project-base/server-skynet-source-3rd-jemalloc
2
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
ab5e3790f6
server-skynet-source-3rd-je.../INSTALL
Jason Evans cbf3a6d703 Move centralized chunk management into arenas. 
Migrate all centralized data structures related to huge allocations and
recyclable chunks into arena_t, so that each arena can manage huge
allocations and recyclable virtual memory completely independently of
other arenas.

Add chunk node caching to arenas, in order to avoid contention on the
base allocator.

Use chunks_rtree to look up huge allocations rather than a red-black
tree.  Maintain a per arena unsorted list of huge allocations (which
will be needed to enumerate huge allocations during arena reset).

Remove the --enable-ivsalloc option, make ivsalloc() always available,
and use it for size queries if --enable-debug is enabled.  The only
practical implications to this removal are that 1) ivsalloc() is now
always available during live debugging (and the underlying radix tree is
available during core-based debugging), and 2) size query validation can
no longer be enabled independent of --enable-debug.

Remove the stats.chunks.{current,total,high} mallctls, and replace their
underlying statistics with simpler atomically updated counters used
exclusively for gdump triggering.  These statistics are no longer very
useful because each arena manages chunks independently, and per arena
statistics provide similar information.

Simplify chunk synchronization code, now that base chunk allocation
cannot cause recursive lock acquisition.
2015-02-12 00:15:56 -08:00

388 lines
15 KiB
Plaintext
Raw Blame History

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-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.
--disable-cc-silence
Disable code that silences non-useful compiler warnings. This is mainly
useful during development when auditing the set of warnings that are being
silenced.
--enable-debug
Enable assertions and validation code. This incurs a substantial
performance hit, but is very useful during application development.
--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.
--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.
--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.
--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-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.
--with-lg-page=<lg-page>
Specify the base 2 log of the system page size. This option is only useful
when cross compiling, since the configure script automatically determines
the host's page size by default.
--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-size-class-group=<lg-size-class-group>
Specify the base 2 log of how many size classes to use for each doubling in
size. By default jemalloc uses <lg-size-class-group>=2, which results in
e.g. the following size classes:
[...], 64,
80, 96, 112, 128,
160, [...]
<lg-size-class-group>=3 results in e.g. the following size classes:
[...], 64,
72, 80, 88, 96, 104, 112, 120, 128,
144, [...]
The minimal <lg-size-class-group>=0 causes jemalloc to only provide size
classes that are powers of 2:
[...],
64,
128,
256,
[...]
An implementation detail currently limits the total number of small size
classes to 255, and a compilation error will result if the
<lg-size-class-group> you specify cannot be supported. The limit is
roughly <lg-size-class-group>=4, depending on page size.
--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 if --with-lg-tiny-min is smaller than
--with-lg-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, assuming
--with-lg-size-class-group=2).
--with-lg-tiny-min=<lg-tiny-min>
Specify the base 2 log of the minimum tiny size class to support. Tiny
size classes are powers of 2 less than the quantum, and are only
incorporated if <lg-tiny-min> is less than <lg-quantum> (see
--with-lg-quantum). Tiny size classes technically violate the C standard
requirement for minimum alignment, and crashes could conceivably result if
the compiler were to generate instructions that made alignment assumptions,
both because illegal instruction traps could result, and because accesses
could straddle page boundaries and cause segmentation faults due to
accessing unmapped addresses.
The default of <lg-tiny-min>=3 works well in practice even on architectures
that technically require 16-byte alignment, probably for the same reason
--with-lg-quantum=3 works. Smaller tiny size classes can, and will, cause
crashes (see https://bugzilla.mozilla.org/show_bug.cgi?id=691003 for an
example).
This option is rarely useful, and is mainly provided as documentation of a
subtle implementation detail. If you do use this option, specify a
value in [3, ..., <lg-quantum>].
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
Reference in New Issue View Git Blame Copy Permalink