our 64-bit values.  Otherwise, we use a plain uint64_t and synchronize
+ * externally.
+ */
+#ifdef JEMALLOC_ATOMIC_U64
+typedef atomic_u64_t arena_stats_u64_t;
+#else
+/* Must hold the arena stats mutex while reading atomically. */
+typedef uint64_t arena_stats_u64_t;
+#endif
+
+typedef struct arena_stats_large_s arena_stats_large_t;
+struct arena_stats_large_s {
+	/*
+	 * Total number of allocation/deallocation requests served directly by
+	 * the arena.
+	 */
+	arena_stats_u64_t	nmalloc;
+	arena_stats_u64_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.
+	 */
+	arena_stats_u64_t	nrequests; /* Partially derived. */
+
+	/* Current number of allocations of this size class. */
+	size_t		curlextents; /* Derived. */
+};
+
+typedef struct arena_stats_decay_s arena_stats_decay_t;
+struct arena_stats_decay_s {
+	/* Total number of purge sweeps. */
+	arena_stats_u64_t	npurge;
+	/* Total number of madvise calls made. */
+	arena_stats_u64_t	nmadvise;
+	/* Total number of pages purged. */
+	arena_stats_u64_t	purged;
+};
+
+/*
+ * Arena stats.  Note that fields marked "derived" are not directly maintained
+ * within the arena code; rather their values are derived during stats merge
+ * requests.
+ */
+typedef struct arena_stats_s arena_stats_t;
+struct arena_stats_s {
+#ifndef JEMALLOC_ATOMIC_U64
+	malloc_mutex_t		mtx;
+#endif
+
+	/* Number of bytes currently mapped, excluding retained memory. */
+	atomic_zu_t		mapped; /* Partially derived. */
+
+	/*
+	 * Number of unused virtual memory bytes currently retained.  Retained
+	 * bytes are technically mapped (though always decommitted or purged),
+	 * but they are excluded from the mapped statistic (above).
+	 */
+	atomic_zu_t		retained; /* Derived. */
+
+	arena_stats_decay_t	decay_dirty;
+	arena_stats_decay_t	decay_muzzy;
+
+	atomic_zu_t		base; /* Derived. */
+	atomic_zu_t		internal;
+	atomic_zu_t		resident; /* Derived. */
+	atomic_zu_t		metadata_thp;
+
+	atomic_zu_t		allocated_large; /* Derived. */
+	arena_stats_u64_t	nmalloc_large; /* Derived. */
+	arena_stats_u64_t	ndalloc_large; /* Derived. */
+	arena_stats_u64_t	nrequests_large; /* Derived. */
+
+	/* Number of bytes cached in tcache associated with this arena. */
+	atomic_zu_t		tcache_bytes; /* Derived. */
+
+	mutex_prof_data_t mutex_prof_data[mutex_prof_num_arena_mutexes];
+
+	/* One element for each large size class. */
+	arena_stats_large_t	lstats[NSIZES - NBINS];
+
+	/* Arena uptime. */
+	nstime_t		uptime;
+};
+
+static inline bool
+arena_stats_init(UNUSED tsdn_t *tsdn, arena_stats_t *arena_stats) {
+	if (config_debug) {
+		for (size_t i = 0; i < sizeof(arena_stats_t); i++) {
+			assert(((char *)arena_stats)[i] == 0);
+		}
+	}
+#ifndef JEMALLOC_ATOMIC_U64
+	if (malloc_mutex_init(&arena_stats->mtx, "arena_stats",
+	    WITNESS_RANK_ARENA_STATS, malloc_mutex_rank_exclusive)) {
+		return true;
+	}
+#endif
+	/* Memory is zeroed, so there is no need to clear stats. */
+	return false;
+}
+
+static inline void
+arena_stats_lock(tsdn_t *tsdn, arena_stats_t *arena_stats) {
+#ifndef JEMALLOC_ATOMIC_U64
+	malloc_mutex_lock(tsdn, &arena_stats->mtx);
+#endif
+}
+
+static inline void
+arena_stats_unlock(tsdn_t *tsdn, arena_stats_t *arena_stats) {
+#ifndef JEMALLOC_ATOMIC_U64
+	malloc_mutex_unlock(tsdn, &arena_stats->mtx);
+#endif
+}
+
+static inline uint64_t
+arena_stats_read_u64(tsdn_t *tsdn, arena_stats_t *arena_stats,
+    arena_stats_u64_t *p) {
+#ifdef JEMALLOC_ATOMIC_U64
+	return atomic_load_u64(p, ATOMIC_RELAXED);
+#else
+	malloc_mutex_assert_owner(tsdn, &arena_stats->mtx);
+	return *p;
+#endif
+}
+
+static inline void
+arena_stats_add_u64(tsdn_t *tsdn, arena_stats_t *arena_stats,
+    arena_stats_u64_t *p, uint64_t x) {
+#ifdef JEMALLOC_ATOMIC_U64
+	atomic_fetch_add_u64(p, x, ATOMIC_RELAXED);
+#else
+	malloc_mutex_assert_owner(tsdn, &arena_stats->mtx);
+	*p += x;
+#endif
+}
+
+UNUSED static inline void
+arena_stats_sub_u64(tsdn_t *tsdn, arena_stats_t *arena_stats,
+    arena_stats_u64_t *p, uint64_t x) {
+#ifdef JEMALLOC_ATOMIC_U64
+	UNUSED uint64_t r = atomic_fetch_sub_u64(p, x, ATOMIC_RELAXED);
+	assert(r - x <= r);
+#else
+	malloc_mutex_assert_owner(tsdn, &arena_stats->mtx);
+	*p -= x;
+	assert(*p + x >= *p);
+#endif
+}
+
+/*
+ * Non-atomically sets *dst += src.  *dst needs external synchronization.
+ * This lets us avoid the cost of a fetch_add when its unnecessary (note that
+ * the types here are atomic).
+ */
+static inline void
+arena_stats_accum_u64(arena_stats_u64_t *dst, uint64_t src) {
+#ifdef JEMALLOC_ATOMIC_U64
+	uint64_t cur_dst = atomic_load_u64(dst, ATOMIC_RELAXED);
+	atomic_store_u64(dst, src + cur_dst, ATOMIC_RELAXED);
+#else
+	*dst += src;
+#endif
+}
+
+static inline size_t
+arena_stats_read_zu(tsdn_t *tsdn, arena_stats_t *arena_stats, atomic_zu_t *p) {
+#ifdef JEMALLOC_ATOMIC_U64
+	return atomic_load_zu(p, ATOMIC_RELAXED);
+#else
+	malloc_mutex_assert_owner(tsdn, &arena_stats->mtx);
+	return atomic_load_zu(p, ATOMIC_RELAXED);
+#endif
+}
+
+static inline void
+arena_stats_add_zu(tsdn_t *tsdn, arena_stats_t *arena_stats, atomic_zu_t *p,
+    size_t x) {
+#ifdef JEMALLOC_ATOMIC_U64
+	atomic_fetch_add_zu(p, x, ATOMIC_RELAXED);
+#else
+	malloc_mutex_assert_owner(tsdn, &arena_stats->mtx);
+	size_t cur = atomic_load_zu(p, ATOMIC_RELAXED);
+	atomic_store_zu(p, cur + x, ATOMIC_RELAXED);
+#endif
+}
+
+static inline void
+arena_stats_sub_zu(tsdn_t *tsdn, arena_stats_t *arena_stats, atomic_zu_t *p,
+    size_t x) {
+#ifdef JEMALLOC_ATOMIC_U64
+	UNUSED size_t r = atomic_fetch_sub_zu(p, x, ATOMIC_RELAXED);
+	assert(r - x <= r);
+#else
+	malloc_mutex_assert_owner(tsdn, &arena_stats->mtx);
+	size_t cur = atomic_load_zu(p, ATOMIC_RELAXED);
+	atomic_store_zu(p, cur - x, ATOMIC_RELAXED);
+#endif
+}
+
+/* Like the _u64 variant, needs an externally synchronized *dst. */
+static inline void
+arena_stats_accum_zu(atomic_zu_t *dst, size_t src) {
+	size_t cur_dst = atomic_load_zu(dst, ATOMIC_RELAXED);
+	atomic_store_zu(dst, src + cur_dst, ATOMIC_RELAXED);
+}
+
+static inline void
+arena_stats_large_nrequests_add(tsdn_t *tsdn, arena_stats_t *arena_stats,
+    szind_t szind, uint64_t nrequests) {
+	arena_stats_lock(tsdn, arena_stats);
+	arena_stats_add_u64(tsdn, arena_stats, &arena_stats->lstats[szind -
+	    NBINS].nrequests, nrequests);
+	arena_stats_unlock(tsdn, arena_stats);
+}
+
+static inline void
+arena_stats_mapped_add(tsdn_t *tsdn, arena_stats_t *arena_stats, size_t size) {
+	arena_stats_lock(tsdn, arena_stats);
+	arena_stats_add_zu(tsdn, arena_stats, &arena_stats->mapped, size);
+	arena_stats_unlock(tsdn, arena_stats);
+}
+
+
+#endif /* JEMALLOC_INTERNAL_ARENA_STATS_H */

include/jemalloc/internal/arena_structs_a.h

@@ -0,0 +1,11 @@
+#ifndef JEMALLOC_INTERNAL_ARENA_STRUCTS_A_H
+#define JEMALLOC_INTERNAL_ARENA_STRUCTS_A_H
+
+#include "jemalloc/internal/bitmap.h"
+
+struct arena_slab_data_s {
+	/* Per region allocated/deallocated bitmap. */
+	bitmap_t	bitmap[BITMAP_GROUPS_MAX];
+};
+
+#endif /* JEMALLOC_INTERNAL_ARENA_STRUCTS_A_H */

include/jemalloc/internal/arena_structs_b.h

@@ -0,0 +1,229 @@
+#ifndef JEMALLOC_INTERNAL_ARENA_STRUCTS_B_H
+#define JEMALLOC_INTERNAL_ARENA_STRUCTS_B_H
+
+#include "jemalloc/internal/arena_stats.h"
+#include "jemalloc/internal/atomic.h"
+#include "jemalloc/internal/bin.h"
+#include "jemalloc/internal/bitmap.h"
+#include "jemalloc/internal/extent_dss.h"
+#include "jemalloc/internal/jemalloc_internal_types.h"
+#include "jemalloc/internal/mutex.h"
+#include "jemalloc/internal/nstime.h"
+#include "jemalloc/internal/ql.h"
+#include "jemalloc/internal/size_classes.h"
+#include "jemalloc/internal/smoothstep.h"
+#include "jemalloc/internal/ticker.h"
+
+struct arena_decay_s {
+	/* Synchronizes all non-atomic fields. */
+	malloc_mutex_t		mtx;
+	/*
+	 * True if a thread is currently purging the extents associated with
+	 * this decay structure.
+	 */
+	bool			purging;
+	/*
+	 * Approximate time in milliseconds from the creation of a set of unused
+	 * dirty pages until an equivalent set of unused dirty pages is purged
+	 * and/or reused.
+	 */
+	atomic_zd_t		time_ms;
+	/* time / SMOOTHSTEP_NSTEPS. */
+	nstime_t		interval;
+	/*
+	 * Time at which the current decay interval logically started.  We do
+	 * not actually advance to a new epoch until sometime after it starts
+	 * because of scheduling and computation delays, and it is even possible
+	 * to completely skip epochs.  In all cases, during epoch advancement we
+	 * merge all relevant activity into the most recently recorded epoch.
+	 */
+	nstime_t		epoch;
+	/* Deadline randomness generator. */
+	uint64_t		jitter_state;
+	/*
+	 * Deadline for current epoch.  This is the sum of interval and per
+	 * epoch jitter which is a uniform random variable in [0..interval).
+	 * Epochs always advance by precise multiples of interval, but we
+	 * randomize the deadline to reduce the likelihood of arenas purging in
+	 * lockstep.
+	 */
+	nstime_t		deadline;
+	/*
+	 * Number of unpurged pages at beginning of current epoch.  During epoch
+	 * advancement we use the delta between arena->decay_*.nunpurged and
+	 * extents_npages_get(&arena->extents_*) to determine how many dirty
+	 * pages, if any, were generated.
+	 */
+	size_t			nunpurged;
+	/*
+	 * Trailing log of how many unused dirty pages were generated during
+	 * each of the past SMOOTHSTEP_NSTEPS decay epochs, where the last
+	 * element is the most recent epoch.  Corresponding epoch times are
+	 * relative to epoch.
+	 */
+	size_t			backlog[SMOOTHSTEP_NSTEPS];
+
+	/*
+	 * Pointer to associated stats.  These stats are embedded directly in
+	 * the arena's stats due to how stats structures are shared between the
+	 * arena and ctl code.
+	 *
+	 * Synchronization: Same as associated arena's stats field. */
+	arena_stats_decay_t	*stats;
+	/* Peak number of pages in associated extents.  Used for debug only. */
+	uint64_t		ceil_npages;
+};
+
+struct arena_s {
+	/*
+	 * Number of threads currently assigned to this arena.  Each thread has
+	 * two distinct assignments, one for application-serving allocation, and
+	 * the other for internal metadata allocation.  Internal metadata must
+	 * not be allocated from arenas explicitly created via the arenas.create
+	 * mallctl, because the arena.<i>.reset mallctl indiscriminately
+	 * discards all allocations for the affected arena.
+	 *
+	 *   0: Application allocation.
+	 *   1: Internal metadata allocation.
+	 *
+	 * Synchronization: atomic.
+	 */
+	atomic_u_t		nthreads[2];
+
+	/*
+	 * When percpu_arena is enabled, to amortize the cost of reading /
+	 * updating the current CPU id, track the most recent thread accessing
+	 * this arena, and only read CPU if there is a mismatch.
+	 */
+	tsdn_t		*last_thd;
+
+	/* Synchronization: internal. */
+	arena_stats_t		stats;
+
+	/*
+	 * Lists of tcaches and cache_bin_array_descriptors for extant threads
+	 * associated with this arena.  Stats from these are merged
+	 * incrementally, and at exit if opt_stats_print is enabled.
+	 *
+	 * Synchronization: tcache_ql_mtx.
+	 */
+	ql_head(tcache_t)			tcache_ql;
+	ql_head(cache_bin_array_descriptor_t)	cache_bin_array_descriptor_ql;
+	malloc_mutex_t				tcache_ql_mtx;
+
+	/* Synchronization: internal. */
+	prof_accum_t		prof_accum;
+	uint64_t		prof_accumbytes;
+
+	/*
+	 * PRNG state for cache index randomization of large allocation base
+	 * pointers.
+	 *
+	 * Synchronization: atomic.
+	 */
+	atomic_zu_t		offset_state;
+
+	/*
+	 * Extent serial number generator state.
+	 *
+	 * Synchronization: atomic.
+	 */
+	atomic_zu_t		extent_sn_next;
+
+	/*
+	 * Represents a dss_prec_t, but atomically.
+	 *
+	 * Synchronization: atomic.
+	 */
+	atomic_u_t		dss_prec;
+
+	/*
+	 * Number of pages in active extents.
+	 *
+	 * Synchronization: atomic.
+	 */
+	atomic_zu_t		nactive;
+
+	/*
+	 * Extant large allocations.
+	 *
+	 * Synchronization: large_mtx.
+	 */
+	extent_list_t		large;
+	/* Synchronizes all large allocation/update/deallocation. */
+	malloc_mutex_t		large_mtx;
+
+	/*
+	 * Collections of extents that were previously allocated.  These are
+	 * used when allocating extents, in an attempt to re-use address space.
+	 *
+	 * Synchronization: internal.
+	 */
+	extents_t		extents_dirty;
+	extents_t		extents_muzzy;
+	extents_t		extents_retained;
+
+	/*
+	 * Decay-based purging state, responsible for scheduling extent state
+	 * transitions.
+	 *
+	 * Synchronization: internal.
+	 */
+	arena_decay_t		decay_dirty; /* dirty --> muzzy */
+	arena_decay_t		decay_muzzy; /* muzzy --> retained */
+
+	/*
+	 * Next extent size class in a growing series to use when satisfying a
+	 * request via the extent hooks (only if opt_retain).  This limits the
+	 * number of disjoint virtual memory ranges so that extent merging can
+	 * be effective even if multiple arenas' extent allocation requests are
+	 * highly interleaved.
+	 *
+	 * retain_grow_limit is the max allowed size ind to expand (unless the
+	 * required size is greater).  Default is no limit, and controlled
+	 * through mallctl only.
+	 *
+	 * Synchronization: extent_grow_mtx
+	 */
+	pszind_t		extent_grow_next;
+	pszind_t		retain_grow_limit;
+	malloc_mutex_t		extent_grow_mtx;
+
+	/*
+	 * Available extent structures that were allocated via
+	 * base_alloc_extent().
+	 *
+	 * Synchronization: extent_avail_mtx.
+	 */
+	extent_tree_t		extent_avail;
+	malloc_mutex_t		extent_avail_mtx;
+
+	/*
+	 * bins is used to store heaps of free regions.
+	 *
+	 * Synchronization: internal.
+	 */
+	bin_t			bins[NBINS];
+
+	/*
+	 * Base allocator, from which arena metadata are allocated.
+	 *
+	 * Synchronization: internal.
+	 */
+	base_t			*base;
+	/* Used to determine uptime.  Read-only after initialization. */
+	nstime_t		create_time;
+};
+
+/* Used in conjunction with tsd for fast arena-related context lookup. */
+struct arena_tdata_s {
+	ticker_t		decay_ticker;
+};
+
+/* Used to pass rtree lookup context down the path. */
+struct alloc_ctx_s {
+	szind_t szind;
+	bool slab;
+};
+
+#endif /* JEMALLOC_INTERNAL_ARENA_STRUCTS_B_H */

include/jemalloc/internal/arena_types.h

@@ -0,0 +1,43 @@
+#ifndef JEMALLOC_INTERNAL_ARENA_TYPES_H
+#define JEMALLOC_INTERNAL_ARENA_TYPES_H
+
+/* Maximum number of regions in one slab. */
+#define LG_SLAB_MAXREGS		(LG_PAGE - LG_TINY_MIN)
+#define SLAB_MAXREGS		(1U << LG_SLAB_MAXREGS)
+
+/* Default decay times in milliseconds. */
+#define DIRTY_DECAY_MS_DEFAULT	ZD(10 * 1000)
+#define MUZZY_DECAY_MS_DEFAULT	ZD(10 * 1000)
+/* Number of event ticks between time checks. */
+#define DECAY_NTICKS_PER_UPDATE	1000
+
+typedef struct arena_slab_data_s arena_slab_data_t;
+typedef struct arena_decay_s arena_decay_t;
+typedef struct arena_s arena_t;
+typedef struct arena_tdata_s arena_tdata_t;
+typedef struct alloc_ctx_s alloc_ctx_t;
+
+typedef enum {
+	percpu_arena_mode_names_base   = 0, /* Used for options processing. */
+
+	/*
+	 * *_uninit are used only during bootstrapping, and must correspond
+	 * to initialized variant plus percpu_arena_mode_enabled_base.
+	 */
+	percpu_arena_uninit            = 0,
+	per_phycpu_arena_uninit        = 1,
+
+	/* All non-disabled modes must come after percpu_arena_disabled. */
+	percpu_arena_disabled          = 2,
+
+	percpu_arena_mode_names_limit  = 3, /* Used for options processing. */
+	percpu_arena_mode_enabled_base = 3,
+
+	percpu_arena                   = 3,
+	per_phycpu_arena               = 4  /* Hyper threads share arena. */
+} percpu_arena_mode_t;
+
+#define PERCPU_ARENA_ENABLED(m)	((m) >= percpu_arena_mode_enabled_base)
+#define PERCPU_ARENA_DEFAULT	percpu_arena_disabled
+
+#endif /* JEMALLOC_INTERNAL_ARENA_TYPES_H */

include/jemalloc/internal/assert.h

@@ -0,0 +1,56 @@
+#include "jemalloc/internal/malloc_io.h"
+#include "jemalloc/internal/util.h"
+
+/*
+ * Define a custom assert() in order to reduce the chances of deadlock during
+ * assertion failure.
+ */
+#ifndef assert
+#define assert(e) do {							\
+	if (unlikely(config_debug && !(e))) {				\
+		malloc_printf(						\
+		    "<jemalloc>: %s:%d: Failed assertion: \"%s\"\n",	\
+		    __FILE__, __LINE__, #e);				\
+		abort();						\
+	}								\
+} while (0)
+#endif
+
+#ifndef not_reached
+#define not_reached() do {						\
+	if (config_debug) {						\
+		malloc_printf(						\
+		    "<jemalloc>: %s:%d: Unreachable code reached\n",	\
+		    __FILE__, __LINE__);				\
+		abort();						\
+	}								\
+	unreachable();							\
+} 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
+
+#ifndef assert_not_implemented
+#define assert_not_implemented(e) do {					\
+	if (unlikely(config_debug && !(e))) {				\
+		not_implemented();					\
+	}								\
+} while (0)
+#endif
+
+/* Use to assert a particular configuration, e.g., cassert(config_debug). */
+#ifndef cassert
+#define cassert(c) do {							\
+	if (unlikely(!(c))) {						\
+		not_reached();						\
+	}								\
+} while (0)
+#endif

include/jemalloc/internal/atomic.h

@@ -1,304 +1,77 @@
-/******************************************************************************/
-#ifdef JEMALLOC_H_TYPES
+#ifndef JEMALLOC_INTERNAL_ATOMIC_H
+#define JEMALLOC_INTERNAL_ATOMIC_H
 
-#endif /* JEMALLOC_H_TYPES */
-/******************************************************************************/
-#ifdef JEMALLOC_H_STRUCTS
+#define ATOMIC_INLINE static inline
 
-#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. */
-#if (LG_SIZEOF_PTR == 3 || LG_SIZEOF_INT == 3)
-#  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(_MSC_VER))
-JEMALLOC_INLINE uint64_t
-atomic_add_uint64(uint64_t *p, uint64_t x)
-{
-
-	return (InterlockedExchangeAdd64(p, x));
-}
-
-JEMALLOC_INLINE uint64_t
-atomic_sub_uint64(uint64_t *p, uint64_t x)
-{
-
-	return (InterlockedExchangeAdd64(p, -((int64_t)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(JEMALLOC_ATOMIC9))
-JEMALLOC_INLINE uint64_t
-atomic_add_uint64(uint64_t *p, uint64_t x)
-{
-
-	/*
-	 * atomic_fetchadd_64() doesn't exist, but we only ever use this
-	 * function on LP64 systems, so atomic_fetchadd_long() will do.
-	 */
-	assert(sizeof(uint64_t) == sizeof(unsigned long));
-
-	return (atomic_fetchadd_long(p, (unsigned long)x) + x);
-}
-
-JEMALLOC_INLINE uint64_t
-atomic_sub_uint64(uint64_t *p, uint64_t x)
-{
-
-	assert(sizeof(uint64_t) == sizeof(unsigned long));
-
-	return (atomic_fetchadd_long(p, (unsigned long)(-(long)x)) - 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
-#    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(_MSC_VER))
-JEMALLOC_INLINE uint32_t
-atomic_add_uint32(uint32_t *p, uint32_t x)
-{
-
-	return (InterlockedExchangeAdd(p, x));
-}
-
-JEMALLOC_INLINE uint32_t
-atomic_sub_uint32(uint32_t *p, uint32_t x)
-{
-
-	return (InterlockedExchangeAdd(p, -((int32_t)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(JEMALLOC_ATOMIC9))
-JEMALLOC_INLINE uint32_t
-atomic_add_uint32(uint32_t *p, uint32_t x)
-{
-
-	return (atomic_fetchadd_32(p, x) + x);
-}
-
-JEMALLOC_INLINE uint32_t
-atomic_sub_uint32(uint32_t *p, uint32_t x)
-{
-
-	return (atomic_fetchadd_32(p, (uint32_t)(-(int32_t)x)) - 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));
-}
+#if defined(JEMALLOC_GCC_ATOMIC_ATOMICS)
+#  include "jemalloc/internal/atomic_gcc_atomic.h"
+#elif defined(JEMALLOC_GCC_SYNC_ATOMICS)
+#  include "jemalloc/internal/atomic_gcc_sync.h"
+#elif defined(_MSC_VER)
+#  include "jemalloc/internal/atomic_msvc.h"
+#elif defined(JEMALLOC_C11_ATOMICS)
+#  include "jemalloc/internal/atomic_c11.h"
 #else
-#  error "Missing implementation for 32-bit atomic operations"
+#  error "Don't have atomics implemented on this platform."
 #endif
 
-/******************************************************************************/
-/* size_t operations. */
-JEMALLOC_INLINE size_t
-atomic_add_z(size_t *p, size_t x)
-{
+/*
+ * This header gives more or less a backport of C11 atomics. The user can write
+ * JEMALLOC_GENERATE_ATOMICS(type, short_type, lg_sizeof_type); to generate
+ * counterparts of the C11 atomic functions for type, as so:
+ *   JEMALLOC_GENERATE_ATOMICS(int *, pi, 3);
+ * and then write things like:
+ *   int *some_ptr;
+ *   atomic_pi_t atomic_ptr_to_int;
+ *   atomic_store_pi(&atomic_ptr_to_int, some_ptr, ATOMIC_RELAXED);
+ *   int *prev_value = atomic_exchange_pi(&ptr_to_int, NULL, ATOMIC_ACQ_REL);
+ *   assert(some_ptr == prev_value);
+ * and expect things to work in the obvious way.
+ *
+ * Also included (with naming differences to avoid conflicts with the standard
+ * library):
+ *   atomic_fence(atomic_memory_order_t) (mimics C11's atomic_thread_fence).
+ *   ATOMIC_INIT (mimics C11's ATOMIC_VAR_INIT).
+ */
 
-#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
-}
+/*
+ * Pure convenience, so that we don't have to type "atomic_memory_order_"
+ * quite so often.
+ */
+#define ATOMIC_RELAXED atomic_memory_order_relaxed
+#define ATOMIC_ACQUIRE atomic_memory_order_acquire
+#define ATOMIC_RELEASE atomic_memory_order_release
+#define ATOMIC_ACQ_REL atomic_memory_order_acq_rel
+#define ATOMIC_SEQ_CST atomic_memory_order_seq_cst
 
-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
-}
-/******************************************************************************/
+/*
+ * Not all platforms have 64-bit atomics.  If we do, this #define exposes that
+ * fact.
+ */
+#if (LG_SIZEOF_PTR == 3 || LG_SIZEOF_INT == 3)
+#  define JEMALLOC_ATOMIC_U64
 #endif
 
-#endif /* JEMALLOC_H_INLINES */
-/******************************************************************************/
+JEMALLOC_GENERATE_ATOMICS(void *, p, LG_SIZEOF_PTR)
+
+/*
+ * There's no actual guarantee that sizeof(bool) == 1, but it's true on the only
+ * platform that actually needs to know the size, MSVC.
+ */
+JEMALLOC_GENERATE_ATOMICS(bool, b, 0)
+
+JEMALLOC_GENERATE_INT_ATOMICS(unsigned, u, LG_SIZEOF_INT)
+
+JEMALLOC_GENERATE_INT_ATOMICS(size_t, zu, LG_SIZEOF_PTR)
+
+JEMALLOC_GENERATE_INT_ATOMICS(ssize_t, zd, LG_SIZEOF_PTR)
+
+JEMALLOC_GENERATE_INT_ATOMICS(uint32_t, u32, 2)
+
+#ifdef JEMALLOC_ATOMIC_U64
+JEMALLOC_GENERATE_INT_ATOMICS(uint64_t, u64, 3)
+#endif
+
+#undef ATOMIC_INLINE
+
+#endif /* JEMALLOC_INTERNAL_ATOMIC_H */

include/jemalloc/internal/atomic_c11.h

@@ -0,0 +1,97 @@
+#ifndef JEMALLOC_INTERNAL_ATOMIC_C11_H
+#define JEMALLOC_INTERNAL_ATOMIC_C11_H
+
+#include <stdatomic.h>
+
+#define ATOMIC_INIT(...) ATOMIC_VAR_INIT(__VA_ARGS__)
+
+#define atomic_memory_order_t memory_order
+#define atomic_memory_order_relaxed memory_order_relaxed
+#define atomic_memory_order_acquire memory_order_acquire
+#define atomic_memory_order_release memory_order_release
+#define atomic_memory_order_acq_rel memory_order_acq_rel
+#define atomic_memory_order_seq_cst memory_order_seq_cst
+
+#define atomic_fence atomic_thread_fence
+
+#define JEMALLOC_GENERATE_ATOMICS(type, short_type,			\
+    /* unused */ lg_size)						\
+typedef _Atomic(type) atomic_##short_type##_t;				\
+									\
+ATOMIC_INLINE type							\
+atomic_load_##short_type(const atomic_##short_type##_t *a,		\
+    atomic_memory_order_t mo) {						\
+	/*								\
+	 * A strict interpretation of the C standard prevents		\
+	 * atomic_load from taking a const argument, but it's		\
+	 * convenient for our purposes. This cast is a workaround.	\
+	 */								\
+	atomic_##short_type##_t* a_nonconst =				\
+	    (atomic_##short_type##_t*)a;				\
+	return atomic_load_explicit(a_nonconst, mo);			\
+}									\
+									\
+ATOMIC_INLINE void							\
+atomic_store_##short_type(atomic_##short_type##_t *a,			\
+    type val, atomic_memory_order_t mo) {				\
+	atomic_store_explicit(a, val, mo);				\
+}									\
+									\
+ATOMIC_INLINE type							\
+atomic_exchange_##short_type(atomic_##short_type##_t *a, type val,	\
+    atomic_memory_order_t mo) {						\
+	return atomic_exchange_explicit(a, val, mo);			\
+}									\
+									\
+ATOMIC_INLINE bool							\
+atomic_compare_exchange_weak_##short_type(atomic_##short_type##_t *a,	\
+    type *expected, type desired, atomic_memory_order_t success_mo,	\
+    atomic_memory_order_t failure_mo) {					\
+	return atomic_compare_exchange_weak_explicit(a, expected,	\
+	    desired, success_mo, failure_mo);				\
+}									\
+									\
+ATOMIC_INLINE bool							\
+atomic_compare_exchange_strong_##short_type(atomic_##short_type##_t *a,	\
+    type *expected, type desired, atomic_memory_order_t success_mo,	\
+    atomic_memory_order_t failure_mo) {					\
+	return atomic_compare_exchange_strong_explicit(a, expected,	\
+	    desired, success_mo, failure_mo);				\
+}
+
+/*
+ * Integral types have some special operations available that non-integral ones
+ * lack.
+ */
+#define JEMALLOC_GENERATE_INT_ATOMICS(type, short_type, 		\
+    /* unused */ lg_size)						\
+JEMALLOC_GENERATE_ATOMICS(type, short_type, /* unused */ lg_size)	\
+									\
+ATOMIC_INLINE type							\
+atomic_fetch_add_##short_type(atomic_##short_type##_t *a,		\
+    type val, atomic_memory_order_t mo) {				\
+	return atomic_fetch_add_explicit(a, val, mo);			\
+}									\
+									\
+ATOMIC_INLINE type							\
+atomic_fetch_sub_##short_type(atomic_##short_type##_t *a,		\
+    type val, atomic_memory_order_t mo) {				\
+	return atomic_fetch_sub_explicit(a, val, mo);			\
+}									\
+ATOMIC_INLINE type							\
+atomic_fetch_and_##short_type(atomic_##short_type##_t *a,		\
+    type val, atomic_memory_order_t mo) {				\
+	return atomic_fetch_and_explicit(a, val, mo);			\
+}									\
+ATOMIC_INLINE type							\
+atomic_fetch_or_##short_type(atomic_##short_type##_t *a,		\
+    type val, atomic_memory_order_t mo) {				\
+	return atomic_fetch_or_explicit(a, val, mo);			\
+}									\
+ATOMIC_INLINE type							\
+atomic_fetch_xor_##short_type(atomic_##short_type##_t *a,		\
+    type val, atomic_memory_order_t mo) {				\
+	return atomic_fetch_xor_explicit(a, val, mo);			\
+}
+
+#endif /* JEMALLOC_INTERNAL_ATOMIC_C11_H */

include/jemalloc/internal/atomic_gcc_atomic.h

@@ -0,0 +1,127 @@
+#ifndef JEMALLOC_INTERNAL_ATOMIC_GCC_ATOMIC_H
+#define JEMALLOC_INTERNAL_ATOMIC_GCC_ATOMIC_H
+
+#include "jemalloc/internal/assert.h"
+
+#define ATOMIC_INIT(...) {__VA_ARGS__}
+
+typedef enum {
+	atomic_memory_order_relaxed,
+	atomic_memory_order_acquire,
+	atomic_memory_order_release,
+	atomic_memory_order_acq_rel,
+	atomic_memory_order_seq_cst
+} atomic_memory_order_t;
+
+ATOMIC_INLINE int
+atomic_enum_to_builtin(atomic_memory_order_t mo) {
+	switch (mo) {
+	case atomic_memory_order_relaxed:
+		return __ATOMIC_RELAXED;
+	case atomic_memory_order_acquire:
+		return __ATOMIC_ACQUIRE;
+	case atomic_memory_order_release:
+		return __ATOMIC_RELEASE;
+	case atomic_memory_order_acq_rel:
+		return __ATOMIC_ACQ_REL;
+	case atomic_memory_order_seq_cst:
+		return __ATOMIC_SEQ_CST;
+	}
+	/* Can't happen; the switch is exhaustive. */
+	not_reached();
+}
+
+ATOMIC_INLINE void
+atomic_fence(atomic_memory_order_t mo) {
+	__atomic_thread_fence(atomic_enum_to_builtin(mo));
+}
+
+#define JEMALLOC_GENERATE_ATOMICS(type, short_type,			\
+    /* unused */ lg_size)						\
+typedef struct {							\
+	type repr;							\
+} atomic_##short_type##_t;						\
+									\
+ATOMIC_INLINE type							\
+atomic_load_##short_type(const atomic_##short_type##_t *a,		\
+    atomic_memory_order_t mo) {						\
+	type result;							\
+	__atomic_load(&a->repr, &result, atomic_enum_to_builtin(mo));	\
+	return result;							\
+}									\
+									\
+ATOMIC_INLINE void							\
+atomic_store_##short_type(atomic_##short_type##_t *a, type val,		\
+    atomic_memory_order_t mo) {						\
+	__atomic_store(&a->repr, &val, atomic_enum_to_builtin(mo));	\
+}									\
+									\
+ATOMIC_INLINE type							\
+atomic_exchange_##short_type(atomic_##short_type##_t *a, type val,	\
+    atomic_memory_order_t mo) {						\
+	type result;							\
+	__atomic_exchange(&a->repr, &val, &result,			\
+	    atomic_enum_to_builtin(mo));				\
+	return result;							\
+}									\
+									\
+ATOMIC_INLINE bool							\
+atomic_compare_exchange_weak_##short_type(atomic_##short_type##_t *a,	\
+    type *expected, type desired, atomic_memory_order_t success_mo,	\
+    atomic_memory_order_t failure_mo) {					\
+	return __atomic_compare_exchange(&a->repr, expected, &desired,	\
+	    true, atomic_enum_to_builtin(success_mo),			\
+	    atomic_enum_to_builtin(failure_mo));			\
+}									\
+									\
+ATOMIC_INLINE bool							\
+atomic_compare_exchange_strong_##short_type(atomic_##short_type##_t *a,	\
+    type *expected, type desired, atomic_memory_order_t success_mo,	\
+    atomic_memory_order_t failure_mo) {					\
+	return __atomic_compare_exchange(&a->repr, expected, &desired,	\
+	    false,							\
+	    atomic_enum_to_builtin(success_mo),				\
+	    atomic_enum_to_builtin(failure_mo));			\
+}
+
+
+#define JEMALLOC_GENERATE_INT_ATOMICS(type, short_type,			\
+    /* unused */ lg_size)						\
+JEMALLOC_GENERATE_ATOMICS(type, short_type, /* unused */ lg_size)	\
+									\
+ATOMIC_INLINE type							\
+atomic_fetch_add_##short_type(atomic_##short_type##_t *a, type val,	\
+    atomic_memory_order_t mo) {						\
+	return __atomic_fetch_add(&a->repr, val,			\
+	    atomic_enum_to_builtin(mo));				\
+}									\
+									\
+ATOMIC_INLINE type							\
+atomic_fetch_sub_##short_type(atomic_##short_type##_t *a, type val,	\
+    atomic_memory_order_t mo) {						\
+	return __atomic_fetch_sub(&a->repr, val,			\
+	    atomic_enum_to_builtin(mo));				\
+}									\
+									\
+ATOMIC_INLINE type							\
+atomic_fetch_and_##short_type(atomic_##short_type##_t *a, type val,	\
+    atomic_memory_order_t mo) {						\
+	return __atomic_fetch_and(&a->repr, val,			\
+	    atomic_enum_to_builtin(mo));				\
+}									\
+									\
+ATOMIC_INLINE type							\
+atomic_fetch_or_##short_type(atomic_##short_type##_t *a, type val,	\
+    atomic_memory_order_t mo) {						\
+	return __atomic_fetch_or(&a->repr, val,				\
+	    atomic_enum_to_builtin(mo));				\
+}									\
+									\
+ATOMIC_INLINE type							\
+atomic_fetch_xor_##short_type(atomic_##short_type##_t *a, type val,	\
+    atomic_memory_order_t mo) {						\
+	return __atomic_fetch_xor(&a->repr, val,			\
+	    atomic_enum_to_builtin(mo));				\
+}
+
+#endif /* JEMALLOC_INTERNAL_ATOMIC_GCC_ATOMIC_H */

include/jemalloc/internal/atomic_gcc_sync.h

@@ -0,0 +1,191 @@
+#ifndef JEMALLOC_INTERNAL_ATOMIC_GCC_SYNC_H
+#define JEMALLOC_INTERNAL_ATOMIC_GCC_SYNC_H
+
+#define ATOMIC_INIT(...) {__VA_ARGS__}
+
+typedef enum {
+	atomic_memory_order_relaxed,
+	atomic_memory_order_acquire,
+	atomic_memory_order_release,
+	atomic_memory_order_acq_rel,
+	atomic_memory_order_seq_cst
+} atomic_memory_order_t;
+
+ATOMIC_INLINE void
+atomic_fence(atomic_memory_order_t mo) {
+	/* Easy cases first: no barrier, and full barrier. */
+	if (mo == atomic_memory_order_relaxed) {
+		asm volatile("" ::: "memory");
+		return;
+	}
+	if (mo == atomic_memory_order_seq_cst) {
+		asm volatile("" ::: "memory");
+		__sync_synchronize();
+		asm volatile("" ::: "memory");
+		return;
+	}
+	asm volatile("" ::: "memory");
+#  if defined(__i386__) || defined(__x86_64__)
+	/* This is implicit on x86. */
+#  elif defined(__ppc__)
+	asm volatile("lwsync");
+#  elif defined(__sparc__) && defined(__arch64__)
+	if (mo == atomic_memory_order_acquire) {
+		asm volatile("membar #LoadLoad | #LoadStore");
+	} else if (mo == atomic_memory_order_release) {
+		asm volatile("membar #LoadStore | #StoreStore");
+	} else {
+		asm volatile("membar #LoadLoad | #LoadStore | #StoreStore");
+	}
+#  else
+	__sync_synchronize();
+#  endif
+	asm volatile("" ::: "memory");
+}
+
+/*
+ * A correct implementation of seq_cst loads and stores on weakly ordered
+ * architectures could do either of the following:
+ *   1. store() is weak-fence -> store -> strong fence, load() is load ->
+ *      strong-fence.
+ *   2. store() is strong-fence -> store, load() is strong-fence -> load ->
+ *      weak-fence.
+ * The tricky thing is, load() and store() above can be the load or store
+ * portions of a gcc __sync builtin, so we have to follow GCC's lead, which
+ * means going with strategy 2.
+ * On strongly ordered architectures, the natural strategy is to stick a strong
+ * fence after seq_cst stores, and have naked loads.  So we want the strong
+ * fences in different places on different architectures.
+ * atomic_pre_sc_load_fence and atomic_post_sc_store_fence allow us to
+ * accomplish this.
+ */
+
+ATOMIC_INLINE void
+atomic_pre_sc_load_fence() {
+#  if defined(__i386__) || defined(__x86_64__) ||			\
+    (defined(__sparc__) && defined(__arch64__))
+	atomic_fence(atomic_memory_order_relaxed);
+#  else
+	atomic_fence(atomic_memory_order_seq_cst);
+#  endif
+}
+
+ATOMIC_INLINE void
+atomic_post_sc_store_fence() {
+#  if defined(__i386__) || defined(__x86_64__) ||			\
+    (defined(__sparc__) && defined(__arch64__))
+	atomic_fence(atomic_memory_order_seq_cst);
+#  else
+	atomic_fence(atomic_memory_order_relaxed);
+#  endif
+
+}
+
+#define JEMALLOC_GENERATE_ATOMICS(type, short_type,			\
+    /* unused */ lg_size)						\
+typedef struct {							\
+	type volatile repr;						\
+} atomic_##short_type##_t;						\
+									\
+ATOMIC_INLINE type							\
+atomic_load_##short_type(const atomic_##short_type##_t *a,		\
+    atomic_memory_order_t mo) {						\
+	if (mo == atomic_memory_order_seq_cst) {			\
+		atomic_pre_sc_load_fence();				\
+	}								\
+	type result = a->repr;						\
+	if (mo != atomic_memory_order_relaxed) {			\
+		atomic_fence(atomic_memory_order_acquire);		\
+	}								\
+	return result;							\
+}									\
+									\
+ATOMIC_INLINE void							\
+atomic_store_##short_type(atomic_##short_type##_t *a,			\
+    type val, atomic_memory_order_t mo) {				\
+	if (mo != atomic_memory_order_relaxed) {			\
+		atomic_fence(atomic_memory_order_release);		\
+	}								\
+	a->repr = val;							\
+	if (mo == atomic_memory_order_seq_cst) {			\
+		atomic_post_sc_store_fence();				\
+	}								\
+}									\
+									\
+ATOMIC_INLINE type							\
+atomic_exchange_##short_type(atomic_##short_type##_t *a, type val,	\
+    atomic_memory_order_t mo) {						\
+	/*								\
+	 * Because of FreeBSD, we care about gcc 4.2, which doesn't have\
+	 * an atomic exchange builtin.  We fake it with a CAS loop.	\
+	 */								\
+	while (true) {							\
+		type old = a->repr;					\
+		if (__sync_bool_compare_and_swap(&a->repr, old, val)) {	\
+			return old;					\
+		}							\
+	}								\
+}									\
+									\
+ATOMIC_INLINE bool							\
+atomic_compare_exchange_weak_##short_type(atomic_##short_type##_t *a,	\
+    type *expected, type desired, atomic_memory_order_t success_mo,	\
+    atomic_memory_order_t failure_mo) {					\
+	type prev = __sync_val_compare_and_swap(&a->repr, *expected,	\
+	    desired);							\
+	if (prev == *expected) {					\
+		return true;						\
+	} else {							\
+		*expected = prev;					\
+		return false;						\
+	}								\
+}									\
+ATOMIC_INLINE bool							\
+atomic_compare_exchange_strong_##short_type(atomic_##short_type##_t *a,	\
+    type *expected, type desired, atomic_memory_order_t success_mo,	\
+    atomic_memory_order_t failure_mo) {					\
+	type prev = __sync_val_compare_and_swap(&a->repr, *expected,	\
+	    desired);							\
+	if (prev == *expected) {					\
+		return true;						\
+	} else {							\
+		*expected = prev;					\
+		return false;						\
+	}								\
+}
+
+#define JEMALLOC_GENERATE_INT_ATOMICS(type, short_type,			\
+    /* unused */ lg_size)						\
+JEMALLOC_GENERATE_ATOMICS(type, short_type, /* unused */ lg_size)	\
+									\
+ATOMIC_INLINE type							\
+atomic_fetch_add_##short_type(atomic_##short_type##_t *a, type val,	\
+    atomic_memory_order_t mo) {						\
+	return __sync_fetch_and_add(&a->repr, val);			\
+}									\
+									\
+ATOMIC_INLINE type							\
+atomic_fetch_sub_##short_type(atomic_##short_type##_t *a, type val,	\
+    atomic_memory_order_t mo) {						\
+	return __sync_fetch_and_sub(&a->repr, val);			\
+}									\
+									\
+ATOMIC_INLINE type							\
+atomic_fetch_and_##short_type(atomic_##short_type##_t *a, type val,	\
+    atomic_memory_order_t mo) {						\
+	return __sync_fetch_and_and(&a->repr, val);			\
+}									\
+									\
+ATOMIC_INLINE type							\
+atomic_fetch_or_##short_type(atomic_##short_type##_t *a, type val,	\
+    atomic_memory_order_t mo) {						\
+	return __sync_fetch_and_or(&a->repr, val);			\
+}									\
+									\
+ATOMIC_INLINE type							\
+atomic_fetch_xor_##short_type(atomic_##short_type##_t *a, type val,	\
+    atomic_memory_order_t mo) {						\
+	return __sync_fetch_and_xor(&a->repr, val);			\
+}
+
+#endif /* JEMALLOC_INTERNAL_ATOMIC_GCC_SYNC_H */

include/jemalloc/internal/atomic_msvc.h

@@ -0,0 +1,158 @@
+#ifndef JEMALLOC_INTERNAL_ATOMIC_MSVC_H
+#define JEMALLOC_INTERNAL_ATOMIC_MSVC_H
+
+#define ATOMIC_INIT(...) {__VA_ARGS__}
+
+typedef enum {
+	atomic_memory_order_relaxed,
+	atomic_memory_order_acquire,
+	atomic_memory_order_release,
+	atomic_memory_order_acq_rel,
+	atomic_memory_order_seq_cst
+} atomic_memory_order_t;
+
+typedef char atomic_repr_0_t;
+typedef short atomic_repr_1_t;
+typedef long atomic_repr_2_t;
+typedef __int64 atomic_repr_3_t;
+
+ATOMIC_INLINE void
+atomic_fence(atomic_memory_order_t mo) {
+	_ReadWriteBarrier();
+#  if defined(_M_ARM) || defined(_M_ARM64)
+	/* ARM needs a barrier for everything but relaxed. */
+	if (mo != atomic_memory_order_relaxed) {
+		MemoryBarrier();
+	}
+#  elif defined(_M_IX86) || defined (_M_X64)
+	/* x86 needs a barrier only for seq_cst. */
+	if (mo == atomic_memory_order_seq_cst) {
+		MemoryBarrier();
+	}
+#  else
+#  error "Don't know how to create atomics for this platform for MSVC."
+#  endif
+	_ReadWriteBarrier();
+}
+
+#define ATOMIC_INTERLOCKED_REPR(lg_size) atomic_repr_ ## lg_size ## _t
+
+#define ATOMIC_CONCAT(a, b) ATOMIC_RAW_CONCAT(a, b)
+#define ATOMIC_RAW_CONCAT(a, b) a ## b
+
+#define ATOMIC_INTERLOCKED_NAME(base_name, lg_size) ATOMIC_CONCAT(	\
+    base_name, ATOMIC_INTERLOCKED_SUFFIX(lg_size))
+
+#define ATOMIC_INTERLOCKED_SUFFIX(lg_size)				\
+    ATOMIC_CONCAT(ATOMIC_INTERLOCKED_SUFFIX_, lg_size)
+
+#define ATOMIC_INTERLOCKED_SUFFIX_0 8
+#define ATOMIC_INTERLOCKED_SUFFIX_1 16
+#define ATOMIC_INTERLOCKED_SUFFIX_2
+#define ATOMIC_INTERLOCKED_SUFFIX_3 64
+
+#define JEMALLOC_GENERATE_ATOMICS(type, short_type, lg_size)		\
+typedef struct {							\
+	ATOMIC_INTERLOCKED_REPR(lg_size) repr;				\
+} atomic_##short_type##_t;						\
+									\
+ATOMIC_INLINE type							\
+atomic_load_##short_type(const atomic_##short_type##_t *a,		\
+    atomic_memory_order_t mo) {						\
+	ATOMIC_INTERLOCKED_REPR(lg_size) ret = a->repr;			\
+	if (mo != atomic_memory_order_relaxed) {			\
+		atomic_fence(atomic_memory_order_acquire);		\
+	}								\
+	return (type) ret;						\
+}									\
+									\
+ATOMIC_INLINE void							\
+atomic_store_##short_type(atomic_##short_type##_t *a,			\
+    type val, atomic_memory_order_t mo) {				\
+	if (mo != atomic_memory_order_relaxed) {			\
+		atomic_fence(atomic_memory_order_release);		\
+	}								\
+	a->repr = (ATOMIC_INTERLOCKED_REPR(lg_size)) val;		\
+	if (mo == atomic_memory_order_seq_cst) {			\
+		atomic_fence(atomic_memory_order_seq_cst);		\
+	}								\
+}									\
+									\
+ATOMIC_INLINE type							\
+atomic_exchange_##short_type(atomic_##short_type##_t *a, type val,	\
+    atomic_memory_order_t mo) {						\
+	return (type)ATOMIC_INTERLOCKED_NAME(_InterlockedExchange,	\
+	    lg_size)(&a->repr, (ATOMIC_INTERLOCKED_REPR(lg_size))val);	\
+}									\
+									\
+ATOMIC_INLINE bool							\
+atomic_compare_exchange_weak_##short_type(atomic_##short_type##_t *a,	\
+    type *expected, type desired, atomic_memory_order_t success_mo,	\
+    atomic_memory_order_t failure_mo) {					\
+	ATOMIC_INTERLOCKED_REPR(lg_size) e =				\
+	    (ATOMIC_INTERLOCKED_REPR(lg_size))*expected;		\
+	ATOMIC_INTERLOCKED_REPR(lg_size) d =				\
+	    (ATOMIC_INTERLOCKED_REPR(lg_size))desired;			\
+	ATOMIC_INTERLOCKED_REPR(lg_size) old =				\
+	    ATOMIC_INTERLOCKED_NAME(_InterlockedCompareExchange, 	\
+		lg_size)(&a->repr, d, e);				\
+	if (old == e) {							\
+		return true;						\
+	} else {							\
+		*expected = (type)old;					\
+		return false;						\
+	}								\
+}									\
+									\
+ATOMIC_INLINE bool							\
+atomic_compare_exchange_strong_##short_type(atomic_##short_type##_t *a,	\
+    type *expected, type desired, atomic_memory_order_t success_mo,	\
+    atomic_memory_order_t failure_mo) {					\
+	/* We implement the weak version with strong semantics. */	\
+	return atomic_compare_exchange_weak_##short_type(a, expected,	\
+	    desired, success_mo, failure_mo);				\
+}
+
+
+#define JEMALLOC_GENERATE_INT_ATOMICS(type, short_type, lg_size)	\
+JEMALLOC_GENERATE_ATOMICS(type, short_type, lg_size)			\
+									\
+ATOMIC_INLINE type							\
+atomic_fetch_add_##short_type(atomic_##short_type##_t *a,		\
+    type val, atomic_memory_order_t mo) {				\
+	return (type)ATOMIC_INTERLOCKED_NAME(_InterlockedExchangeAdd,	\
+	    lg_size)(&a->repr, (ATOMIC_INTERLOCKED_REPR(lg_size))val);	\
+}									\
+									\
+ATOMIC_INLINE type							\
+atomic_fetch_sub_##short_type(atomic_##short_type##_t *a,		\
+    type val, atomic_memory_order_t mo) {				\
+	/*								\
+	 * MSVC warns on negation of unsigned operands, but for us it	\
+	 * gives exactly the right semantics (MAX_TYPE + 1 - operand).	\
+	 */								\
+	__pragma(warning(push))						\
+	__pragma(warning(disable: 4146))				\
+	return atomic_fetch_add_##short_type(a, -val, mo);		\
+	__pragma(warning(pop))						\
+}									\
+ATOMIC_INLINE type							\
+atomic_fetch_and_##short_type(atomic_##short_type##_t *a,		\
+    type val, atomic_memory_order_t mo) {				\
+	return (type)ATOMIC_INTERLOCKED_NAME(_InterlockedAnd, lg_size)(	\
+	    &a->repr, (ATOMIC_INTERLOCKED_REPR(lg_size))val);		\
+}									\
+ATOMIC_INLINE type							\
+atomic_fetch_or_##short_type(atomic_##short_type##_t *a,		\
+    type val, atomic_memory_order_t mo) {				\
+	return (type)ATOMIC_INTERLOCKED_NAME(_InterlockedOr, lg_size)(	\
+	    &a->repr, (ATOMIC_INTERLOCKED_REPR(lg_size))val);		\
+}									\
+ATOMIC_INLINE type							\
+atomic_fetch_xor_##short_type(atomic_##short_type##_t *a,		\
+    type val, atomic_memory_order_t mo) {				\
+	return (type)ATOMIC_INTERLOCKED_NAME(_InterlockedXor, lg_size)(	\
+	    &a->repr, (ATOMIC_INTERLOCKED_REPR(lg_size))val);		\
+}
+
+#endif /* JEMALLOC_INTERNAL_ATOMIC_MSVC_H */

include/jemalloc/internal/background_thread_externs.h

@@ -0,0 +1,33 @@
+#ifndef JEMALLOC_INTERNAL_BACKGROUND_THREAD_EXTERNS_H
+#define JEMALLOC_INTERNAL_BACKGROUND_THREAD_EXTERNS_H
+
+extern bool opt_background_thread;
+extern size_t opt_max_background_threads;
+extern malloc_mutex_t background_thread_lock;
+extern atomic_b_t background_thread_enabled_state;
+extern size_t n_background_threads;
+extern size_t max_background_threads;
+extern background_thread_info_t *background_thread_info;
+extern bool can_enable_background_thread;
+
+bool background_thread_create(tsd_t *tsd, unsigned arena_ind);
+bool background_threads_enable(tsd_t *tsd);
+bool background_threads_disable(tsd_t *tsd);
+void background_thread_interval_check(tsdn_t *tsdn, arena_t *arena,
+    arena_decay_t *decay, size_t npages_new);
+void background_thread_prefork0(tsdn_t *tsdn);
+void background_thread_prefork1(tsdn_t *tsdn);
+void background_thread_postfork_parent(tsdn_t *tsdn);
+void background_thread_postfork_child(tsdn_t *tsdn);
+bool background_thread_stats_read(tsdn_t *tsdn,
+    background_thread_stats_t *stats);
+void background_thread_ctl_init(tsdn_t *tsdn);
+
+#ifdef JEMALLOC_PTHREAD_CREATE_WRAPPER
+extern int pthread_create_wrapper(pthread_t *__restrict, const pthread_attr_t *,
+    void *(*)(void *), void *__restrict);
+#endif
+bool background_thread_boot0(void);
+bool background_thread_boot1(tsdn_t *tsdn);
+
+#endif /* JEMALLOC_INTERNAL_BACKGROUND_THREAD_EXTERNS_H */

include/jemalloc/internal/background_thread_inlines.h

@@ -0,0 +1,57 @@
+#ifndef JEMALLOC_INTERNAL_BACKGROUND_THREAD_INLINES_H
+#define JEMALLOC_INTERNAL_BACKGROUND_THREAD_INLINES_H
+
+JEMALLOC_ALWAYS_INLINE bool
+background_thread_enabled(void) {
+	return atomic_load_b(&background_thread_enabled_state, ATOMIC_RELAXED);
+}
+
+JEMALLOC_ALWAYS_INLINE void
+background_thread_enabled_set(tsdn_t *tsdn, bool state) {
+	malloc_mutex_assert_owner(tsdn, &background_thread_lock);
+	atomic_store_b(&background_thread_enabled_state, state, ATOMIC_RELAXED);
+}
+
+JEMALLOC_ALWAYS_INLINE background_thread_info_t *
+arena_background_thread_info_get(arena_t *arena) {
+	unsigned arena_ind = arena_ind_get(arena);
+	return &background_thread_info[arena_ind % ncpus];
+}
+
+JEMALLOC_ALWAYS_INLINE uint64_t
+background_thread_wakeup_time_get(background_thread_info_t *info) {
+	uint64_t next_wakeup = nstime_ns(&info->next_wakeup);
+	assert(atomic_load_b(&info->indefinite_sleep, ATOMIC_ACQUIRE) ==
+	    (next_wakeup == BACKGROUND_THREAD_INDEFINITE_SLEEP));
+	return next_wakeup;
+}
+
+JEMALLOC_ALWAYS_INLINE void
+background_thread_wakeup_time_set(tsdn_t *tsdn, background_thread_info_t *info,
+    uint64_t wakeup_time) {
+	malloc_mutex_assert_owner(tsdn, &info->mtx);
+	atomic_store_b(&info->indefinite_sleep,
+	    wakeup_time == BACKGROUND_THREAD_INDEFINITE_SLEEP, ATOMIC_RELEASE);
+	nstime_init(&info->next_wakeup, wakeup_time);
+}
+
+JEMALLOC_ALWAYS_INLINE bool
+background_thread_indefinite_sleep(background_thread_info_t *info) {
+	return atomic_load_b(&info->indefinite_sleep, ATOMIC_ACQUIRE);
+}
+
+JEMALLOC_ALWAYS_INLINE void
+arena_background_thread_inactivity_check(tsdn_t *tsdn, arena_t *arena,
+    bool is_background_thread) {
+	if (!background_thread_enabled() || is_background_thread) {
+		return;
+	}
+	background_thread_info_t *info =
+	    arena_background_thread_info_get(arena);
+	if (background_thread_indefinite_sleep(info)) {
+		background_thread_interval_check(tsdn, arena,
+		    &arena->decay_dirty, 0);
+	}
+}
+
+#endif /* JEMALLOC_INTERNAL_BACKGROUND_THREAD_INLINES_H */

include/jemalloc/internal/background_thread_structs.h

@@ -0,0 +1,53 @@
+#ifndef JEMALLOC_INTERNAL_BACKGROUND_THREAD_STRUCTS_H
+#define JEMALLOC_INTERNAL_BACKGROUND_THREAD_STRUCTS_H
+
+/* This file really combines "structs" and "types", but only transitionally. */
+
+#if defined(JEMALLOC_BACKGROUND_THREAD) || defined(JEMALLOC_LAZY_LOCK)
+#  define JEMALLOC_PTHREAD_CREATE_WRAPPER
+#endif
+
+#define BACKGROUND_THREAD_INDEFINITE_SLEEP UINT64_MAX
+#define MAX_BACKGROUND_THREAD_LIMIT MALLOCX_ARENA_LIMIT
+
+typedef enum {
+	background_thread_stopped,
+	background_thread_started,
+	/* Thread waits on the global lock when paused (for arena_reset). */
+	background_thread_paused,
+} background_thread_state_t;
+
+struct background_thread_info_s {
+#ifdef JEMALLOC_BACKGROUND_THREAD
+	/* Background thread is pthread specific. */
+	pthread_t		thread;
+	pthread_cond_t		cond;
+#endif
+	malloc_mutex_t		mtx;
+	background_thread_state_t	state;
+	/* When true, it means no wakeup scheduled. */
+	atomic_b_t		indefinite_sleep;
+	/* Next scheduled wakeup time (absolute time in ns). */
+	nstime_t		next_wakeup;
+	/*
+	 *  Since the last background thread run, newly added number of pages
+	 *  that need to be purged by the next wakeup.  This is adjusted on
+	 *  epoch advance, and is used to determine whether we should signal the
+	 *  background thread to wake up earlier.
+	 */
+	size_t			npages_to_purge_new;
+	/* Stats: total number of runs since started. */
+	uint64_t		tot_n_runs;
+	/* Stats: total sleep time since started. */
+	nstime_t		tot_sleep_time;
+};
+typedef struct background_thread_info_s background_thread_info_t;
+
+struct background_thread_stats_s {
+	size_t num_threads;
+	uint64_t num_runs;
+	nstime_t run_interval;
+};
+typedef struct background_thread_stats_s background_thread_stats_t;
+
+#endif /* JEMALLOC_INTERNAL_BACKGROUND_THREAD_STRUCTS_H */

include/jemalloc/internal/base.h

@@ -1,26 +0,0 @@
-/******************************************************************************/
-#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 */
-/******************************************************************************/

include/jemalloc/internal/base_externs.h

@@ -0,0 +1,22 @@
+#ifndef JEMALLOC_INTERNAL_BASE_EXTERNS_H
+#define JEMALLOC_INTERNAL_BASE_EXTERNS_H
+
+extern metadata_thp_mode_t opt_metadata_thp;
+extern const char *metadata_thp_mode_names[];
+
+base_t *b0get(void);
+base_t *base_new(tsdn_t *tsdn, unsigned ind, extent_hooks_t *extent_hooks);
+void base_delete(tsdn_t *tsdn, base_t *base);
+extent_hooks_t *base_extent_hooks_get(base_t *base);
+extent_hooks_t *base_extent_hooks_set(base_t *base,
+    extent_hooks_t *extent_hooks);
+void *base_alloc(tsdn_t *tsdn, base_t *base, size_t size, size_t alignment);
+extent_t *base_alloc_extent(tsdn_t *tsdn, base_t *base);
+void base_stats_get(tsdn_t *tsdn, base_t *base, size_t *allocated,
+    size_t *resident, size_t *mapped, size_t *n_thp);
+void base_prefork(tsdn_t *tsdn, base_t *base);
+void base_postfork_parent(tsdn_t *tsdn, base_t *base);
+void base_postfork_child(tsdn_t *tsdn, base_t *base);
+bool base_boot(tsdn_t *tsdn);
+
+#endif /* JEMALLOC_INTERNAL_BASE_EXTERNS_H */

include/jemalloc/internal/base_inlines.h

@@ -0,0 +1,13 @@
+#ifndef JEMALLOC_INTERNAL_BASE_INLINES_H
+#define JEMALLOC_INTERNAL_BASE_INLINES_H
+
+static inline unsigned
+base_ind_get(const base_t *base) {
+	return base->ind;
+}
+
+static inline bool
+metadata_thp_enabled(void) {
+	return (opt_metadata_thp != metadata_thp_disabled);
+}
+#endif /* JEMALLOC_INTERNAL_BASE_INLINES_H */

include/jemalloc/internal/base_structs.h

@@ -0,0 +1,59 @@
+#ifndef JEMALLOC_INTERNAL_BASE_STRUCTS_H
+#define JEMALLOC_INTERNAL_BASE_STRUCTS_H
+
+#include "jemalloc/internal/jemalloc_internal_types.h"
+#include "jemalloc/internal/mutex.h"
+#include "jemalloc/internal/size_classes.h"
+
+/* Embedded at the beginning of every block of base-managed virtual memory. */
+struct base_block_s {
+	/* Total size of block's virtual memory mapping. */
+	size_t		size;
+
+	/* Next block in list of base's blocks. */
+	base_block_t	*next;
+
+	/* Tracks unused trailing space. */
+	extent_t	extent;
+};
+
+struct base_s {
+	/* Associated arena's index within the arenas array. */
+	unsigned	ind;
+
+	/*
+	 * User-configurable extent hook functions.  Points to an
+	 * extent_hooks_t.
+	 */
+	atomic_p_t	extent_hooks;
+
+	/* Protects base_alloc() and base_stats_get() operations. */
+	malloc_mutex_t	mtx;
+
+	/* Using THP when true (metadata_thp auto mode). */
+	bool		auto_thp_switched;
+	/*
+	 * Most recent size class in the series of increasingly large base
+	 * extents.  Logarithmic spacing between subsequent allocations ensures
+	 * that the total number of distinct mappings remains small.
+	 */
+	pszind_t	pind_last;
+
+	/* Serial number generation state. */
+	size_t		extent_sn_next;
+
+	/* Chain of all blocks associated with base. */
+	base_block_t	*blocks;
+
+	/* Heap of extents that track unused trailing space within blocks. */
+	extent_heap_t	avail[NSIZES];
+
+	/* Stats, only maintained if config_stats. */
+	size_t		allocated;
+	size_t		resident;
+	size_t		mapped;
+	/* Number of THP regions touched. */
+	size_t		n_thp;
+};
+
+#endif /* JEMALLOC_INTERNAL_BASE_STRUCTS_H */

include/jemalloc/internal/base_types.h

@@ -0,0 +1,33 @@
+#ifndef JEMALLOC_INTERNAL_BASE_TYPES_H
+#define JEMALLOC_INTERNAL_BASE_TYPES_H
+
+typedef struct base_block_s base_block_t;
+typedef struct base_s base_t;
+
+#define METADATA_THP_DEFAULT metadata_thp_disabled
+
+/*
+ * In auto mode, arenas switch to huge pages for the base allocator on the
+ * second base block.  a0 switches to thp on the 5th block (after 20 megabytes
+ * of metadata), since more metadata (e.g. rtree nodes) come from a0's base.
+ */
+
+#define BASE_AUTO_THP_THRESHOLD    2
+#define BASE_AUTO_THP_THRESHOLD_A0 5
+
+typedef enum {
+	metadata_thp_disabled   = 0,
+	/*
+	 * Lazily enable hugepage for metadata. To avoid high RSS caused by THP
+	 * + low usage arena (i.e. THP becomes a significant percentage), the
+	 * "auto" option only starts using THP after a base allocator used up
+	 * the first THP region.  Starting from the second hugepage (in a single
+	 * arena), "auto" behaves the same as "always", i.e. madvise hugepage
+	 * right away.
+	 */
+	metadata_thp_auto       = 1,
+	metadata_thp_always     = 2,
+	metadata_thp_mode_limit = 3
+} metadata_thp_mode_t;
+
+#endif /* JEMALLOC_INTERNAL_BASE_TYPES_H */

include/jemalloc/internal/bin.h

@@ -0,0 +1,106 @@
+#ifndef JEMALLOC_INTERNAL_BIN_H
+#define JEMALLOC_INTERNAL_BIN_H
+
+#include "jemalloc/internal/extent_types.h"
+#include "jemalloc/internal/extent_structs.h"
+#include "jemalloc/internal/mutex.h"
+#include "jemalloc/internal/bin_stats.h"
+
+/*
+ * A bin contains a set of extents that are currently being used for slab
+ * allocations.
+ */
+
+/*
+ * 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 slab has the following layout:
+ *
+ *   /--------------------\
+ *   | region 0           |
+ *   |--------------------|
+ *   | region 1           |
+ *   |--------------------|
+ *   | ...                |
+ *   | ...                |
+ *   | ...                |
+ *   |--------------------|
+ *   | region nregs-1     |
+ *   \--------------------/
+ */
+typedef struct bin_info_s bin_info_t;
+struct bin_info_s {
+	/* Size of regions in a slab for this bin's size class. */
+	size_t			reg_size;
+
+	/* Total size of a slab for this bin's size class. */
+	size_t			slab_size;
+
+	/* Total number of regions in a slab for this bin's size class. */
+	uint32_t		nregs;
+
+	/*
+	 * Metadata used to manipulate bitmaps for slabs associated with this
+	 * bin.
+	 */
+	bitmap_info_t		bitmap_info;
+};
+
+extern const bin_info_t bin_infos[NBINS];
+
+
+typedef struct bin_s bin_t;
+struct bin_s {
+	/* All operations on bin_t fields require lock ownership. */
+	malloc_mutex_t		lock;
+
+	/*
+	 * Current slab being used to service allocations of this bin's size
+	 * class.  slabcur is independent of slabs_{nonfull,full}; whenever
+	 * slabcur is reassigned, the previous slab must be deallocated or
+	 * inserted into slabs_{nonfull,full}.
+	 */
+	extent_t		*slabcur;
+
+	/*
+	 * Heap of non-full slabs.  This heap is used to assure that new
+	 * allocations come from the non-full slab that is oldest/lowest in
+	 * memory.
+	 */
+	extent_heap_t		slabs_nonfull;
+
+	/* List used to track full slabs. */
+	extent_list_t		slabs_full;
+
+	/* Bin statistics. */
+	bin_stats_t	stats;
+};
+
+/* Initializes a bin to empty.  Returns true on error. */
+bool bin_init(bin_t *bin);
+
+/* Forking. */
+void bin_prefork(tsdn_t *tsdn, bin_t *bin);
+void bin_postfork_parent(tsdn_t *tsdn, bin_t *bin);
+void bin_postfork_child(tsdn_t *tsdn, bin_t *bin);
+
+/* Stats. */
+static inline void
+bin_stats_merge(tsdn_t *tsdn, bin_stats_t *dst_bin_stats, bin_t *bin) {
+	malloc_mutex_lock(tsdn, &bin->lock);
+	malloc_mutex_prof_read(tsdn, &dst_bin_stats->mutex_data, &bin->lock);
+	dst_bin_stats->nmalloc += bin->stats.nmalloc;
+	dst_bin_stats->ndalloc += bin->stats.ndalloc;
+	dst_bin_stats->nrequests += bin->stats.nrequests;
+	dst_bin_stats->curregs += bin->stats.curregs;
+	dst_bin_stats->nfills += bin->stats.nfills;
+	dst_bin_stats->nflushes += bin->stats.nflushes;
+	dst_bin_stats->nslabs += bin->stats.nslabs;
+	dst_bin_stats->reslabs += bin->stats.reslabs;
+	dst_bin_stats->curslabs += bin->stats.curslabs;
+	malloc_mutex_unlock(tsdn, &bin->lock);
+}
+
+#endif /* JEMALLOC_INTERNAL_BIN_H */

include/jemalloc/internal/bin_stats.h

@@ -0,0 +1,51 @@
+#ifndef JEMALLOC_INTERNAL_BIN_STATS_H
+#define JEMALLOC_INTERNAL_BIN_STATS_H
+
+#include "jemalloc/internal/mutex_prof.h"
+
+typedef struct bin_stats_s bin_stats_t;
+struct bin_stats_s {
+	/*
+	 * 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;
+
+	/*
+	 * Current number of regions of this size class, including regions
+	 * currently cached by tcache.
+	 */
+	size_t		curregs;
+
+	/* Number of tcache fills from this bin. */
+	uint64_t	nfills;
+
+	/* Number of tcache flushes to this bin. */
+	uint64_t	nflushes;
+
+	/* Total number of slabs created for this bin's size class. */
+	uint64_t	nslabs;
+
+	/*
+	 * Total number of slabs reused by extracting them from the slabs heap
+	 * for this bin's size class.
+	 */
+	uint64_t	reslabs;
+
+	/* Current number of slabs in this bin. */
+	size_t		curslabs;
+
+	mutex_prof_data_t mutex_data;
+};
+
+#endif /* JEMALLOC_INTERNAL_BIN_STATS_H */

include/jemalloc/internal/bit_util.h

@@ -0,0 +1,165 @@
+#ifndef JEMALLOC_INTERNAL_BIT_UTIL_H
+#define JEMALLOC_INTERNAL_BIT_UTIL_H
+
+#include "jemalloc/internal/assert.h"
+
+#define BIT_UTIL_INLINE static inline
+
+/* Sanity check. */
+#if !defined(JEMALLOC_INTERNAL_FFSLL) || !defined(JEMALLOC_INTERNAL_FFSL) \
+    || !defined(JEMALLOC_INTERNAL_FFS)
+#  error JEMALLOC_INTERNAL_FFS{,L,LL} should have been defined by configure
+#endif
+
+
+BIT_UTIL_INLINE unsigned
+ffs_llu(unsigned long long bitmap) {
+	return JEMALLOC_INTERNAL_FFSLL(bitmap);
+}
+
+BIT_UTIL_INLINE unsigned
+ffs_lu(unsigned long bitmap) {
+	return JEMALLOC_INTERNAL_FFSL(bitmap);
+}
+
+BIT_UTIL_INLINE unsigned
+ffs_u(unsigned bitmap) {
+	return JEMALLOC_INTERNAL_FFS(bitmap);
+}
+
+BIT_UTIL_INLINE unsigned
+ffs_zu(size_t bitmap) {
+#if LG_SIZEOF_PTR == LG_SIZEOF_INT
+	return ffs_u(bitmap);
+#elif LG_SIZEOF_PTR == LG_SIZEOF_LONG
+	return ffs_lu(bitmap);
+#elif LG_SIZEOF_PTR == LG_SIZEOF_LONG_LONG
+	return ffs_llu(bitmap);
+#else
+#error No implementation for size_t ffs()
+#endif
+}
+
+BIT_UTIL_INLINE unsigned
+ffs_u64(uint64_t bitmap) {
+#if LG_SIZEOF_LONG == 3
+	return ffs_lu(bitmap);
+#elif LG_SIZEOF_LONG_LONG == 3
+	return ffs_llu(bitmap);
+#else
+#error No implementation for 64-bit ffs()
+#endif
+}
+
+BIT_UTIL_INLINE unsigned
+ffs_u32(uint32_t bitmap) {
+#if LG_SIZEOF_INT == 2
+	return ffs_u(bitmap);
+#else
+#error No implementation for 32-bit ffs()
+#endif
+	return ffs_u(bitmap);
+}
+
+BIT_UTIL_INLINE uint64_t
+pow2_ceil_u64(uint64_t x) {
+	x--;
+	x |= x >> 1;
+	x |= x >> 2;
+	x |= x >> 4;
+	x |= x >> 8;
+	x |= x >> 16;
+	x |= x >> 32;
+	x++;
+	return x;
+}
+
+BIT_UTIL_INLINE uint32_t
+pow2_ceil_u32(uint32_t x) {
+	x--;
+	x |= x >> 1;
+	x |= x >> 2;
+	x |= x >> 4;
+	x |= x >> 8;
+	x |= x >> 16;
+	x++;
+	return x;
+}
+
+/* Compute the smallest power of 2 that is >= x. */
+BIT_UTIL_INLINE size_t
+pow2_ceil_zu(size_t x) {
+#if (LG_SIZEOF_PTR == 3)
+	return pow2_ceil_u64(x);
+#else
+	return pow2_ceil_u32(x);
+#endif
+}
+
+#if (defined(__i386__) || defined(__amd64__) || defined(__x86_64__))
+BIT_UTIL_INLINE unsigned
+lg_floor(size_t x) {
+	size_t ret;
+	assert(x != 0);
+
+	asm ("bsr %1, %0"
+	    : "=r"(ret) // Outputs.
+	    : "r"(x)    // Inputs.
+	    );
+	assert(ret < UINT_MAX);
+	return (unsigned)ret;
+}
+#elif (defined(_MSC_VER))
+BIT_UTIL_INLINE unsigned
+lg_floor(size_t x) {
+	unsigned long ret;
+
+	assert(x != 0);
+
+#if (LG_SIZEOF_PTR == 3)
+	_BitScanReverse64(&ret, x);
+#elif (LG_SIZEOF_PTR == 2)
+	_BitScanReverse(&ret, x);
+#else
+#  error "Unsupported type size for lg_floor()"
+#endif
+	assert(ret < UINT_MAX);
+	return (unsigned)ret;
+}
+#elif (defined(JEMALLOC_HAVE_BUILTIN_CLZ))
+BIT_UTIL_INLINE unsigned
+lg_floor(size_t x) {
+	assert(x != 0);
+
+#if (LG_SIZEOF_PTR == LG_SIZEOF_INT)
+	return ((8 << LG_SIZEOF_PTR) - 1) - __builtin_clz(x);
+#elif (LG_SIZEOF_PTR == LG_SIZEOF_LONG)
+	return ((8 << LG_SIZEOF_PTR) - 1) - __builtin_clzl(x);
+#else
+#  error "Unsupported type size for lg_floor()"
+#endif
+}
+#else
+BIT_UTIL_INLINE unsigned
+lg_floor(size_t x) {
+	assert(x != 0);
+
+	x |= (x >> 1);
+	x |= (x >> 2);
+	x |= (x >> 4);
+	x |= (x >> 8);
+	x |= (x >> 16);
+#if (LG_SIZEOF_PTR == 3)
+	x |= (x >> 32);
+#endif
+	if (x == SIZE_T_MAX) {
+		return (8 << LG_SIZEOF_PTR) - 1;
+	}
+	x++;
+	return ffs_zu(x) - 2;
+}
+#endif
+
+#undef BIT_UTIL_INLINE
+
+#endif /* JEMALLOC_INTERNAL_BIT_UTIL_H */

include/jemalloc/internal/bitmap.h

@@ -1,37 +1,159 @@
-/******************************************************************************/
-#ifdef JEMALLOC_H_TYPES
+#ifndef JEMALLOC_INTERNAL_BITMAP_H
+#define JEMALLOC_INTERNAL_BITMAP_H
+
+#include "jemalloc/internal/arena_types.h"
+#include "jemalloc/internal/bit_util.h"
+#include "jemalloc/internal/size_classes.h"
+
+typedef unsigned long bitmap_t;
+#define LG_SIZEOF_BITMAP	LG_SIZEOF_LONG
 
 /* 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
+#if LG_SLAB_MAXREGS > LG_CEIL_NSIZES
+/* Maximum bitmap bit count is determined by maximum regions per slab. */
+#  define LG_BITMAP_MAXBITS	LG_SLAB_MAXREGS
+#else
+/* Maximum bitmap bit count is determined by number of extent size classes. */
+#  define LG_BITMAP_MAXBITS	LG_CEIL_NSIZES
+#endif
+#define BITMAP_MAXBITS		(ZU(1) << LG_BITMAP_MAXBITS)
 
 /* 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)
+#define LG_BITMAP_GROUP_NBITS		(LG_SIZEOF_BITMAP + 3)
+#define BITMAP_GROUP_NBITS		(1U << 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)
+/*
+ * Do some analysis on how big the bitmap is before we use a tree.  For a brute
+ * force linear search, if we would have to call ffs_lu() more than 2^3 times,
+ * use a tree instead.
+ */
+#if LG_BITMAP_MAXBITS - LG_BITMAP_GROUP_NBITS > 3
+#  define BITMAP_USE_TREE
+#endif
 
-#endif /* JEMALLOC_H_TYPES */
-/******************************************************************************/
-#ifdef JEMALLOC_H_STRUCTS
+/* Number of groups required to store a given number of bits. */
+#define BITMAP_BITS2GROUPS(nbits)					\
+    (((nbits) + BITMAP_GROUP_NBITS_MASK) >> LG_BITMAP_GROUP_NBITS)
 
-struct bitmap_level_s {
+/*
+ * Number of groups required at a particular level for a given number of bits.
+ */
+#define BITMAP_GROUPS_L0(nbits)						\
+    BITMAP_BITS2GROUPS(nbits)
+#define BITMAP_GROUPS_L1(nbits)						\
+    BITMAP_BITS2GROUPS(BITMAP_BITS2GROUPS(nbits))
+#define BITMAP_GROUPS_L2(nbits)						\
+    BITMAP_BITS2GROUPS(BITMAP_BITS2GROUPS(BITMAP_BITS2GROUPS((nbits))))
+#define BITMAP_GROUPS_L3(nbits)						\
+    BITMAP_BITS2GROUPS(BITMAP_BITS2GROUPS(BITMAP_BITS2GROUPS(		\
+	BITMAP_BITS2GROUPS((nbits)))))
+#define BITMAP_GROUPS_L4(nbits)						\
+    BITMAP_BITS2GROUPS(BITMAP_BITS2GROUPS(BITMAP_BITS2GROUPS(		\
+	BITMAP_BITS2GROUPS(BITMAP_BITS2GROUPS((nbits))))))
+
+/*
+ * Assuming the number of levels, number of groups required for a given number
+ * of bits.
+ */
+#define BITMAP_GROUPS_1_LEVEL(nbits)					\
+    BITMAP_GROUPS_L0(nbits)
+#define BITMAP_GROUPS_2_LEVEL(nbits)					\
+    (BITMAP_GROUPS_1_LEVEL(nbits) + BITMAP_GROUPS_L1(nbits))
+#define BITMAP_GROUPS_3_LEVEL(nbits)					\
+    (BITMAP_GROUPS_2_LEVEL(nbits) + BITMAP_GROUPS_L2(nbits))
+#define BITMAP_GROUPS_4_LEVEL(nbits)					\
+    (BITMAP_GROUPS_3_LEVEL(nbits) + BITMAP_GROUPS_L3(nbits))
+#define BITMAP_GROUPS_5_LEVEL(nbits)					\
+    (BITMAP_GROUPS_4_LEVEL(nbits) + BITMAP_GROUPS_L4(nbits))
+
+/*
+ * Maximum number of groups required to support LG_BITMAP_MAXBITS.
+ */
+#ifdef BITMAP_USE_TREE
+
+#if LG_BITMAP_MAXBITS <= LG_BITMAP_GROUP_NBITS
+#  define BITMAP_GROUPS(nbits)	BITMAP_GROUPS_1_LEVEL(nbits)
+#  define BITMAP_GROUPS_MAX	BITMAP_GROUPS_1_LEVEL(BITMAP_MAXBITS)
+#elif LG_BITMAP_MAXBITS <= LG_BITMAP_GROUP_NBITS * 2
+#  define BITMAP_GROUPS(nbits)	BITMAP_GROUPS_2_LEVEL(nbits)
+#  define BITMAP_GROUPS_MAX	BITMAP_GROUPS_2_LEVEL(BITMAP_MAXBITS)
+#elif LG_BITMAP_MAXBITS <= LG_BITMAP_GROUP_NBITS * 3
+#  define BITMAP_GROUPS(nbits)	BITMAP_GROUPS_3_LEVEL(nbits)
+#  define BITMAP_GROUPS_MAX	BITMAP_GROUPS_3_LEVEL(BITMAP_MAXBITS)
+#elif LG_BITMAP_MAXBITS <= LG_BITMAP_GROUP_NBITS * 4
+#  define BITMAP_GROUPS(nbits)	BITMAP_GROUPS_4_LEVEL(nbits)
+#  define BITMAP_GROUPS_MAX	BITMAP_GROUPS_4_LEVEL(BITMAP_MAXBITS)
+#elif LG_BITMAP_MAXBITS <= LG_BITMAP_GROUP_NBITS * 5
+#  define BITMAP_GROUPS(nbits)	BITMAP_GROUPS_5_LEVEL(nbits)
+#  define BITMAP_GROUPS_MAX	BITMAP_GROUPS_5_LEVEL(BITMAP_MAXBITS)
+#else
+#  error "Unsupported bitmap size"
+#endif
+
+/*
+ * Maximum number of levels possible.  This could be statically computed based
+ * on LG_BITMAP_MAXBITS:
+ *
+ * #define BITMAP_MAX_LEVELS \
+ *     (LG_BITMAP_MAXBITS / LG_SIZEOF_BITMAP) \
+ *     + !!(LG_BITMAP_MAXBITS % LG_SIZEOF_BITMAP)
+ *
+ * However, that would not allow the generic BITMAP_INFO_INITIALIZER() macro, so
+ * instead hardcode BITMAP_MAX_LEVELS to the largest number supported by the
+ * various cascading macros.  The only additional cost this incurs is some
+ * unused trailing entries in bitmap_info_t structures; the bitmaps themselves
+ * are not impacted.
+ */
+#define BITMAP_MAX_LEVELS	5
+
+#define BITMAP_INFO_INITIALIZER(nbits) {				\
+	/* nbits. */							\
+	nbits,								\
+	/* nlevels. */							\
+	(BITMAP_GROUPS_L0(nbits) > BITMAP_GROUPS_L1(nbits)) +		\
+	    (BITMAP_GROUPS_L1(nbits) > BITMAP_GROUPS_L2(nbits)) +	\
+	    (BITMAP_GROUPS_L2(nbits) > BITMAP_GROUPS_L3(nbits)) +	\
+	    (BITMAP_GROUPS_L3(nbits) > BITMAP_GROUPS_L4(nbits)) + 1,	\
+	/* levels. */							\
+	{								\
+		{0},							\
+		{BITMAP_GROUPS_L0(nbits)},				\
+		{BITMAP_GROUPS_L1(nbits) + BITMAP_GROUPS_L0(nbits)},	\
+		{BITMAP_GROUPS_L2(nbits) + BITMAP_GROUPS_L1(nbits) +	\
+		    BITMAP_GROUPS_L0(nbits)},				\
+		{BITMAP_GROUPS_L3(nbits) + BITMAP_GROUPS_L2(nbits) +	\
+		    BITMAP_GROUPS_L1(nbits) + BITMAP_GROUPS_L0(nbits)},	\
+		{BITMAP_GROUPS_L4(nbits) + BITMAP_GROUPS_L3(nbits) +	\
+		     BITMAP_GROUPS_L2(nbits) + BITMAP_GROUPS_L1(nbits)	\
+		     + BITMAP_GROUPS_L0(nbits)}				\
+	}								\
+}
+
+#else /* BITMAP_USE_TREE */
+
+#define BITMAP_GROUPS(nbits)	BITMAP_BITS2GROUPS(nbits)
+#define BITMAP_GROUPS_MAX	BITMAP_BITS2GROUPS(BITMAP_MAXBITS)
+
+#define BITMAP_INFO_INITIALIZER(nbits) {				\
+	/* nbits. */							\
+	nbits,								\
+	/* ngroups. */							\
+	BITMAP_BITS2GROUPS(nbits)					\
+}
+
+#endif /* BITMAP_USE_TREE */
+
+typedef struct bitmap_level_s {
 	/* Offset of this level's groups within the array of groups. */
 	size_t group_offset;
-};
+} bitmap_level_t;
 
-struct bitmap_info_s {
+typedef struct bitmap_info_s {
 	/* Logical number of bits in bitmap (stored at bottom level). */
 	size_t nbits;
 
+#ifdef BITMAP_USE_TREE
 	/* Number of levels necessary for nbits. */
 	unsigned nlevels;
 
@@ -40,67 +162,62 @@ struct bitmap_info_s {
 	 * bottom to top (e.g. the bottom level is stored in levels[0]).
 	 */
 	bitmap_level_t levels[BITMAP_MAX_LEVELS+1];
-};
+#else /* BITMAP_USE_TREE */
+	/* Number of groups necessary for nbits. */
+	size_t ngroups;
+#endif /* BITMAP_USE_TREE */
+} bitmap_info_t;
 
-#endif /* JEMALLOC_H_STRUCTS */
-/******************************************************************************/
-#ifdef JEMALLOC_H_EXTERNS
+void bitmap_info_init(bitmap_info_t *binfo, size_t nbits);
+void bitmap_init(bitmap_t *bitmap, const bitmap_info_t *binfo, bool fill);
+size_t bitmap_size(const bitmap_info_t *binfo);
 
-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;
+static inline bool
+bitmap_full(bitmap_t *bitmap, const bitmap_info_t *binfo) {
+#ifdef BITMAP_USE_TREE
+	size_t 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);
+#else
+	size_t i;
+
+	for (i = 0; i < binfo->ngroups; i++) {
+		if (bitmap[i] != 0) {
+			return false;
+		}
+	}
+	return true;
+#endif
 }
 
-JEMALLOC_INLINE bool
-bitmap_get(bitmap_t *bitmap, const bitmap_info_t *binfo, size_t bit)
-{
+static 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))));
+	return !(g & (ZU(1) << (bit & BITMAP_GROUP_NBITS_MASK)));
 }
 
-JEMALLOC_INLINE void
-bitmap_set(bitmap_t *bitmap, const bitmap_info_t *binfo, size_t bit)
-{
+static 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);
+	assert(!bitmap_get(bitmap, binfo, bit));
 	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);
+	assert(g & (ZU(1) << (bit & BITMAP_GROUP_NBITS_MASK)));
+	g ^= ZU(1) << (bit & BITMAP_GROUP_NBITS_MASK);
 	*gp = g;
 	assert(bitmap_get(bitmap, binfo, bit));
+#ifdef BITMAP_USE_TREE
 	/* Propagate group state transitions up the tree. */
 	if (g == 0) {
 		unsigned i;
@@ -109,45 +226,113 @@ bitmap_set(bitmap_t *bitmap, const bitmap_info_t *binfo, size_t bit)
 			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);
+			assert(g & (ZU(1) << (bit & BITMAP_GROUP_NBITS_MASK)));
+			g ^= ZU(1) << (bit & BITMAP_GROUP_NBITS_MASK);
 			*gp = g;
-			if (g != 0)
+			if (g != 0) {
 				break;
+			}
 		}
 	}
+#endif
+}
+
+/* ffu: find first unset >= bit. */
+static inline size_t
+bitmap_ffu(const bitmap_t *bitmap, const bitmap_info_t *binfo, size_t min_bit) {
+	assert(min_bit < binfo->nbits);
+
+#ifdef BITMAP_USE_TREE
+	size_t bit = 0;
+	for (unsigned level = binfo->nlevels; level--;) {
+		size_t lg_bits_per_group = (LG_BITMAP_GROUP_NBITS * (level +
+		    1));
+		bitmap_t group = bitmap[binfo->levels[level].group_offset + (bit
+		    >> lg_bits_per_group)];
+		unsigned group_nmask = (unsigned)(((min_bit > bit) ? (min_bit -
+		    bit) : 0) >> (lg_bits_per_group - LG_BITMAP_GROUP_NBITS));
+		assert(group_nmask <= BITMAP_GROUP_NBITS);
+		bitmap_t group_mask = ~((1LU << group_nmask) - 1);
+		bitmap_t group_masked = group & group_mask;
+		if (group_masked == 0LU) {
+			if (group == 0LU) {
+				return binfo->nbits;
+			}
+			/*
+			 * min_bit was preceded by one or more unset bits in
+			 * this group, but there are no other unset bits in this
+			 * group.  Try again starting at the first bit of the
+			 * next sibling.  This will recurse at most once per
+			 * non-root level.
+			 */
+			size_t sib_base = bit + (ZU(1) << lg_bits_per_group);
+			assert(sib_base > min_bit);
+			assert(sib_base > bit);
+			if (sib_base >= binfo->nbits) {
+				return binfo->nbits;
+			}
+			return bitmap_ffu(bitmap, binfo, sib_base);
+		}
+		bit += ((size_t)(ffs_lu(group_masked) - 1)) <<
+		    (lg_bits_per_group - LG_BITMAP_GROUP_NBITS);
+	}
+	assert(bit >= min_bit);
+	assert(bit < binfo->nbits);
+	return bit;
+#else
+	size_t i = min_bit >> LG_BITMAP_GROUP_NBITS;
+	bitmap_t g = bitmap[i] & ~((1LU << (min_bit & BITMAP_GROUP_NBITS_MASK))
+	    - 1);
+	size_t bit;
+	do {
+		bit = ffs_lu(g);
+		if (bit != 0) {
+			return (i << LG_BITMAP_GROUP_NBITS) + (bit - 1);
+		}
+		i++;
+		g = bitmap[i];
+	} while (i < binfo->ngroups);
+	return binfo->nbits;
+#endif
 }
 
 /* sfu: set first unset. */
-JEMALLOC_INLINE size_t
-bitmap_sfu(bitmap_t *bitmap, const bitmap_info_t *binfo)
-{
+static 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);
+	assert(!bitmap_full(bitmap, binfo));
 
+#ifdef BITMAP_USE_TREE
 	i = binfo->nlevels - 1;
 	g = bitmap[binfo->levels[i].group_offset];
-	bit = ffsl(g) - 1;
+	bit = ffs_lu(g) - 1;
 	while (i > 0) {
 		i--;
 		g = bitmap[binfo->levels[i].group_offset + bit];
-		bit = (bit << LG_BITMAP_GROUP_NBITS) + (ffsl(g) - 1);
+		bit = (bit << LG_BITMAP_GROUP_NBITS) + (ffs_lu(g) - 1);
 	}
-
+#else
+	i = 0;
+	g = bitmap[0];
+	while ((bit = ffs_lu(g)) == 0) {
+		i++;
+		g = bitmap[i];
+	}
+	bit = (i << LG_BITMAP_GROUP_NBITS) + (bit - 1);
+#endif
 	bitmap_set(bitmap, binfo, bit);
-	return (bit);
+	return bit;
 }
 
-JEMALLOC_INLINE void
-bitmap_unset(bitmap_t *bitmap, const bitmap_info_t *binfo, size_t bit)
-{
+static 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;
+	UNUSED bool propagate;
 
 	assert(bit < binfo->nbits);
 	assert(bitmap_get(bitmap, binfo, bit));
@@ -155,10 +340,11 @@ bitmap_unset(bitmap_t *bitmap, const bitmap_info_t *binfo, size_t bit)
 	gp = &bitmap[goff];
 	g = *gp;
 	propagate = (g == 0);
-	assert((g & (1LU << (bit & BITMAP_GROUP_NBITS_MASK))) == 0);
-	g ^= 1LU << (bit & BITMAP_GROUP_NBITS_MASK);
+	assert((g & (ZU(1) << (bit & BITMAP_GROUP_NBITS_MASK))) == 0);
+	g ^= ZU(1) << (bit & BITMAP_GROUP_NBITS_MASK);
 	*gp = g;
-	assert(bitmap_get(bitmap, binfo, bit) == false);
+	assert(!bitmap_get(bitmap, binfo, bit));
+#ifdef BITMAP_USE_TREE
 	/* Propagate group state transitions up the tree. */
 	if (propagate) {
 		unsigned i;
@@ -168,17 +354,16 @@ bitmap_unset(bitmap_t *bitmap, const bitmap_info_t *binfo, size_t bit)
 			gp = &bitmap[binfo->levels[i].group_offset + goff];
 			g = *gp;
 			propagate = (g == 0);
-			assert((g & (1LU << (bit & BITMAP_GROUP_NBITS_MASK)))
+			assert((g & (ZU(1) << (bit & BITMAP_GROUP_NBITS_MASK)))
 			    == 0);
-			g ^= 1LU << (bit & BITMAP_GROUP_NBITS_MASK);
+			g ^= ZU(1) << (bit & BITMAP_GROUP_NBITS_MASK);
 			*gp = g;
-			if (propagate == false)
+			if (!propagate) {
 				break;
+			}
 		}
 	}
+#endif /* BITMAP_USE_TREE */
 }
 
-#endif
-
-#endif /* JEMALLOC_H_INLINES */
-/******************************************************************************/
+#endif /* JEMALLOC_INTERNAL_BITMAP_H */

include/jemalloc/internal/cache_bin.h

@@ -0,0 +1,114 @@
+#ifndef JEMALLOC_INTERNAL_CACHE_BIN_H
+#define JEMALLOC_INTERNAL_CACHE_BIN_H
+
+#include "jemalloc/internal/ql.h"
+
+/*
+ * The cache_bins are the mechanism that the tcache and the arena use to
+ * communicate.  The tcache fills from and flushes to the arena by passing a
+ * cache_bin_t to fill/flush.  When the arena needs to pull stats from the
+ * tcaches associated with it, it does so by iterating over its
+ * cache_bin_array_descriptor_t objects and reading out per-bin stats it
+ * contains.  This makes it so that the arena need not know about the existence
+ * of the tcache at all.
+ */
+
+
+/*
+ * The count of the number of cached allocations in a bin.  We make this signed
+ * so that negative numbers can encode "invalid" states (e.g. a low water mark
+ * of -1 for a cache that has been depleted).
+ */
+typedef int32_t cache_bin_sz_t;
+
+typedef struct cache_bin_stats_s cache_bin_stats_t;
+struct cache_bin_stats_s {
+	/*
+	 * Number of allocation requests that corresponded to the size of this
+	 * bin.
+	 */
+	uint64_t nrequests;
+};
+
+/*
+ * Read-only information associated with each element of tcache_t's tbins array
+ * is stored separately, mainly to reduce memory usage.
+ */
+typedef struct cache_bin_info_s cache_bin_info_t;
+struct cache_bin_info_s {
+	/* Upper limit on ncached. */
+	cache_bin_sz_t ncached_max;
+};
+
+typedef struct cache_bin_s cache_bin_t;
+struct cache_bin_s {
+	/* Min # cached since last GC. */
+	cache_bin_sz_t low_water;
+	/* # of cached objects. */
+	cache_bin_sz_t ncached;
+	/*
+	 * ncached and stats are both modified frequently.  Let's keep them
+	 * close so that they have a higher chance of being on the same
+	 * cacheline, thus less write-backs.
+	 */
+	cache_bin_stats_t tstats;
+	/*
+	 * Stack of available objects.
+	 *
+	 * To make use of adjacent cacheline prefetch, the items in the avail
+	 * stack goes to higher address for newer allocations.  avail points
+	 * just above the available space, which means that
+	 * avail[-ncached, ... -1] are available items and the lowest item will
+	 * be allocated first.
+	 */
+	void **avail;
+};
+
+typedef struct cache_bin_array_descriptor_s cache_bin_array_descriptor_t;
+struct cache_bin_array_descriptor_s {
+	/*
+	 * The arena keeps a list of the cache bins associated with it, for
+	 * stats collection.
+	 */
+	ql_elm(cache_bin_array_descriptor_t) link;
+	/* Pointers to the tcache bins. */
+	cache_bin_t *bins_small;
+	cache_bin_t *bins_large;
+};
+
+static inline void
+cache_bin_array_descriptor_init(cache_bin_array_descriptor_t *descriptor,
+    cache_bin_t *bins_small, cache_bin_t *bins_large) {
+	ql_elm_new(descriptor, link);
+	descriptor->bins_small = bins_small;
+	descriptor->bins_large = bins_large;
+}
+
+JEMALLOC_ALWAYS_INLINE void *
+cache_bin_alloc_easy(cache_bin_t *bin, bool *success) {
+	void *ret;
+
+	if (unlikely(bin->ncached == 0)) {
+		bin->low_water = -1;
+		*success = false;
+		return NULL;
+	}
+	/*
+	 * success (instead of ret) should be checked upon the return of this
+	 * function.  We avoid checking (ret == NULL) because there is never a
+	 * null stored on the avail stack (which is unknown to the compiler),
+	 * and eagerly checking ret would cause pipeline stall (waiting for the
+	 * cacheline).
+	 */
+	*success = true;
+	ret = *(bin->avail - bin->ncached);
+	bin->ncached--;
+
+	if (unlikely(bin->ncached < bin->low_water)) {
+		bin->low_water = bin->ncached;
+	}
+
+	return ret;
+}
+
+#endif /* JEMALLOC_INTERNAL_CACHE_BIN_H */

include/jemalloc/internal/chunk.h

@@ -1,63 +0,0 @@
-/******************************************************************************/
-#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;
-extern const char	*opt_dss;
-
-/* 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,
-    dss_prec_t dss_prec);
-void	chunk_unmap(void *chunk, size_t size);
-void	chunk_dealloc(void *chunk, size_t size, bool unmap);
-bool	chunk_boot(void);
-void	chunk_prefork(void);
-void	chunk_postfork_parent(void);
-void	chunk_postfork_child(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"

include/jemalloc/internal/chunk_dss.h

@@ -1,38 +0,0 @@
-/******************************************************************************/
-#ifdef JEMALLOC_H_TYPES
-
-typedef enum {
-	dss_prec_disabled  = 0,
-	dss_prec_primary   = 1,
-	dss_prec_secondary = 2,
-
-	dss_prec_limit     = 3
-} dss_prec_t;
-#define	DSS_PREC_DEFAULT	dss_prec_secondary
-#define	DSS_DEFAULT		"secondary"
-
-#endif /* JEMALLOC_H_TYPES */
-/******************************************************************************/
-#ifdef JEMALLOC_H_STRUCTS
-
-extern const char *dss_prec_names[];
-
-#endif /* JEMALLOC_H_STRUCTS */
-/******************************************************************************/
-#ifdef JEMALLOC_H_EXTERNS
-
-dss_prec_t	chunk_dss_prec_get(void);
-bool	chunk_dss_prec_set(dss_prec_t dss_prec);
-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 */
-/******************************************************************************/

include/jemalloc/internal/chunk_mmap.h

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

include/jemalloc/internal/ckh.h

@@ -1,88 +1,101 @@
+#ifndef JEMALLOC_INTERNAL_CKH_H
+#define JEMALLOC_INTERNAL_CKH_H
+
+#include "jemalloc/internal/tsd.h"
+
+/* Cuckoo hashing implementation.  Skip to the end for the interface. */
+
+/******************************************************************************/
+/* INTERNAL DEFINITIONS -- IGNORE */
 /******************************************************************************/
-#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 *, size_t[2]);
-typedef bool ckh_keycomp_t (const void *, const void *);
 
 /* Maintain counters used to get an idea of performance. */
-/* #define	CKH_COUNT */
+/* #define CKH_COUNT */
 /* Print counter values in ckh_delete() (requires CKH_COUNT). */
-/* #define	CKH_VERBOSE */
+/* #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)
+#define LG_CKH_BUCKET_CELLS (LG_CACHELINE - LG_SIZEOF_PTR - 1)
 
-#endif /* JEMALLOC_H_TYPES */
-/******************************************************************************/
-#ifdef JEMALLOC_H_STRUCTS
+/* Typedefs to allow easy function pointer passing. */
+typedef void ckh_hash_t (const void *, size_t[2]);
+typedef bool ckh_keycomp_t (const void *, const void *);
 
 /* Hash table cell. */
-struct ckhc_s {
-	const void	*key;
-	const void	*data;
-};
+typedef struct {
+	const void *key;
+	const void *data;
+} ckhc_t;
 
-struct ckh_s {
+/* The hash table itself. */
+typedef struct {
 #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;
+	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;
+	uint64_t prng_state;
 
 	/* Total number of items. */
-	size_t		count;
+	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;
+	unsigned lg_minbuckets;
+	unsigned lg_curbuckets;
 
 	/* Hash and comparison functions. */
-	ckh_hash_t	*hash;
-	ckh_keycomp_t	*keycomp;
+	ckh_hash_t *hash;
+	ckh_keycomp_t *keycomp;
 
 	/* Hash table with 2^lg_curbuckets buckets. */
-	ckhc_t		*tab;
-};
+	ckhc_t *tab;
+} ckh_t;
 
-#endif /* JEMALLOC_H_STRUCTS */
 /******************************************************************************/
-#ifdef JEMALLOC_H_EXTERNS
+/* BEGIN PUBLIC API */
+/******************************************************************************/
 
-bool	ckh_new(ckh_t *ckh, size_t minitems, ckh_hash_t *hash,
+/* Lifetime management.  Minitems is the initial capacity. */
+bool ckh_new(tsd_t *tsd, 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 ckh_delete(tsd_t *tsd, ckh_t *ckh);
+
+/* Get the number of elements in the set. */
+size_t ckh_count(ckh_t *ckh);
+
+/*
+ * To iterate over the elements in the table, initialize *tabind to 0 and call
+ * this function until it returns true.  Each call that returns false will
+ * update *key and *data to the next element in the table, assuming the pointers
+ * are non-NULL.
+ */
+bool ckh_iter(ckh_t *ckh, size_t *tabind, void **key, void **data);
+
+/*
+ * Basic hash table operations -- insert, removal, lookup.  For ckh_remove and
+ * ckh_search, key or data can be NULL.  The hash-table only stores pointers to
+ * the key and value, and doesn't do any lifetime management.
+ */
+bool ckh_insert(tsd_t *tsd, ckh_t *ckh, const void *key, const void *data);
+bool ckh_remove(tsd_t *tsd, 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, size_t r_hash[2]);
-bool	ckh_string_keycomp(const void *k1, const void *k2);
-void	ckh_pointer_hash(const void *key, size_t r_hash[2]);
-bool	ckh_pointer_keycomp(const void *k1, const void *k2);
+bool ckh_search(ckh_t *ckh, const void *searchkey, void **key, void **data);
 
-#endif /* JEMALLOC_H_EXTERNS */
-/******************************************************************************/
-#ifdef JEMALLOC_H_INLINES
+/* Some useful hash and comparison functions for strings and pointers. */
+void ckh_string_hash(const void *key, size_t r_hash[2]);
+bool ckh_string_keycomp(const void *k1, const void *k2);
+void ckh_pointer_hash(const void *key, size_t r_hash[2]);
+bool ckh_pointer_keycomp(const void *k1, const void *k2);
 
-#endif /* JEMALLOC_H_INLINES */
-/******************************************************************************/
+#endif /* JEMALLOC_INTERNAL_CKH_H */

include/jemalloc/internal/ctl.h

@@ -1,87 +1,107 @@
-/******************************************************************************/
-#ifdef JEMALLOC_H_TYPES
+#ifndef JEMALLOC_INTERNAL_CTL_H
+#define JEMALLOC_INTERNAL_CTL_H
 
-typedef struct ctl_node_s ctl_node_t;
-typedef struct ctl_named_node_s ctl_named_node_t;
-typedef struct ctl_indexed_node_s ctl_indexed_node_t;
-typedef struct ctl_arena_stats_s ctl_arena_stats_t;
-typedef struct ctl_stats_s ctl_stats_t;
+#include "jemalloc/internal/jemalloc_internal_types.h"
+#include "jemalloc/internal/malloc_io.h"
+#include "jemalloc/internal/mutex_prof.h"
+#include "jemalloc/internal/ql.h"
+#include "jemalloc/internal/size_classes.h"
+#include "jemalloc/internal/stats.h"
 
-#endif /* JEMALLOC_H_TYPES */
-/******************************************************************************/
-#ifdef JEMALLOC_H_STRUCTS
+/* Maximum ctl tree depth. */
+#define CTL_MAX_DEPTH	7
 
-struct ctl_node_s {
-	bool			named;
-};
+typedef struct ctl_node_s {
+	bool named;
+} ctl_node_t;
 
-struct ctl_named_node_s {
-	struct ctl_node_s	node;
-	const char		*name;
+typedef struct ctl_named_node_s {
+	ctl_node_t node;
+	const char *name;
 	/* If (nchildren == 0), this is a terminal node. */
-	unsigned		nchildren;
-	const			ctl_node_t *children;
-	int			(*ctl)(const size_t *, size_t, void *, size_t *,
-	    void *, size_t);
-};
+	size_t nchildren;
+	const ctl_node_t *children;
+	int (*ctl)(tsd_t *, const size_t *, size_t, void *, size_t *, void *,
+	    size_t);
+} ctl_named_node_t;
 
-struct ctl_indexed_node_s {
-	struct ctl_node_s	node;
-	const ctl_named_node_t	*(*index)(const size_t *, size_t, size_t);
-};
+typedef struct ctl_indexed_node_s {
+	struct ctl_node_s node;
+	const ctl_named_node_t *(*index)(tsdn_t *, const size_t *, size_t,
+	    size_t);
+} ctl_indexed_node_t;
 
-struct ctl_arena_stats_s {
-	bool			initialized;
-	unsigned		nthreads;
-	const char		*dss;
-	size_t			pactive;
-	size_t			pdirty;
-	arena_stats_t		astats;
+typedef struct ctl_arena_stats_s {
+	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;
+	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. */
+	bin_stats_t bstats[NBINS];
+	arena_stats_large_t lstats[NSIZES - NBINS];
+} ctl_arena_stats_t;
+
+typedef struct ctl_stats_s {
+	size_t allocated;
+	size_t active;
+	size_t metadata;
+	size_t metadata_thp;
+	size_t resident;
+	size_t mapped;
+	size_t retained;
+
+	background_thread_stats_t background_thread;
+	mutex_prof_data_t mutex_prof_data[mutex_prof_num_global_mutexes];
+} ctl_stats_t;
+
+typedef struct ctl_arena_s ctl_arena_t;
+struct ctl_arena_s {
+	unsigned arena_ind;
+	bool initialized;
+	ql_elm(ctl_arena_t) destroyed_link;
+
+	/* Basic stats, supported even if !config_stats. */
+	unsigned nthreads;
+	const char *dss;
+	ssize_t dirty_decay_ms;
+	ssize_t muzzy_decay_ms;
+	size_t pactive;
+	size_t pdirty;
+	size_t pmuzzy;
+
+	/* NULL if !config_stats. */
+	ctl_arena_stats_t *astats;
 };
 
-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;
-	unsigned		narenas;
-	ctl_arena_stats_t	*arenas;	/* (narenas + 1) elements. */
-};
+typedef struct ctl_arenas_s {
+	uint64_t epoch;
+	unsigned narenas;
+	ql_head(ctl_arena_t) destroyed;
 
-#endif /* JEMALLOC_H_STRUCTS */
-/******************************************************************************/
-#ifdef JEMALLOC_H_EXTERNS
+	/*
+	 * Element 0 corresponds to merged stats for extant arenas (accessed via
+	 * MALLCTL_ARENAS_ALL), element 1 corresponds to merged stats for
+	 * destroyed arenas (accessed via MALLCTL_ARENAS_DESTROYED), and the
+	 * remaining MALLOCX_ARENA_LIMIT elements correspond to arenas.
+	 */
+	ctl_arena_t *arenas[2 + MALLOCX_ARENA_LIMIT];
+} ctl_arenas_t;
 
-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,
+int ctl_byname(tsd_t *tsd, const char *name, void *oldp, size_t *oldlenp,
     void *newp, size_t newlen);
-bool	ctl_boot(void);
-void	ctl_prefork(void);
-void	ctl_postfork_parent(void);
-void	ctl_postfork_child(void);
+int ctl_nametomib(tsd_t *tsd, const char *name, size_t *mibp, size_t *miblenp);
 
-#define	xmallctl(name, oldp, oldlenp, newp, newlen) do {		\
+int ctl_bymib(tsd_t *tsd, const size_t *mib, size_t miblen, void *oldp,
+    size_t *oldlenp, void *newp, size_t newlen);
+bool ctl_boot(void);
+void ctl_prefork(tsdn_t *tsdn);
+void ctl_postfork_parent(tsdn_t *tsdn);
+void ctl_postfork_child(tsdn_t *tsdn);
+
+#define xmallctl(name, oldp, oldlenp, newp, newlen) do {		\
 	if (je_mallctl(name, oldp, oldlenp, newp, newlen)		\
 	    != 0) {							\
 		malloc_printf(						\
@@ -91,7 +111,7 @@ void	ctl_postfork_child(void);
 	}								\
 } while (0)
 
-#define	xmallctlnametomib(name, mibp, miblenp) do {			\
+#define xmallctlnametomib(name, mibp, miblenp) do {			\
 	if (je_mallctlnametomib(name, mibp, miblenp) != 0) {		\
 		malloc_printf("<jemalloc>: Failure in "			\
 		    "xmallctlnametomib(\"%s\", ...)\n", name);		\
@@ -99,7 +119,7 @@ void	ctl_postfork_child(void);
 	}								\
 } while (0)
 
-#define	xmallctlbymib(mib, miblen, oldp, oldlenp, newp, newlen) do {	\
+#define xmallctlbymib(mib, miblen, oldp, oldlenp, newp, newlen) do {	\
 	if (je_mallctlbymib(mib, miblen, oldp, oldlenp, newp,		\
 	    newlen) != 0) {						\
 		malloc_write(						\
@@ -108,10 +128,4 @@ void	ctl_postfork_child(void);
 	}								\
 } while (0)
 
-#endif /* JEMALLOC_H_EXTERNS */
-/******************************************************************************/
-#ifdef JEMALLOC_H_INLINES
-
-#endif /* JEMALLOC_H_INLINES */
-/******************************************************************************/
-
+#endif /* JEMALLOC_INTERNAL_CTL_H */

include/jemalloc/internal/div.h

@@ -0,0 +1,41 @@
+#ifndef JEMALLOC_INTERNAL_DIV_H
+#define JEMALLOC_INTERNAL_DIV_H
+
+#include "jemalloc/internal/assert.h"
+
+/*
+ * This module does the division that computes the index of a region in a slab,
+ * given its offset relative to the base.
+ * That is, given a divisor d, an n = i * d (all integers), we'll return i.
+ * We do some pre-computation to do this more quickly than a CPU division
+ * instruction.
+ * We bound n < 2^32, and don't support dividing by one.
+ */
+
+typedef struct div_info_s div_info_t;
+struct div_info_s {
+	uint32_t magic;
+#ifdef JEMALLOC_DEBUG
+	size_t d;
+#endif
+};
+
+void div_init(div_info_t *div_info, size_t divisor);
+
+static inline size_t
+div_compute(div_info_t *div_info, size_t n) {
+	assert(n <= (uint32_t)-1);
+	/*
+	 * This generates, e.g. mov; imul; shr on x86-64. On a 32-bit machine,
+	 * the compilers I tried were all smart enough to turn this into the
+	 * appropriate "get the high 32 bits of the result of a multiply" (e.g.
+	 * mul; mov edx eax; on x86, umull on arm, etc.).
+	 */
+	size_t i = ((uint64_t)n * (uint64_t)div_info->magic) >> 32;
+#ifdef JEMALLOC_DEBUG
+	assert(i * div_info->d == n);
+#endif
+	return i;
+}
+
+#endif /* JEMALLOC_INTERNAL_DIV_H */

include/jemalloc/internal/emitter.h

@@ -0,0 +1,435 @@
+#ifndef JEMALLOC_INTERNAL_EMITTER_H
+#define JEMALLOC_INTERNAL_EMITTER_H
+
+#include "jemalloc/internal/ql.h"
+
+typedef enum emitter_output_e emitter_output_t;
+enum emitter_output_e {
+	emitter_output_json,
+	emitter_output_table
+};
+
+typedef enum emitter_justify_e emitter_justify_t;
+enum emitter_justify_e {
+	emitter_justify_left,
+	emitter_justify_right,
+	/* Not for users; just to pass to internal functions. */
+	emitter_justify_none
+};
+
+typedef enum emitter_type_e emitter_type_t;
+enum emitter_type_e {
+	emitter_type_bool,
+	emitter_type_int,
+	emitter_type_unsigned,
+	emitter_type_uint32,
+	emitter_type_uint64,
+	emitter_type_size,
+	emitter_type_ssize,
+	emitter_type_string,
+	/*
+	 * A title is a column title in a table; it's just a string, but it's
+	 * not quoted.
+	 */
+	emitter_type_title,
+};
+
+typedef struct emitter_col_s emitter_col_t;
+struct emitter_col_s {
+	/* Filled in by the user. */
+	emitter_justify_t justify;
+	int width;
+	emitter_type_t type;
+	union {
+		bool bool_val;
+		int int_val;
+		unsigned unsigned_val;
+		uint32_t uint32_val;
+		uint64_t uint64_val;
+		size_t size_val;
+		ssize_t ssize_val;
+		const char *str_val;
+	};
+
+	/* Filled in by initialization. */
+	ql_elm(emitter_col_t) link;
+};
+
+typedef struct emitter_row_s emitter_row_t;
+struct emitter_row_s {
+	ql_head(emitter_col_t) cols;
+};
+
+static inline void
+emitter_row_init(emitter_row_t *row) {
+	ql_new(&row->cols);
+}
+
+static inline void
+emitter_col_init(emitter_col_t *col, emitter_row_t *row) {
+	ql_elm_new(col, link);
+	ql_tail_insert(&row->cols, col, link);
+}
+
+typedef struct emitter_s emitter_t;
+struct emitter_s {
+	emitter_output_t output;
+	/* The output information. */
+	void (*write_cb)(void *, const char *);
+	void *cbopaque;
+	int nesting_depth;
+	/* True if we've already emitted a value at the given depth. */
+	bool item_at_depth;
+};
+
+static inline void
+emitter_init(emitter_t *emitter, emitter_output_t emitter_output,
+    void (*write_cb)(void *, const char *), void *cbopaque) {
+	emitter->output = emitter_output;
+	emitter->write_cb = write_cb;
+	emitter->cbopaque = cbopaque;
+	emitter->item_at_depth = false;
+	emitter->nesting_depth = 0;
+}
+
+/* Internal convenience function.  Write to the emitter the given string. */
+JEMALLOC_FORMAT_PRINTF(2, 3)
+static inline void
+emitter_printf(emitter_t *emitter, const char *format, ...) {
+	va_list ap;
+
+	va_start(ap, format);
+	malloc_vcprintf(emitter->write_cb, emitter->cbopaque, format, ap);
+	va_end(ap);
+}
+
+/* Write to the emitter the given string, but only in table mode. */
+JEMALLOC_FORMAT_PRINTF(2, 3)
+static inline void
+emitter_table_printf(emitter_t *emitter, const char *format, ...) {
+	if (emitter->output == emitter_output_table) {
+		va_list ap;
+		va_start(ap, format);
+		malloc_vcprintf(emitter->write_cb, emitter->cbopaque, format, ap);
+		va_end(ap);
+	}
+}
+
+static inline void
+emitter_gen_fmt(char *out_fmt, size_t out_size, const char *fmt_specifier,
+    emitter_justify_t justify, int width) {
+	size_t written;
+	if (justify == emitter_justify_none) {
+		written = malloc_snprintf(out_fmt, out_size,
+		    "%%%s", fmt_specifier);
+	} else if (justify == emitter_justify_left) {
+		written = malloc_snprintf(out_fmt, out_size,
+		    "%%-%d%s", width, fmt_specifier);
+	} else {
+		written = malloc_snprintf(out_fmt, out_size,
+		    "%%%d%s", width, fmt_specifier);
+	}
+	/* Only happens in case of bad format string, which *we* choose. */
+	assert(written <  out_size);
+}
+
+/*
+ * Internal.  Emit the given value type in the relevant encoding (so that the
+ * bool true gets mapped to json "true", but the string "true" gets mapped to
+ * json "\"true\"", for instance.
+ *
+ * Width is ignored if justify is emitter_justify_none.
+ */
+static inline void
+emitter_print_value(emitter_t *emitter, emitter_justify_t justify, int width,
+    emitter_type_t value_type, const void *value) {
+	size_t str_written;
+#define BUF_SIZE 256
+#define FMT_SIZE 10
+	/*
+	 * We dynamically generate a format string to emit, to let us use the
+	 * snprintf machinery.  This is kinda hacky, but gets the job done
+	 * quickly without having to think about the various snprintf edge
+	 * cases.
+	 */
+	char fmt[FMT_SIZE];
+	char buf[BUF_SIZE];
+
+#define EMIT_SIMPLE(type, format)					\
+	emitter_gen_fmt(fmt, FMT_SIZE, format, justify, width);		\
+	emitter_printf(emitter, fmt, *(const type *)value);		\
+
+	switch (value_type) {
+	case emitter_type_bool:
+		emitter_gen_fmt(fmt, FMT_SIZE, "s", justify, width);
+		emitter_printf(emitter, fmt, *(const bool *)value ?
+		    "true" : "false");
+		break;
+	case emitter_type_int:
+		EMIT_SIMPLE(int, "d")
+		break;
+	case emitter_type_unsigned:
+		EMIT_SIMPLE(unsigned, "u")
+		break;
+	case emitter_type_ssize:
+		EMIT_SIMPLE(ssize_t, "zd")
+		break;
+	case emitter_type_size:
+		EMIT_SIMPLE(size_t, "zu")
+		break;
+	case emitter_type_string:
+		str_written = malloc_snprintf(buf, BUF_SIZE, "\"%s\"",
+		    *(const char *const *)value);
+		/*
+		 * We control the strings we output; we shouldn't get anything
+		 * anywhere near the fmt size.
+		 */
+		assert(str_written < BUF_SIZE);
+		emitter_gen_fmt(fmt, FMT_SIZE, "s", justify, width);
+		emitter_printf(emitter, fmt, buf);
+		break;
+	case emitter_type_uint32:
+		EMIT_SIMPLE(uint32_t, FMTu32)
+		break;
+	case emitter_type_uint64:
+		EMIT_SIMPLE(uint64_t, FMTu64)
+		break;
+	case emitter_type_title:
+		EMIT_SIMPLE(char *const, "s");
+		break;
+	default:
+		unreachable();
+	}
+#undef BUF_SIZE
+#undef FMT_SIZE
+}
+
+
+/* Internal functions.  In json mode, tracks nesting state. */
+static inline void
+emitter_nest_inc(emitter_t *emitter) {
+	emitter->nesting_depth++;
+	emitter->item_at_depth = false;
+}
+
+static inline void
+emitter_nest_dec(emitter_t *emitter) {
+	emitter->nesting_depth--;
+	emitter->item_at_depth = true;
+}
+
+static inline void
+emitter_indent(emitter_t *emitter) {
+	int amount = emitter->nesting_depth;
+	const char *indent_str;
+	if (emitter->output == emitter_output_json) {
+		indent_str = "\t";
+	} else {
+		amount *= 2;
+		indent_str = " ";
+	}
+	for (int i = 0; i < amount; i++) {
+		emitter_printf(emitter, "%s", indent_str);
+	}
+}
+
+static inline void
+emitter_json_key_prefix(emitter_t *emitter) {
+	emitter_printf(emitter, "%s\n", emitter->item_at_depth ? "," : "");
+	emitter_indent(emitter);
+}
+
+static inline void
+emitter_begin(emitter_t *emitter) {
+	if (emitter->output == emitter_output_json) {
+		assert(emitter->nesting_depth == 0);
+		emitter_printf(emitter, "{");
+		emitter_nest_inc(emitter);
+	} else {
+		// tabular init
+		emitter_printf(emitter, "%s", "");
+	}
+}
+
+static inline void
+emitter_end(emitter_t *emitter) {
+	if (emitter->output == emitter_output_json) {
+		assert(emitter->nesting_depth == 1);
+		emitter_nest_dec(emitter);
+		emitter_printf(emitter, "\n}\n");
+	}
+}
+
+/*
+ * Note emits a different kv pair as well, but only in table mode.  Omits the
+ * note if table_note_key is NULL.
+ */
+static inline void
+emitter_kv_note(emitter_t *emitter, const char *json_key, const char *table_key,
+    emitter_type_t value_type, const void *value,
+    const char *table_note_key, emitter_type_t table_note_value_type,
+    const void *table_note_value) {
+	if (emitter->output == emitter_output_json) {
+		assert(emitter->nesting_depth > 0);
+		emitter_json_key_prefix(emitter);
+		emitter_printf(emitter, "\"%s\": ", json_key);
+		emitter_print_value(emitter, emitter_justify_none, -1,
+		    value_type, value);
+	} else {
+		emitter_indent(emitter);
+		emitter_printf(emitter, "%s: ", table_key);
+		emitter_print_value(emitter, emitter_justify_none, -1,
+		    value_type, value);
+		if (table_note_key != NULL) {
+			emitter_printf(emitter, " (%s: ", table_note_key);
+			emitter_print_value(emitter, emitter_justify_none, -1,
+			    table_note_value_type, table_note_value);
+			emitter_printf(emitter, ")");
+		}
+		emitter_printf(emitter, "\n");
+	}
+	emitter->item_at_depth = true;
+}
+
+static inline void
+emitter_kv(emitter_t *emitter, const char *json_key, const char *table_key,
+    emitter_type_t value_type, const void *value) {
+	emitter_kv_note(emitter, json_key, table_key, value_type, value, NULL,
+	    emitter_type_bool, NULL);
+}
+
+static inline void
+emitter_json_kv(emitter_t *emitter, const char *json_key,
+    emitter_type_t value_type, const void *value) {
+	if (emitter->output == emitter_output_json) {
+		emitter_kv(emitter, json_key, NULL, value_type, value);
+	}
+}
+
+static inline void
+emitter_table_kv(emitter_t *emitter, const char *table_key,
+    emitter_type_t value_type, const void *value) {
+	if (emitter->output == emitter_output_table) {
+		emitter_kv(emitter, NULL, table_key, value_type, value);
+	}
+}
+
+static inline void
+emitter_dict_begin(emitter_t *emitter, const char *json_key,
+    const char *table_header) {
+	if (emitter->output == emitter_output_json) {
+		emitter_json_key_prefix(emitter);
+		emitter_printf(emitter, "\"%s\": {", json_key);
+		emitter_nest_inc(emitter);
+	} else {
+		emitter_indent(emitter);
+		emitter_printf(emitter, "%s\n", table_header);
+		emitter_nest_inc(emitter);
+	}
+}
+
+static inline void
+emitter_dict_end(emitter_t *emitter) {
+	if (emitter->output == emitter_output_json) {
+		assert(emitter->nesting_depth > 0);
+		emitter_nest_dec(emitter);
+		emitter_printf(emitter, "\n");
+		emitter_indent(emitter);
+		emitter_printf(emitter, "}");
+	} else {
+		emitter_nest_dec(emitter);
+	}
+}
+
+static inline void
+emitter_json_dict_begin(emitter_t *emitter, const char *json_key) {
+	if (emitter->output == emitter_output_json) {
+		emitter_dict_begin(emitter, json_key, NULL);
+	}
+}
+
+static inline void
+emitter_json_dict_end(emitter_t *emitter) {
+	if (emitter->output == emitter_output_json) {
+		emitter_dict_end(emitter);
+	}
+}
+
+static inline void
+emitter_table_dict_begin(emitter_t *emitter, const char *table_key) {
+	if (emitter->output == emitter_output_table) {
+		emitter_dict_begin(emitter, NULL, table_key);
+	}
+}
+
+static inline void
+emitter_table_dict_end(emitter_t *emitter) {
+	if (emitter->output == emitter_output_table) {
+		emitter_dict_end(emitter);
+	}
+}
+
+static inline void
+emitter_json_arr_begin(emitter_t *emitter, const char *json_key) {
+	if (emitter->output == emitter_output_json) {
+		emitter_json_key_prefix(emitter);
+		emitter_printf(emitter, "\"%s\": [", json_key);
+		emitter_nest_inc(emitter);
+	}
+}
+
+static inline void
+emitter_json_arr_end(emitter_t *emitter) {
+	if (emitter->output == emitter_output_json) {
+		assert(emitter->nesting_depth > 0);
+		emitter_nest_dec(emitter);
+		emitter_printf(emitter, "\n");
+		emitter_indent(emitter);
+		emitter_printf(emitter, "]");
+	}
+}
+
+static inline void
+emitter_json_arr_obj_begin(emitter_t *emitter) {
+	if (emitter->output == emitter_output_json) {
+		emitter_json_key_prefix(emitter);
+		emitter_printf(emitter, "{");
+		emitter_nest_inc(emitter);
+	}
+}
+
+static inline void
+emitter_json_arr_obj_end(emitter_t *emitter) {
+	if (emitter->output == emitter_output_json) {
+		assert(emitter->nesting_depth > 0);
+		emitter_nest_dec(emitter);
+		emitter_printf(emitter, "\n");
+		emitter_indent(emitter);
+		emitter_printf(emitter, "}");
+	}
+}
+
+static inline void
+emitter_json_arr_value(emitter_t *emitter, emitter_type_t value_type,
+    const void *value) {
+	if (emitter->output == emitter_output_json) {
+		emitter_json_key_prefix(emitter);
+		emitter_print_value(emitter, emitter_justify_none, -1,
+		    value_type, value);
+	}
+}
+
+static inline void
+emitter_table_row(emitter_t *emitter, emitter_row_t *row) {
+	if (emitter->output != emitter_output_table) {
+		return;
+	}
+	emitter_col_t *col;
+	ql_foreach(col, &row->cols, link) {
+		emitter_print_value(emitter, col->justify, col->width,
+		    col->type, (const void *)&col->bool_val);
+	}
+	emitter_table_printf(emitter, "\n");
+}
+
+#endif /* JEMALLOC_INTERNAL_EMITTER_H */

include/jemalloc/internal/extent.h

@@ -1,46 +0,0 @@
-/******************************************************************************/
-#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;
-
-	/* True if zero-filled; used by chunk recycling code. */
-	bool			zeroed;
-};
-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 */
-/******************************************************************************/
-

include/jemalloc/internal/extent_dss.h

@@ -0,0 +1,26 @@
+#ifndef JEMALLOC_INTERNAL_EXTENT_DSS_H
+#define JEMALLOC_INTERNAL_EXTENT_DSS_H
+
+typedef enum {
+	dss_prec_disabled  = 0,
+	dss_prec_primary   = 1,
+	dss_prec_secondary = 2,
+
+	dss_prec_limit     = 3
+} dss_prec_t;
+#define DSS_PREC_DEFAULT dss_prec_secondary
+#define DSS_DEFAULT "secondary"
+
+extern const char *dss_prec_names[];
+
+extern const char *opt_dss;
+
+dss_prec_t extent_dss_prec_get(void);
+bool extent_dss_prec_set(dss_prec_t dss_prec);
+void *extent_alloc_dss(tsdn_t *tsdn, arena_t *arena, void *new_addr,
+    size_t size, size_t alignment, bool *zero, bool *commit);
+bool extent_in_dss(void *addr);
+bool extent_dss_mergeable(void *addr_a, void *addr_b);
+void extent_dss_boot(void);
+
+#endif /* JEMALLOC_INTERNAL_EXTENT_DSS_H */

include/jemalloc/internal/extent_externs.h

@@ -0,0 +1,73 @@
+#ifndef JEMALLOC_INTERNAL_EXTENT_EXTERNS_H
+#define JEMALLOC_INTERNAL_EXTENT_EXTERNS_H
+
+#include "jemalloc/internal/mutex.h"
+#include "jemalloc/internal/mutex_pool.h"
+#include "jemalloc/internal/ph.h"
+#include "jemalloc/internal/rtree.h"
+
+extern size_t opt_lg_extent_max_active_fit;
+
+extern rtree_t extents_rtree;
+extern const extent_hooks_t extent_hooks_default;
+extern mutex_pool_t extent_mutex_pool;
+
+extent_t *extent_alloc(tsdn_t *tsdn, arena_t *arena);
+void extent_dalloc(tsdn_t *tsdn, arena_t *arena, extent_t *extent);
+
+extent_hooks_t *extent_hooks_get(arena_t *arena);
+extent_hooks_t *extent_hooks_set(tsd_t *tsd, arena_t *arena,
+    extent_hooks_t *extent_hooks);
+
+#ifdef JEMALLOC_JET
+size_t extent_size_quantize_floor(size_t size);
+size_t extent_size_quantize_ceil(size_t size);
+#endif
+
+rb_proto(, extent_avail_, extent_tree_t, extent_t)
+ph_proto(, extent_heap_, extent_heap_t, extent_t)
+
+bool extents_init(tsdn_t *tsdn, extents_t *extents, extent_state_t state,
+    bool delay_coalesce);
+extent_state_t extents_state_get(const extents_t *extents);
+size_t extents_npages_get(extents_t *extents);
+extent_t *extents_alloc(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, extents_t *extents, void *new_addr,
+    size_t size, size_t pad, size_t alignment, bool slab, szind_t szind,
+    bool *zero, bool *commit);
+void extents_dalloc(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, extents_t *extents, extent_t *extent);
+extent_t *extents_evict(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, extents_t *extents, size_t npages_min);
+void extents_prefork(tsdn_t *tsdn, extents_t *extents);
+void extents_postfork_parent(tsdn_t *tsdn, extents_t *extents);
+void extents_postfork_child(tsdn_t *tsdn, extents_t *extents);
+extent_t *extent_alloc_wrapper(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, void *new_addr, size_t size, size_t pad,
+    size_t alignment, bool slab, szind_t szind, bool *zero, bool *commit);
+void extent_dalloc_gap(tsdn_t *tsdn, arena_t *arena, extent_t *extent);
+void extent_dalloc_wrapper(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, extent_t *extent);
+void extent_destroy_wrapper(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, extent_t *extent);
+bool extent_commit_wrapper(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, extent_t *extent, size_t offset,
+    size_t length);
+bool extent_decommit_wrapper(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, extent_t *extent, size_t offset,
+    size_t length);
+bool extent_purge_lazy_wrapper(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, extent_t *extent, size_t offset,
+    size_t length);
+bool extent_purge_forced_wrapper(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, extent_t *extent, size_t offset,
+    size_t length);
+extent_t *extent_split_wrapper(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, extent_t *extent, size_t size_a,
+    szind_t szind_a, bool slab_a, size_t size_b, szind_t szind_b, bool slab_b);
+bool extent_merge_wrapper(tsdn_t *tsdn, arena_t *arena,
+    extent_hooks_t **r_extent_hooks, extent_t *a, extent_t *b);
+
+bool extent_boot(void);
+
+#endif /* JEMALLOC_INTERNAL_EXTENT_EXTERNS_H */

include/jemalloc/internal/extent_inlines.h

@@ -0,0 +1,433 @@
+#ifndef JEMALLOC_INTERNAL_EXTENT_INLINES_H
+#define JEMALLOC_INTERNAL_EXTENT_INLINES_H
+
+#include "jemalloc/internal/mutex.h"
+#include "jemalloc/internal/mutex_pool.h"
+#include "jemalloc/internal/pages.h"
+#include "jemalloc/internal/prng.h"
+#include "jemalloc/internal/ql.h"
+#include "jemalloc/internal/sz.h"
+
+static inline void
+extent_lock(tsdn_t *tsdn, extent_t *extent) {
+	assert(extent != NULL);
+	mutex_pool_lock(tsdn, &extent_mutex_pool, (uintptr_t)extent);
+}
+
+static inline void
+extent_unlock(tsdn_t *tsdn, extent_t *extent) {
+	assert(extent != NULL);
+	mutex_pool_unlock(tsdn, &extent_mutex_pool, (uintptr_t)extent);
+}
+
+static inline void
+extent_lock2(tsdn_t *tsdn, extent_t *extent1, extent_t *extent2) {
+	assert(extent1 != NULL && extent2 != NULL);
+	mutex_pool_lock2(tsdn, &extent_mutex_pool, (uintptr_t)extent1,
+	    (uintptr_t)extent2);
+}
+
+static inline void
+extent_unlock2(tsdn_t *tsdn, extent_t *extent1, extent_t *extent2) {
+	assert(extent1 != NULL && extent2 != NULL);
+	mutex_pool_unlock2(tsdn, &extent_mutex_pool, (uintptr_t)extent1,
+	    (uintptr_t)extent2);
+}
+
+static inline arena_t *
+extent_arena_get(const extent_t *extent) {
+	unsigned arena_ind = (unsigned)((extent->e_bits &
+	    EXTENT_BITS_ARENA_MASK) >> EXTENT_BITS_ARENA_SHIFT);
+	/*
+	 * The following check is omitted because we should never actually read
+	 * a NULL arena pointer.
+	 */
+	if (false && arena_ind >= MALLOCX_ARENA_LIMIT) {
+		return NULL;
+	}
+	assert(arena_ind < MALLOCX_ARENA_LIMIT);
+	return (arena_t *)atomic_load_p(&arenas[arena_ind], ATOMIC_ACQUIRE);
+}
+
+static inline szind_t
+extent_szind_get_maybe_invalid(const extent_t *extent) {
+	szind_t szind = (szind_t)((extent->e_bits & EXTENT_BITS_SZIND_MASK) >>
+	    EXTENT_BITS_SZIND_SHIFT);
+	assert(szind <= NSIZES);
+	return szind;
+}
+
+static inline szind_t
+extent_szind_get(const extent_t *extent) {
+	szind_t szind = extent_szind_get_maybe_invalid(extent);
+	assert(szind < NSIZES); /* Never call when "invalid". */
+	return szind;
+}
+
+static inline size_t
+extent_usize_get(const extent_t *extent) {
+	return sz_index2size(extent_szind_get(extent));
+}
+
+static inline size_t
+extent_sn_get(const extent_t *extent) {
+	return (size_t)((extent->e_bits & EXTENT_BITS_SN_MASK) >>
+	    EXTENT_BITS_SN_SHIFT);
+}
+
+static inline extent_state_t
+extent_state_get(const extent_t *extent) {
+	return (extent_state_t)((extent->e_bits & EXTENT_BITS_STATE_MASK) >>
+	    EXTENT_BITS_STATE_SHIFT);
+}
+
+static inline bool
+extent_zeroed_get(const extent_t *extent) {
+	return (bool)((extent->e_bits & EXTENT_BITS_ZEROED_MASK) >>
+	    EXTENT_BITS_ZEROED_SHIFT);
+}
+
+static inline bool
+extent_committed_get(const extent_t *extent) {
+	return (bool)((extent->e_bits & EXTENT_BITS_COMMITTED_MASK) >>
+	    EXTENT_BITS_COMMITTED_SHIFT);
+}
+
+static inline bool
+extent_dumpable_get(const extent_t *extent) {
+	return (bool)((extent->e_bits & EXTENT_BITS_DUMPABLE_MASK) >>
+	    EXTENT_BITS_DUMPABLE_SHIFT);
+}
+
+static inline bool
+extent_slab_get(const extent_t *extent) {
+	return (bool)((extent->e_bits & EXTENT_BITS_SLAB_MASK) >>
+	    EXTENT_BITS_SLAB_SHIFT);
+}
+
+static inline unsigned
+extent_nfree_get(const extent_t *extent) {
+	assert(extent_slab_get(extent));
+	return (unsigned)((extent->e_bits & EXTENT_BITS_NFREE_MASK) >>
+	    EXTENT_BITS_NFREE_SHIFT);
+}
+
+static inline void *
+extent_base_get(const extent_t *extent) {
+	assert(extent->e_addr == PAGE_ADDR2BASE(extent->e_addr) ||
+	    !extent_slab_get(extent));
+	return PAGE_ADDR2BASE(extent->e_addr);
+}
+
+static inline void *
+extent_addr_get(const extent_t *extent) {
+	assert(extent->e_addr == PAGE_ADDR2BASE(extent->e_addr) ||
+	    !extent_slab_get(extent));
+	return extent->e_addr;
+}
+
+static inline size_t
+extent_size_get(const extent_t *extent) {
+	return (extent->e_size_esn & EXTENT_SIZE_MASK);
+}
+
+static inline size_t
+extent_esn_get(const extent_t *extent) {
+	return (extent->e_size_esn & EXTENT_ESN_MASK);
+}
+
+static inline size_t
+extent_bsize_get(const extent_t *extent) {
+	return extent->e_bsize;
+}
+
+static inline void *
+extent_before_get(const extent_t *extent) {
+	return (void *)((uintptr_t)extent_base_get(extent) - PAGE);
+}
+
+static inline void *
+extent_last_get(const extent_t *extent) {
+	return (void *)((uintptr_t)extent_base_get(extent) +
+	    extent_size_get(extent) - PAGE);
+}
+
+static inline void *
+extent_past_get(const extent_t *extent) {
+	return (void *)((uintptr_t)extent_base_get(extent) +
+	    extent_size_get(extent));
+}
+
+static inline arena_slab_data_t *
+extent_slab_data_get(extent_t *extent) {
+	assert(extent_slab_get(extent));
+	return &extent->e_slab_data;
+}
+
+static inline const arena_slab_data_t *
+extent_slab_data_get_const(const extent_t *extent) {
+	assert(extent_slab_get(extent));
+	return &extent->e_slab_data;
+}
+
+static inline prof_tctx_t *
+extent_prof_tctx_get(const extent_t *extent) {
+	return (prof_tctx_t *)atomic_load_p(&extent->e_prof_tctx,
+	    ATOMIC_ACQUIRE);
+}
+
+static inline void
+extent_arena_set(extent_t *extent, arena_t *arena) {
+	unsigned arena_ind = (arena != NULL) ? arena_ind_get(arena) : ((1U <<
+	    MALLOCX_ARENA_BITS) - 1);
+	extent->e_bits = (extent->e_bits & ~EXTENT_BITS_ARENA_MASK) |
+	    ((uint64_t)arena_ind << EXTENT_BITS_ARENA_SHIFT);
+}
+
+static inline void
+extent_addr_set(extent_t *extent, void *addr) {
+	extent->e_addr = addr;
+}
+
+static inline void
+extent_addr_randomize(UNUSED tsdn_t *tsdn, extent_t *extent, size_t alignment) {
+	assert(extent_base_get(extent) == extent_addr_get(extent));
+
+	if (alignment < PAGE) {
+		unsigned lg_range = LG_PAGE -
+		    lg_floor(CACHELINE_CEILING(alignment));
+		size_t r;
+		if (!tsdn_null(tsdn)) {
+			tsd_t *tsd = tsdn_tsd(tsdn);
+			r = (size_t)prng_lg_range_u64(
+			    tsd_offset_statep_get(tsd), lg_range);
+		} else {
+			r = prng_lg_range_zu(
+			    &extent_arena_get(extent)->offset_state,
+			    lg_range, true);
+		}
+		uintptr_t random_offset = ((uintptr_t)r) << (LG_PAGE -
+		    lg_range);
+		extent->e_addr = (void *)((uintptr_t)extent->e_addr +
+		    random_offset);
+		assert(ALIGNMENT_ADDR2BASE(extent->e_addr, alignment) ==
+		    extent->e_addr);
+	}
+}
+
+static inline void
+extent_size_set(extent_t *extent, size_t size) {
+	assert((size & ~EXTENT_SIZE_MASK) == 0);
+	extent->e_size_esn = size | (extent->e_size_esn & ~EXTENT_SIZE_MASK);
+}
+
+static inline void
+extent_esn_set(extent_t *extent, size_t esn) {
+	extent->e_size_esn = (extent->e_size_esn & ~EXTENT_ESN_MASK) | (esn &
+	    EXTENT_ESN_MASK);
+}
+
+static inline void
+extent_bsize_set(extent_t *extent, size_t bsize) {
+	extent->e_bsize = bsize;
+}
+
+static inline void
+extent_szind_set(extent_t *extent, szind_t szind) {
+	assert(szind <= NSIZES); /* NSIZES means "invalid". */
+	extent->e_bits = (extent->e_bits & ~EXTENT_BITS_SZIND_MASK) |
+	    ((uint64_t)szind << EXTENT_BITS_SZIND_SHIFT);
+}
+
+static inline void
+extent_nfree_set(extent_t *extent, unsigned nfree) {
+	assert(extent_slab_get(extent));
+	extent->e_bits = (extent->e_bits & ~EXTENT_BITS_NFREE_MASK) |
+	    ((uint64_t)nfree << EXTENT_BITS_NFREE_SHIFT);
+}
+
+static inline void
+extent_nfree_inc(extent_t *extent) {
+	assert(extent_slab_get(extent));
+	extent->e_bits += ((uint64_t)1U << EXTENT_BITS_NFREE_SHIFT);
+}
+
+static inline void
+extent_nfree_dec(extent_t *extent) {
+	assert(extent_slab_get(extent));
+	extent->e_bits -= ((uint64_t)1U << EXTENT_BITS_NFREE_SHIFT);
+}
+
+static inline void
+extent_sn_set(extent_t *extent, size_t sn) {
+	extent->e_bits = (extent->e_bits & ~EXTENT_BITS_SN_MASK) |
+	    ((uint64_t)sn << EXTENT_BITS_SN_SHIFT);
+}
+
+static inline void
+extent_state_set(extent_t *extent, extent_state_t state) {
+	extent->e_bits = (extent->e_bits & ~EXTENT_BITS_STATE_MASK) |
+	    ((uint64_t)state << EXTENT_BITS_STATE_SHIFT);
+}
+
+static inline void
+extent_zeroed_set(extent_t *extent, bool zeroed) {
+	extent->e_bits = (extent->e_bits & ~EXTENT_BITS_ZEROED_MASK) |
+	    ((uint64_t)zeroed << EXTENT_BITS_ZEROED_SHIFT);
+}
+
+static inline void
+extent_committed_set(extent_t *extent, bool committed) {
+	extent->e_bits = (extent->e_bits & ~EXTENT_BITS_COMMITTED_MASK) |
+	    ((uint64_t)committed << EXTENT_BITS_COMMITTED_SHIFT);
+}
+
+static inline void
+extent_dumpable_set(extent_t *extent, bool dumpable) {
+	extent->e_bits = (extent->e_bits & ~EXTENT_BITS_DUMPABLE_MASK) |
+	    ((uint64_t)dumpable << EXTENT_BITS_DUMPABLE_SHIFT);
+}
+
+static inline void
+extent_slab_set(extent_t *extent, bool slab) {
+	extent->e_bits = (extent->e_bits & ~EXTENT_BITS_SLAB_MASK) |
+	    ((uint64_t)slab << EXTENT_BITS_SLAB_SHIFT);
+}
+
+static inline void
+extent_prof_tctx_set(extent_t *extent, prof_tctx_t *tctx) {
+	atomic_store_p(&extent->e_prof_tctx, tctx, ATOMIC_RELEASE);
+}
+
+static inline void
+extent_init(extent_t *extent, arena_t *arena, void *addr, size_t size,
+    bool slab, szind_t szind, size_t sn, extent_state_t state, bool zeroed,
+    bool committed, bool dumpable) {
+	assert(addr == PAGE_ADDR2BASE(addr) || !slab);
+
+	extent_arena_set(extent, arena);
+	extent_addr_set(extent, addr);
+	extent_size_set(extent, size);
+	extent_slab_set(extent, slab);
+	extent_szind_set(extent, szind);
+	extent_sn_set(extent, sn);
+	extent_state_set(extent, state);
+	extent_zeroed_set(extent, zeroed);
+	extent_committed_set(extent, committed);
+	extent_dumpable_set(extent, dumpable);
+	ql_elm_new(extent, ql_link);
+	if (config_prof) {
+		extent_prof_tctx_set(extent, NULL);
+	}
+}
+
+static inline void
+extent_binit(extent_t *extent, void *addr, size_t bsize, size_t sn) {
+	extent_arena_set(extent, NULL);
+	extent_addr_set(extent, addr);
+	extent_bsize_set(extent, bsize);
+	extent_slab_set(extent, false);
+	extent_szind_set(extent, NSIZES);
+	extent_sn_set(extent, sn);
+	extent_state_set(extent, extent_state_active);
+	extent_zeroed_set(extent, true);
+	extent_committed_set(extent, true);
+	extent_dumpable_set(extent, true);
+}
+
+static inline void
+extent_list_init(extent_list_t *list) {
+	ql_new(list);
+}
+
+static inline extent_t *
+extent_list_first(const extent_list_t *list) {
+	return ql_first(list);
+}
+
+static inline extent_t *
+extent_list_last(const extent_list_t *list) {
+	return ql_last(list, ql_link);
+}
+
+static inline void
+extent_list_append(extent_list_t *list, extent_t *extent) {
+	ql_tail_insert(list, extent, ql_link);
+}
+
+static inline void
+extent_list_prepend(extent_list_t *list, extent_t *extent) {
+	ql_head_insert(list, extent, ql_link);
+}
+
+static inline void
+extent_list_replace(extent_list_t *list, extent_t *to_remove,
+    extent_t *to_insert) {
+	ql_after_insert(to_remove, to_insert, ql_link);
+	ql_remove(list, to_remove, ql_link);
+}
+
+static inline void
+extent_list_remove(extent_list_t *list, extent_t *extent) {
+	ql_remove(list, extent, ql_link);
+}
+
+static inline int
+extent_sn_comp(const extent_t *a, const extent_t *b) {
+	size_t a_sn = extent_sn_get(a);
+	size_t b_sn = extent_sn_get(b);
+
+	return (a_sn > b_sn) - (a_sn < b_sn);
+}
+
+static inline int
+extent_esn_comp(const extent_t *a, const extent_t *b) {
+	size_t a_esn = extent_esn_get(a);
+	size_t b_esn = extent_esn_get(b);
+
+	return (a_esn > b_esn) - (a_esn < b_esn);
+}
+
+static inline int
+extent_ad_comp(const extent_t *a, const extent_t *b) {
+	uintptr_t a_addr = (uintptr_t)extent_addr_get(a);
+	uintptr_t b_addr = (uintptr_t)extent_addr_get(b);
+
+	return (a_addr > b_addr) - (a_addr < b_addr);
+}
+
+static inline int
+extent_ead_comp(const extent_t *a, const extent_t *b) {
+	uintptr_t a_eaddr = (uintptr_t)a;
+	uintptr_t b_eaddr = (uintptr_t)b;
+
+	return (a_eaddr > b_eaddr) - (a_eaddr < b_eaddr);
+}
+
+static inline int
+extent_snad_comp(const extent_t *a, const extent_t *b) {
+	int ret;
+
+	ret = extent_sn_comp(a, b);
+	if (ret != 0) {
+		return ret;
+	}
+
+	ret = extent_ad_comp(a, b);
+	return ret;
+}
+
+static inline int
+extent_esnead_comp(const extent_t *a, const extent_t *b) {
+	int ret;
+
+	ret = extent_esn_comp(a, b);
+	if (ret != 0) {
+		return ret;
+	}
+
+	ret = extent_ead_comp(a, b);
+	return ret;
+}
+
+#endif /* JEMALLOC_INTERNAL_EXTENT_INLINES_H */

include/jemalloc/internal/extent_mmap.h

@@ -0,0 +1,10 @@
+#ifndef JEMALLOC_INTERNAL_EXTENT_MMAP_EXTERNS_H
+#define JEMALLOC_INTERNAL_EXTENT_MMAP_EXTERNS_H
+
+extern bool opt_retain;
+
+void *extent_alloc_mmap(void *new_addr, size_t size, size_t alignment,
+    bool *zero, bool *commit);
+bool extent_dalloc_mmap(void *addr, size_t size);
+
+#endif /* JEMALLOC_INTERNAL_EXTENT_MMAP_EXTERNS_H */

include/jemalloc/internal/extent_structs.h

@@ -0,0 +1,219 @@
+#ifndef JEMALLOC_INTERNAL_EXTENT_STRUCTS_H
+#define JEMALLOC_INTERNAL_EXTENT_STRUCTS_H
+
+#include "jemalloc/internal/atomic.h"
+#include "jemalloc/internal/bitmap.h"
+#include "jemalloc/internal/mutex.h"
+#include "jemalloc/internal/ql.h"
+#include "jemalloc/internal/ph.h"
+#include "jemalloc/internal/size_classes.h"
+
+typedef enum {
+	extent_state_active   = 0,
+	extent_state_dirty    = 1,
+	extent_state_muzzy    = 2,
+	extent_state_retained = 3
+} extent_state_t;
+
+/* Extent (span of pages).  Use accessor functions for e_* fields. */
+struct extent_s {
+	/*
+	 * Bitfield containing several fields:
+	 *
+	 * a: arena_ind
+	 * b: slab
+	 * c: committed
+	 * d: dumpable
+	 * z: zeroed
+	 * t: state
+	 * i: szind
+	 * f: nfree
+	 * n: sn
+	 *
+	 * nnnnnnnn ... nnnnffff ffffffii iiiiiitt zdcbaaaa aaaaaaaa
+	 *
+	 * arena_ind: Arena from which this extent came, or all 1 bits if
+	 *            unassociated.
+	 *
+	 * slab: The slab flag indicates whether the extent is used for a slab
+	 *       of small regions.  This helps differentiate small size classes,
+	 *       and it indicates whether interior pointers can be looked up via
+	 *       iealloc().
+	 *
+	 * committed: The committed flag indicates whether physical memory is
+	 *            committed to the extent, whether explicitly or implicitly
+	 *            as on a system that overcommits and satisfies physical
+	 *            memory needs on demand via soft page faults.
+	 *
+	 * dumpable: The dumpable flag indicates whether or not we've set the
+	 *           memory in question to be dumpable.  Note that this
+	 *           interacts somewhat subtly with user-specified extent hooks,
+	 *           since we don't know if *they* are fiddling with
+	 *           dumpability (in which case, we don't want to undo whatever
+	 *           they're doing).  To deal with this scenario, we:
+	 *             - Make dumpable false only for memory allocated with the
+	 *               default hooks.
+	 *             - Only allow memory to go from non-dumpable to dumpable,
+	 *               and only once.
+	 *             - Never make the OS call to allow dumping when the
+	 *               dumpable bit is already set.
+	 *           These three constraints mean that we will never
+	 *           accidentally dump user memory that the user meant to set
+	 *           nondumpable with their extent hooks.
+	 *
+	 *
+	 * zeroed: The zeroed flag is used by extent recycling code to track
+	 *         whether memory is zero-filled.
+	 *
+	 * state: The state flag is an extent_state_t.
+	 *
+	 * szind: The szind flag indicates usable size class index for
+	 *        allocations residing in this extent, regardless of whether the
+	 *        extent is a slab.  Extent size and usable size often differ
+	 *        even for non-slabs, either due to sz_large_pad or promotion of
+	 *        sampled small regions.
+	 *
+	 * nfree: Number of free regions in slab.
+	 *
+	 * sn: Serial number (potentially non-unique).
+	 *
+	 *     Serial numbers may wrap around if !opt_retain, but as long as
+	 *     comparison functions fall back on address comparison for equal
+	 *     serial numbers, stable (if imperfect) ordering is maintained.
+	 *
+	 *     Serial numbers may not be unique even in the absence of
+	 *     wrap-around, e.g. when splitting an extent and assigning the same
+	 *     serial number to both resulting adjacent extents.
+	 */
+	uint64_t		e_bits;
+#define MASK(CURRENT_FIELD_WIDTH, CURRENT_FIELD_SHIFT) ((((((uint64_t)0x1U) << (CURRENT_FIELD_WIDTH)) - 1)) << (CURRENT_FIELD_SHIFT))
+
+#define EXTENT_BITS_ARENA_WIDTH  MALLOCX_ARENA_BITS
+#define EXTENT_BITS_ARENA_SHIFT  0
+#define EXTENT_BITS_ARENA_MASK  MASK(EXTENT_BITS_ARENA_WIDTH, EXTENT_BITS_ARENA_SHIFT)
+
+#define EXTENT_BITS_SLAB_WIDTH  1
+#define EXTENT_BITS_SLAB_SHIFT  (EXTENT_BITS_ARENA_WIDTH + EXTENT_BITS_ARENA_SHIFT)
+#define EXTENT_BITS_SLAB_MASK  MASK(EXTENT_BITS_SLAB_WIDTH, EXTENT_BITS_SLAB_SHIFT)
+
+#define EXTENT_BITS_COMMITTED_WIDTH  1
+#define EXTENT_BITS_COMMITTED_SHIFT  (EXTENT_BITS_SLAB_WIDTH + EXTENT_BITS_SLAB_SHIFT)
+#define EXTENT_BITS_COMMITTED_MASK  MASK(EXTENT_BITS_COMMITTED_WIDTH, EXTENT_BITS_COMMITTED_SHIFT)
+
+#define EXTENT_BITS_DUMPABLE_WIDTH  1
+#define EXTENT_BITS_DUMPABLE_SHIFT  (EXTENT_BITS_COMMITTED_WIDTH + EXTENT_BITS_COMMITTED_SHIFT)
+#define EXTENT_BITS_DUMPABLE_MASK  MASK(EXTENT_BITS_DUMPABLE_WIDTH, EXTENT_BITS_DUMPABLE_SHIFT)
+
+#define EXTENT_BITS_ZEROED_WIDTH  1
+#define EXTENT_BITS_ZEROED_SHIFT  (EXTENT_BITS_DUMPABLE_WIDTH + EXTENT_BITS_DUMPABLE_SHIFT)
+#define EXTENT_BITS_ZEROED_MASK  MASK(EXTENT_BITS_ZEROED_WIDTH, EXTENT_BITS_ZEROED_SHIFT)
+
+#define EXTENT_BITS_STATE_WIDTH  2
+#define EXTENT_BITS_STATE_SHIFT  (EXTENT_BITS_ZEROED_WIDTH + EXTENT_BITS_ZEROED_SHIFT)
+#define EXTENT_BITS_STATE_MASK  MASK(EXTENT_BITS_STATE_WIDTH, EXTENT_BITS_STATE_SHIFT)
+
+#define EXTENT_BITS_SZIND_WIDTH  LG_CEIL_NSIZES
+#define EXTENT_BITS_SZIND_SHIFT  (EXTENT_BITS_STATE_WIDTH + EXTENT_BITS_STATE_SHIFT)
+#define EXTENT_BITS_SZIND_MASK  MASK(EXTENT_BITS_SZIND_WIDTH, EXTENT_BITS_SZIND_SHIFT)
+
+#define EXTENT_BITS_NFREE_WIDTH  (LG_SLAB_MAXREGS + 1)
+#define EXTENT_BITS_NFREE_SHIFT  (EXTENT_BITS_SZIND_WIDTH + EXTENT_BITS_SZIND_SHIFT)
+#define EXTENT_BITS_NFREE_MASK  MASK(EXTENT_BITS_NFREE_WIDTH, EXTENT_BITS_NFREE_SHIFT)
+
+#define EXTENT_BITS_SN_SHIFT  (EXTENT_BITS_NFREE_WIDTH + EXTENT_BITS_NFREE_SHIFT)
+#define EXTENT_BITS_SN_MASK  (UINT64_MAX << EXTENT_BITS_SN_SHIFT)
+
+	/* Pointer to the extent that this structure is responsible for. */
+	void			*e_addr;
+
+	union {
+		/*
+		 * Extent size and serial number associated with the extent
+		 * structure (different than the serial number for the extent at
+		 * e_addr).
+		 *
+		 * ssssssss [...] ssssssss ssssnnnn nnnnnnnn
+		 */
+		size_t			e_size_esn;
+	#define EXTENT_SIZE_MASK	((size_t)~(PAGE-1))
+	#define EXTENT_ESN_MASK		((size_t)PAGE-1)
+		/* Base extent size, which may not be a multiple of PAGE. */
+		size_t			e_bsize;
+	};
+
+	/*
+	 * List linkage, used by a variety of lists:
+	 * - bin_t's slabs_full
+	 * - extents_t's LRU
+	 * - stashed dirty extents
+	 * - arena's large allocations
+	 */
+	ql_elm(extent_t)	ql_link;
+
+	/*
+	 * Linkage for per size class sn/address-ordered heaps, and
+	 * for extent_avail
+	 */
+	phn(extent_t)		ph_link;
+
+	union {
+		/* Small region slab metadata. */
+		arena_slab_data_t	e_slab_data;
+
+		/*
+		 * Profile counters, used for large objects.  Points to a
+		 * prof_tctx_t.
+		 */
+		atomic_p_t		e_prof_tctx;
+	};
+};
+typedef ql_head(extent_t) extent_list_t;
+typedef ph(extent_t) extent_tree_t;
+typedef ph(extent_t) extent_heap_t;
+
+/* Quantized collection of extents, with built-in LRU queue. */
+struct extents_s {
+	malloc_mutex_t		mtx;
+
+	/*
+	 * Quantized per size class heaps of extents.
+	 *
+	 * Synchronization: mtx.
+	 */
+	extent_heap_t		heaps[NPSIZES+1];
+
+	/*
+	 * Bitmap for which set bits correspond to non-empty heaps.
+	 *
+	 * Synchronization: mtx.
+	 */
+	bitmap_t		bitmap[BITMAP_GROUPS(NPSIZES+1)];
+
+	/*
+	 * LRU of all extents in heaps.
+	 *
+	 * Synchronization: mtx.
+	 */
+	extent_list_t		lru;
+
+	/*
+	 * Page sum for all extents in heaps.
+	 *
+	 * The synchronization here is a little tricky.  Modifications to npages
+	 * must hold mtx, but reads need not (though, a reader who sees npages
+	 * without holding the mutex can't assume anything about the rest of the
+	 * state of the extents_t).
+	 */
+	atomic_zu_t		npages;
+
+	/* All stored extents must be in the same state. */
+	extent_state_t		state;
+
+	/*
+	 * If true, delay coalescing until eviction; otherwise coalesce during
+	 * deallocation.
+	 */
+	bool			delay_coalesce;
+};
+
+#endif /* JEMALLOC_INTERNAL_EXTENT_STRUCTS_H */

include/jemalloc/internal/extent_types.h

@@ -0,0 +1,17 @@
+#ifndef JEMALLOC_INTERNAL_EXTENT_TYPES_H
+#define JEMALLOC_INTERNAL_EXTENT_TYPES_H
+
+typedef struct extent_s extent_t;
+typedef struct extents_s extents_t;
+
+#define EXTENT_HOOKS_INITIALIZER	NULL
+
+#define EXTENT_GROW_MAX_PIND (NPSIZES - 1)
+
+/*
+ * When reuse (and split) an active extent, (1U << opt_lg_extent_max_active_fit)
+ * is the max ratio between the size of the active extent and the new extent.
+ */
+#define LG_EXTENT_MAX_ACTIVE_FIT_DEFAULT 6
+
+#endif /* JEMALLOC_INTERNAL_EXTENT_TYPES_H */

include/jemalloc/internal/hash.h

@@ -1,92 +1,76 @@
+#ifndef JEMALLOC_INTERNAL_HASH_H
+#define JEMALLOC_INTERNAL_HASH_H
+
+#include "jemalloc/internal/assert.h"
+
 /*
  * The following hash function is based on MurmurHash3, placed into the public
- * domain by Austin Appleby.  See http://code.google.com/p/smhasher/ for
+ * domain by Austin Appleby.  See https://github.com/aappleby/smhasher for
  * details.
  */
-/******************************************************************************/
-#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
-uint32_t	hash_x86_32(const void *key, int len, uint32_t seed);
-void	hash_x86_128(const void *key, const int len, uint32_t seed,
-    uint64_t r_out[2]);
-void	hash_x64_128(const void *key, const int len, const uint32_t seed,
-    uint64_t r_out[2]);
-void	hash(const void *key, size_t len, const uint32_t seed,
-    size_t r_hash[2]);
-#endif
-
-#if (defined(JEMALLOC_ENABLE_INLINE) || defined(JEMALLOC_HASH_C_))
 /******************************************************************************/
 /* Internal implementation. */
-JEMALLOC_INLINE uint32_t
-hash_rotl_32(uint32_t x, int8_t r)
-{
-
-	return (x << r) | (x >> (32 - r));
+static inline uint32_t
+hash_rotl_32(uint32_t x, int8_t r) {
+	return ((x << r) | (x >> (32 - r)));
 }
 
-JEMALLOC_INLINE uint64_t
-hash_rotl_64(uint64_t x, int8_t r)
-{
-	return (x << r) | (x >> (64 - r));
+static inline uint64_t
+hash_rotl_64(uint64_t x, int8_t r) {
+	return ((x << r) | (x >> (64 - r)));
 }
 
-JEMALLOC_INLINE uint32_t
-hash_get_block_32(const uint32_t *p, int i)
-{
+static inline uint32_t
+hash_get_block_32(const uint32_t *p, int i) {
+	/* Handle unaligned read. */
+	if (unlikely((uintptr_t)p & (sizeof(uint32_t)-1)) != 0) {
+		uint32_t ret;
 
-	return (p[i]);
+		memcpy(&ret, (uint8_t *)(p + i), sizeof(uint32_t));
+		return ret;
+	}
+
+	return p[i];
 }
 
-JEMALLOC_INLINE uint64_t
-hash_get_block_64(const uint64_t *p, int i)
-{
+static inline uint64_t
+hash_get_block_64(const uint64_t *p, int i) {
+	/* Handle unaligned read. */
+	if (unlikely((uintptr_t)p & (sizeof(uint64_t)-1)) != 0) {
+		uint64_t ret;
 
-	return (p[i]);
+		memcpy(&ret, (uint8_t *)(p + i), sizeof(uint64_t));
+		return ret;
+	}
+
+	return p[i];
 }
 
-JEMALLOC_INLINE uint32_t
-hash_fmix_32(uint32_t h)
-{
-
+static inline uint32_t
+hash_fmix_32(uint32_t h) {
 	h ^= h >> 16;
 	h *= 0x85ebca6b;
 	h ^= h >> 13;
 	h *= 0xc2b2ae35;
 	h ^= h >> 16;
 
-	return (h);
+	return h;
 }
 
-JEMALLOC_INLINE uint64_t
-hash_fmix_64(uint64_t k)
-{
-
+static inline uint64_t
+hash_fmix_64(uint64_t k) {
 	k ^= k >> 33;
-	k *= QU(0xff51afd7ed558ccdLLU);
+	k *= KQU(0xff51afd7ed558ccd);
 	k ^= k >> 33;
-	k *= QU(0xc4ceb9fe1a85ec53LLU);
+	k *= KQU(0xc4ceb9fe1a85ec53);
 	k ^= k >> 33;
 
-	return (k);
+	return k;
 }
 
-JEMALLOC_INLINE uint32_t
-hash_x86_32(const void *key, int len, uint32_t seed)
-{
+static inline uint32_t
+hash_x86_32(const void *key, int len, uint32_t seed) {
 	const uint8_t *data = (const uint8_t *) key;
 	const int nblocks = len / 4;
 
@@ -132,13 +116,12 @@ hash_x86_32(const void *key, int len, uint32_t seed)
 
 	h1 = hash_fmix_32(h1);
 
-	return (h1);
+	return h1;
 }
 
-UNUSED JEMALLOC_INLINE void
+UNUSED static inline void
 hash_x86_128(const void *key, const int len, uint32_t seed,
-    uint64_t r_out[2])
-{
+    uint64_t r_out[2]) {
 	const uint8_t * data = (const uint8_t *) key;
 	const int nblocks = len / 16;
 
@@ -237,18 +220,17 @@ hash_x86_128(const void *key, const int len, uint32_t seed,
 	r_out[1] = (((uint64_t) h4) << 32) | h3;
 }
 
-UNUSED JEMALLOC_INLINE void
+UNUSED static inline void
 hash_x64_128(const void *key, const int len, const uint32_t seed,
-    uint64_t r_out[2])
-{
+    uint64_t r_out[2]) {
 	const uint8_t *data = (const uint8_t *) key;
 	const int nblocks = len / 16;
 
 	uint64_t h1 = seed;
 	uint64_t h2 = seed;
 
-	const uint64_t c1 = QU(0x87c37b91114253d5LLU);
-	const uint64_t c2 = QU(0x4cf5ad432745937fLLU);
+	const uint64_t c1 = KQU(0x87c37b91114253d5);
+	const uint64_t c2 = KQU(0x4cf5ad432745937f);
 
 	/* body */
 	{
@@ -278,22 +260,22 @@ hash_x64_128(const void *key, const int len, const uint32_t seed,
 		uint64_t k2 = 0;
 
 		switch (len & 15) {
-		case 15: k2 ^= ((uint64_t)(tail[14])) << 48;
-		case 14: k2 ^= ((uint64_t)(tail[13])) << 40;
-		case 13: k2 ^= ((uint64_t)(tail[12])) << 32;
-		case 12: k2 ^= ((uint64_t)(tail[11])) << 24;
-		case 11: k2 ^= ((uint64_t)(tail[10])) << 16;
-		case 10: k2 ^= ((uint64_t)(tail[ 9])) << 8;
+		case 15: k2 ^= ((uint64_t)(tail[14])) << 48; /* falls through */
+		case 14: k2 ^= ((uint64_t)(tail[13])) << 40; /* falls through */
+		case 13: k2 ^= ((uint64_t)(tail[12])) << 32; /* falls through */
+		case 12: k2 ^= ((uint64_t)(tail[11])) << 24; /* falls through */
+		case 11: k2 ^= ((uint64_t)(tail[10])) << 16; /* falls through */
+		case 10: k2 ^= ((uint64_t)(tail[ 9])) << 8;  /* falls through */
 		case  9: k2 ^= ((uint64_t)(tail[ 8])) << 0;
 			k2 *= c2; k2 = hash_rotl_64(k2, 33); k2 *= c1; h2 ^= k2;
-
-		case  8: k1 ^= ((uint64_t)(tail[ 7])) << 56;
-		case  7: k1 ^= ((uint64_t)(tail[ 6])) << 48;
-		case  6: k1 ^= ((uint64_t)(tail[ 5])) << 40;
-		case  5: k1 ^= ((uint64_t)(tail[ 4])) << 32;
-		case  4: k1 ^= ((uint64_t)(tail[ 3])) << 24;
-		case  3: k1 ^= ((uint64_t)(tail[ 2])) << 16;
-		case  2: k1 ^= ((uint64_t)(tail[ 1])) << 8;
+			/* falls through */
+		case  8: k1 ^= ((uint64_t)(tail[ 7])) << 56; /* falls through */
+		case  7: k1 ^= ((uint64_t)(tail[ 6])) << 48; /* falls through */
+		case  6: k1 ^= ((uint64_t)(tail[ 5])) << 40; /* falls through */
+		case  5: k1 ^= ((uint64_t)(tail[ 4])) << 32; /* falls through */
+		case  4: k1 ^= ((uint64_t)(tail[ 3])) << 24; /* falls through */
+		case  3: k1 ^= ((uint64_t)(tail[ 2])) << 16; /* falls through */
+		case  2: k1 ^= ((uint64_t)(tail[ 1])) << 8;  /* falls through */
 		case  1: k1 ^= ((uint64_t)(tail[ 0])) << 0;
 			k1 *= c1; k1 = hash_rotl_64(k1, 31); k1 *= c2; h1 ^= k1;
 		}
@@ -317,19 +299,20 @@ hash_x64_128(const void *key, const int len, const uint32_t seed,
 
 /******************************************************************************/
 /* API. */
-JEMALLOC_INLINE void
-hash(const void *key, size_t len, const uint32_t seed, size_t r_hash[2])
-{
+static inline void
+hash(const void *key, size_t len, const uint32_t seed, size_t r_hash[2]) {
+	assert(len <= INT_MAX); /* Unfortunate implementation limitation. */
+
 #if (LG_SIZEOF_PTR == 3 && !defined(JEMALLOC_BIG_ENDIAN))
-	hash_x64_128(key, len, seed, (uint64_t *)r_hash);
+	hash_x64_128(key, (int)len, seed, (uint64_t *)r_hash);
 #else
-	uint64_t hashes[2];
-	hash_x86_128(key, len, seed, hashes);
-	r_hash[0] = (size_t)hashes[0];
-	r_hash[1] = (size_t)hashes[1];
+	{
+		uint64_t hashes[2];
+		hash_x86_128(key, (int)len, seed, hashes);
+		r_hash[0] = (size_t)hashes[0];
+		r_hash[1] = (size_t)hashes[1];
+	}
 #endif
 }
-#endif
 
-#endif /* JEMALLOC_H_INLINES */
-/******************************************************************************/
+#endif /* JEMALLOC_INTERNAL_HASH_H */

include/jemalloc/internal/hooks.h

@@ -0,0 +1,19 @@
+#ifndef JEMALLOC_INTERNAL_HOOKS_H
+#define JEMALLOC_INTERNAL_HOOKS_H
+
+extern JEMALLOC_EXPORT void (*hooks_arena_new_hook)();
+extern JEMALLOC_EXPORT void (*hooks_libc_hook)();
+
+#define JEMALLOC_HOOK(fn, hook) ((void)(hook != NULL && (hook(), 0)), fn)
+
+#define open JEMALLOC_HOOK(open, hooks_libc_hook)
+#define read JEMALLOC_HOOK(read, hooks_libc_hook)
+#define write JEMALLOC_HOOK(write, hooks_libc_hook)
+#define readlink JEMALLOC_HOOK(readlink, hooks_libc_hook)
+#define close JEMALLOC_HOOK(close, hooks_libc_hook)
+#define creat JEMALLOC_HOOK(creat, hooks_libc_hook)
+#define secure_getenv JEMALLOC_HOOK(secure_getenv, hooks_libc_hook)
+/* Note that this is undef'd and re-define'd in src/prof.c. */
+#define _Unwind_Backtrace JEMALLOC_HOOK(_Unwind_Backtrace, hooks_libc_hook)
+
+#endif /* JEMALLOC_INTERNAL_HOOKS_H */

include/jemalloc/internal/huge.h

@@ -1,46 +0,0 @@
-/******************************************************************************/
-#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, dss_prec_t dss_prec);
-void	*huge_palloc(size_t size, size_t alignment, bool zero,
-    dss_prec_t dss_prec);
-bool	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, bool try_tcache_dalloc, dss_prec_t dss_prec);
-#ifdef JEMALLOC_JET
-typedef void (huge_dalloc_junk_t)(void *, size_t);
-extern huge_dalloc_junk_t *huge_dalloc_junk;
-#endif
-void	huge_dalloc(void *ptr, bool unmap);
-size_t	huge_salloc(const void *ptr);
-dss_prec_t	huge_dss_prec_get(arena_t *arena);
-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 */
-/******************************************************************************/

include/jemalloc/internal/jemalloc_internal_decls.h

@@ -0,0 +1,91 @@
+#ifndef JEMALLOC_INTERNAL_DECLS_H
+#define JEMALLOC_INTERNAL_DECLS_H
+
+#include <math.h>
+#ifdef _WIN32
+#  include <windows.h>
+#  include "msvc_compat/windows_extra.h"
+#  ifdef _WIN64
+#    if LG_VADDR <= 32
+#      error Generate the headers using x64 vcargs
+#    endif
+#  else
+#    if LG_VADDR > 32
+#      undef LG_VADDR
+#      define LG_VADDR 32
+#    endif
+#  endif
+#else
+#  include <sys/param.h>
+#  include <sys/mman.h>
+#  if !defined(__pnacl__) && !defined(__native_client__)
+#    include <sys/syscall.h>
+#    if !defined(SYS_write) && defined(__NR_write)
+#      define SYS_write __NR_write
+#    endif
+#    if defined(SYS_open) && defined(__aarch64__)
+       /* Android headers may define SYS_open to __NR_open even though
+        * __NR_open may not exist on AArch64 (superseded by __NR_openat). */
+#      undef SYS_open
+#    endif
+#    include <sys/uio.h>
+#  endif
+#  include <pthread.h>
+#  include <signal.h>
+#  ifdef JEMALLOC_OS_UNFAIR_LOCK
+#    include <os/lock.h>
+#  endif
+#  ifdef JEMALLOC_GLIBC_MALLOC_HOOK
+#    include <sched.h>
+#  endif
+#  include <errno.h>
+#  include <sys/time.h>
+#  include <time.h>
+#  ifdef JEMALLOC_HAVE_MACH_ABSOLUTE_TIME
+#    include <mach/mach_time.h>
+#  endif
+#endif
+#include <sys/types.h>
+
+#include <limits.h>
+#ifndef SIZE_T_MAX
+#  define SIZE_T_MAX	SIZE_MAX
+#endif
+#ifndef SSIZE_MAX
+#  define SSIZE_MAX	((ssize_t)(SIZE_T_MAX >> 1))
+#endif
+#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 <string.h>
+#include <strings.h>
+#include <ctype.h>
+#ifdef _MSC_VER
+#  include <io.h>
+typedef intptr_t ssize_t;
+#  define PATH_MAX 1024
+#  define STDERR_FILENO 2
+#  define __func__ __FUNCTION__
+#  ifdef JEMALLOC_HAS_RESTRICT
+#    define restrict __restrict
+#  endif
+/* Disable warnings about deprecated system functions. */
+#  pragma warning(disable: 4996)
+#if _MSC_VER < 1800
+static int
+isblank(int c) {
+	return (c == '\t' || c == ' ');
+}
+#endif
+#else
+#  include <unistd.h>
+#endif
+#include <fcntl.h>
+
+#endif /* JEMALLOC_INTERNAL_H */

include/jemalloc/internal/jemalloc_internal_defs.h.in

@@ -1,5 +1,5 @@
 #ifndef JEMALLOC_INTERNAL_DEFS_H_
-#define	JEMALLOC_INTERNAL_DEFS_H_
+#define JEMALLOC_INTERNAL_DEFS_H_
 /*
  * 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
@@ -8,6 +8,18 @@
 #undef JEMALLOC_PREFIX
 #undef JEMALLOC_CPREFIX
 
+/*
+ * Define overrides for non-standard allocator-related functions if they are
+ * present on the system.
+ */
+#undef JEMALLOC_OVERRIDE___LIBC_CALLOC
+#undef JEMALLOC_OVERRIDE___LIBC_FREE
+#undef JEMALLOC_OVERRIDE___LIBC_MALLOC
+#undef JEMALLOC_OVERRIDE___LIBC_MEMALIGN
+#undef JEMALLOC_OVERRIDE___LIBC_REALLOC
+#undef JEMALLOC_OVERRIDE___LIBC_VALLOC
+#undef JEMALLOC_OVERRIDE___POSIX_MEMALIGN
+
 /*
  * JEMALLOC_PRIVATE_NAMESPACE is used as a prefix for all library-private APIs.
  * For shared libraries, symbol visibility mechanisms prevent these symbols
@@ -21,21 +33,30 @@
  * order to yield to another virtual CPU.
  */
 #undef CPU_SPINWAIT
-
-/* Defined if the equivalent of FreeBSD's atomic(9) functions are available. */
-#undef JEMALLOC_ATOMIC9
+/* 1 if CPU_SPINWAIT is defined, 0 otherwise. */
+#undef HAVE_CPU_SPINWAIT
 
 /*
- * Defined if OSAtomic*() functions are available, as provided by Darwin, and
- * documented in the atomic(3) manual page.
+ * Number of significant bits in virtual addresses.  This may be less than the
+ * total number of bits in a pointer, e.g. on x64, for which the uppermost 16
+ * bits are the same as bit 47.
  */
-#undef JEMALLOC_OSATOMIC
+#undef LG_VADDR
+
+/* Defined if C11 atomics are available. */
+#undef JEMALLOC_C11_ATOMICS
+
+/* Defined if GCC __atomic atomics are available. */
+#undef JEMALLOC_GCC_ATOMIC_ATOMICS
+
+/* Defined if GCC __sync atomics are available. */
+#undef JEMALLOC_GCC_SYNC_ATOMICS
 
 /*
  * 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)
+ * functions are defined in libgcc instead of being inlines).
  */
 #undef JE_FORCE_SYNC_COMPARE_AND_SWAP_4
 
@@ -43,16 +64,60 @@
  * 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)
+ * functions are defined in libgcc instead of being inlines).
  */
 #undef JE_FORCE_SYNC_COMPARE_AND_SWAP_8
 
+/*
+ * Defined if __builtin_clz() and __builtin_clzl() are available.
+ */
+#undef JEMALLOC_HAVE_BUILTIN_CLZ
+
+/*
+ * Defined if os_unfair_lock_*() functions are available, as provided by Darwin.
+ */
+#undef JEMALLOC_OS_UNFAIR_LOCK
+
 /*
  * Defined if OSSpin*() functions are available, as provided by Darwin, and
  * documented in the spinlock(3) manual page.
  */
 #undef JEMALLOC_OSSPIN
 
+/* Defined if syscall(2) is usable. */
+#undef JEMALLOC_USE_SYSCALL
+
+/*
+ * Defined if secure_getenv(3) is available.
+ */
+#undef JEMALLOC_HAVE_SECURE_GETENV
+
+/*
+ * Defined if issetugid(2) is available.
+ */
+#undef JEMALLOC_HAVE_ISSETUGID
+
+/* Defined if pthread_atfork(3) is available. */
+#undef JEMALLOC_HAVE_PTHREAD_ATFORK
+
+/* Defined if pthread_setname_np(3) is available. */
+#undef JEMALLOC_HAVE_PTHREAD_SETNAME_NP
+
+/*
+ * Defined if clock_gettime(CLOCK_MONOTONIC_COARSE, ...) is available.
+ */
+#undef JEMALLOC_HAVE_CLOCK_MONOTONIC_COARSE
+
+/*
+ * Defined if clock_gettime(CLOCK_MONOTONIC, ...) is available.
+ */
+#undef JEMALLOC_HAVE_CLOCK_MONOTONIC
+
+/*
+ * Defined if mach_absolute_time() is available.
+ */
+#undef JEMALLOC_HAVE_MACH_ABSOLUTE_TIME
+
 /*
  * Defined if _malloc_thread_cleanup() exists.  At least in the case of
  * FreeBSD, pthread_key_create() allocates, which if used during malloc
@@ -76,18 +141,9 @@
  */
 #undef JEMALLOC_MUTEX_INIT_CB
 
-/* Defined if sbrk() is supported. */
-#undef JEMALLOC_HAVE_SBRK
-
 /* Non-empty if the tls_model attribute is supported. */
 #undef JEMALLOC_TLS_MODEL
 
-/* JEMALLOC_CC_SILENCE enables code that silences unuseful compiler warnings. */
-#undef JEMALLOC_CC_SILENCE
-
-/* JEMALLOC_CODE_COVERAGE enables test code coverage analysis. */
-#undef JEMALLOC_CODE_COVERAGE
-
 /*
  * JEMALLOC_DEBUG enables assertions and other sanity checks, and disables
  * inline functions.
@@ -110,81 +166,140 @@
 #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.
- */
-#undef JEMALLOC_TCACHE
-
-/*
- * JEMALLOC_DSS enables use of sbrk(2) to allocate chunks from the data storage
+ * JEMALLOC_DSS enables use of sbrk(2) to allocate extents from the data storage
  * segment (DSS).
  */
 #undef JEMALLOC_DSS
 
-/* Support memory filling (junk/zero/quarantine/redzone). */
+/* Support memory filling (junk/zero). */
 #undef JEMALLOC_FILL
 
 /* Support utrace(2)-based tracing. */
 #undef JEMALLOC_UTRACE
 
-/* Support Valgrind. */
-#undef JEMALLOC_VALGRIND
-
 /* Support optional abort() on OOM. */
 #undef JEMALLOC_XMALLOC
 
 /* Support lazy locking (avoid locking unless a second thread is launched). */
 #undef JEMALLOC_LAZY_LOCK
 
-/* One page is 2^STATIC_PAGE_SHIFT bytes. */
-#undef STATIC_PAGE_SHIFT
+/*
+ * Minimum allocation alignment is 2^LG_QUANTUM bytes (ignoring tiny size
+ * classes).
+ */
+#undef LG_QUANTUM
+
+/* One page is 2^LG_PAGE bytes. */
+#undef LG_PAGE
 
 /*
- * If defined, use munmap() to unmap freed chunks, rather than storing them for
- * later reuse.  This is disabled by default on Linux because common sequences
- * of mmap()/munmap() calls will cause virtual memory map holes.
+ * One huge page is 2^LG_HUGEPAGE bytes.  Note that this is defined even if the
+ * system does not explicitly support huge pages; system calls that require
+ * explicit huge page support are separately configured.
  */
-#undef JEMALLOC_MUNMAP
+#undef LG_HUGEPAGE
 
 /*
- * If defined, use mremap(...MREMAP_FIXED...) for huge realloc().  This is
- * disabled by default because it is Linux-specific and it will cause virtual
- * memory map holes, much like munmap(2) does.
+ * If defined, adjacent virtual memory mappings with identical attributes
+ * automatically coalesce, and they fragment when changes are made to subranges.
+ * This is the normal order of things for mmap()/munmap(), but on Windows
+ * VirtualAlloc()/VirtualFree() operations must be precisely matched, i.e.
+ * mappings do *not* coalesce/fragment.
  */
-#undef JEMALLOC_MREMAP
+#undef JEMALLOC_MAPS_COALESCE
+
+/*
+ * If defined, retain memory for later reuse by default rather than using e.g.
+ * munmap() to unmap freed extents.  This is enabled on 64-bit Linux because
+ * common sequences of mmap()/munmap() calls will cause virtual memory map
+ * holes.
+ */
+#undef JEMALLOC_RETAIN
 
 /* TLS is used to map arenas and magazine caches to threads. */
 #undef JEMALLOC_TLS
 
 /*
- * JEMALLOC_IVSALLOC enables ivsalloc(), which verifies that pointers reside
- * within jemalloc-owned chunks before dereferencing them.
+ * Used to mark unreachable code to quiet "end of non-void" compiler warnings.
+ * Don't use this directly; instead use unreachable() from util.h
  */
-#undef JEMALLOC_IVSALLOC
+#undef JEMALLOC_INTERNAL_UNREACHABLE
+
+/*
+ * ffs*() functions to use for bitmapping.  Don't use these directly; instead,
+ * use ffs_*() from util.h.
+ */
+#undef JEMALLOC_INTERNAL_FFSLL
+#undef JEMALLOC_INTERNAL_FFSL
+#undef JEMALLOC_INTERNAL_FFS
+
+/*
+ * If defined, explicitly attempt to more uniformly distribute large allocation
+ * pointer alignments across all cache indices.
+ */
+#undef JEMALLOC_CACHE_OBLIVIOUS
+
+/*
+ * If defined, enable logging facilities.  We make this a configure option to
+ * avoid taking extra branches everywhere.
+ */
+#undef JEMALLOC_LOG
 
 /*
  * Darwin (OS X) uses zones to work around Mach-O symbol override shortcomings.
  */
 #undef JEMALLOC_ZONE
-#undef JEMALLOC_ZONE_VERSION
+
+/*
+ * Methods for determining whether the OS overcommits.
+ * JEMALLOC_PROC_SYS_VM_OVERCOMMIT_MEMORY: Linux's
+ *                                         /proc/sys/vm.overcommit_memory file.
+ * JEMALLOC_SYSCTL_VM_OVERCOMMIT: FreeBSD's vm.overcommit sysctl.
+ */
+#undef JEMALLOC_SYSCTL_VM_OVERCOMMIT
+#undef JEMALLOC_PROC_SYS_VM_OVERCOMMIT_MEMORY
+
+/* Defined if madvise(2) is available. */
+#undef JEMALLOC_HAVE_MADVISE
+
+/*
+ * Defined if transparent huge pages are supported via the MADV_[NO]HUGEPAGE
+ * arguments to madvise(2).
+ */
+#undef JEMALLOC_HAVE_MADVISE_HUGE
 
 /*
  * Methods for purging unused pages differ between operating systems.
  *
- *   madvise(..., MADV_DONTNEED) : On Linux, this immediately discards pages,
+ *   madvise(..., MADV_FREE) : This marks pages as being unused, such that they
+ *                             will be discarded rather than swapped out.
+ *   madvise(..., MADV_DONTNEED) : If JEMALLOC_PURGE_MADVISE_DONTNEED_ZEROS is
+ *                                 defined, 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.
+ *                                 the address region is later touched;
+ *                                 otherwise this behaves similarly to
+ *                                 MADV_FREE, though typically with higher
+ *                                 system overhead.
  */
-#undef JEMALLOC_PURGE_MADVISE_DONTNEED
 #undef JEMALLOC_PURGE_MADVISE_FREE
+#undef JEMALLOC_PURGE_MADVISE_DONTNEED
+#undef JEMALLOC_PURGE_MADVISE_DONTNEED_ZEROS
+
+/* Defined if madvise(2) is available but MADV_FREE is not (x86 Linux only). */
+#undef JEMALLOC_DEFINE_MADVISE_FREE
 
 /*
- * Define if operating system has alloca.h header.
+ * Defined if MADV_DO[NT]DUMP is supported as an argument to madvise.
  */
+#undef JEMALLOC_MADVISE_DONTDUMP
+
+/*
+ * Defined if transparent huge pages (THPs) are supported via the
+ * MADV_[NO]HUGEPAGE arguments to madvise(2), and THP support is enabled.
+ */
+#undef JEMALLOC_THP
+
+/* Define if operating system has alloca.h header. */
 #undef JEMALLOC_HAS_ALLOCA_H
 
 /* C99 restrict keyword supported. */
@@ -199,7 +314,53 @@
 /* sizeof(long) == 2^LG_SIZEOF_LONG. */
 #undef LG_SIZEOF_LONG
 
+/* sizeof(long long) == 2^LG_SIZEOF_LONG_LONG. */
+#undef LG_SIZEOF_LONG_LONG
+
 /* sizeof(intmax_t) == 2^LG_SIZEOF_INTMAX_T. */
 #undef LG_SIZEOF_INTMAX_T
 
+/* glibc malloc hooks (__malloc_hook, __realloc_hook, __free_hook). */
+#undef JEMALLOC_GLIBC_MALLOC_HOOK
+
+/* glibc memalign hook. */
+#undef JEMALLOC_GLIBC_MEMALIGN_HOOK
+
+/* pthread support */
+#undef JEMALLOC_HAVE_PTHREAD
+
+/* dlsym() support */
+#undef JEMALLOC_HAVE_DLSYM
+
+/* Adaptive mutex support in pthreads. */
+#undef JEMALLOC_HAVE_PTHREAD_MUTEX_ADAPTIVE_NP
+
+/* GNU specific sched_getcpu support */
+#undef JEMALLOC_HAVE_SCHED_GETCPU
+
+/* GNU specific sched_setaffinity support */
+#undef JEMALLOC_HAVE_SCHED_SETAFFINITY
+
+/*
+ * If defined, all the features necessary for background threads are present.
+ */
+#undef JEMALLOC_BACKGROUND_THREAD
+
+/*
+ * If defined, jemalloc symbols are not exported (doesn't work when
+ * JEMALLOC_PREFIX is not defined).
+ */
+#undef JEMALLOC_EXPORT
+
+/* config.malloc_conf options string. */
+#undef JEMALLOC_CONFIG_MALLOC_CONF
+
+/* If defined, jemalloc takes the malloc/free/etc. symbol names. */
+#undef JEMALLOC_IS_MALLOC
+
+/*
+ * Defined if strerror_r returns char * if _GNU_SOURCE is defined.
+ */
+#undef JEMALLOC_STRERROR_R_RETURNS_CHAR_WITH_GNU_SOURCE
+
 #endif /* JEMALLOC_INTERNAL_DEFS_H_ */

include/jemalloc/internal/jemalloc_internal_externs.h

@@ -0,0 +1,53 @@
+#ifndef JEMALLOC_INTERNAL_EXTERNS_H
+#define JEMALLOC_INTERNAL_EXTERNS_H
+
+#include "jemalloc/internal/atomic.h"
+#include "jemalloc/internal/size_classes.h"
+#include "jemalloc/internal/tsd_types.h"
+
+/* TSD checks this to set thread local slow state accordingly. */
+extern bool malloc_slow;
+
+/* Run-time options. */
+extern bool opt_abort;
+extern bool opt_abort_conf;
+extern const char *opt_junk;
+extern bool opt_junk_alloc;
+extern bool opt_junk_free;
+extern bool opt_utrace;
+extern bool opt_xmalloc;
+extern bool opt_zero;
+extern unsigned opt_narenas;
+
+/* Number of CPUs. */
+extern unsigned ncpus;
+
+/* Number of arenas used for automatic multiplexing of threads and arenas. */
+extern unsigned narenas_auto;
+
+/*
+ * 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 atomic_p_t arenas[];
+
+void *a0malloc(size_t size);
+void a0dalloc(void *ptr);
+void *bootstrap_malloc(size_t size);
+void *bootstrap_calloc(size_t num, size_t size);
+void bootstrap_free(void *ptr);
+void arena_set(unsigned ind, arena_t *arena);
+unsigned narenas_total_get(void);
+arena_t *arena_init(tsdn_t *tsdn, unsigned ind, extent_hooks_t *extent_hooks);
+arena_tdata_t *arena_tdata_get_hard(tsd_t *tsd, unsigned ind);
+arena_t *arena_choose_hard(tsd_t *tsd, bool internal);
+void arena_migrate(tsd_t *tsd, unsigned oldind, unsigned newind);
+void iarena_cleanup(tsd_t *tsd);
+void arena_cleanup(tsd_t *tsd);
+void arenas_tdata_cleanup(tsd_t *tsd);
+void jemalloc_prefork(void);
+void jemalloc_postfork_parent(void);
+void jemalloc_postfork_child(void);
+bool malloc_initialized(void);
+
+#endif /* JEMALLOC_INTERNAL_EXTERNS_H */

include/jemalloc/internal/jemalloc_internal_includes.h

@@ -0,0 +1,94 @@
+#ifndef JEMALLOC_INTERNAL_INCLUDES_H
+#define JEMALLOC_INTERNAL_INCLUDES_H
+
+/*
+ * jemalloc can conceptually be broken into components (arena, tcache, etc.),
+ * but there are circular dependencies that cannot be broken without
+ * substantial performance degradation.
+ *
+ * Historically, we dealt with this by each header into four sections (types,
+ * structs, externs, and inlines), and included each header file multiple times
+ * in this file, picking out the portion we want on each pass using the
+ * following #defines:
+ *   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.
+ *
+ * We're moving toward a world in which the dependencies are explicit; each file
+ * will #include the headers it depends on (rather than relying on them being
+ * implicitly available via this file including every header file in the
+ * project).
+ *
+ * We're now in an intermediate state: we've broken up the header files to avoid
+ * having to include each one multiple times, but have not yet moved the
+ * dependency information into the header files (i.e. we still rely on the
+ * ordering in this file to ensure all a header's dependencies are available in
+ * its translation unit).  Each component is now broken up into multiple header
+ * files, corresponding to the sections above (e.g. instead of "foo.h", we now
+ * have "foo_types.h", "foo_structs.h", "foo_externs.h", "foo_inlines.h").
+ *
+ * Those files which have been converted to explicitly include their
+ * inter-component dependencies are now in the initial HERMETIC HEADERS
+ * section.  All headers may still rely on jemalloc_preamble.h (which, by fiat,
+ * must be included first in every translation unit) for system headers and
+ * global jemalloc definitions, however.
+ */
+
+/******************************************************************************/
+/* TYPES */
+/******************************************************************************/
+
+#include "jemalloc/internal/extent_types.h"
+#include "jemalloc/internal/base_types.h"
+#include "jemalloc/internal/arena_types.h"
+#include "jemalloc/internal/tcache_types.h"
+#include "jemalloc/internal/prof_types.h"
+
+/******************************************************************************/
+/* STRUCTS */
+/******************************************************************************/
+
+#include "jemalloc/internal/arena_structs_a.h"
+#include "jemalloc/internal/extent_structs.h"
+#include "jemalloc/internal/base_structs.h"
+#include "jemalloc/internal/prof_structs.h"
+#include "jemalloc/internal/arena_structs_b.h"
+#include "jemalloc/internal/tcache_structs.h"
+#include "jemalloc/internal/background_thread_structs.h"
+
+/******************************************************************************/
+/* EXTERNS */
+/******************************************************************************/
+
+#include "jemalloc/internal/jemalloc_internal_externs.h"
+#include "jemalloc/internal/extent_externs.h"
+#include "jemalloc/internal/base_externs.h"
+#include "jemalloc/internal/arena_externs.h"
+#include "jemalloc/internal/large_externs.h"
+#include "jemalloc/internal/tcache_externs.h"
+#include "jemalloc/internal/prof_externs.h"
+#include "jemalloc/internal/background_thread_externs.h"
+
+/******************************************************************************/
+/* INLINES */
+/******************************************************************************/
+
+#include "jemalloc/internal/jemalloc_internal_inlines_a.h"
+#include "jemalloc/internal/base_inlines.h"
+/*
+ * Include portions of arena code interleaved with tcache code in order to
+ * resolve circular dependencies.
+ */
+#include "jemalloc/internal/prof_inlines_a.h"
+#include "jemalloc/internal/arena_inlines_a.h"
+#include "jemalloc/internal/extent_inlines.h"
+#include "jemalloc/internal/jemalloc_internal_inlines_b.h"
+#include "jemalloc/internal/tcache_inlines.h"
+#include "jemalloc/internal/arena_inlines_b.h"
+#include "jemalloc/internal/jemalloc_internal_inlines_c.h"
+#include "jemalloc/internal/prof_inlines_b.h"
+#include "jemalloc/internal/background_thread_inlines.h"
+
+#endif /* JEMALLOC_INTERNAL_INCLUDES_H */

include/jemalloc/internal/jemalloc_internal_inlines_a.h

@@ -0,0 +1,172 @@
+#ifndef JEMALLOC_INTERNAL_INLINES_A_H
+#define JEMALLOC_INTERNAL_INLINES_A_H
+
+#include "jemalloc/internal/atomic.h"
+#include "jemalloc/internal/bit_util.h"
+#include "jemalloc/internal/jemalloc_internal_types.h"
+#include "jemalloc/internal/size_classes.h"
+#include "jemalloc/internal/ticker.h"
+
+JEMALLOC_ALWAYS_INLINE malloc_cpuid_t
+malloc_getcpu(void) {
+	assert(have_percpu_arena);
+#if defined(JEMALLOC_HAVE_SCHED_GETCPU)
+	return (malloc_cpuid_t)sched_getcpu();
+#else
+	not_reached();
+	return -1;
+#endif
+}
+
+/* Return the chosen arena index based on current cpu. */
+JEMALLOC_ALWAYS_INLINE unsigned
+percpu_arena_choose(void) {
+	assert(have_percpu_arena && PERCPU_ARENA_ENABLED(opt_percpu_arena));
+
+	malloc_cpuid_t cpuid = malloc_getcpu();
+	assert(cpuid >= 0);
+
+	unsigned arena_ind;
+	if ((opt_percpu_arena == percpu_arena) || ((unsigned)cpuid < ncpus /
+	    2)) {
+		arena_ind = cpuid;
+	} else {
+		assert(opt_percpu_arena == per_phycpu_arena);
+		/* Hyper threads on the same physical CPU share arena. */
+		arena_ind = cpuid - ncpus / 2;
+	}
+
+	return arena_ind;
+}
+
+/* Return the limit of percpu auto arena range, i.e. arenas[0...ind_limit). */
+JEMALLOC_ALWAYS_INLINE unsigned
+percpu_arena_ind_limit(percpu_arena_mode_t mode) {
+	assert(have_percpu_arena && PERCPU_ARENA_ENABLED(mode));
+	if (mode == per_phycpu_arena && ncpus > 1) {
+		if (ncpus % 2) {
+			/* This likely means a misconfig. */
+			return ncpus / 2 + 1;
+		}
+		return ncpus / 2;
+	} else {
+		return ncpus;
+	}
+}
+
+static inline arena_tdata_t *
+arena_tdata_get(tsd_t *tsd, unsigned ind, bool refresh_if_missing) {
+	arena_tdata_t *tdata;
+	arena_tdata_t *arenas_tdata = tsd_arenas_tdata_get(tsd);
+
+	if (unlikely(arenas_tdata == NULL)) {
+		/* arenas_tdata hasn't been initialized yet. */
+		return arena_tdata_get_hard(tsd, ind);
+	}
+	if (unlikely(ind >= tsd_narenas_tdata_get(tsd))) {
+		/*
+		 * ind is invalid, cache is old (too small), or tdata to be
+		 * initialized.
+		 */
+		return (refresh_if_missing ? arena_tdata_get_hard(tsd, ind) :
+		    NULL);
+	}
+
+	tdata = &arenas_tdata[ind];
+	if (likely(tdata != NULL) || !refresh_if_missing) {
+		return tdata;
+	}
+	return arena_tdata_get_hard(tsd, ind);
+}
+
+static inline arena_t *
+arena_get(tsdn_t *tsdn, unsigned ind, bool init_if_missing) {
+	arena_t *ret;
+
+	assert(ind < MALLOCX_ARENA_LIMIT);
+
+	ret = (arena_t *)atomic_load_p(&arenas[ind], ATOMIC_ACQUIRE);
+	if (unlikely(ret == NULL)) {
+		if (init_if_missing) {
+			ret = arena_init(tsdn, ind,
+			    (extent_hooks_t *)&extent_hooks_default);
+		}
+	}
+	return ret;
+}
+
+static inline ticker_t *
+decay_ticker_get(tsd_t *tsd, unsigned ind) {
+	arena_tdata_t *tdata;
+
+	tdata = arena_tdata_get(tsd, ind, true);
+	if (unlikely(tdata == NULL)) {
+		return NULL;
+	}
+	return &tdata->decay_ticker;
+}
+
+JEMALLOC_ALWAYS_INLINE cache_bin_t *
+tcache_small_bin_get(tcache_t *tcache, szind_t binind) {
+	assert(binind < NBINS);
+	return &tcache->bins_small[binind];
+}
+
+JEMALLOC_ALWAYS_INLINE cache_bin_t *
+tcache_large_bin_get(tcache_t *tcache, szind_t binind) {
+	assert(binind >= NBINS &&binind < nhbins);
+	return &tcache->bins_large[binind - NBINS];
+}
+
+JEMALLOC_ALWAYS_INLINE bool
+tcache_available(tsd_t *tsd) {
+	/*
+	 * Thread specific auto tcache might be unavailable if: 1) during tcache
+	 * initialization, or 2) disabled through thread.tcache.enabled mallctl
+	 * or config options.  This check covers all cases.
+	 */
+	if (likely(tsd_tcache_enabled_get(tsd))) {
+		/* Associated arena == NULL implies tcache init in progress. */
+		assert(tsd_tcachep_get(tsd)->arena == NULL ||
+		    tcache_small_bin_get(tsd_tcachep_get(tsd), 0)->avail !=
+		    NULL);
+		return true;
+	}
+
+	return false;
+}
+
+JEMALLOC_ALWAYS_INLINE tcache_t *
+tcache_get(tsd_t *tsd) {
+	if (!tcache_available(tsd)) {
+		return NULL;
+	}
+
+	return tsd_tcachep_get(tsd);
+}
+
+static inline void
+pre_reentrancy(tsd_t *tsd, arena_t *arena) {
+	/* arena is the current context.  Reentry from a0 is not allowed. */
+	assert(arena != arena_get(tsd_tsdn(tsd), 0, false));
+
+	bool fast = tsd_fast(tsd);
+	assert(tsd_reentrancy_level_get(tsd) < INT8_MAX);
+	++*tsd_reentrancy_levelp_get(tsd);
+	if (fast) {
+		/* Prepare slow path for reentrancy. */
+		tsd_slow_update(tsd);
+		assert(tsd->state == tsd_state_nominal_slow);
+	}
+}
+
+static inline void
+post_reentrancy(tsd_t *tsd) {
+	int8_t *reentrancy_level = tsd_reentrancy_levelp_get(tsd);
+	assert(*reentrancy_level > 0);
+	if (--*reentrancy_level == 0) {
+		tsd_slow_update(tsd);
+	}
+}
+
+#endif /* JEMALLOC_INTERNAL_INLINES_A_H */

include/jemalloc/internal/jemalloc_internal_inlines_b.h

@@ -0,0 +1,86 @@
+#ifndef JEMALLOC_INTERNAL_INLINES_B_H
+#define JEMALLOC_INTERNAL_INLINES_B_H
+
+#include "jemalloc/internal/rtree.h"
+
+/* Choose an arena based on a per-thread value. */
+static inline arena_t *
+arena_choose_impl(tsd_t *tsd, arena_t *arena, bool internal) {
+	arena_t *ret;
+
+	if (arena != NULL) {
+		return arena;
+	}
+
+	/* During reentrancy, arena 0 is the safest bet. */
+	if (unlikely(tsd_reentrancy_level_get(tsd) > 0)) {
+		return arena_get(tsd_tsdn(tsd), 0, true);
+	}
+
+	ret = internal ? tsd_iarena_get(tsd) : tsd_arena_get(tsd);
+	if (unlikely(ret == NULL)) {
+		ret = arena_choose_hard(tsd, internal);
+		assert(ret);
+		if (tcache_available(tsd)) {
+			tcache_t *tcache = tcache_get(tsd);
+			if (tcache->arena != NULL) {
+				/* See comments in tcache_data_init().*/
+				assert(tcache->arena ==
+				    arena_get(tsd_tsdn(tsd), 0, false));
+				if (tcache->arena != ret) {
+					tcache_arena_reassociate(tsd_tsdn(tsd),
+					    tcache, ret);
+				}
+			} else {
+				tcache_arena_associate(tsd_tsdn(tsd), tcache,
+				    ret);
+			}
+		}
+	}
+
+	/*
+	 * Note that for percpu arena, if the current arena is outside of the
+	 * auto percpu arena range, (i.e. thread is assigned to a manually
+	 * managed arena), then percpu arena is skipped.
+	 */
+	if (have_percpu_arena && PERCPU_ARENA_ENABLED(opt_percpu_arena) &&
+	    !internal && (arena_ind_get(ret) <
+	    percpu_arena_ind_limit(opt_percpu_arena)) && (ret->last_thd !=
+	    tsd_tsdn(tsd))) {
+		unsigned ind = percpu_arena_choose();
+		if (arena_ind_get(ret) != ind) {
+			percpu_arena_update(tsd, ind);
+			ret = tsd_arena_get(tsd);
+		}
+		ret->last_thd = tsd_tsdn(tsd);
+	}
+
+	return ret;
+}
+
+static inline arena_t *
+arena_choose(tsd_t *tsd, arena_t *arena) {
+	return arena_choose_impl(tsd, arena, false);
+}
+
+static inline arena_t *
+arena_ichoose(tsd_t *tsd, arena_t *arena) {
+	return arena_choose_impl(tsd, arena, true);
+}
+
+static inline bool
+arena_is_auto(arena_t *arena) {
+	assert(narenas_auto > 0);
+	return (arena_ind_get(arena) < narenas_auto);
+}
+
+JEMALLOC_ALWAYS_INLINE extent_t *
+iealloc(tsdn_t *tsdn, const void *ptr) {
+	rtree_ctx_t rtree_ctx_fallback;
+	rtree_ctx_t *rtree_ctx = tsdn_rtree_ctx(tsdn, &rtree_ctx_fallback);
+
+	return rtree_extent_read(tsdn, &extents_rtree, rtree_ctx,
+	    (uintptr_t)ptr, true);
+}
+
+#endif /* JEMALLOC_INTERNAL_INLINES_B_H */

include/jemalloc/internal/jemalloc_internal_inlines_c.h

@@ -0,0 +1,218 @@
+#ifndef JEMALLOC_INTERNAL_INLINES_C_H
+#define JEMALLOC_INTERNAL_INLINES_C_H
+
+#include "jemalloc/internal/jemalloc_internal_types.h"
+#include "jemalloc/internal/sz.h"
+#include "jemalloc/internal/witness.h"
+
+/*
+ * Translating the names of the 'i' functions:
+ *   Abbreviations used in the first part of the function name (before
+ *   alloc/dalloc) describe what that function accomplishes:
+ *     a: arena (query)
+ *     s: size (query, or sized deallocation)
+ *     e: extent (query)
+ *     p: aligned (allocates)
+ *     vs: size (query, without knowing that the pointer is into the heap)
+ *     r: rallocx implementation
+ *     x: xallocx implementation
+ *   Abbreviations used in the second part of the function name (after
+ *   alloc/dalloc) describe the arguments it takes
+ *     z: whether to return zeroed memory
+ *     t: accepts a tcache_t * parameter
+ *     m: accepts an arena_t * parameter
+ */
+
+JEMALLOC_ALWAYS_INLINE arena_t *
+iaalloc(tsdn_t *tsdn, const void *ptr) {
+	assert(ptr != NULL);
+
+	return arena_aalloc(tsdn, ptr);
+}
+
+JEMALLOC_ALWAYS_INLINE size_t
+isalloc(tsdn_t *tsdn, const void *ptr) {
+	assert(ptr != NULL);
+
+	return arena_salloc(tsdn, ptr);
+}
+
+JEMALLOC_ALWAYS_INLINE void *
+iallocztm(tsdn_t *tsdn, size_t size, szind_t ind, bool zero, tcache_t *tcache,
+    bool is_internal, arena_t *arena, bool slow_path) {
+	void *ret;
+
+	assert(size != 0);
+	assert(!is_internal || tcache == NULL);
+	assert(!is_internal || arena == NULL || arena_is_auto(arena));
+	if (!tsdn_null(tsdn) && tsd_reentrancy_level_get(tsdn_tsd(tsdn)) == 0) {
+		witness_assert_depth_to_rank(tsdn_witness_tsdp_get(tsdn),
+		    WITNESS_RANK_CORE, 0);
+	}
+
+	ret = arena_malloc(tsdn, arena, size, ind, zero, tcache, slow_path);
+	if (config_stats && is_internal && likely(ret != NULL)) {
+		arena_internal_add(iaalloc(tsdn, ret), isalloc(tsdn, ret));
+	}
+	return ret;
+}
+
+JEMALLOC_ALWAYS_INLINE void *
+ialloc(tsd_t *tsd, size_t size, szind_t ind, bool zero, bool slow_path) {
+	return iallocztm(tsd_tsdn(tsd), size, ind, zero, tcache_get(tsd), false,
+	    NULL, slow_path);
+}
+
+JEMALLOC_ALWAYS_INLINE void *
+ipallocztm(tsdn_t *tsdn, size_t usize, size_t alignment, bool zero,
+    tcache_t *tcache, bool is_internal, arena_t *arena) {
+	void *ret;
+
+	assert(usize != 0);
+	assert(usize == sz_sa2u(usize, alignment));
+	assert(!is_internal || tcache == NULL);
+	assert(!is_internal || arena == NULL || arena_is_auto(arena));
+	witness_assert_depth_to_rank(tsdn_witness_tsdp_get(tsdn),
+	    WITNESS_RANK_CORE, 0);
+
+	ret = arena_palloc(tsdn, arena, usize, alignment, zero, tcache);
+	assert(ALIGNMENT_ADDR2BASE(ret, alignment) == ret);
+	if (config_stats && is_internal && likely(ret != NULL)) {
+		arena_internal_add(iaalloc(tsdn, ret), isalloc(tsdn, ret));
+	}
+	return ret;
+}
+
+JEMALLOC_ALWAYS_INLINE void *
+ipalloct(tsdn_t *tsdn, size_t usize, size_t alignment, bool zero,
+    tcache_t *tcache, arena_t *arena) {
+	return ipallocztm(tsdn, usize, alignment, zero, tcache, false, arena);
+}
+
+JEMALLOC_ALWAYS_INLINE void *
+ipalloc(tsd_t *tsd, size_t usize, size_t alignment, bool zero) {
+	return ipallocztm(tsd_tsdn(tsd), usize, alignment, zero,
+	    tcache_get(tsd), false, NULL);
+}
+
+JEMALLOC_ALWAYS_INLINE size_t
+ivsalloc(tsdn_t *tsdn, const void *ptr) {
+	return arena_vsalloc(tsdn, ptr);
+}
+
+JEMALLOC_ALWAYS_INLINE void
+idalloctm(tsdn_t *tsdn, void *ptr, tcache_t *tcache, alloc_ctx_t *alloc_ctx,
+    bool is_internal, bool slow_path) {
+	assert(ptr != NULL);
+	assert(!is_internal || tcache == NULL);
+	assert(!is_internal || arena_is_auto(iaalloc(tsdn, ptr)));
+	witness_assert_depth_to_rank(tsdn_witness_tsdp_get(tsdn),
+	    WITNESS_RANK_CORE, 0);
+	if (config_stats && is_internal) {
+		arena_internal_sub(iaalloc(tsdn, ptr), isalloc(tsdn, ptr));
+	}
+	if (!is_internal && !tsdn_null(tsdn) &&
+	    tsd_reentrancy_level_get(tsdn_tsd(tsdn)) != 0) {
+		assert(tcache == NULL);
+	}
+	arena_dalloc(tsdn, ptr, tcache, alloc_ctx, slow_path);
+}
+
+JEMALLOC_ALWAYS_INLINE void
+idalloc(tsd_t *tsd, void *ptr) {
+	idalloctm(tsd_tsdn(tsd), ptr, tcache_get(tsd), NULL, false, true);
+}
+
+JEMALLOC_ALWAYS_INLINE void
+isdalloct(tsdn_t *tsdn, void *ptr, size_t size, tcache_t *tcache,
+    alloc_ctx_t *alloc_ctx, bool slow_path) {
+	witness_assert_depth_to_rank(tsdn_witness_tsdp_get(tsdn),
+	    WITNESS_RANK_CORE, 0);
+	arena_sdalloc(tsdn, ptr, size, tcache, alloc_ctx, slow_path);
+}
+
+JEMALLOC_ALWAYS_INLINE void *
+iralloct_realign(tsdn_t *tsdn, void *ptr, size_t oldsize, size_t size,
+    size_t extra, size_t alignment, bool zero, tcache_t *tcache,
+    arena_t *arena) {
+	witness_assert_depth_to_rank(tsdn_witness_tsdp_get(tsdn),
+	    WITNESS_RANK_CORE, 0);
+	void *p;
+	size_t usize, copysize;
+
+	usize = sz_sa2u(size + extra, alignment);
+	if (unlikely(usize == 0 || usize > LARGE_MAXCLASS)) {
+		return NULL;
+	}
+	p = ipalloct(tsdn, usize, alignment, zero, tcache, arena);
+	if (p == NULL) {
+		if (extra == 0) {
+			return NULL;
+		}
+		/* Try again, without extra this time. */
+		usize = sz_sa2u(size, alignment);
+		if (unlikely(usize == 0 || usize > LARGE_MAXCLASS)) {
+			return NULL;
+		}
+		p = ipalloct(tsdn, usize, alignment, zero, tcache, arena);
+		if (p == 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(p, ptr, copysize);
+	isdalloct(tsdn, ptr, oldsize, tcache, NULL, true);
+	return p;
+}
+
+JEMALLOC_ALWAYS_INLINE void *
+iralloct(tsdn_t *tsdn, void *ptr, size_t oldsize, size_t size, size_t alignment,
+    bool zero, tcache_t *tcache, arena_t *arena) {
+	assert(ptr != NULL);
+	assert(size != 0);
+	witness_assert_depth_to_rank(tsdn_witness_tsdp_get(tsdn),
+	    WITNESS_RANK_CORE, 0);
+
+	if (alignment != 0 && ((uintptr_t)ptr & ((uintptr_t)alignment-1))
+	    != 0) {
+		/*
+		 * Existing object alignment is inadequate; allocate new space
+		 * and copy.
+		 */
+		return iralloct_realign(tsdn, ptr, oldsize, size, 0, alignment,
+		    zero, tcache, arena);
+	}
+
+	return arena_ralloc(tsdn, arena, ptr, oldsize, size, alignment, zero,
+	    tcache);
+}
+
+JEMALLOC_ALWAYS_INLINE void *
+iralloc(tsd_t *tsd, void *ptr, size_t oldsize, size_t size, size_t alignment,
+    bool zero) {
+	return iralloct(tsd_tsdn(tsd), ptr, oldsize, size, alignment, zero,
+	    tcache_get(tsd), NULL);
+}
+
+JEMALLOC_ALWAYS_INLINE bool
+ixalloc(tsdn_t *tsdn, void *ptr, size_t oldsize, size_t size, size_t extra,
+    size_t alignment, bool zero) {
+	assert(ptr != NULL);
+	assert(size != 0);
+	witness_assert_depth_to_rank(tsdn_witness_tsdp_get(tsdn),
+	    WITNESS_RANK_CORE, 0);
+
+	if (alignment != 0 && ((uintptr_t)ptr & ((uintptr_t)alignment-1))
+	    != 0) {
+		/* Existing object alignment is inadequate. */
+		return true;
+	}
+
+	return arena_ralloc_no_move(tsdn, ptr, oldsize, size, extra, zero);
+}
+
+#endif /* JEMALLOC_INTERNAL_INLINES_C_H */

include/jemalloc/internal/jemalloc_internal_macros.h

@@ -1,51 +1,43 @@
-/*
- * JEMALLOC_ALWAYS_INLINE and JEMALLOC_INLINE are used within header files for
- * functions that are static inline functions if inlining is enabled, and
- * single-definition library-private functions if inlining is disabled.
- *
- * JEMALLOC_ALWAYS_INLINE_C and JEMALLOC_INLINE_C are for use in .c files, in
- * which case the denoted functions are always static, regardless of whether
- * inlining is enabled.
- */
-#if defined(JEMALLOC_DEBUG) || defined(JEMALLOC_CODE_COVERAGE)
-   /* Disable inlining to make debugging/profiling easier. */
-#  define JEMALLOC_ALWAYS_INLINE
-#  define JEMALLOC_ALWAYS_INLINE_C static
-#  define JEMALLOC_INLINE
-#  define JEMALLOC_INLINE_C static
-#  define inline
+#ifndef JEMALLOC_INTERNAL_MACROS_H
+#define JEMALLOC_INTERNAL_MACROS_H
+
+#ifdef JEMALLOC_DEBUG
+#  define JEMALLOC_ALWAYS_INLINE static inline
 #else
-#  define JEMALLOC_ENABLE_INLINE
-#  ifdef JEMALLOC_HAVE_ATTR
-#    define JEMALLOC_ALWAYS_INLINE \
-	 static inline JEMALLOC_ATTR(unused) JEMALLOC_ATTR(always_inline)
-#    define JEMALLOC_ALWAYS_INLINE_C \
-	 static inline JEMALLOC_ATTR(always_inline)
-#  else
-#    define JEMALLOC_ALWAYS_INLINE static inline
-#    define JEMALLOC_ALWAYS_INLINE_C static inline
-#  endif
-#  define JEMALLOC_INLINE static inline
-#  define JEMALLOC_INLINE_C static inline
-#  ifdef _MSC_VER
-#    define inline _inline
-#  endif
+#  define JEMALLOC_ALWAYS_INLINE JEMALLOC_ATTR(always_inline) static inline
+#endif
+#ifdef _MSC_VER
+#  define inline _inline
 #endif
 
-#ifdef JEMALLOC_CC_SILENCE
-#  define UNUSED JEMALLOC_ATTR(unused)
-#else
-#  define UNUSED
-#endif
+#define UNUSED JEMALLOC_ATTR(unused)
 
-#define	ZU(z)	((size_t)z)
-#define	QU(q)	((uint64_t)q)
-#define	QI(q)	((int64_t)q)
+#define ZU(z)	((size_t)z)
+#define ZD(z)	((ssize_t)z)
+#define QU(q)	((uint64_t)q)
+#define QD(q)	((int64_t)q)
+
+#define KZU(z)	ZU(z##ULL)
+#define KZD(z)	ZD(z##LL)
+#define KQU(q)	QU(q##ULL)
+#define KQD(q)	QI(q##LL)
 
 #ifndef __DECONST
 #  define	__DECONST(type, var)	((type)(uintptr_t)(const void *)(var))
 #endif
 
-#ifndef JEMALLOC_HAS_RESTRICT
+#if !defined(JEMALLOC_HAS_RESTRICT) || defined(__cplusplus)
 #  define restrict
 #endif
+
+/* Various function pointers are statick and immutable except during testing. */
+#ifdef JEMALLOC_JET
+#  define JET_MUTABLE
+#else
+#  define JET_MUTABLE const
+#endif
+
+#define JEMALLOC_VA_ARGS_HEAD(head, ...) head
+#define JEMALLOC_VA_ARGS_TAIL(head, ...) __VA_ARGS__
+
+#endif /* JEMALLOC_INTERNAL_MACROS_H */

include/jemalloc/internal/jemalloc_internal_types.h

@@ -0,0 +1,185 @@
+#ifndef JEMALLOC_INTERNAL_TYPES_H
+#define JEMALLOC_INTERNAL_TYPES_H
+
+/* Page size index type. */
+typedef unsigned pszind_t;
+
+/* Size class index type. */
+typedef unsigned szind_t;
+
+/* Processor / core id type. */
+typedef int malloc_cpuid_t;
+
+/*
+ * Flags bits:
+ *
+ * a: arena
+ * t: tcache
+ * 0: unused
+ * z: zero
+ * n: alignment
+ *
+ * aaaaaaaa aaaatttt tttttttt 0znnnnnn
+ */
+#define MALLOCX_ARENA_BITS	12
+#define MALLOCX_TCACHE_BITS	12
+#define MALLOCX_LG_ALIGN_BITS	6
+#define MALLOCX_ARENA_SHIFT	20
+#define MALLOCX_TCACHE_SHIFT	8
+#define MALLOCX_ARENA_MASK \
+    (((1 << MALLOCX_ARENA_BITS) - 1) << MALLOCX_ARENA_SHIFT)
+/* NB: Arena index bias decreases the maximum number of arenas by 1. */
+#define MALLOCX_ARENA_LIMIT	((1 << MALLOCX_ARENA_BITS) - 1)
+#define MALLOCX_TCACHE_MASK \
+    (((1 << MALLOCX_TCACHE_BITS) - 1) << MALLOCX_TCACHE_SHIFT)
+#define MALLOCX_TCACHE_MAX	((1 << MALLOCX_TCACHE_BITS) - 3)
+#define MALLOCX_LG_ALIGN_MASK	((1 << MALLOCX_LG_ALIGN_BITS) - 1)
+/* Use MALLOCX_ALIGN_GET() if alignment may not be specified in flags. */
+#define MALLOCX_ALIGN_GET_SPECIFIED(flags)				\
+    (ZU(1) << (flags & MALLOCX_LG_ALIGN_MASK))
+#define MALLOCX_ALIGN_GET(flags)					\
+    (MALLOCX_ALIGN_GET_SPECIFIED(flags) & (SIZE_T_MAX-1))
+#define MALLOCX_ZERO_GET(flags)						\
+    ((bool)(flags & MALLOCX_ZERO))
+
+#define MALLOCX_TCACHE_GET(flags)					\
+    (((unsigned)((flags & MALLOCX_TCACHE_MASK) >> MALLOCX_TCACHE_SHIFT)) - 2)
+#define MALLOCX_ARENA_GET(flags)					\
+    (((unsigned)(((unsigned)flags) >> MALLOCX_ARENA_SHIFT)) - 1)
+
+/* Smallest size class to support. */
+#define TINY_MIN		(1U << LG_TINY_MIN)
+
+/*
+ * Minimum allocation alignment is 2^LG_QUANTUM bytes (ignoring tiny size
+ * classes).
+ */
+#ifndef LG_QUANTUM
+#  if (defined(__i386__) || defined(_M_IX86))
+#    define LG_QUANTUM		4
+#  endif
+#  ifdef __ia64__
+#    define LG_QUANTUM		4
+#  endif
+#  ifdef __alpha__
+#    define LG_QUANTUM		4
+#  endif
+#  if (defined(__sparc64__) || defined(__sparcv9) || defined(__sparc_v9__))
+#    define LG_QUANTUM		4
+#  endif
+#  if (defined(__amd64__) || defined(__x86_64__) || defined(_M_X64))
+#    define LG_QUANTUM		4
+#  endif
+#  ifdef __arm__
+#    define LG_QUANTUM		3
+#  endif
+#  ifdef __aarch64__
+#    define LG_QUANTUM		4
+#  endif
+#  ifdef __hppa__
+#    define LG_QUANTUM		4
+#  endif
+#  ifdef __m68k__
+#    define LG_QUANTUM		3
+#  endif
+#  ifdef __mips__
+#    define LG_QUANTUM		3
+#  endif
+#  ifdef __nios2__
+#    define LG_QUANTUM		3
+#  endif
+#  ifdef __or1k__
+#    define LG_QUANTUM		3
+#  endif
+#  ifdef __powerpc__
+#    define LG_QUANTUM		4
+#  endif
+#  if defined(__riscv) || defined(__riscv__)
+#    define LG_QUANTUM		4
+#  endif
+#  ifdef __s390__
+#    define LG_QUANTUM		4
+#  endif
+#  if (defined (__SH3E__) || defined(__SH4_SINGLE__) || defined(__SH4__) || \
+	defined(__SH4_SINGLE_ONLY__))
+#    define LG_QUANTUM		4
+#  endif
+#  ifdef __tile__
+#    define LG_QUANTUM		4
+#  endif
+#  ifdef __le32__
+#    define LG_QUANTUM		4
+#  endif
+#  ifndef LG_QUANTUM
+#    error "Unknown minimum alignment for architecture; specify via "
+	 "--with-lg-quantum"
+#  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.
+ *
+ * CACHELINE cannot be based on LG_CACHELINE because __declspec(align()) can
+ * only handle raw constants.
+ */
+#define LG_CACHELINE		6
+#define CACHELINE		64
+#define CACHELINE_MASK		(CACHELINE - 1)
+
+/* Return the smallest cacheline multiple that is >= s. */
+#define CACHELINE_CEILING(s)						\
+	(((s) + CACHELINE_MASK) & ~CACHELINE_MASK)
+
+/* Return the nearest aligned address at or below a. */
+#define ALIGNMENT_ADDR2BASE(a, alignment)				\
+	((void *)((uintptr_t)(a) & ((~(alignment)) + 1)))
+
+/* 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)) + 1))
+
+/* Declare a variable-length array. */
+#if __STDC_VERSION__ < 199901L
+#  ifdef _MSC_VER
+#    include <malloc.h>
+#    define alloca _alloca
+#  else
+#    ifdef JEMALLOC_HAS_ALLOCA_H
+#      include <alloca.h>
+#    else
+#      include <stdlib.h>
+#    endif
+#  endif
+#  define VARIABLE_ARRAY(type, name, count) \
+	type *name = alloca(sizeof(type) * (count))
+#else
+#  define VARIABLE_ARRAY(type, name, count) type name[(count)]
+#endif
+
+#endif /* JEMALLOC_INTERNAL_TYPES_H */

include/jemalloc/internal/jemalloc_preamble.h.in

@@ -0,0 +1,194 @@
+#ifndef JEMALLOC_PREAMBLE_H
+#define JEMALLOC_PREAMBLE_H
+
+#include "jemalloc_internal_defs.h"
+#include "jemalloc/internal/jemalloc_internal_decls.h"
+
+#ifdef JEMALLOC_UTRACE
+#include <sys/ktrace.h>
+#endif
+
+#define JEMALLOC_NO_DEMANGLE
+#ifdef JEMALLOC_JET
+#  undef JEMALLOC_IS_MALLOC
+#  define JEMALLOC_N(n) jet_##n
+#  include "jemalloc/internal/public_namespace.h"
+#  define JEMALLOC_NO_RENAME
+#  include "../jemalloc@install_suffix@.h"
+#  undef JEMALLOC_NO_RENAME
+#else
+#  define JEMALLOC_N(n) @private_namespace@##n
+#  include "../jemalloc@install_suffix@.h"
+#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>
+#endif
+
+#include "jemalloc/internal/jemalloc_internal_macros.h"
+
+/*
+ * Note that the ordering matters here; the hook itself is name-mangled.  We
+ * want the inclusion of hooks to happen early, so that we hook as much as
+ * possible.
+ */
+#ifndef JEMALLOC_NO_PRIVATE_NAMESPACE
+#  ifndef JEMALLOC_JET
+#    include "jemalloc/internal/private_namespace.h"
+#  else
+#    include "jemalloc/internal/private_namespace_jet.h"
+#  endif
+#endif
+#include "jemalloc/internal/hooks.h"
+
+#ifdef JEMALLOC_DEFINE_MADVISE_FREE
+#  define JEMALLOC_MADV_FREE 8
+#endif
+
+static const bool config_debug =
+#ifdef JEMALLOC_DEBUG
+    true
+#else
+    false
+#endif
+    ;
+static const bool have_dss =
+#ifdef JEMALLOC_DSS
+    true
+#else
+    false
+#endif
+    ;
+static const bool have_madvise_huge =
+#ifdef JEMALLOC_HAVE_MADVISE_HUGE
+    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 char * const config_malloc_conf = JEMALLOC_CONFIG_MALLOC_CONF;
+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 maps_coalesce =
+#ifdef JEMALLOC_MAPS_COALESCE
+    true
+#else
+    false
+#endif
+    ;
+static const bool config_stats =
+#ifdef JEMALLOC_STATS
+    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_xmalloc =
+#ifdef JEMALLOC_XMALLOC
+    true
+#else
+    false
+#endif
+    ;
+static const bool config_cache_oblivious =
+#ifdef JEMALLOC_CACHE_OBLIVIOUS
+    true
+#else
+    false
+#endif
+    ;
+/*
+ * Undocumented, for jemalloc development use only at the moment.  See the note
+ * in jemalloc/internal/log.h.
+ */
+static const bool config_log =
+#ifdef JEMALLOC_LOG
+    true
+#else
+    false
+#endif
+    ;
+#ifdef JEMALLOC_HAVE_SCHED_GETCPU
+/* Currently percpu_arena depends on sched_getcpu. */
+#define JEMALLOC_PERCPU_ARENA
+#endif
+static const bool have_percpu_arena =
+#ifdef JEMALLOC_PERCPU_ARENA
+    true
+#else
+    false
+#endif
+    ;
+/*
+ * Undocumented, and not recommended; the application should take full
+ * responsibility for tracking provenance.
+ */
+static const bool force_ivsalloc =
+#ifdef JEMALLOC_FORCE_IVSALLOC
+    true
+#else
+    false
+#endif
+    ;
+static const bool have_background_thread =
+#ifdef JEMALLOC_BACKGROUND_THREAD
+    true
+#else
+    false
+#endif
+    ;
+
+#endif /* JEMALLOC_PREAMBLE_H */

include/jemalloc/internal/large_externs.h

@@ -0,0 +1,26 @@
+#ifndef JEMALLOC_INTERNAL_LARGE_EXTERNS_H
+#define JEMALLOC_INTERNAL_LARGE_EXTERNS_H
+
+void *large_malloc(tsdn_t *tsdn, arena_t *arena, size_t usize, bool zero);
+void *large_palloc(tsdn_t *tsdn, arena_t *arena, size_t usize, size_t alignment,
+    bool zero);
+bool large_ralloc_no_move(tsdn_t *tsdn, extent_t *extent, size_t usize_min,
+    size_t usize_max, bool zero);
+void *large_ralloc(tsdn_t *tsdn, arena_t *arena, extent_t *extent, size_t usize,
+    size_t alignment, bool zero, tcache_t *tcache);
+
+typedef void (large_dalloc_junk_t)(void *, size_t);
+extern large_dalloc_junk_t *JET_MUTABLE large_dalloc_junk;
+
+typedef void (large_dalloc_maybe_junk_t)(void *, size_t);
+extern large_dalloc_maybe_junk_t *JET_MUTABLE large_dalloc_maybe_junk;
+
+void large_dalloc_prep_junked_locked(tsdn_t *tsdn, extent_t *extent);
+void large_dalloc_finish(tsdn_t *tsdn, extent_t *extent);
+void large_dalloc(tsdn_t *tsdn, extent_t *extent);
+size_t large_salloc(tsdn_t *tsdn, const extent_t *extent);
+prof_tctx_t *large_prof_tctx_get(tsdn_t *tsdn, const extent_t *extent);
+void large_prof_tctx_set(tsdn_t *tsdn, extent_t *extent, prof_tctx_t *tctx);
+void large_prof_tctx_reset(tsdn_t *tsdn, extent_t *extent);
+
+#endif /* JEMALLOC_INTERNAL_LARGE_EXTERNS_H */

include/jemalloc/internal/log.h

@@ -0,0 +1,115 @@
+#ifndef JEMALLOC_INTERNAL_LOG_H
+#define JEMALLOC_INTERNAL_LOG_H
+
+#include "jemalloc/internal/atomic.h"
+#include "jemalloc/internal/malloc_io.h"
+#include "jemalloc/internal/mutex.h"
+
+#ifdef JEMALLOC_LOG
+#  define JEMALLOC_LOG_VAR_BUFSIZE 1000
+#else
+#  define JEMALLOC_LOG_VAR_BUFSIZE 1
+#endif
+
+#define JEMALLOC_LOG_BUFSIZE 4096
+
+/*
+ * The log malloc_conf option is a '|'-delimited list of log_var name segments
+ * which should be logged.  The names are themselves hierarchical, with '.' as
+ * the delimiter (a "segment" is just a prefix in the log namespace).  So, if
+ * you have:
+ *
+ * log("arena", "log msg for arena"); // 1
+ * log("arena.a", "log msg for arena.a"); // 2
+ * log("arena.b", "log msg for arena.b"); // 3
+ * log("arena.a.a", "log msg for arena.a.a"); // 4
+ * log("extent.a", "log msg for extent.a"); // 5
+ * log("extent.b", "log msg for extent.b"); // 6
+ *
+ * And your malloc_conf option is "log=arena.a|extent", then lines 2, 4, 5, and
+ * 6 will print at runtime.  You can enable logging from all log vars by
+ * writing "log=.".
+ *
+ * None of this should be regarded as a stable API for right now.  It's intended
+ * as a debugging interface, to let us keep around some of our printf-debugging
+ * statements.
+ */
+
+extern char log_var_names[JEMALLOC_LOG_VAR_BUFSIZE];
+extern atomic_b_t log_init_done;
+
+typedef struct log_var_s log_var_t;
+struct log_var_s {
+	/*
+	 * Lowest bit is "inited", second lowest is "enabled".  Putting them in
+	 * a single word lets us avoid any fences on weak architectures.
+	 */
+	atomic_u_t state;
+	const char *name;
+};
+
+#define LOG_NOT_INITIALIZED 0U
+#define LOG_INITIALIZED_NOT_ENABLED 1U
+#define LOG_ENABLED 2U
+
+#define LOG_VAR_INIT(name_str) {ATOMIC_INIT(LOG_NOT_INITIALIZED), name_str}
+
+/*
+ * Returns the value we should assume for state (which is not necessarily
+ * accurate; if logging is done before logging has finished initializing, then
+ * we default to doing the safe thing by logging everything).
+ */
+unsigned log_var_update_state(log_var_t *log_var);
+
+/* We factor out the metadata management to allow us to test more easily. */
+#define log_do_begin(log_var)						\
+if (config_log) {							\
+	unsigned log_state = atomic_load_u(&(log_var).state,		\
+	    ATOMIC_RELAXED);						\
+	if (unlikely(log_state == LOG_NOT_INITIALIZED)) {		\
+		log_state = log_var_update_state(&(log_var));		\
+		assert(log_state != LOG_NOT_INITIALIZED);		\
+	}								\
+	if (log_state == LOG_ENABLED) {					\
+		{
+			/* User code executes here. */
+#define log_do_end(log_var)						\
+		}							\
+	}								\
+}
+
+/*
+ * MSVC has some preprocessor bugs in its expansion of __VA_ARGS__ during
+ * preprocessing.  To work around this, we take all potential extra arguments in
+ * a var-args functions.  Since a varargs macro needs at least one argument in
+ * the "...", we accept the format string there, and require that the first
+ * argument in this "..." is a const char *.
+ */
+static inline void
+log_impl_varargs(const char *name, ...) {
+	char buf[JEMALLOC_LOG_BUFSIZE];
+	va_list ap;
+
+	va_start(ap, name);
+	const char *format = va_arg(ap, const char *);
+	size_t dst_offset = 0;
+	dst_offset += malloc_snprintf(buf, JEMALLOC_LOG_BUFSIZE, "%s: ", name);
+	dst_offset += malloc_vsnprintf(buf + dst_offset,
+	    JEMALLOC_LOG_BUFSIZE - dst_offset, format, ap);
+	dst_offset += malloc_snprintf(buf + dst_offset,
+	    JEMALLOC_LOG_BUFSIZE - dst_offset, "\n");
+	va_end(ap);
+
+	malloc_write(buf);
+}
+
+/* Call as log("log.var.str", "format_string %d", arg_for_format_string); */
+#define LOG(log_var_str, ...)						\
+do {									\
+	static log_var_t log_var = LOG_VAR_INIT(log_var_str);		\
+	log_do_begin(log_var)						\
+		log_impl_varargs((log_var).name, __VA_ARGS__);		\
+	log_do_end(log_var)						\
+} while (0)
+
+#endif /* JEMALLOC_INTERNAL_LOG_H */

include/jemalloc/internal/malloc_io.h

@@ -0,0 +1,102 @@
+#ifndef JEMALLOC_INTERNAL_MALLOC_IO_H
+#define JEMALLOC_INTERNAL_MALLOC_IO_H
+
+#ifdef _WIN32
+#  ifdef _WIN64
+#    define FMT64_PREFIX "ll"
+#    define FMTPTR_PREFIX "ll"
+#  else
+#    define FMT64_PREFIX "ll"
+#    define FMTPTR_PREFIX ""
+#  endif
+#  define FMTd32 "d"
+#  define FMTu32 "u"
+#  define FMTx32 "x"
+#  define FMTd64 FMT64_PREFIX "d"
+#  define FMTu64 FMT64_PREFIX "u"
+#  define FMTx64 FMT64_PREFIX "x"
+#  define FMTdPTR FMTPTR_PREFIX "d"
+#  define FMTuPTR FMTPTR_PREFIX "u"
+#  define FMTxPTR FMTPTR_PREFIX "x"
+#else
+#  include <inttypes.h>
+#  define FMTd32 PRId32
+#  define FMTu32 PRIu32
+#  define FMTx32 PRIx32
+#  define FMTd64 PRId64
+#  define FMTu64 PRIu64
+#  define FMTx64 PRIx64
+#  define FMTdPTR PRIdPTR
+#  define FMTuPTR PRIuPTR
+#  define FMTxPTR PRIxPTR
+#endif
+
+/* 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
+
+int buferror(int err, char *buf, size_t buflen);
+uintmax_t malloc_strtoumax(const char *restrict nptr, char **restrict endptr,
+    int base);
+void malloc_write(const char *s);
+
+/*
+ * malloc_vsnprintf() supports a subset of snprintf(3) that avoids floating
+ * point math.
+ */
+size_t malloc_vsnprintf(char *str, size_t size, const char *format,
+    va_list ap);
+size_t malloc_snprintf(char *str, size_t size, const char *format, ...)
+    JEMALLOC_FORMAT_PRINTF(3, 4);
+/*
+ * The caller can set write_cb and cbopaque to null to choose to print with the
+ * je_malloc_message hook.
+ */
+void malloc_vcprintf(void (*write_cb)(void *, const char *), void *cbopaque,
+    const char *format, va_list ap);
+void malloc_cprintf(void (*write_cb)(void *, const char *), void *cbopaque,
+    const char *format, ...) JEMALLOC_FORMAT_PRINTF(3, 4);
+void malloc_printf(const char *format, ...) JEMALLOC_FORMAT_PRINTF(1, 2);
+
+static inline ssize_t
+malloc_write_fd(int fd, const void *buf, size_t count) {
+#if defined(JEMALLOC_USE_SYSCALL) && defined(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.
+	 *
+	 * syscall() returns long or int, depending on platform, so capture the
+	 * result in the widest plausible type to avoid compiler warnings.
+	 */
+	long result = syscall(SYS_write, fd, buf, count);
+#else
+	ssize_t result = (ssize_t)write(fd, buf,
+#ifdef _WIN32
+	    (unsigned int)
+#endif
+	    count);
+#endif
+	return (ssize_t)result;
+}
+
+static inline ssize_t
+malloc_read_fd(int fd, void *buf, size_t count) {
+#if defined(JEMALLOC_USE_SYSCALL) && defined(SYS_read)
+	long result = syscall(SYS_read, fd, buf, count);
+#else
+	ssize_t result = read(fd, buf,
+#ifdef _WIN32
+	    (unsigned int)
+#endif
+	    count);
+#endif
+	return (ssize_t)result;
+}
+
+#endif /* JEMALLOC_INTERNAL_MALLOC_IO_H */

include/jemalloc/internal/mb.h

@@ -1,115 +0,0 @@
-/******************************************************************************/
-#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 */
-/******************************************************************************/

include/jemalloc/internal/mutex.h

@@ -1,45 +1,123 @@
-/******************************************************************************/
-#ifdef JEMALLOC_H_TYPES
+#ifndef JEMALLOC_INTERNAL_MUTEX_H
+#define JEMALLOC_INTERNAL_MUTEX_H
+
+#include "jemalloc/internal/atomic.h"
+#include "jemalloc/internal/mutex_prof.h"
+#include "jemalloc/internal/tsd.h"
+#include "jemalloc/internal/witness.h"
+
+typedef enum {
+	/* Can only acquire one mutex of a given witness rank at a time. */
+	malloc_mutex_rank_exclusive,
+	/*
+	 * Can acquire multiple mutexes of the same witness rank, but in
+	 * address-ascending order only.
+	 */
+	malloc_mutex_address_ordered
+} malloc_mutex_lock_order_t;
 
 typedef struct malloc_mutex_s malloc_mutex_t;
-
-#ifdef _WIN32
-#  define MALLOC_MUTEX_INITIALIZER
-#elif (defined(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 {
+	union {
+		struct {
+			/*
+			 * prof_data is defined first to reduce cacheline
+			 * bouncing: the data is not touched by the mutex holder
+			 * during unlocking, while might be modified by
+			 * contenders.  Having it before the mutex itself could
+			 * avoid prefetching a modified cacheline (for the
+			 * unlocking thread).
+			 */
+			mutex_prof_data_t	prof_data;
 #ifdef _WIN32
-	CRITICAL_SECTION	lock;
+#  if _WIN32_WINNT >= 0x0600
+			SRWLOCK         	lock;
+#  else
+			CRITICAL_SECTION	lock;
+#  endif
+#elif (defined(JEMALLOC_OS_UNFAIR_LOCK))
+			os_unfair_lock		lock;
 #elif (defined(JEMALLOC_OSSPIN))
-	OSSpinLock		lock;
+			OSSpinLock		lock;
 #elif (defined(JEMALLOC_MUTEX_INIT_CB))
-	pthread_mutex_t		lock;
-	malloc_mutex_t		*postponed_next;
+			pthread_mutex_t		lock;
+			malloc_mutex_t		*postponed_next;
 #else
-	pthread_mutex_t		lock;
+			pthread_mutex_t		lock;
+#endif
+		};
+		/*
+		 * We only touch witness when configured w/ debug.  However we
+		 * keep the field in a union when !debug so that we don't have
+		 * to pollute the code base with #ifdefs, while avoid paying the
+		 * memory cost.
+		 */
+#if !defined(JEMALLOC_DEBUG)
+		witness_t			witness;
+		malloc_mutex_lock_order_t	lock_order;
+#endif
+	};
+
+#if defined(JEMALLOC_DEBUG)
+	witness_t			witness;
+	malloc_mutex_lock_order_t	lock_order;
 #endif
 };
 
-#endif /* JEMALLOC_H_STRUCTS */
-/******************************************************************************/
-#ifdef JEMALLOC_H_EXTERNS
+/*
+ * Based on benchmark results, a fixed spin with this amount of retries works
+ * well for our critical sections.
+ */
+#define MALLOC_MUTEX_MAX_SPIN 250
+
+#ifdef _WIN32
+#  if _WIN32_WINNT >= 0x0600
+#    define MALLOC_MUTEX_LOCK(m)    AcquireSRWLockExclusive(&(m)->lock)
+#    define MALLOC_MUTEX_UNLOCK(m)  ReleaseSRWLockExclusive(&(m)->lock)
+#    define MALLOC_MUTEX_TRYLOCK(m) (!TryAcquireSRWLockExclusive(&(m)->lock))
+#  else
+#    define MALLOC_MUTEX_LOCK(m)    EnterCriticalSection(&(m)->lock)
+#    define MALLOC_MUTEX_UNLOCK(m)  LeaveCriticalSection(&(m)->lock)
+#    define MALLOC_MUTEX_TRYLOCK(m) (!TryEnterCriticalSection(&(m)->lock))
+#  endif
+#elif (defined(JEMALLOC_OS_UNFAIR_LOCK))
+#    define MALLOC_MUTEX_LOCK(m)    os_unfair_lock_lock(&(m)->lock)
+#    define MALLOC_MUTEX_UNLOCK(m)  os_unfair_lock_unlock(&(m)->lock)
+#    define MALLOC_MUTEX_TRYLOCK(m) (!os_unfair_lock_trylock(&(m)->lock))
+#elif (defined(JEMALLOC_OSSPIN))
+#    define MALLOC_MUTEX_LOCK(m)    OSSpinLockLock(&(m)->lock)
+#    define MALLOC_MUTEX_UNLOCK(m)  OSSpinLockUnlock(&(m)->lock)
+#    define MALLOC_MUTEX_TRYLOCK(m) (!OSSpinLockTry(&(m)->lock))
+#else
+#    define MALLOC_MUTEX_LOCK(m)    pthread_mutex_lock(&(m)->lock)
+#    define MALLOC_MUTEX_UNLOCK(m)  pthread_mutex_unlock(&(m)->lock)
+#    define MALLOC_MUTEX_TRYLOCK(m) (pthread_mutex_trylock(&(m)->lock) != 0)
+#endif
+
+#define LOCK_PROF_DATA_INITIALIZER					\
+    {NSTIME_ZERO_INITIALIZER, NSTIME_ZERO_INITIALIZER, 0, 0, 0,		\
+	    ATOMIC_INIT(0), 0, NULL, 0}
+
+#ifdef _WIN32
+#  define MALLOC_MUTEX_INITIALIZER
+#elif (defined(JEMALLOC_OS_UNFAIR_LOCK))
+#  define MALLOC_MUTEX_INITIALIZER					\
+     {{{LOCK_PROF_DATA_INITIALIZER, OS_UNFAIR_LOCK_INIT}},		\
+      WITNESS_INITIALIZER("mutex", WITNESS_RANK_OMIT)}
+#elif (defined(JEMALLOC_OSSPIN))
+#  define MALLOC_MUTEX_INITIALIZER					\
+     {{{LOCK_PROF_DATA_INITIALIZER, 0}},				\
+      WITNESS_INITIALIZER("mutex", WITNESS_RANK_OMIT)}
+#elif (defined(JEMALLOC_MUTEX_INIT_CB))
+#  define MALLOC_MUTEX_INITIALIZER					\
+     {{{LOCK_PROF_DATA_INITIALIZER, PTHREAD_MUTEX_INITIALIZER, NULL}},	\
+      WITNESS_INITIALIZER("mutex", WITNESS_RANK_OMIT)}
+#else
+#    define MALLOC_MUTEX_TYPE PTHREAD_MUTEX_DEFAULT
+#    define MALLOC_MUTEX_INITIALIZER					\
+       {{{LOCK_PROF_DATA_INITIALIZER, PTHREAD_MUTEX_INITIALIZER}},	\
+        WITNESS_INITIALIZER("mutex", WITNESS_RANK_OMIT)}
+#endif
 
 #ifdef JEMALLOC_LAZY_LOCK
 extern bool isthreaded;
@@ -48,52 +126,123 @@ extern bool isthreaded;
 #  define isthreaded true
 #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);
+bool malloc_mutex_init(malloc_mutex_t *mutex, const char *name,
+    witness_rank_t rank, malloc_mutex_lock_order_t lock_order);
+void malloc_mutex_prefork(tsdn_t *tsdn, malloc_mutex_t *mutex);
+void malloc_mutex_postfork_parent(tsdn_t *tsdn, malloc_mutex_t *mutex);
+void malloc_mutex_postfork_child(tsdn_t *tsdn, malloc_mutex_t *mutex);
+bool malloc_mutex_boot(void);
+void malloc_mutex_prof_data_reset(tsdn_t *tsdn, malloc_mutex_t *mutex);
 
-#endif /* JEMALLOC_H_EXTERNS */
-/******************************************************************************/
-#ifdef JEMALLOC_H_INLINES
+void malloc_mutex_lock_slow(malloc_mutex_t *mutex);
 
-#ifndef JEMALLOC_ENABLE_INLINE
-void	malloc_mutex_lock(malloc_mutex_t *mutex);
-void	malloc_mutex_unlock(malloc_mutex_t *mutex);
-#endif
+static inline void
+malloc_mutex_lock_final(malloc_mutex_t *mutex) {
+	MALLOC_MUTEX_LOCK(mutex);
+}
 
-#if (defined(JEMALLOC_ENABLE_INLINE) || defined(JEMALLOC_MUTEX_C_))
-JEMALLOC_INLINE void
-malloc_mutex_lock(malloc_mutex_t *mutex)
-{
+static inline bool
+malloc_mutex_trylock_final(malloc_mutex_t *mutex) {
+	return MALLOC_MUTEX_TRYLOCK(mutex);
+}
 
-	if (isthreaded) {
-#ifdef _WIN32
-		EnterCriticalSection(&mutex->lock);
-#elif (defined(JEMALLOC_OSSPIN))
-		OSSpinLockLock(&mutex->lock);
-#else
-		pthread_mutex_lock(&mutex->lock);
-#endif
+static inline void
+mutex_owner_stats_update(tsdn_t *tsdn, malloc_mutex_t *mutex) {
+	if (config_stats) {
+		mutex_prof_data_t *data = &mutex->prof_data;
+		data->n_lock_ops++;
+		if (data->prev_owner != tsdn) {
+			data->prev_owner = tsdn;
+			data->n_owner_switches++;
+		}
 	}
 }
 
-JEMALLOC_INLINE void
-malloc_mutex_unlock(malloc_mutex_t *mutex)
-{
-
+/* Trylock: return false if the lock is successfully acquired. */
+static inline bool
+malloc_mutex_trylock(tsdn_t *tsdn, malloc_mutex_t *mutex) {
+	witness_assert_not_owner(tsdn_witness_tsdp_get(tsdn), &mutex->witness);
 	if (isthreaded) {
-#ifdef _WIN32
-		LeaveCriticalSection(&mutex->lock);
-#elif (defined(JEMALLOC_OSSPIN))
-		OSSpinLockUnlock(&mutex->lock);
-#else
-		pthread_mutex_unlock(&mutex->lock);
-#endif
+		if (malloc_mutex_trylock_final(mutex)) {
+			return true;
+		}
+		mutex_owner_stats_update(tsdn, mutex);
+	}
+	witness_lock(tsdn_witness_tsdp_get(tsdn), &mutex->witness);
+
+	return false;
+}
+
+/* Aggregate lock prof data. */
+static inline void
+malloc_mutex_prof_merge(mutex_prof_data_t *sum, mutex_prof_data_t *data) {
+	nstime_add(&sum->tot_wait_time, &data->tot_wait_time);
+	if (nstime_compare(&sum->max_wait_time, &data->max_wait_time) < 0) {
+		nstime_copy(&sum->max_wait_time, &data->max_wait_time);
+	}
+
+	sum->n_wait_times += data->n_wait_times;
+	sum->n_spin_acquired += data->n_spin_acquired;
+
+	if (sum->max_n_thds < data->max_n_thds) {
+		sum->max_n_thds = data->max_n_thds;
+	}
+	uint32_t cur_n_waiting_thds = atomic_load_u32(&sum->n_waiting_thds,
+	    ATOMIC_RELAXED);
+	uint32_t new_n_waiting_thds = cur_n_waiting_thds + atomic_load_u32(
+	    &data->n_waiting_thds, ATOMIC_RELAXED);
+	atomic_store_u32(&sum->n_waiting_thds, new_n_waiting_thds,
+	    ATOMIC_RELAXED);
+	sum->n_owner_switches += data->n_owner_switches;
+	sum->n_lock_ops += data->n_lock_ops;
+}
+
+static inline void
+malloc_mutex_lock(tsdn_t *tsdn, malloc_mutex_t *mutex) {
+	witness_assert_not_owner(tsdn_witness_tsdp_get(tsdn), &mutex->witness);
+	if (isthreaded) {
+		if (malloc_mutex_trylock_final(mutex)) {
+			malloc_mutex_lock_slow(mutex);
+		}
+		mutex_owner_stats_update(tsdn, mutex);
+	}
+	witness_lock(tsdn_witness_tsdp_get(tsdn), &mutex->witness);
+}
+
+static inline void
+malloc_mutex_unlock(tsdn_t *tsdn, malloc_mutex_t *mutex) {
+	witness_unlock(tsdn_witness_tsdp_get(tsdn), &mutex->witness);
+	if (isthreaded) {
+		MALLOC_MUTEX_UNLOCK(mutex);
 	}
 }
-#endif
 
-#endif /* JEMALLOC_H_INLINES */
-/******************************************************************************/
+static inline void
+malloc_mutex_assert_owner(tsdn_t *tsdn, malloc_mutex_t *mutex) {
+	witness_assert_owner(tsdn_witness_tsdp_get(tsdn), &mutex->witness);
+}
+
+static inline void
+malloc_mutex_assert_not_owner(tsdn_t *tsdn, malloc_mutex_t *mutex) {
+	witness_assert_not_owner(tsdn_witness_tsdp_get(tsdn), &mutex->witness);
+}
+
+/* Copy the prof data from mutex for processing. */
+static inline void
+malloc_mutex_prof_read(tsdn_t *tsdn, mutex_prof_data_t *data,
+    malloc_mutex_t *mutex) {
+	mutex_prof_data_t *source = &mutex->prof_data;
+	/* Can only read holding the mutex. */
+	malloc_mutex_assert_owner(tsdn, mutex);
+
+	/*
+	 * Not *really* allowed (we shouldn't be doing non-atomic loads of
+	 * atomic data), but the mutex protection makes this safe, and writing
+	 * a member-for-member copy is tedious for this situation.
+	 */
+	*data = *source;
+	/* n_wait_thds is not reported (modified w/o locking). */
+	atomic_store_u32(&data->n_waiting_thds, 0, ATOMIC_RELAXED);
+}
+
+#endif /* JEMALLOC_INTERNAL_MUTEX_H */

include/jemalloc/internal/mutex_pool.h

@@ -0,0 +1,94 @@
+#ifndef JEMALLOC_INTERNAL_MUTEX_POOL_H
+#define JEMALLOC_INTERNAL_MUTEX_POOL_H
+
+#include "jemalloc/internal/hash.h"
+#include "jemalloc/internal/mutex.h"
+#include "jemalloc/internal/witness.h"
+
+/* We do mod reductions by this value, so it should be kept a power of 2. */
+#define MUTEX_POOL_SIZE 256
+
+typedef struct mutex_pool_s mutex_pool_t;
+struct mutex_pool_s {
+	malloc_mutex_t mutexes[MUTEX_POOL_SIZE];
+};
+
+bool mutex_pool_init(mutex_pool_t *pool, const char *name, witness_rank_t rank);
+
+/* Internal helper - not meant to be called outside this module. */
+static inline malloc_mutex_t *
+mutex_pool_mutex(mutex_pool_t *pool, uintptr_t key) {
+	size_t hash_result[2];
+	hash(&key, sizeof(key), 0xd50dcc1b, hash_result);
+	return &pool->mutexes[hash_result[0] % MUTEX_POOL_SIZE];
+}
+
+static inline void
+mutex_pool_assert_not_held(tsdn_t *tsdn, mutex_pool_t *pool) {
+	for (int i = 0; i < MUTEX_POOL_SIZE; i++) {
+		malloc_mutex_assert_not_owner(tsdn, &pool->mutexes[i]);
+	}
+}
+
+/*
+ * Note that a mutex pool doesn't work exactly the way an embdedded mutex would.
+ * You're not allowed to acquire mutexes in the pool one at a time.  You have to
+ * acquire all the mutexes you'll need in a single function call, and then
+ * release them all in a single function call.
+ */
+
+static inline void
+mutex_pool_lock(tsdn_t *tsdn, mutex_pool_t *pool, uintptr_t key) {
+	mutex_pool_assert_not_held(tsdn, pool);
+
+	malloc_mutex_t *mutex = mutex_pool_mutex(pool, key);
+	malloc_mutex_lock(tsdn, mutex);
+}
+
+static inline void
+mutex_pool_unlock(tsdn_t *tsdn, mutex_pool_t *pool, uintptr_t key) {
+	malloc_mutex_t *mutex = mutex_pool_mutex(pool, key);
+	malloc_mutex_unlock(tsdn, mutex);
+
+	mutex_pool_assert_not_held(tsdn, pool);
+}
+
+static inline void
+mutex_pool_lock2(tsdn_t *tsdn, mutex_pool_t *pool, uintptr_t key1,
+    uintptr_t key2) {
+	mutex_pool_assert_not_held(tsdn, pool);
+
+	malloc_mutex_t *mutex1 = mutex_pool_mutex(pool, key1);
+	malloc_mutex_t *mutex2 = mutex_pool_mutex(pool, key2);
+	if ((uintptr_t)mutex1 < (uintptr_t)mutex2) {
+		malloc_mutex_lock(tsdn, mutex1);
+		malloc_mutex_lock(tsdn, mutex2);
+	} else if ((uintptr_t)mutex1 == (uintptr_t)mutex2) {
+		malloc_mutex_lock(tsdn, mutex1);
+	} else {
+		malloc_mutex_lock(tsdn, mutex2);
+		malloc_mutex_lock(tsdn, mutex1);
+	}
+}
+
+static inline void
+mutex_pool_unlock2(tsdn_t *tsdn, mutex_pool_t *pool, uintptr_t key1,
+    uintptr_t key2) {
+	malloc_mutex_t *mutex1 = mutex_pool_mutex(pool, key1);
+	malloc_mutex_t *mutex2 = mutex_pool_mutex(pool, key2);
+	if (mutex1 == mutex2) {
+		malloc_mutex_unlock(tsdn, mutex1);
+	} else {
+		malloc_mutex_unlock(tsdn, mutex1);
+		malloc_mutex_unlock(tsdn, mutex2);
+	}
+
+	mutex_pool_assert_not_held(tsdn, pool);
+}
+
+static inline void
+mutex_pool_assert_owner(tsdn_t *tsdn, mutex_pool_t *pool, uintptr_t key) {
+	malloc_mutex_assert_owner(tsdn, mutex_pool_mutex(pool, key));
+}
+
+#endif /* JEMALLOC_INTERNAL_MUTEX_POOL_H */

include/jemalloc/internal/mutex_prof.h

@@ -0,0 +1,99 @@
+#ifndef JEMALLOC_INTERNAL_MUTEX_PROF_H
+#define JEMALLOC_INTERNAL_MUTEX_PROF_H
+
+#include "jemalloc/internal/atomic.h"
+#include "jemalloc/internal/nstime.h"
+#include "jemalloc/internal/tsd_types.h"
+
+#define MUTEX_PROF_GLOBAL_MUTEXES					\
+    OP(background_thread)						\
+    OP(ctl)								\
+    OP(prof)
+
+typedef enum {
+#define OP(mtx) global_prof_mutex_##mtx,
+	MUTEX_PROF_GLOBAL_MUTEXES
+#undef OP
+	mutex_prof_num_global_mutexes
+} mutex_prof_global_ind_t;
+
+#define MUTEX_PROF_ARENA_MUTEXES					\
+    OP(large)								\
+    OP(extent_avail)							\
+    OP(extents_dirty)							\
+    OP(extents_muzzy)							\
+    OP(extents_retained)						\
+    OP(decay_dirty)							\
+    OP(decay_muzzy)							\
+    OP(base)								\
+    OP(tcache_list)
+
+typedef enum {
+#define OP(mtx) arena_prof_mutex_##mtx,
+	MUTEX_PROF_ARENA_MUTEXES
+#undef OP
+	mutex_prof_num_arena_mutexes
+} mutex_prof_arena_ind_t;
+
+#define MUTEX_PROF_UINT64_COUNTERS					\
+    OP(num_ops, uint64_t, "n_lock_ops")					\
+    OP(num_wait, uint64_t, "n_waiting")					\
+    OP(num_spin_acq, uint64_t, "n_spin_acq")				\
+    OP(num_owner_switch, uint64_t, "n_owner_switch")			\
+    OP(total_wait_time, uint64_t, "total_wait_ns")			\
+    OP(max_wait_time, uint64_t, "max_wait_ns")
+
+#define MUTEX_PROF_UINT32_COUNTERS					\
+    OP(max_num_thds, uint32_t, "max_n_thds")
+
+#define MUTEX_PROF_COUNTERS						\
+		MUTEX_PROF_UINT64_COUNTERS				\
+		MUTEX_PROF_UINT32_COUNTERS
+
+#define OP(counter, type, human) mutex_counter_##counter,
+
+#define COUNTER_ENUM(counter_list, t)					\
+		typedef enum {						\
+			counter_list					\
+			mutex_prof_num_##t##_counters			\
+		} mutex_prof_##t##_counter_ind_t;
+
+COUNTER_ENUM(MUTEX_PROF_UINT64_COUNTERS, uint64_t)
+COUNTER_ENUM(MUTEX_PROF_UINT32_COUNTERS, uint32_t)
+
+#undef COUNTER_ENUM
+#undef OP
+
+typedef struct {
+	/*
+	 * Counters touched on the slow path, i.e. when there is lock
+	 * contention.  We update them once we have the lock.
+	 */
+	/* Total time (in nano seconds) spent waiting on this mutex. */
+	nstime_t		tot_wait_time;
+	/* Max time (in nano seconds) spent on a single lock operation. */
+	nstime_t		max_wait_time;
+	/* # of times have to wait for this mutex (after spinning). */
+	uint64_t		n_wait_times;
+	/* # of times acquired the mutex through local spinning. */
+	uint64_t		n_spin_acquired;
+	/* Max # of threads waiting for the mutex at the same time. */
+	uint32_t		max_n_thds;
+	/* Current # of threads waiting on the lock.  Atomic synced. */
+	atomic_u32_t		n_waiting_thds;
+
+	/*
+	 * Data touched on the fast path.  These are modified right after we
+	 * grab the lock, so it's placed closest to the end (i.e. right before
+	 * the lock) so that we have a higher chance of them being on the same
+	 * cacheline.
+	 */
+	/* # of times the mutex holder is different than the previous one. */
+	uint64_t		n_owner_switches;
+	/* Previous mutex holder, to facilitate n_owner_switches. */
+	tsdn_t			*prev_owner;
+	/* # of lock() operations in total. */
+	uint64_t		n_lock_ops;
+} mutex_prof_data_t;
+
+#endif /* JEMALLOC_INTERNAL_MUTEX_PROF_H */

include/jemalloc/internal/nstime.h

@@ -0,0 +1,34 @@
+#ifndef JEMALLOC_INTERNAL_NSTIME_H
+#define JEMALLOC_INTERNAL_NSTIME_H
+
+/* Maximum supported number of seconds (~584 years). */
+#define NSTIME_SEC_MAX KQU(18446744072)
+#define NSTIME_ZERO_INITIALIZER {0}
+
+typedef struct {
+	uint64_t ns;
+} nstime_t;
+
+void nstime_init(nstime_t *time, uint64_t ns);
+void nstime_init2(nstime_t *time, uint64_t sec, uint64_t nsec);
+uint64_t nstime_ns(const nstime_t *time);
+uint64_t nstime_sec(const nstime_t *time);
+uint64_t nstime_msec(const nstime_t *time);
+uint64_t nstime_nsec(const nstime_t *time);
+void nstime_copy(nstime_t *time, const nstime_t *source);
+int nstime_compare(const nstime_t *a, const nstime_t *b);
+void nstime_add(nstime_t *time, const nstime_t *addend);
+void nstime_iadd(nstime_t *time, uint64_t addend);
+void nstime_subtract(nstime_t *time, const nstime_t *subtrahend);
+void nstime_isubtract(nstime_t *time, uint64_t subtrahend);
+void nstime_imultiply(nstime_t *time, uint64_t multiplier);
+void nstime_idivide(nstime_t *time, uint64_t divisor);
+uint64_t nstime_divide(const nstime_t *time, const nstime_t *divisor);
+
+typedef bool (nstime_monotonic_t)(void);
+extern nstime_monotonic_t *JET_MUTABLE nstime_monotonic;
+
+typedef bool (nstime_update_t)(nstime_t *);
+extern nstime_update_t *JET_MUTABLE nstime_update;
+
+#endif /* JEMALLOC_INTERNAL_NSTIME_H */

include/jemalloc/internal/pages.h

@@ -0,0 +1,88 @@
+#ifndef JEMALLOC_INTERNAL_PAGES_EXTERNS_H
+#define JEMALLOC_INTERNAL_PAGES_EXTERNS_H
+
+/* Page size.  LG_PAGE is determined by the configure script. */
+#ifdef PAGE_MASK
+#  undef PAGE_MASK
+#endif
+#define PAGE		((size_t)(1U << LG_PAGE))
+#define PAGE_MASK	((size_t)(PAGE - 1))
+/* Return the page base address for the page containing address a. */
+#define PAGE_ADDR2BASE(a)						\
+	((void *)((uintptr_t)(a) & ~PAGE_MASK))
+/* Return the smallest pagesize multiple that is >= s. */
+#define PAGE_CEILING(s)							\
+	(((s) + PAGE_MASK) & ~PAGE_MASK)
+
+/* Huge page size.  LG_HUGEPAGE is determined by the configure script. */
+#define HUGEPAGE	((size_t)(1U << LG_HUGEPAGE))
+#define HUGEPAGE_MASK	((size_t)(HUGEPAGE - 1))
+/* Return the huge page base address for the huge page containing address a. */
+#define HUGEPAGE_ADDR2BASE(a)						\
+	((void *)((uintptr_t)(a) & ~HUGEPAGE_MASK))
+/* Return the smallest pagesize multiple that is >= s. */
+#define HUGEPAGE_CEILING(s)						\
+	(((s) + HUGEPAGE_MASK) & ~HUGEPAGE_MASK)
+
+/* PAGES_CAN_PURGE_LAZY is defined if lazy purging is supported. */
+#if defined(_WIN32) || defined(JEMALLOC_PURGE_MADVISE_FREE)
+#  define PAGES_CAN_PURGE_LAZY
+#endif
+/*
+ * PAGES_CAN_PURGE_FORCED is defined if forced purging is supported.
+ *
+ * The only supported way to hard-purge on Windows is to decommit and then
+ * re-commit, but doing so is racy, and if re-commit fails it's a pain to
+ * propagate the "poisoned" memory state.  Since we typically decommit as the
+ * next step after purging on Windows anyway, there's no point in adding such
+ * complexity.
+ */
+#if !defined(_WIN32) && ((defined(JEMALLOC_PURGE_MADVISE_DONTNEED) && \
+    defined(JEMALLOC_PURGE_MADVISE_DONTNEED_ZEROS)) || \
+    defined(JEMALLOC_MAPS_COALESCE))
+#  define PAGES_CAN_PURGE_FORCED
+#endif
+
+static const bool pages_can_purge_lazy =
+#ifdef PAGES_CAN_PURGE_LAZY
+    true
+#else
+    false
+#endif
+    ;
+static const bool pages_can_purge_forced =
+#ifdef PAGES_CAN_PURGE_FORCED
+    true
+#else
+    false
+#endif
+    ;
+
+typedef enum {
+	thp_mode_default       = 0, /* Do not change hugepage settings. */
+	thp_mode_always        = 1, /* Always set MADV_HUGEPAGE. */
+	thp_mode_never         = 2, /* Always set MADV_NOHUGEPAGE. */
+
+	thp_mode_names_limit   = 3, /* Used for option processing. */
+	thp_mode_not_supported = 3  /* No THP support detected. */
+} thp_mode_t;
+
+#define THP_MODE_DEFAULT thp_mode_default
+extern thp_mode_t opt_thp;
+extern thp_mode_t init_system_thp_mode; /* Initial system wide state. */
+extern const char *thp_mode_names[];
+
+void *pages_map(void *addr, size_t size, size_t alignment, bool *commit);
+void pages_unmap(void *addr, size_t size);
+bool pages_commit(void *addr, size_t size);
+bool pages_decommit(void *addr, size_t size);
+bool pages_purge_lazy(void *addr, size_t size);
+bool pages_purge_forced(void *addr, size_t size);
+bool pages_huge(void *addr, size_t size);
+bool pages_nohuge(void *addr, size_t size);
+bool pages_dontdump(void *addr, size_t size);
+bool pages_dodump(void *addr, size_t size);
+bool pages_boot(void);
+void pages_set_thp_state (void *ptr, size_t size);
+
+#endif /* JEMALLOC_INTERNAL_PAGES_EXTERNS_H */

include/jemalloc/internal/ph.h

@@ -0,0 +1,391 @@
+/*
+ * A Pairing Heap implementation.
+ *
+ * "The Pairing Heap: A New Form of Self-Adjusting Heap"
+ * https://www.cs.cmu.edu/~sleator/papers/pairing-heaps.pdf
+ *
+ * With auxiliary twopass list, described in a follow on paper.
+ *
+ * "Pairing Heaps: Experiments and Analysis"
+ * http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.106.2988&rep=rep1&type=pdf
+ *
+ *******************************************************************************
+ */
+
+#ifndef PH_H_
+#define PH_H_
+
+/* Node structure. */
+#define phn(a_type)							\
+struct {								\
+	a_type	*phn_prev;						\
+	a_type	*phn_next;						\
+	a_type	*phn_lchild;						\
+}
+
+/* Root structure. */
+#define ph(a_type)							\
+struct {								\
+	a_type	*ph_root;						\
+}
+
+/* Internal utility macros. */
+#define phn_lchild_get(a_type, a_field, a_phn)				\
+	(a_phn->a_field.phn_lchild)
+#define phn_lchild_set(a_type, a_field, a_phn, a_lchild) do {		\
+	a_phn->a_field.phn_lchild = a_lchild;				\
+} while (0)
+
+#define phn_next_get(a_type, a_field, a_phn)				\
+	(a_phn->a_field.phn_next)
+#define phn_prev_set(a_type, a_field, a_phn, a_prev) do {		\
+	a_phn->a_field.phn_prev = a_prev;				\
+} while (0)
+
+#define phn_prev_get(a_type, a_field, a_phn)				\
+	(a_phn->a_field.phn_prev)
+#define phn_next_set(a_type, a_field, a_phn, a_next) do {		\
+	a_phn->a_field.phn_next = a_next;				\
+} while (0)
+
+#define phn_merge_ordered(a_type, a_field, a_phn0, a_phn1, a_cmp) do {	\
+	a_type *phn0child;						\
+									\
+	assert(a_phn0 != NULL);						\
+	assert(a_phn1 != NULL);						\
+	assert(a_cmp(a_phn0, a_phn1) <= 0);				\
+									\
+	phn_prev_set(a_type, a_field, a_phn1, a_phn0);			\
+	phn0child = phn_lchild_get(a_type, a_field, a_phn0);		\
+	phn_next_set(a_type, a_field, a_phn1, phn0child);		\
+	if (phn0child != NULL) {					\
+		phn_prev_set(a_type, a_field, phn0child, a_phn1);	\
+	}								\
+	phn_lchild_set(a_type, a_field, a_phn0, a_phn1);		\
+} while (0)
+
+#define phn_merge(a_type, a_field, a_phn0, a_phn1, a_cmp, r_phn) do {	\
+	if (a_phn0 == NULL) {						\
+		r_phn = a_phn1;						\
+	} else if (a_phn1 == NULL) {					\
+		r_phn = a_phn0;						\
+	} else if (a_cmp(a_phn0, a_phn1) < 0) {				\
+		phn_merge_ordered(a_type, a_field, a_phn0, a_phn1,	\
+		    a_cmp);						\
+		r_phn = a_phn0;						\
+	} else {							\
+		phn_merge_ordered(a_type, a_field, a_phn1, a_phn0,	\
+		    a_cmp);						\
+		r_phn = a_phn1;						\
+	}								\
+} while (0)
+
+#define ph_merge_siblings(a_type, a_field, a_phn, a_cmp, r_phn) do {	\
+	a_type *head = NULL;						\
+	a_type *tail = NULL;						\
+	a_type *phn0 = a_phn;						\
+	a_type *phn1 = phn_next_get(a_type, a_field, phn0);		\
+									\
+	/*								\
+	 * Multipass merge, wherein the first two elements of a FIFO	\
+	 * are repeatedly merged, and each result is appended to the	\
+	 * singly linked FIFO, until the FIFO contains only a single	\
+	 * element.  We start with a sibling list but no reference to	\
+	 * its tail, so we do a single pass over the sibling list to	\
+	 * populate the FIFO.						\
+	 */								\
+	if (phn1 != NULL) {						\
+		a_type *phnrest = phn_next_get(a_type, a_field, phn1);	\
+		if (phnrest != NULL) {					\
+			phn_prev_set(a_type, a_field, phnrest, NULL);	\
+		}							\
+		phn_prev_set(a_type, a_field, phn0, NULL);		\
+		phn_next_set(a_type, a_field, phn0, NULL);		\
+		phn_prev_set(a_type, a_field, phn1, NULL);		\
+		phn_next_set(a_type, a_field, phn1, NULL);		\
+		phn_merge(a_type, a_field, phn0, phn1, a_cmp, phn0);	\
+		head = tail = phn0;					\
+		phn0 = phnrest;						\
+		while (phn0 != NULL) {					\
+			phn1 = phn_next_get(a_type, a_field, phn0);	\
+			if (phn1 != NULL) {				\
+				phnrest = phn_next_get(a_type, a_field,	\
+				    phn1);				\
+				if (phnrest != NULL) {			\
+					phn_prev_set(a_type, a_field,	\
+					    phnrest, NULL);		\
+				}					\
+				phn_prev_set(a_type, a_field, phn0,	\
+				    NULL);				\
+				phn_next_set(a_type, a_field, phn0,	\
+				    NULL);				\
+				phn_prev_set(a_type, a_field, phn1,	\
+				    NULL);				\
+				phn_next_set(a_type, a_field, phn1,	\
+				    NULL);				\
+				phn_merge(a_type, a_field, phn0, phn1,	\
+				    a_cmp, phn0);			\
+				phn_next_set(a_type, a_field, tail,	\
+				    phn0);				\
+				tail = phn0;				\
+				phn0 = phnrest;				\
+			} else {					\
+				phn_next_set(a_type, a_field, tail,	\
+				    phn0);				\
+				tail = phn0;				\
+				phn0 = NULL;				\
+			}						\
+		}							\
+		phn0 = head;						\
+		phn1 = phn_next_get(a_type, a_field, phn0);		\
+		if (phn1 != NULL) {					\
+			while (true) {					\
+				head = phn_next_get(a_type, a_field,	\
+				    phn1);				\
+				assert(phn_prev_get(a_type, a_field,	\
+				    phn0) == NULL);			\
+				phn_next_set(a_type, a_field, phn0,	\
+				    NULL);				\
+				assert(phn_prev_get(a_type, a_field,	\
+				    phn1) == NULL);			\
+				phn_next_set(a_type, a_field, phn1,	\
+				    NULL);				\
+				phn_merge(a_type, a_field, phn0, phn1,	\
+				    a_cmp, phn0);			\
+				if (head == NULL) {			\
+					break;				\
+				}					\
+				phn_next_set(a_type, a_field, tail,	\
+				    phn0);				\
+				tail = phn0;				\
+				phn0 = head;				\
+				phn1 = phn_next_get(a_type, a_field,	\
+				    phn0);				\
+			}						\
+		}							\
+	}								\
+	r_phn = phn0;							\
+} while (0)
+
+#define ph_merge_aux(a_type, a_field, a_ph, a_cmp) do {			\
+	a_type *phn = phn_next_get(a_type, a_field, a_ph->ph_root);	\
+	if (phn != NULL) {						\
+		phn_prev_set(a_type, a_field, a_ph->ph_root, NULL);	\
+		phn_next_set(a_type, a_field, a_ph->ph_root, NULL);	\
+		phn_prev_set(a_type, a_field, phn, NULL);		\
+		ph_merge_siblings(a_type, a_field, phn, a_cmp, phn);	\
+		assert(phn_next_get(a_type, a_field, phn) == NULL);	\
+		phn_merge(a_type, a_field, a_ph->ph_root, phn, a_cmp,	\
+		    a_ph->ph_root);					\
+	}								\
+} while (0)
+
+#define ph_merge_children(a_type, a_field, a_phn, a_cmp, r_phn) do {	\
+	a_type *lchild = phn_lchild_get(a_type, a_field, a_phn);	\
+	if (lchild == NULL) {						\
+		r_phn = NULL;						\
+	} else {							\
+		ph_merge_siblings(a_type, a_field, lchild, a_cmp,	\
+		    r_phn);						\
+	}								\
+} while (0)
+
+/*
+ * The ph_proto() macro generates function prototypes that correspond to the
+ * functions generated by an equivalently parameterized call to ph_gen().
+ */
+#define ph_proto(a_attr, a_prefix, a_ph_type, a_type)			\
+a_attr void	a_prefix##new(a_ph_type *ph);				\
+a_attr bool	a_prefix##empty(a_ph_type *ph);				\
+a_attr a_type	*a_prefix##first(a_ph_type *ph);			\
+a_attr a_type	*a_prefix##any(a_ph_type *ph);				\
+a_attr void	a_prefix##insert(a_ph_type *ph, a_type *phn);		\
+a_attr a_type	*a_prefix##remove_first(a_ph_type *ph);			\
+a_attr a_type	*a_prefix##remove_any(a_ph_type *ph);			\
+a_attr void	a_prefix##remove(a_ph_type *ph, a_type *phn);
+
+/*
+ * The ph_gen() macro generates a type-specific pairing heap implementation,
+ * based on the above cpp macros.
+ */
+#define ph_gen(a_attr, a_prefix, a_ph_type, a_type, a_field, a_cmp)	\
+a_attr void								\
+a_prefix##new(a_ph_type *ph) {						\
+	memset(ph, 0, sizeof(ph(a_type)));				\
+}									\
+a_attr bool								\
+a_prefix##empty(a_ph_type *ph) {					\
+	return (ph->ph_root == NULL);					\
+}									\
+a_attr a_type *								\
+a_prefix##first(a_ph_type *ph) {					\
+	if (ph->ph_root == NULL) {					\
+		return NULL;						\
+	}								\
+	ph_merge_aux(a_type, a_field, ph, a_cmp);			\
+	return ph->ph_root;						\
+}									\
+a_attr a_type *								\
+a_prefix##any(a_ph_type *ph) {						\
+	if (ph->ph_root == NULL) {					\
+		return NULL;						\
+	}								\
+	a_type *aux = phn_next_get(a_type, a_field, ph->ph_root);	\
+	if (aux != NULL) {						\
+		return aux;						\
+	}								\
+	return ph->ph_root;						\
+}									\
+a_attr void								\
+a_prefix##insert(a_ph_type *ph, a_type *phn) {				\
+	memset(&phn->a_field, 0, sizeof(phn(a_type)));			\
+									\
+	/*								\
+	 * Treat the root as an aux list during insertion, and lazily	\
+	 * merge during a_prefix##remove_first().  For elements that	\
+	 * are inserted, then removed via a_prefix##remove() before the	\
+	 * aux list is ever processed, this makes insert/remove		\
+	 * constant-time, whereas eager merging would make insert	\
+	 * O(log n).							\
+	 */								\
+	if (ph->ph_root == NULL) {					\
+		ph->ph_root = phn;					\
+	} else {							\
+		phn_next_set(a_type, a_field, phn, phn_next_get(a_type,	\
+		    a_field, ph->ph_root));				\
+		if (phn_next_get(a_type, a_field, ph->ph_root) !=	\
+		    NULL) {						\
+			phn_prev_set(a_type, a_field,			\
+			    phn_next_get(a_type, a_field, ph->ph_root),	\
+			    phn);					\
+		}							\
+		phn_prev_set(a_type, a_field, phn, ph->ph_root);	\
+		phn_next_set(a_type, a_field, ph->ph_root, phn);	\
+	}								\
+}									\
+a_attr a_type *								\
+a_prefix##remove_first(a_ph_type *ph) {					\
+	a_type *ret;							\
+									\
+	if (ph->ph_root == NULL) {					\
+		return NULL;						\
+	}								\
+	ph_merge_aux(a_type, a_field, ph, a_cmp);			\
+									\
+	ret = ph->ph_root;						\
+									\
+	ph_merge_children(a_type, a_field, ph->ph_root, a_cmp,		\
+	    ph->ph_root);						\
+									\
+	return ret;							\
+}									\
+a_attr a_type *								\
+a_prefix##remove_any(a_ph_type *ph) {					\
+	/*								\
+	 * Remove the most recently inserted aux list element, or the	\
+	 * root if the aux list is empty.  This has the effect of	\
+	 * behaving as a LIFO (and insertion/removal is therefore	\
+	 * constant-time) if a_prefix##[remove_]first() are never	\
+	 * called.							\
+	 */								\
+	if (ph->ph_root == NULL) {					\
+		return NULL;						\
+	}								\
+	a_type *ret = phn_next_get(a_type, a_field, ph->ph_root);	\
+	if (ret != NULL) {						\
+		a_type *aux = phn_next_get(a_type, a_field, ret);	\
+		phn_next_set(a_type, a_field, ph->ph_root, aux);	\
+		if (aux != NULL) {					\
+			phn_prev_set(a_type, a_field, aux,		\
+			    ph->ph_root);				\
+		}							\
+		return ret;						\
+	}								\
+	ret = ph->ph_root;						\
+	ph_merge_children(a_type, a_field, ph->ph_root, a_cmp,		\
+	    ph->ph_root);						\
+	return ret;							\
+}									\
+a_attr void								\
+a_prefix##remove(a_ph_type *ph, a_type *phn) {				\
+	a_type *replace, *parent;					\
+									\
+	if (ph->ph_root == phn) {					\
+		/*							\
+		 * We can delete from aux list without merging it, but	\
+		 * we need to merge if we are dealing with the root	\
+		 * node and it has children.				\
+		 */							\
+		if (phn_lchild_get(a_type, a_field, phn) == NULL) {	\
+			ph->ph_root = phn_next_get(a_type, a_field,	\
+			    phn);					\
+			if (ph->ph_root != NULL) {			\
+				phn_prev_set(a_type, a_field,		\
+				    ph->ph_root, NULL);			\
+			}						\
+			return;						\
+		}							\
+		ph_merge_aux(a_type, a_field, ph, a_cmp);		\
+		if (ph->ph_root == phn) {				\
+			ph_merge_children(a_type, a_field, ph->ph_root,	\
+			    a_cmp, ph->ph_root);			\
+			return;						\
+		}							\
+	}								\
+									\
+	/* Get parent (if phn is leftmost child) before mutating. */	\
+	if ((parent = phn_prev_get(a_type, a_field, phn)) != NULL) {	\
+		if (phn_lchild_get(a_type, a_field, parent) != phn) {	\
+			parent = NULL;					\
+		}							\
+	}								\
+	/* Find a possible replacement node, and link to parent. */	\
+	ph_merge_children(a_type, a_field, phn, a_cmp, replace);	\
+	/* Set next/prev for sibling linked list. */			\
+	if (replace != NULL) {						\
+		if (parent != NULL) {					\
+			phn_prev_set(a_type, a_field, replace, parent);	\
+			phn_lchild_set(a_type, a_field, parent,		\
+			    replace);					\
+		} else {						\
+			phn_prev_set(a_type, a_field, replace,		\
+			    phn_prev_get(a_type, a_field, phn));	\
+			if (phn_prev_get(a_type, a_field, phn) !=	\
+			    NULL) {					\
+				phn_next_set(a_type, a_field,		\
+				    phn_prev_get(a_type, a_field, phn),	\
+				    replace);				\
+			}						\
+		}							\
+		phn_next_set(a_type, a_field, replace,			\
+		    phn_next_get(a_type, a_field, phn));		\
+		if (phn_next_get(a_type, a_field, phn) != NULL) {	\
+			phn_prev_set(a_type, a_field,			\
+			    phn_next_get(a_type, a_field, phn),		\
+			    replace);					\
+		}							\
+	} else {							\
+		if (parent != NULL) {					\
+			a_type *next = phn_next_get(a_type, a_field,	\
+			    phn);					\
+			phn_lchild_set(a_type, a_field, parent, next);	\
+			if (next != NULL) {				\
+				phn_prev_set(a_type, a_field, next,	\
+				    parent);				\
+			}						\
+		} else {						\
+			assert(phn_prev_get(a_type, a_field, phn) !=	\
+			    NULL);					\
+			phn_next_set(a_type, a_field,			\
+			    phn_prev_get(a_type, a_field, phn),		\
+			    phn_next_get(a_type, a_field, phn));	\
+		}							\
+		if (phn_next_get(a_type, a_field, phn) != NULL) {	\
+			phn_prev_set(a_type, a_field,			\
+			    phn_next_get(a_type, a_field, phn),		\
+			    phn_prev_get(a_type, a_field, phn));	\
+		}							\
+	}								\
+}
+
+#endif /* PH_H_ */

include/jemalloc/internal/private_namespace.sh

@@ -1,5 +1,5 @@
 #!/bin/sh
 
-for symbol in `cat $1` ; do
-  echo "#define	${symbol} JEMALLOC_N(${symbol})"
+for symbol in `cat "$@"` ; do
+  echo "#define ${symbol} JEMALLOC_N(${symbol})"
 done

include/jemalloc/internal/private_symbols.sh

@@ -0,0 +1,51 @@
+#!/bin/sh
+#
+# Generate private_symbols[_jet].awk.
+#
+# Usage: private_symbols.sh <sym_prefix> <sym>*
+#
+# <sym_prefix> is typically "" or "_".
+
+sym_prefix=$1
+shift
+
+cat <<EOF
+#!/usr/bin/env awk -f
+
+BEGIN {
+  sym_prefix = "${sym_prefix}"
+  split("\\
+EOF
+
+for public_sym in "$@" ; do
+  cat <<EOF
+        ${sym_prefix}${public_sym} \\
+EOF
+done
+
+cat <<"EOF"
+        ", exported_symbol_names)
+  # Store exported symbol names as keys in exported_symbols.
+  for (i in exported_symbol_names) {
+    exported_symbols[exported_symbol_names[i]] = 1
+  }
+}
+
+# Process 'nm -a <c_source.o>' output.
+#
+# Handle lines like:
+#   0000000000000008 D opt_junk
+#   0000000000007574 T malloc_initialized
+(NF == 3 && $2 ~ /^[ABCDGRSTVW]$/ && !($3 in exported_symbols) && $3 ~ /^[A-Za-z0-9_]+$/) {
+  print substr($3, 1+length(sym_prefix), length($3)-length(sym_prefix))
+}
+
+# Process 'dumpbin /SYMBOLS <c_source.obj>' output.
+#
+# Handle lines like:
+#   353 00008098 SECT4  notype       External     | opt_junk
+#   3F1 00000000 SECT7  notype ()    External     | malloc_initialized
+($3 ~ /^SECT[0-9]+/ && $(NF-2) == "External" && !($NF in exported_symbols)) {
+  print $NF
+}
+EOF

include/jemalloc/internal/private_symbols.txt

@@ -1,413 +0,0 @@
-a0calloc
-a0free
-a0malloc
-arena_alloc_junk_small
-arena_bin_index
-arena_bin_info
-arena_boot
-arena_dalloc
-arena_dalloc_bin
-arena_dalloc_bin_locked
-arena_dalloc_junk_large
-arena_dalloc_junk_small
-arena_dalloc_large
-arena_dalloc_large_locked
-arena_dalloc_small
-arena_dss_prec_get
-arena_dss_prec_set
-arena_malloc
-arena_malloc_large
-arena_malloc_small
-arena_mapbits_allocated_get
-arena_mapbits_binind_get
-arena_mapbits_dirty_get
-arena_mapbits_get
-arena_mapbits_large_binind_set
-arena_mapbits_large_get
-arena_mapbits_large_set
-arena_mapbits_large_size_get
-arena_mapbits_small_runind_get
-arena_mapbits_small_set
-arena_mapbits_unallocated_set
-arena_mapbits_unallocated_size_get
-arena_mapbits_unallocated_size_set
-arena_mapbits_unzeroed_get
-arena_mapbits_unzeroed_set
-arena_mapbitsp_get
-arena_mapbitsp_read
-arena_mapbitsp_write
-arena_mapp_get
-arena_maxclass
-arena_new
-arena_palloc
-arena_postfork_child
-arena_postfork_parent
-arena_prefork
-arena_prof_accum
-arena_prof_accum_impl
-arena_prof_accum_locked
-arena_prof_ctx_get
-arena_prof_ctx_set
-arena_prof_promoted
-arena_ptr_small_binind_get
-arena_purge_all
-arena_quarantine_junk_small
-arena_ralloc
-arena_ralloc_junk_large
-arena_ralloc_no_move
-arena_redzone_corruption
-arena_run_regind
-arena_salloc
-arena_stats_merge
-arena_tcache_fill_small
-arenas
-arenas_booted
-arenas_cleanup
-arenas_extend
-arenas_initialized
-arenas_lock
-arenas_tls
-arenas_tsd
-arenas_tsd_boot
-arenas_tsd_cleanup_wrapper
-arenas_tsd_get
-arenas_tsd_get_wrapper
-arenas_tsd_init_head
-arenas_tsd_set
-atomic_add_u
-atomic_add_uint32
-atomic_add_uint64
-atomic_add_z
-atomic_sub_u
-atomic_sub_uint32
-atomic_sub_uint64
-atomic_sub_z
-base_alloc
-base_boot
-base_calloc
-base_node_alloc
-base_node_dealloc
-base_postfork_child
-base_postfork_parent
-base_prefork
-bitmap_full
-bitmap_get
-bitmap_info_init
-bitmap_info_ngroups
-bitmap_init
-bitmap_set
-bitmap_sfu
-bitmap_size
-bitmap_unset
-bt_init
-buferror
-choose_arena
-choose_arena_hard
-chunk_alloc
-chunk_alloc_dss
-chunk_alloc_mmap
-chunk_boot
-chunk_dealloc
-chunk_dealloc_mmap
-chunk_dss_boot
-chunk_dss_postfork_child
-chunk_dss_postfork_parent
-chunk_dss_prec_get
-chunk_dss_prec_set
-chunk_dss_prefork
-chunk_in_dss
-chunk_npages
-chunk_postfork_child
-chunk_postfork_parent
-chunk_prefork
-chunk_unmap
-chunks_mtx
-chunks_rtree
-chunksize
-chunksize_mask
-ckh_bucket_search
-ckh_count
-ckh_delete
-ckh_evict_reloc_insert
-ckh_insert
-ckh_isearch
-ckh_iter
-ckh_new
-ckh_pointer_hash
-ckh_pointer_keycomp
-ckh_rebuild
-ckh_remove
-ckh_search
-ckh_string_hash
-ckh_string_keycomp
-ckh_try_bucket_insert
-ckh_try_insert
-ctl_boot
-ctl_bymib
-ctl_byname
-ctl_nametomib
-ctl_postfork_child
-ctl_postfork_parent
-ctl_prefork
-dss_prec_names
-extent_tree_ad_first
-extent_tree_ad_insert
-extent_tree_ad_iter
-extent_tree_ad_iter_recurse
-extent_tree_ad_iter_start
-extent_tree_ad_last
-extent_tree_ad_new
-extent_tree_ad_next
-extent_tree_ad_nsearch
-extent_tree_ad_prev
-extent_tree_ad_psearch
-extent_tree_ad_remove
-extent_tree_ad_reverse_iter
-extent_tree_ad_reverse_iter_recurse
-extent_tree_ad_reverse_iter_start
-extent_tree_ad_search
-extent_tree_szad_first
-extent_tree_szad_insert
-extent_tree_szad_iter
-extent_tree_szad_iter_recurse
-extent_tree_szad_iter_start
-extent_tree_szad_last
-extent_tree_szad_new
-extent_tree_szad_next
-extent_tree_szad_nsearch
-extent_tree_szad_prev
-extent_tree_szad_psearch
-extent_tree_szad_remove
-extent_tree_szad_reverse_iter
-extent_tree_szad_reverse_iter_recurse
-extent_tree_szad_reverse_iter_start
-extent_tree_szad_search
-get_errno
-hash
-hash_fmix_32
-hash_fmix_64
-hash_get_block_32
-hash_get_block_64
-hash_rotl_32
-hash_rotl_64
-hash_x64_128
-hash_x86_128
-hash_x86_32
-huge_allocated
-huge_boot
-huge_dalloc
-huge_dalloc_junk
-huge_dss_prec_get
-huge_malloc
-huge_mtx
-huge_ndalloc
-huge_nmalloc
-huge_palloc
-huge_postfork_child
-huge_postfork_parent
-huge_prefork
-huge_prof_ctx_get
-huge_prof_ctx_set
-huge_ralloc
-huge_ralloc_no_move
-huge_salloc
-iallocm
-icalloc
-icalloct
-idalloc
-idalloct
-imalloc
-imalloct
-ipalloc
-ipalloct
-iqalloc
-iqalloct
-iralloc
-iralloct
-iralloct_realign
-isalloc
-isthreaded
-ivsalloc
-ixalloc
-jemalloc_postfork_child
-jemalloc_postfork_parent
-jemalloc_prefork
-malloc_cprintf
-malloc_mutex_init
-malloc_mutex_lock
-malloc_mutex_postfork_child
-malloc_mutex_postfork_parent
-malloc_mutex_prefork
-malloc_mutex_unlock
-malloc_printf
-malloc_snprintf
-malloc_strtoumax
-malloc_tsd_boot
-malloc_tsd_cleanup_register
-malloc_tsd_dalloc
-malloc_tsd_malloc
-malloc_tsd_no_cleanup
-malloc_vcprintf
-malloc_vsnprintf
-malloc_write
-map_bias
-mb_write
-mutex_boot
-narenas_auto
-narenas_total
-narenas_total_get
-ncpus
-nhbins
-opt_abort
-opt_dss
-opt_junk
-opt_lg_chunk
-opt_lg_dirty_mult
-opt_lg_prof_interval
-opt_lg_prof_sample
-opt_lg_tcache_max
-opt_narenas
-opt_prof
-opt_prof_accum
-opt_prof_active
-opt_prof_final
-opt_prof_gdump
-opt_prof_leak
-opt_prof_prefix
-opt_quarantine
-opt_redzone
-opt_stats_print
-opt_tcache
-opt_utrace
-opt_valgrind
-opt_xmalloc
-opt_zero
-p2rz
-pages_purge
-pow2_ceil
-prof_backtrace
-prof_boot0
-prof_boot1
-prof_boot2
-prof_bt_count
-prof_ctx_get
-prof_ctx_set
-prof_dump_open
-prof_free
-prof_gdump
-prof_idump
-prof_interval
-prof_lookup
-prof_malloc
-prof_mdump
-prof_postfork_child
-prof_postfork_parent
-prof_prefork
-prof_promote
-prof_realloc
-prof_sample_accum_update
-prof_sample_threshold_update
-prof_tdata_booted
-prof_tdata_cleanup
-prof_tdata_get
-prof_tdata_init
-prof_tdata_initialized
-prof_tdata_tls
-prof_tdata_tsd
-prof_tdata_tsd_boot
-prof_tdata_tsd_cleanup_wrapper
-prof_tdata_tsd_get
-prof_tdata_tsd_get_wrapper
-prof_tdata_tsd_init_head
-prof_tdata_tsd_set
-quarantine
-quarantine_alloc_hook
-quarantine_boot
-quarantine_booted
-quarantine_cleanup
-quarantine_init
-quarantine_tls
-quarantine_tsd
-quarantine_tsd_boot
-quarantine_tsd_cleanup_wrapper
-quarantine_tsd_get
-quarantine_tsd_get_wrapper
-quarantine_tsd_init_head
-quarantine_tsd_set
-register_zone
-rtree_delete
-rtree_get
-rtree_get_locked
-rtree_new
-rtree_postfork_child
-rtree_postfork_parent
-rtree_prefork
-rtree_set
-s2u
-sa2u
-set_errno
-small_size2bin
-stats_cactive
-stats_cactive_add
-stats_cactive_get
-stats_cactive_sub
-stats_chunks
-stats_print
-tcache_alloc_easy
-tcache_alloc_large
-tcache_alloc_small
-tcache_alloc_small_hard
-tcache_arena_associate
-tcache_arena_dissociate
-tcache_bin_flush_large
-tcache_bin_flush_small
-tcache_bin_info
-tcache_boot0
-tcache_boot1
-tcache_booted
-tcache_create
-tcache_dalloc_large
-tcache_dalloc_small
-tcache_destroy
-tcache_enabled_booted
-tcache_enabled_get
-tcache_enabled_initialized
-tcache_enabled_set
-tcache_enabled_tls
-tcache_enabled_tsd
-tcache_enabled_tsd_boot
-tcache_enabled_tsd_cleanup_wrapper
-tcache_enabled_tsd_get
-tcache_enabled_tsd_get_wrapper
-tcache_enabled_tsd_init_head
-tcache_enabled_tsd_set
-tcache_event
-tcache_event_hard
-tcache_flush
-tcache_get
-tcache_initialized
-tcache_maxclass
-tcache_salloc
-tcache_stats_merge
-tcache_thread_cleanup
-tcache_tls
-tcache_tsd
-tcache_tsd_boot
-tcache_tsd_cleanup_wrapper
-tcache_tsd_get
-tcache_tsd_get_wrapper
-tcache_tsd_init_head
-tcache_tsd_set
-thread_allocated_booted
-thread_allocated_initialized
-thread_allocated_tls
-thread_allocated_tsd
-thread_allocated_tsd_boot
-thread_allocated_tsd_cleanup_wrapper
-thread_allocated_tsd_get
-thread_allocated_tsd_get_wrapper
-thread_allocated_tsd_init_head
-thread_allocated_tsd_set
-tsd_init_check_recursion
-tsd_init_finish
-u2rz

include/jemalloc/internal/private_unnamespace.sh

@@ -1,5 +0,0 @@
-#!/bin/sh
-
-for symbol in `cat $1` ; do
-  echo "#undef ${symbol}"
-done

include/jemalloc/internal/prng.h

@@ -1,5 +1,8 @@
-/******************************************************************************/
-#ifdef JEMALLOC_H_TYPES
+#ifndef JEMALLOC_INTERNAL_PRNG_H
+#define JEMALLOC_INTERNAL_PRNG_H
+
+#include "jemalloc/internal/atomic.h"
+#include "jemalloc/internal/bit_util.h"
 
 /*
  * Simple linear congruential pseudo-random number generator:
@@ -15,46 +18,168 @@
  * 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,
+ * 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 */
+/* INTERNAL DEFINITIONS -- IGNORE */
 /******************************************************************************/
-#ifdef JEMALLOC_H_EXTERNS
+#define PRNG_A_32	UINT32_C(1103515241)
+#define PRNG_C_32	UINT32_C(12347)
 
-#endif /* JEMALLOC_H_EXTERNS */
-/******************************************************************************/
-#ifdef JEMALLOC_H_INLINES
+#define PRNG_A_64	UINT64_C(6364136223846793005)
+#define PRNG_C_64	UINT64_C(1442695040888963407)
+
+JEMALLOC_ALWAYS_INLINE uint32_t
+prng_state_next_u32(uint32_t state) {
+	return (state * PRNG_A_32) + PRNG_C_32;
+}
+
+JEMALLOC_ALWAYS_INLINE uint64_t
+prng_state_next_u64(uint64_t state) {
+	return (state * PRNG_A_64) + PRNG_C_64;
+}
+
+JEMALLOC_ALWAYS_INLINE size_t
+prng_state_next_zu(size_t state) {
+#if LG_SIZEOF_PTR == 2
+	return (state * PRNG_A_32) + PRNG_C_32;
+#elif LG_SIZEOF_PTR == 3
+	return (state * PRNG_A_64) + PRNG_C_64;
+#else
+#error Unsupported pointer size
+#endif
+}
 
-#endif /* JEMALLOC_H_INLINES */
 /******************************************************************************/
+/* BEGIN PUBLIC API */
+/******************************************************************************/
+
+/*
+ * The prng_lg_range functions give a uniform int in the half-open range [0,
+ * 2**lg_range).  If atomic is true, they do so safely from multiple threads.
+ * Multithreaded 64-bit prngs aren't supported.
+ */
+
+JEMALLOC_ALWAYS_INLINE uint32_t
+prng_lg_range_u32(atomic_u32_t *state, unsigned lg_range, bool atomic) {
+	uint32_t ret, state0, state1;
+
+	assert(lg_range > 0);
+	assert(lg_range <= 32);
+
+	state0 = atomic_load_u32(state, ATOMIC_RELAXED);
+
+	if (atomic) {
+		do {
+			state1 = prng_state_next_u32(state0);
+		} while (!atomic_compare_exchange_weak_u32(state, &state0,
+		    state1, ATOMIC_RELAXED, ATOMIC_RELAXED));
+	} else {
+		state1 = prng_state_next_u32(state0);
+		atomic_store_u32(state, state1, ATOMIC_RELAXED);
+	}
+	ret = state1 >> (32 - lg_range);
+
+	return ret;
+}
+
+JEMALLOC_ALWAYS_INLINE uint64_t
+prng_lg_range_u64(uint64_t *state, unsigned lg_range) {
+	uint64_t ret, state1;
+
+	assert(lg_range > 0);
+	assert(lg_range <= 64);
+
+	state1 = prng_state_next_u64(*state);
+	*state = state1;
+	ret = state1 >> (64 - lg_range);
+
+	return ret;
+}
+
+JEMALLOC_ALWAYS_INLINE size_t
+prng_lg_range_zu(atomic_zu_t *state, unsigned lg_range, bool atomic) {
+	size_t ret, state0, state1;
+
+	assert(lg_range > 0);
+	assert(lg_range <= ZU(1) << (3 + LG_SIZEOF_PTR));
+
+	state0 = atomic_load_zu(state, ATOMIC_RELAXED);
+
+	if (atomic) {
+		do {
+			state1 = prng_state_next_zu(state0);
+		} while (atomic_compare_exchange_weak_zu(state, &state0,
+		    state1, ATOMIC_RELAXED, ATOMIC_RELAXED));
+	} else {
+		state1 = prng_state_next_zu(state0);
+		atomic_store_zu(state, state1, ATOMIC_RELAXED);
+	}
+	ret = state1 >> ((ZU(1) << (3 + LG_SIZEOF_PTR)) - lg_range);
+
+	return ret;
+}
+
+/*
+ * The prng_range functions behave like the prng_lg_range, but return a result
+ * in [0, range) instead of [0, 2**lg_range).
+ */
+
+JEMALLOC_ALWAYS_INLINE uint32_t
+prng_range_u32(atomic_u32_t *state, uint32_t range, bool atomic) {
+	uint32_t ret;
+	unsigned lg_range;
+
+	assert(range > 1);
+
+	/* Compute the ceiling of lg(range). */
+	lg_range = ffs_u32(pow2_ceil_u32(range)) - 1;
+
+	/* Generate a result in [0..range) via repeated trial. */
+	do {
+		ret = prng_lg_range_u32(state, lg_range, atomic);
+	} while (ret >= range);
+
+	return ret;
+}
+
+JEMALLOC_ALWAYS_INLINE uint64_t
+prng_range_u64(uint64_t *state, uint64_t range) {
+	uint64_t ret;
+	unsigned lg_range;
+
+	assert(range > 1);
+
+	/* Compute the ceiling of lg(range). */
+	lg_range = ffs_u64(pow2_ceil_u64(range)) - 1;
+
+	/* Generate a result in [0..range) via repeated trial. */
+	do {
+		ret = prng_lg_range_u64(state, lg_range);
+	} while (ret >= range);
+
+	return ret;
+}
+
+JEMALLOC_ALWAYS_INLINE size_t
+prng_range_zu(atomic_zu_t *state, size_t range, bool atomic) {
+	size_t ret;
+	unsigned lg_range;
+
+	assert(range > 1);
+
+	/* Compute the ceiling of lg(range). */
+	lg_range = ffs_u64(pow2_ceil_u64(range)) - 1;
+
+	/* Generate a result in [0..range) via repeated trial. */
+	do {
+		ret = prng_lg_range_zu(state, lg_range, atomic);
+	} while (ret >= range);
+
+	return ret;
+}
+
+#endif /* JEMALLOC_INTERNAL_PRNG_H */

include/jemalloc/internal/prof.h

@@ -1,613 +0,0 @@
-/******************************************************************************/
-#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. */
-#ifdef JEMALLOC_PROF
-#  define PROF_PREFIX_DEFAULT		"jeprof"
-#else
-#  define PROF_PREFIX_DEFAULT		""
-#endif
-#define	LG_PROF_SAMPLE_DEFAULT		19
-#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
-
-/*
- * prof_tdata pointers close to NULL are used to encode state information that
- * is used for cleaning up during thread shutdown.
- */
-#define	PROF_TDATA_STATE_REINCARNATED	((prof_tdata_t *)(uintptr_t)1)
-#define	PROF_TDATA_STATE_PURGATORY	((prof_tdata_t *)(uintptr_t)2)
-#define	PROF_TDATA_STATE_MAX		PROF_TDATA_STATE_PURGATORY
-
-#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 nlimbo, cnt_merged, and cnts_ql. */
-	malloc_mutex_t		*lock;
-
-	/*
-	 * Number of threads that currently cause this ctx to be in a state of
-	 * limbo due to one of:
-	 *   - Initializing per thread counters associated with this ctx.
-	 *   - Preparing to destroy this ctx.
-	 *   - Dumping a heap profile that includes this ctx.
-	 * nlimbo must be 1 (single destroyer) in order to safely destroy the
-	 * ctx.
-	 */
-	unsigned		nlimbo;
-
-	/* 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;
-
-	/* Linkage for list of contexts to be dumped. */
-	ql_elm(prof_ctx_t)	dump_link;
-};
-typedef ql_head(prof_ctx_t) prof_ctx_list_t;
-
-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;
-
-	/* State used to avoid dumping while operating on prof internals. */
-	bool			enq;
-	bool			enq_idump;
-	bool			enq_gdump;
-};
-
-#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_final;       /* Final profile dumping. */
-extern bool	opt_prof_leak;        /* Dump leak summary at exit. */
-extern bool	opt_prof_accum;       /* Report cumulative bytes. */
-extern char	opt_prof_prefix[
-    /* Minimize memory bloat for non-prof builds. */
-#ifdef JEMALLOC_PROF
-    PATH_MAX +
-#endif
-    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);
-#ifdef JEMALLOC_JET
-size_t	prof_bt_count(void);
-typedef int (prof_dump_open_t)(bool, const char *);
-extern prof_dump_open_t *prof_dump_open;
-#endif
-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);
-void	prof_prefork(void);
-void	prof_postfork_parent(void);
-void	prof_postfork_child(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_get(true);				\
-	if ((uintptr_t)prof_tdata <= (uintptr_t)PROF_TDATA_STATE_MAX) {	\
-		if (prof_tdata != NULL)					\
-			ret = (prof_thr_cnt_t *)(uintptr_t)1U;		\
-		else							\
-			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 *)
-
-prof_tdata_t	*prof_tdata_get(bool create);
-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, size_t usize, prof_ctx_t *ctx);
-bool	prof_sample_accum_update(size_t size);
-void	prof_malloc(const void *ptr, size_t usize, prof_thr_cnt_t *cnt);
-void	prof_realloc(const void *ptr, size_t usize, prof_thr_cnt_t *cnt,
-    size_t old_usize, 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 prof_tdata_t *
-prof_tdata_get(bool create)
-{
-	prof_tdata_t *prof_tdata;
-
-	cassert(config_prof);
-
-	prof_tdata = *prof_tdata_tsd_get();
-	if (create && prof_tdata == NULL)
-		prof_tdata = prof_tdata_init();
-
-	return (prof_tdata);
-}
-
-JEMALLOC_INLINE void
-prof_sample_threshold_update(prof_tdata_t *prof_tdata)
-{
-	/*
-	 * The body of this function is compiled out unless heap profiling is
-	 * enabled, so that it is possible to compile jemalloc with floating
-	 * point support completely disabled.  Avoiding floating point code is
-	 * important on memory-constrained systems, but it also enables a
-	 * workaround for versions of glibc that don't properly save/restore
-	 * floating point registers during dynamic lazy symbol loading (which
-	 * internally calls into whatever malloc implementation happens to be
-	 * integrated into the application).  Note that some compilers (e.g.
-	 * gcc 4.8) may use floating point registers for fast memory moves, so
-	 * jemalloc must be compiled with such optimizations disabled (e.g.
-	 * -mno-sse) in order for the workaround to be complete.
-	 */
-#ifdef JEMALLOC_PROF
-	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://luc.devroye.org/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;
-#endif
-}
-
-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, size_t usize, 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, usize, 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_get(false);
-	if ((uintptr_t)prof_tdata <= (uintptr_t)PROF_TDATA_STATE_MAX)
-		return (true);
-
-	/* 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 usize, prof_thr_cnt_t *cnt)
-{
-
-	cassert(config_prof);
-	assert(ptr != NULL);
-	assert(usize == isalloc(ptr, true));
-
-	if (opt_lg_prof_sample != 0) {
-		if (prof_sample_accum_update(usize)) {
-			/*
-			 * 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 usize 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, usize, cnt->ctx);
-
-		cnt->epoch++;
-		/*********/
-		mb_write();
-		/*********/
-		cnt->cnts.curobjs++;
-		cnt->cnts.curbytes += usize;
-		if (opt_prof_accum) {
-			cnt->cnts.accumobjs++;
-			cnt->cnts.accumbytes += usize;
-		}
-		/*********/
-		mb_write();
-		/*********/
-		cnt->epoch++;
-		/*********/
-		mb_write();
-		/*********/
-	} else
-		prof_ctx_set(ptr, usize, (prof_ctx_t *)(uintptr_t)1U);
-}
-
-JEMALLOC_INLINE void
-prof_realloc(const void *ptr, size_t usize, prof_thr_cnt_t *cnt,
-    size_t old_usize, 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(usize == isalloc(ptr, true));
-		if (opt_lg_prof_sample != 0) {
-			if (prof_sample_accum_update(usize)) {
-				/*
-				 * Don't sample.  The usize passed to
-				 * PROF_ALLOC_PREP() was larger than what
-				 * actually got allocated, so a backtrace was
-				 * captured for this allocation, even though
-				 * its actual usize 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_usize;
-			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, usize, cnt->ctx);
-		cnt->epoch++;
-	} else if (ptr != NULL)
-		prof_ctx_set(ptr, usize, (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_usize;
-	}
-	if ((uintptr_t)cnt > (uintptr_t)1U) {
-		cnt->cnts.curobjs++;
-		cnt->cnts.curbytes += usize;
-		if (opt_prof_accum) {
-			cnt->cnts.accumobjs++;
-			cnt->cnts.accumbytes += usize;
-		}
-	}
-	/*********/
-	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) {
-		prof_thr_cnt_t *tcnt;
-		assert(size == isalloc(ptr, true));
-		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 */
-/******************************************************************************/

include/jemalloc/internal/prof_externs.h

@@ -0,0 +1,92 @@
+#ifndef JEMALLOC_INTERNAL_PROF_EXTERNS_H
+#define JEMALLOC_INTERNAL_PROF_EXTERNS_H
+
+#include "jemalloc/internal/mutex.h"
+
+extern malloc_mutex_t	bt2gctx_mtx;
+
+extern bool	opt_prof;
+extern bool	opt_prof_active;
+extern bool	opt_prof_thread_active_init;
+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_final;       /* Final profile dumping. */
+extern bool	opt_prof_leak;        /* Dump leak summary at exit. */
+extern bool	opt_prof_accum;       /* Report cumulative bytes. */
+extern char	opt_prof_prefix[
+    /* Minimize memory bloat for non-prof builds. */
+#ifdef JEMALLOC_PROF
+    PATH_MAX +
+#endif
+    1];
+
+/* Accessed via prof_active_[gs]et{_unlocked,}(). */
+extern bool	prof_active;
+
+/* Accessed via prof_gdump_[gs]et{_unlocked,}(). */
+extern bool	prof_gdump_val;
+
+/*
+ * 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;
+
+/*
+ * Initialized as opt_lg_prof_sample, and potentially modified during profiling
+ * resets.
+ */
+extern size_t	lg_prof_sample;
+
+void prof_alloc_rollback(tsd_t *tsd, prof_tctx_t *tctx, bool updated);
+void prof_malloc_sample_object(tsdn_t *tsdn, const void *ptr, size_t usize,
+    prof_tctx_t *tctx);
+void prof_free_sampled_object(tsd_t *tsd, size_t usize, prof_tctx_t *tctx);
+void bt_init(prof_bt_t *bt, void **vec);
+void prof_backtrace(prof_bt_t *bt);
+prof_tctx_t *prof_lookup(tsd_t *tsd, prof_bt_t *bt);
+#ifdef JEMALLOC_JET
+size_t prof_tdata_count(void);
+size_t prof_bt_count(void);
+#endif
+typedef int (prof_dump_open_t)(bool, const char *);
+extern prof_dump_open_t *JET_MUTABLE prof_dump_open;
+
+typedef bool (prof_dump_header_t)(tsdn_t *, bool, const prof_cnt_t *);
+extern prof_dump_header_t *JET_MUTABLE prof_dump_header;
+#ifdef JEMALLOC_JET
+void prof_cnt_all(uint64_t *curobjs, uint64_t *curbytes, uint64_t *accumobjs,
+    uint64_t *accumbytes);
+#endif
+bool prof_accum_init(tsdn_t *tsdn, prof_accum_t *prof_accum);
+void prof_idump(tsdn_t *tsdn);
+bool prof_mdump(tsd_t *tsd, const char *filename);
+void prof_gdump(tsdn_t *tsdn);
+prof_tdata_t *prof_tdata_init(tsd_t *tsd);
+prof_tdata_t *prof_tdata_reinit(tsd_t *tsd, prof_tdata_t *tdata);
+void prof_reset(tsd_t *tsd, size_t lg_sample);
+void prof_tdata_cleanup(tsd_t *tsd);
+bool prof_active_get(tsdn_t *tsdn);
+bool prof_active_set(tsdn_t *tsdn, bool active);
+const char *prof_thread_name_get(tsd_t *tsd);
+int prof_thread_name_set(tsd_t *tsd, const char *thread_name);
+bool prof_thread_active_get(tsd_t *tsd);
+bool prof_thread_active_set(tsd_t *tsd, bool active);
+bool prof_thread_active_init_get(tsdn_t *tsdn);
+bool prof_thread_active_init_set(tsdn_t *tsdn, bool active_init);
+bool prof_gdump_get(tsdn_t *tsdn);
+bool prof_gdump_set(tsdn_t *tsdn, bool active);
+void prof_boot0(void);
+void prof_boot1(void);
+bool prof_boot2(tsd_t *tsd);
+void prof_prefork0(tsdn_t *tsdn);
+void prof_prefork1(tsdn_t *tsdn);
+void prof_postfork_parent(tsdn_t *tsdn);
+void prof_postfork_child(tsdn_t *tsdn);
+void prof_sample_threshold_update(prof_tdata_t *tdata);
+
+#endif /* JEMALLOC_INTERNAL_PROF_EXTERNS_H */

include/jemalloc/internal/prof_inlines_a.h

@@ -0,0 +1,83 @@
+#ifndef JEMALLOC_INTERNAL_PROF_INLINES_A_H
+#define JEMALLOC_INTERNAL_PROF_INLINES_A_H
+
+#include "jemalloc/internal/mutex.h"
+
+static inline bool
+prof_accum_add(tsdn_t *tsdn, prof_accum_t *prof_accum, uint64_t accumbytes) {
+	cassert(config_prof);
+
+	bool overflow;
+	uint64_t a0, a1;
+
+	/*
+	 * If the application allocates fast enough (and/or if idump is slow
+	 * enough), extreme overflow here (a1 >= prof_interval * 2) can cause
+	 * idump trigger coalescing.  This is an intentional mechanism that
+	 * avoids rate-limiting allocation.
+	 */
+#ifdef JEMALLOC_ATOMIC_U64
+	a0 = atomic_load_u64(&prof_accum->accumbytes, ATOMIC_RELAXED);
+	do {
+		a1 = a0 + accumbytes;
+		assert(a1 >= a0);
+		overflow = (a1 >= prof_interval);
+		if (overflow) {
+			a1 %= prof_interval;
+		}
+	} while (!atomic_compare_exchange_weak_u64(&prof_accum->accumbytes, &a0,
+	    a1, ATOMIC_RELAXED, ATOMIC_RELAXED));
+#else
+	malloc_mutex_lock(tsdn, &prof_accum->mtx);
+	a0 = prof_accum->accumbytes;
+	a1 = a0 + accumbytes;
+	overflow = (a1 >= prof_interval);
+	if (overflow) {
+		a1 %= prof_interval;
+	}
+	prof_accum->accumbytes = a1;
+	malloc_mutex_unlock(tsdn, &prof_accum->mtx);
+#endif
+	return overflow;
+}
+
+static inline void
+prof_accum_cancel(tsdn_t *tsdn, prof_accum_t *prof_accum, size_t usize) {
+	cassert(config_prof);
+
+	/*
+	 * Cancel out as much of the excessive prof_accumbytes increase as
+	 * possible without underflowing.  Interval-triggered dumps occur
+	 * slightly more often than intended as a result of incomplete
+	 * canceling.
+	 */
+	uint64_t a0, a1;
+#ifdef JEMALLOC_ATOMIC_U64
+	a0 = atomic_load_u64(&prof_accum->accumbytes, ATOMIC_RELAXED);
+	do {
+		a1 = (a0 >= LARGE_MINCLASS - usize) ?  a0 - (LARGE_MINCLASS -
+		    usize) : 0;
+	} while (!atomic_compare_exchange_weak_u64(&prof_accum->accumbytes, &a0,
+	    a1, ATOMIC_RELAXED, ATOMIC_RELAXED));
+#else
+	malloc_mutex_lock(tsdn, &prof_accum->mtx);
+	a0 = prof_accum->accumbytes;
+	a1 = (a0 >= LARGE_MINCLASS - usize) ?  a0 - (LARGE_MINCLASS - usize) :
+	    0;
+	prof_accum->accumbytes = a1;
+	malloc_mutex_unlock(tsdn, &prof_accum->mtx);
+#endif
+}
+
+JEMALLOC_ALWAYS_INLINE bool
+prof_active_get_unlocked(void) {
+	/*
+	 * Even if opt_prof is true, sampling can be temporarily disabled by
+	 * setting prof_active to false.  No locking is used when reading
+	 * prof_active in the fast path, so there are no guarantees regarding
+	 * how long it will take for all threads to notice state changes.
+	 */
+	return prof_active;
+}
+
+#endif /* JEMALLOC_INTERNAL_PROF_INLINES_A_H */

include/jemalloc/internal/prof_inlines_b.h

@@ -0,0 +1,206 @@
+#ifndef JEMALLOC_INTERNAL_PROF_INLINES_B_H
+#define JEMALLOC_INTERNAL_PROF_INLINES_B_H
+
+#include "jemalloc/internal/sz.h"
+
+JEMALLOC_ALWAYS_INLINE bool
+prof_gdump_get_unlocked(void) {
+	/*
+	 * No locking is used when reading prof_gdump_val in the fast path, so
+	 * there are no guarantees regarding how long it will take for all
+	 * threads to notice state changes.
+	 */
+	return prof_gdump_val;
+}
+
+JEMALLOC_ALWAYS_INLINE prof_tdata_t *
+prof_tdata_get(tsd_t *tsd, bool create) {
+	prof_tdata_t *tdata;
+
+	cassert(config_prof);
+
+	tdata = tsd_prof_tdata_get(tsd);
+	if (create) {
+		if (unlikely(tdata == NULL)) {
+			if (tsd_nominal(tsd)) {
+				tdata = prof_tdata_init(tsd);
+				tsd_prof_tdata_set(tsd, tdata);
+			}
+		} else if (unlikely(tdata->expired)) {
+			tdata = prof_tdata_reinit(tsd, tdata);
+			tsd_prof_tdata_set(tsd, tdata);
+		}
+		assert(tdata == NULL || tdata->attached);
+	}
+
+	return tdata;
+}
+
+JEMALLOC_ALWAYS_INLINE prof_tctx_t *
+prof_tctx_get(tsdn_t *tsdn, const void *ptr, alloc_ctx_t *alloc_ctx) {
+	cassert(config_prof);
+	assert(ptr != NULL);
+
+	return arena_prof_tctx_get(tsdn, ptr, alloc_ctx);
+}
+
+JEMALLOC_ALWAYS_INLINE void
+prof_tctx_set(tsdn_t *tsdn, const void *ptr, size_t usize,
+    alloc_ctx_t *alloc_ctx, prof_tctx_t *tctx) {
+	cassert(config_prof);
+	assert(ptr != NULL);
+
+	arena_prof_tctx_set(tsdn, ptr, usize, alloc_ctx, tctx);
+}
+
+JEMALLOC_ALWAYS_INLINE void
+prof_tctx_reset(tsdn_t *tsdn, const void *ptr, prof_tctx_t *tctx) {
+	cassert(config_prof);
+	assert(ptr != NULL);
+
+	arena_prof_tctx_reset(tsdn, ptr, tctx);
+}
+
+JEMALLOC_ALWAYS_INLINE bool
+prof_sample_accum_update(tsd_t *tsd, size_t usize, bool update,
+    prof_tdata_t **tdata_out) {
+	prof_tdata_t *tdata;
+
+	cassert(config_prof);
+
+	tdata = prof_tdata_get(tsd, true);
+	if (unlikely((uintptr_t)tdata <= (uintptr_t)PROF_TDATA_STATE_MAX)) {
+		tdata = NULL;
+	}
+
+	if (tdata_out != NULL) {
+		*tdata_out = tdata;
+	}
+
+	if (unlikely(tdata == NULL)) {
+		return true;
+	}
+
+	if (likely(tdata->bytes_until_sample >= usize)) {
+		if (update) {
+			tdata->bytes_until_sample -= usize;
+		}
+		return true;
+	} else {
+		if (tsd_reentrancy_level_get(tsd) > 0) {
+			return true;
+		}
+		/* Compute new sample threshold. */
+		if (update) {
+			prof_sample_threshold_update(tdata);
+		}
+		return !tdata->active;
+	}
+}
+
+JEMALLOC_ALWAYS_INLINE prof_tctx_t *
+prof_alloc_prep(tsd_t *tsd, size_t usize, bool prof_active, bool update) {
+	prof_tctx_t *ret;
+	prof_tdata_t *tdata;
+	prof_bt_t bt;
+
+	assert(usize == sz_s2u(usize));
+
+	if (!prof_active || likely(prof_sample_accum_update(tsd, usize, update,
+	    &tdata))) {
+		ret = (prof_tctx_t *)(uintptr_t)1U;
+	} else {
+		bt_init(&bt, tdata->vec);
+		prof_backtrace(&bt);
+		ret = prof_lookup(tsd, &bt);
+	}
+
+	return ret;
+}
+
+JEMALLOC_ALWAYS_INLINE void
+prof_malloc(tsdn_t *tsdn, const void *ptr, size_t usize, alloc_ctx_t *alloc_ctx,
+    prof_tctx_t *tctx) {
+	cassert(config_prof);
+	assert(ptr != NULL);
+	assert(usize == isalloc(tsdn, ptr));
+
+	if (unlikely((uintptr_t)tctx > (uintptr_t)1U)) {
+		prof_malloc_sample_object(tsdn, ptr, usize, tctx);
+	} else {
+		prof_tctx_set(tsdn, ptr, usize, alloc_ctx,
+		    (prof_tctx_t *)(uintptr_t)1U);
+	}
+}
+
+JEMALLOC_ALWAYS_INLINE void
+prof_realloc(tsd_t *tsd, const void *ptr, size_t usize, prof_tctx_t *tctx,
+    bool prof_active, bool updated, const void *old_ptr, size_t old_usize,
+    prof_tctx_t *old_tctx) {
+	bool sampled, old_sampled, moved;
+
+	cassert(config_prof);
+	assert(ptr != NULL || (uintptr_t)tctx <= (uintptr_t)1U);
+
+	if (prof_active && !updated && ptr != NULL) {
+		assert(usize == isalloc(tsd_tsdn(tsd), ptr));
+		if (prof_sample_accum_update(tsd, usize, true, NULL)) {
+			/*
+			 * Don't sample.  The usize passed to prof_alloc_prep()
+			 * was larger than what actually got allocated, so a
+			 * backtrace was captured for this allocation, even
+			 * though its actual usize was insufficient to cross the
+			 * sample threshold.
+			 */
+			prof_alloc_rollback(tsd, tctx, true);
+			tctx = (prof_tctx_t *)(uintptr_t)1U;
+		}
+	}
+
+	sampled = ((uintptr_t)tctx > (uintptr_t)1U);
+	old_sampled = ((uintptr_t)old_tctx > (uintptr_t)1U);
+	moved = (ptr != old_ptr);
+
+	if (unlikely(sampled)) {
+		prof_malloc_sample_object(tsd_tsdn(tsd), ptr, usize, tctx);
+	} else if (moved) {
+		prof_tctx_set(tsd_tsdn(tsd), ptr, usize, NULL,
+		    (prof_tctx_t *)(uintptr_t)1U);
+	} else if (unlikely(old_sampled)) {
+		/*
+		 * prof_tctx_set() would work for the !moved case as well, but
+		 * prof_tctx_reset() is slightly cheaper, and the proper thing
+		 * to do here in the presence of explicit knowledge re: moved
+		 * state.
+		 */
+		prof_tctx_reset(tsd_tsdn(tsd), ptr, tctx);
+	} else {
+		assert((uintptr_t)prof_tctx_get(tsd_tsdn(tsd), ptr, NULL) ==
+		    (uintptr_t)1U);
+	}
+
+	/*
+	 * The prof_free_sampled_object() call must come after the
+	 * prof_malloc_sample_object() call, because tctx and old_tctx may be
+	 * the same, in which case reversing the call order could cause the tctx
+	 * to be prematurely destroyed as a side effect of momentarily zeroed
+	 * counters.
+	 */
+	if (unlikely(old_sampled)) {
+		prof_free_sampled_object(tsd, old_usize, old_tctx);
+	}
+}
+
+JEMALLOC_ALWAYS_INLINE void
+prof_free(tsd_t *tsd, const void *ptr, size_t usize, alloc_ctx_t *alloc_ctx) {
+	prof_tctx_t *tctx = prof_tctx_get(tsd_tsdn(tsd), ptr, alloc_ctx);
+
+	cassert(config_prof);
+	assert(usize == isalloc(tsd_tsdn(tsd), ptr));
+
+	if (unlikely((uintptr_t)tctx > (uintptr_t)1U)) {
+		prof_free_sampled_object(tsd, usize, tctx);
+	}
+}
+
+#endif /* JEMALLOC_INTERNAL_PROF_INLINES_B_H */

include/jemalloc/internal/prof_structs.h

@@ -0,0 +1,201 @@
+#ifndef JEMALLOC_INTERNAL_PROF_STRUCTS_H
+#define JEMALLOC_INTERNAL_PROF_STRUCTS_H
+
+#include "jemalloc/internal/ckh.h"
+#include "jemalloc/internal/mutex.h"
+#include "jemalloc/internal/prng.h"
+#include "jemalloc/internal/rb.h"
+
+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	max;
+} prof_unwind_data_t;
+#endif
+
+struct prof_accum_s {
+#ifndef JEMALLOC_ATOMIC_U64
+	malloc_mutex_t	mtx;
+	uint64_t	accumbytes;
+#else
+	atomic_u64_t	accumbytes;
+#endif
+};
+
+struct prof_cnt_s {
+	/* Profiling counters. */
+	uint64_t	curobjs;
+	uint64_t	curbytes;
+	uint64_t	accumobjs;
+	uint64_t	accumbytes;
+};
+
+typedef enum {
+	prof_tctx_state_initializing,
+	prof_tctx_state_nominal,
+	prof_tctx_state_dumping,
+	prof_tctx_state_purgatory /* Dumper must finish destroying. */
+} prof_tctx_state_t;
+
+struct prof_tctx_s {
+	/* Thread data for thread that performed the allocation. */
+	prof_tdata_t		*tdata;
+
+	/*
+	 * Copy of tdata->thr_{uid,discrim}, necessary because tdata may be
+	 * defunct during teardown.
+	 */
+	uint64_t		thr_uid;
+	uint64_t		thr_discrim;
+
+	/* Profiling counters, protected by tdata->lock. */
+	prof_cnt_t		cnts;
+
+	/* Associated global context. */
+	prof_gctx_t		*gctx;
+
+	/*
+	 * UID that distinguishes multiple tctx's created by the same thread,
+	 * but coexisting in gctx->tctxs.  There are two ways that such
+	 * coexistence can occur:
+	 * - A dumper thread can cause a tctx to be retained in the purgatory
+	 *   state.
+	 * - Although a single "producer" thread must create all tctx's which
+	 *   share the same thr_uid, multiple "consumers" can each concurrently
+	 *   execute portions of prof_tctx_destroy().  prof_tctx_destroy() only
+	 *   gets called once each time cnts.cur{objs,bytes} drop to 0, but this
+	 *   threshold can be hit again before the first consumer finishes
+	 *   executing prof_tctx_destroy().
+	 */
+	uint64_t		tctx_uid;
+
+	/* Linkage into gctx's tctxs. */
+	rb_node(prof_tctx_t)	tctx_link;
+
+	/*
+	 * True during prof_alloc_prep()..prof_malloc_sample_object(), prevents
+	 * sample vs destroy race.
+	 */
+	bool			prepared;
+
+	/* Current dump-related state, protected by gctx->lock. */
+	prof_tctx_state_t	state;
+
+	/*
+	 * Copy of cnts snapshotted during early dump phase, protected by
+	 * dump_mtx.
+	 */
+	prof_cnt_t		dump_cnts;
+};
+typedef rb_tree(prof_tctx_t) prof_tctx_tree_t;
+
+struct prof_gctx_s {
+	/* Protects nlimbo, cnt_summed, and tctxs. */
+	malloc_mutex_t		*lock;
+
+	/*
+	 * Number of threads that currently cause this gctx to be in a state of
+	 * limbo due to one of:
+	 *   - Initializing this gctx.
+	 *   - Initializing per thread counters associated with this gctx.
+	 *   - Preparing to destroy this gctx.
+	 *   - Dumping a heap profile that includes this gctx.
+	 * nlimbo must be 1 (single destroyer) in order to safely destroy the
+	 * gctx.
+	 */
+	unsigned		nlimbo;
+
+	/*
+	 * Tree of profile counters, one for each thread that has allocated in
+	 * this context.
+	 */
+	prof_tctx_tree_t	tctxs;
+
+	/* Linkage for tree of contexts to be dumped. */
+	rb_node(prof_gctx_t)	dump_link;
+
+	/* Temporary storage for summation during dump. */
+	prof_cnt_t		cnt_summed;
+
+	/* Associated backtrace. */
+	prof_bt_t		bt;
+
+	/* Backtrace vector, variable size, referred to by bt. */
+	void			*vec[1];
+};
+typedef rb_tree(prof_gctx_t) prof_gctx_tree_t;
+
+struct prof_tdata_s {
+	malloc_mutex_t		*lock;
+
+	/* Monotonically increasing unique thread identifier. */
+	uint64_t		thr_uid;
+
+	/*
+	 * Monotonically increasing discriminator among tdata structures
+	 * associated with the same thr_uid.
+	 */
+	uint64_t		thr_discrim;
+
+	/* Included in heap profile dumps if non-NULL. */
+	char			*thread_name;
+
+	bool			attached;
+	bool			expired;
+
+	rb_node(prof_tdata_t)	tdata_link;
+
+	/*
+	 * Counter used to initialize prof_tctx_t's tctx_uid.  No locking is
+	 * necessary when incrementing this field, because only one thread ever
+	 * does so.
+	 */
+	uint64_t		tctx_uid_next;
+
+	/*
+	 * Hash of (prof_bt_t *)-->(prof_tctx_t *).  Each thread tracks
+	 * backtraces for which it has non-zero allocation/deallocation counters
+	 * associated with thread-specific prof_tctx_t objects.  Other threads
+	 * may write to prof_tctx_t contents when freeing associated objects.
+	 */
+	ckh_t			bt2tctx;
+
+	/* Sampling state. */
+	uint64_t		prng_state;
+	uint64_t		bytes_until_sample;
+
+	/* State used to avoid dumping while operating on prof internals. */
+	bool			enq;
+	bool			enq_idump;
+	bool			enq_gdump;
+
+	/*
+	 * Set to true during an early dump phase for tdata's which are
+	 * currently being dumped.  New threads' tdata's have this initialized
+	 * to false so that they aren't accidentally included in later dump
+	 * phases.
+	 */
+	bool			dumping;
+
+	/*
+	 * True if profiling is active for this tdata's thread
+	 * (thread.prof.active mallctl).
+	 */
+	bool			active;
+
+	/* Temporary storage for summation during dump. */
+	prof_cnt_t		cnt_summed;
+
+	/* Backtrace vector, used for calls to prof_backtrace(). */
+	void			*vec[PROF_BT_MAX];
+};
+typedef rb_tree(prof_tdata_t) prof_tdata_tree_t;
+
+#endif /* JEMALLOC_INTERNAL_PROF_STRUCTS_H */

include/jemalloc/internal/prof_types.h

@@ -0,0 +1,56 @@
+#ifndef JEMALLOC_INTERNAL_PROF_TYPES_H
+#define JEMALLOC_INTERNAL_PROF_TYPES_H
+
+typedef struct prof_bt_s prof_bt_t;
+typedef struct prof_accum_s prof_accum_t;
+typedef struct prof_cnt_s prof_cnt_t;
+typedef struct prof_tctx_s prof_tctx_t;
+typedef struct prof_gctx_s prof_gctx_t;
+typedef struct prof_tdata_s prof_tdata_t;
+
+/* Option defaults. */
+#ifdef JEMALLOC_PROF
+#  define PROF_PREFIX_DEFAULT		"jeprof"
+#else
+#  define PROF_PREFIX_DEFAULT		""
+#endif
+#define LG_PROF_SAMPLE_DEFAULT		19
+#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
+
+/* 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 gctx's.  No space is allocated for these
+ * unless profiling is enabled, so it's okay to over-provision.
+ */
+#define PROF_NCTX_LOCKS			1024
+
+/*
+ * Number of mutexes shared among all tdata's.  No space is allocated for these
+ * unless profiling is enabled, so it's okay to over-provision.
+ */
+#define PROF_NTDATA_LOCKS		256
+
+/*
+ * prof_tdata pointers close to NULL are used to encode state information that
+ * is used for cleaning up during thread shutdown.
+ */
+#define PROF_TDATA_STATE_REINCARNATED	((prof_tdata_t *)(uintptr_t)1)
+#define PROF_TDATA_STATE_PURGATORY	((prof_tdata_t *)(uintptr_t)2)
+#define PROF_TDATA_STATE_MAX		PROF_TDATA_STATE_PURGATORY
+
+#endif /* JEMALLOC_INTERNAL_PROF_TYPES_H */

include/jemalloc/internal/public_namespace.sh

@@ -2,5 +2,5 @@
 
 for nm in `cat $1` ; do
   n=`echo ${nm} |tr ':' ' ' |awk '{print $1}'`
-  echo "#define	je_${n} JEMALLOC_N(${n})"
+  echo "#define je_${n} JEMALLOC_N(${n})"
 done

include/jemalloc/internal/ql.h

@@ -1,61 +1,64 @@
-/*
- * List definitions.
- */
-#define	ql_head(a_type)							\
+#ifndef JEMALLOC_INTERNAL_QL_H
+#define JEMALLOC_INTERNAL_QL_H
+
+#include "jemalloc/internal/qr.h"
+
+/* List definitions. */
+#define ql_head(a_type)							\
 struct {								\
 	a_type *qlh_first;						\
 }
 
-#define	ql_head_initializer(a_head) {NULL}
+#define ql_head_initializer(a_head) {NULL}
 
-#define	ql_elm(a_type)	qr(a_type)
+#define ql_elm(a_type)	qr(a_type)
 
 /* List functions. */
-#define	ql_new(a_head) do {						\
+#define ql_new(a_head) do {						\
 	(a_head)->qlh_first = NULL;					\
 } while (0)
 
-#define	ql_elm_new(a_elm, a_field) qr_new((a_elm), a_field)
+#define ql_elm_new(a_elm, a_field) qr_new((a_elm), a_field)
 
-#define	ql_first(a_head) ((a_head)->qlh_first)
+#define ql_first(a_head) ((a_head)->qlh_first)
 
-#define	ql_last(a_head, a_field)					\
+#define ql_last(a_head, a_field)					\
 	((ql_first(a_head) != NULL)					\
 	    ? qr_prev(ql_first(a_head), a_field) : NULL)
 
-#define	ql_next(a_head, a_elm, a_field)					\
+#define ql_next(a_head, a_elm, a_field)					\
 	((ql_last(a_head, a_field) != (a_elm))				\
 	    ? qr_next((a_elm), a_field)	: NULL)
 
-#define	ql_prev(a_head, a_elm, a_field)					\
+#define ql_prev(a_head, a_elm, a_field)					\
 	((ql_first(a_head) != (a_elm)) ? qr_prev((a_elm), a_field)	\
 				       : NULL)
 
-#define	ql_before_insert(a_head, a_qlelm, a_elm, a_field) do {		\
+#define ql_before_insert(a_head, a_qlelm, a_elm, a_field) do {		\
 	qr_before_insert((a_qlelm), (a_elm), a_field);			\
 	if (ql_first(a_head) == (a_qlelm)) {				\
 		ql_first(a_head) = (a_elm);				\
 	}								\
 } while (0)
 
-#define	ql_after_insert(a_qlelm, a_elm, a_field)			\
+#define ql_after_insert(a_qlelm, a_elm, a_field)			\
 	qr_after_insert((a_qlelm), (a_elm), a_field)
 
-#define	ql_head_insert(a_head, a_elm, a_field) do {			\
+#define ql_head_insert(a_head, a_elm, a_field) do {			\
 	if (ql_first(a_head) != NULL) {					\
 		qr_before_insert(ql_first(a_head), (a_elm), a_field);	\
 	}								\
 	ql_first(a_head) = (a_elm);					\
 } while (0)
 
-#define	ql_tail_insert(a_head, a_elm, a_field) do {			\
+#define ql_tail_insert(a_head, a_elm, a_field) do {			\
 	if (ql_first(a_head) != NULL) {					\
 		qr_before_insert(ql_first(a_head), (a_elm), a_field);	\
 	}								\
 	ql_first(a_head) = qr_next((a_elm), a_field);			\
 } while (0)
 
-#define	ql_remove(a_head, a_elm, a_field) do {				\
+#define ql_remove(a_head, a_elm, a_field) do {				\
 	if (ql_first(a_head) == (a_elm)) {				\
 		ql_first(a_head) = qr_next(ql_first(a_head), a_field);	\
 	}								\
@@ -66,18 +69,20 @@ struct {								\
 	}								\
 } while (0)
 
-#define	ql_head_remove(a_head, a_type, a_field) do {			\
+#define ql_head_remove(a_head, a_type, a_field) do {			\
 	a_type *t = ql_first(a_head);					\
 	ql_remove((a_head), t, a_field);				\
 } while (0)
 
-#define	ql_tail_remove(a_head, a_type, a_field) do {			\
+#define ql_tail_remove(a_head, a_type, a_field) do {			\
 	a_type *t = ql_last(a_head, a_field);				\
 	ql_remove((a_head), t, a_field);				\
 } while (0)
 
-#define	ql_foreach(a_var, a_head, a_field)				\
+#define ql_foreach(a_var, a_head, a_field)				\
 	qr_foreach((a_var), ql_first(a_head), a_field)
 
-#define	ql_reverse_foreach(a_var, a_head, a_field)			\
+#define ql_reverse_foreach(a_var, a_head, a_field)			\
 	qr_reverse_foreach((a_var), ql_first(a_head), a_field)
+
+#endif /* JEMALLOC_INTERNAL_QL_H */