glibc/malloc/arena.c

1025 lines
30 KiB
C
Raw Normal View History

/* Malloc implementation for multiple threads without lock contention.
Copyright (C) 2001-2023 Free Software Foundation, Inc.
This file is part of the GNU C Library.
The GNU C Library is free software; you can redistribute it and/or
2002-08-26 Brian Youmans <3diff@gnu.org> * crypt/crypt.c: Changed copying permission notice to Lesser GPL from Library GPL. * crypt/crypt_util.c: Likewise. * crypt/ufc.c: Likewise. * elf/dl-conflict.c: Likewise. * elf/dl-iteratephdr.c: Likewise. * iconv/iconvconfig.h: Likewise. * linuxthreads/Examples/ex10.c: Likewise. * linuxthreads/Examples/ex11.c: Likewise. * linuxthreads/Examples/ex13.c: Likewise. * linuxthreads/Examples/ex8.c: Likewise. * linuxthreads/Examples/ex9.c: Likewise. * linuxthreads/barrier.c: Likewise. * linuxthreads/events.c: Likewise. * linuxthreads/lockfile.c: Likewise. * linuxthreads/no-tsd.c: Likewise. * linuxthreads/pt-machine.c: Likewise. * linuxthreads/ptclock_gettime.c: Likewise. * linuxthreads/ptclock_settime.c: Likewise. * linuxthreads/rwlock.c: Likewise. * linuxthreads/sysdeps/alpha/pspinlock.c: Likewise. * linuxthreads/sysdeps/alpha/pt-machine.h: Likewise. * linuxthreads/sysdeps/arm/pspinlock.c: Likewise. * linuxthreads/sysdeps/arm/pt-machine.h: Likewise. * linuxthreads/sysdeps/cris/pspinlock.c: Likewise. * linuxthreads/sysdeps/cris/pt-machine.h: Likewise. * linuxthreads/sysdeps/hppa/pspinlock.c: Likewise. * linuxthreads/sysdeps/hppa/pt-machine.h: Likewise. * linuxthreads/sysdeps/i386/i686/pt-machine.h: Likewise. * linuxthreads/sysdeps/i386/pspinlock.c: Likewise. * linuxthreads/sysdeps/i386/pt-machine.h: Likewise. * linuxthreads/sysdeps/i386/useldt.h: Likewise. * linuxthreads/sysdeps/ia64/pspinlock.c: Likewise. * linuxthreads/sysdeps/ia64/pt-machine.h: Likewise. * linuxthreads/sysdeps/m68k/pspinlock.c: Likewise. * linuxthreads/sysdeps/m68k/pt-machine.h: Likewise. * linuxthreads/sysdeps/mips/pspinlock.c: Likewise. * linuxthreads/sysdeps/mips/pt-machine.h: Likewise. * linuxthreads/sysdeps/powerpc/pspinlock.c: Likewise. * linuxthreads/sysdeps/powerpc/pt-machine.h: Likewise. * linuxthreads/sysdeps/pthread/bits/initspin.h: Likewise. * linuxthreads/sysdeps/pthread/bits/libc-lock.h: Likewise. * linuxthreads/sysdeps/pthread/bits/libc-tsd.h: Likewise. * linuxthreads/sysdeps/pthread/getcpuclockid.c: Likewise. * linuxthreads/sysdeps/pthread/posix-timer.h: Likewise. * linuxthreads/sysdeps/pthread/timer_create.c: Likewise. * linuxthreads/sysdeps/pthread/timer_delete.c: Likewise. * linuxthreads/sysdeps/pthread/timer_getoverr.c: Likewise. * linuxthreads/sysdeps/pthread/timer_gettime.c: Likewise. * linuxthreads/sysdeps/pthread/timer_routines.c: Likewise. * linuxthreads/sysdeps/pthread/timer_settime.c: Likewise. * linuxthreads/sysdeps/pthread/tst-timer.c: Likewise. * linuxthreads/sysdeps/s390/pspinlock.c: Likewise. * linuxthreads/sysdeps/s390/s390-32/pt-machine.h: Likewise. * linuxthreads/sysdeps/s390/s390-64/pt-machine.h: Likewise. * linuxthreads/sysdeps/sh/pspinlock.c: Likewise. * linuxthreads/sysdeps/sh/pt-machine.h: Likewise. * linuxthreads/sysdeps/sparc/sparc32/pspinlock.c: Likewise. * linuxthreads/sysdeps/sparc/sparc32/pt-machine.h: Likewise. * linuxthreads/sysdeps/sparc/sparc32/sparcv9/pspinlock.c: Likewise. * linuxthreads/sysdeps/sparc/sparc64/pspinlock.c: Likewise. * linuxthreads/sysdeps/sparc/sparc64/pt-machine.h: Likewise. * linuxthreads/sysdeps/unix/sysv/linux/bits/local_lim.h: Likewise. * linuxthreads/sysdeps/unix/sysv/linux/bits/posix_opt.h: Likewise. * linuxthreads/sysdeps/unix/sysv/linux/bits/sigthread.h: Likewise. * linuxthreads/sysdeps/unix/sysv/linux/hppa/bits/initspin.h: Likewise. * linuxthreads/sysdeps/unix/sysv/linux/i386/bits/posix_opt.h: Likewise. * linuxthreads/tststack.c: Likewise. * linuxthreads/unload.c: Likewise. * linuxthreads/weaks.c: Likewise. * linuxthreads/wrapsyscall.c: Likewise. * malloc/arena.c: Likewise. * malloc/hooks.c: Likewise. * malloc/malloc.c: Likewise. * posix/glob/Makefile.ami: Likewise. * posix/glob/Makefile.in: Likewise. * stdlib/gmp-impl.h: Likewise. * stdlib/gmp.h: Likewise. * sysdeps/generic/dl-iteratephdr-static.c: Likewise. * sysdeps/generic/strnlen.c: Likewise. * sysdeps/mach/hurd/powerpc/bits/sigcontext.h: Likewise. * sysdeps/mach/hurd/recvmsg.c: Likewise. * sysdeps/mach/hurd/sendmsg.c: Likewise. * sysdeps/mach/hurd/spawni.c: Likewise. * sysdeps/mach/powerpc/machine-sp.h: Likewise. * sysdeps/mach/powerpc/sysdep.h: Likewise. * sysdeps/mach/powerpc/thread_state.h: Likewise. * sysdeps/unix/bsd/bsd4.4/bits/socket.h: Likewise. * sysdeps/unix/sysv/linux/ia64/dl-iteratephdr-static.c: Likewise. * sysdeps/x86_64/gmp-mparam.h: Likewise.
2002-08-26 22:40:48 +00:00
modify it under the terms of the GNU Lesser General Public License as
published by the Free Software Foundation; either version 2.1 of the
License, or (at your option) any later version.
The GNU C Library is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
2002-08-26 Brian Youmans <3diff@gnu.org> * crypt/crypt.c: Changed copying permission notice to Lesser GPL from Library GPL. * crypt/crypt_util.c: Likewise. * crypt/ufc.c: Likewise. * elf/dl-conflict.c: Likewise. * elf/dl-iteratephdr.c: Likewise. * iconv/iconvconfig.h: Likewise. * linuxthreads/Examples/ex10.c: Likewise. * linuxthreads/Examples/ex11.c: Likewise. * linuxthreads/Examples/ex13.c: Likewise. * linuxthreads/Examples/ex8.c: Likewise. * linuxthreads/Examples/ex9.c: Likewise. * linuxthreads/barrier.c: Likewise. * linuxthreads/events.c: Likewise. * linuxthreads/lockfile.c: Likewise. * linuxthreads/no-tsd.c: Likewise. * linuxthreads/pt-machine.c: Likewise. * linuxthreads/ptclock_gettime.c: Likewise. * linuxthreads/ptclock_settime.c: Likewise. * linuxthreads/rwlock.c: Likewise. * linuxthreads/sysdeps/alpha/pspinlock.c: Likewise. * linuxthreads/sysdeps/alpha/pt-machine.h: Likewise. * linuxthreads/sysdeps/arm/pspinlock.c: Likewise. * linuxthreads/sysdeps/arm/pt-machine.h: Likewise. * linuxthreads/sysdeps/cris/pspinlock.c: Likewise. * linuxthreads/sysdeps/cris/pt-machine.h: Likewise. * linuxthreads/sysdeps/hppa/pspinlock.c: Likewise. * linuxthreads/sysdeps/hppa/pt-machine.h: Likewise. * linuxthreads/sysdeps/i386/i686/pt-machine.h: Likewise. * linuxthreads/sysdeps/i386/pspinlock.c: Likewise. * linuxthreads/sysdeps/i386/pt-machine.h: Likewise. * linuxthreads/sysdeps/i386/useldt.h: Likewise. * linuxthreads/sysdeps/ia64/pspinlock.c: Likewise. * linuxthreads/sysdeps/ia64/pt-machine.h: Likewise. * linuxthreads/sysdeps/m68k/pspinlock.c: Likewise. * linuxthreads/sysdeps/m68k/pt-machine.h: Likewise. * linuxthreads/sysdeps/mips/pspinlock.c: Likewise. * linuxthreads/sysdeps/mips/pt-machine.h: Likewise. * linuxthreads/sysdeps/powerpc/pspinlock.c: Likewise. * linuxthreads/sysdeps/powerpc/pt-machine.h: Likewise. * linuxthreads/sysdeps/pthread/bits/initspin.h: Likewise. * linuxthreads/sysdeps/pthread/bits/libc-lock.h: Likewise. * linuxthreads/sysdeps/pthread/bits/libc-tsd.h: Likewise. * linuxthreads/sysdeps/pthread/getcpuclockid.c: Likewise. * linuxthreads/sysdeps/pthread/posix-timer.h: Likewise. * linuxthreads/sysdeps/pthread/timer_create.c: Likewise. * linuxthreads/sysdeps/pthread/timer_delete.c: Likewise. * linuxthreads/sysdeps/pthread/timer_getoverr.c: Likewise. * linuxthreads/sysdeps/pthread/timer_gettime.c: Likewise. * linuxthreads/sysdeps/pthread/timer_routines.c: Likewise. * linuxthreads/sysdeps/pthread/timer_settime.c: Likewise. * linuxthreads/sysdeps/pthread/tst-timer.c: Likewise. * linuxthreads/sysdeps/s390/pspinlock.c: Likewise. * linuxthreads/sysdeps/s390/s390-32/pt-machine.h: Likewise. * linuxthreads/sysdeps/s390/s390-64/pt-machine.h: Likewise. * linuxthreads/sysdeps/sh/pspinlock.c: Likewise. * linuxthreads/sysdeps/sh/pt-machine.h: Likewise. * linuxthreads/sysdeps/sparc/sparc32/pspinlock.c: Likewise. * linuxthreads/sysdeps/sparc/sparc32/pt-machine.h: Likewise. * linuxthreads/sysdeps/sparc/sparc32/sparcv9/pspinlock.c: Likewise. * linuxthreads/sysdeps/sparc/sparc64/pspinlock.c: Likewise. * linuxthreads/sysdeps/sparc/sparc64/pt-machine.h: Likewise. * linuxthreads/sysdeps/unix/sysv/linux/bits/local_lim.h: Likewise. * linuxthreads/sysdeps/unix/sysv/linux/bits/posix_opt.h: Likewise. * linuxthreads/sysdeps/unix/sysv/linux/bits/sigthread.h: Likewise. * linuxthreads/sysdeps/unix/sysv/linux/hppa/bits/initspin.h: Likewise. * linuxthreads/sysdeps/unix/sysv/linux/i386/bits/posix_opt.h: Likewise. * linuxthreads/tststack.c: Likewise. * linuxthreads/unload.c: Likewise. * linuxthreads/weaks.c: Likewise. * linuxthreads/wrapsyscall.c: Likewise. * malloc/arena.c: Likewise. * malloc/hooks.c: Likewise. * malloc/malloc.c: Likewise. * posix/glob/Makefile.ami: Likewise. * posix/glob/Makefile.in: Likewise. * stdlib/gmp-impl.h: Likewise. * stdlib/gmp.h: Likewise. * sysdeps/generic/dl-iteratephdr-static.c: Likewise. * sysdeps/generic/strnlen.c: Likewise. * sysdeps/mach/hurd/powerpc/bits/sigcontext.h: Likewise. * sysdeps/mach/hurd/recvmsg.c: Likewise. * sysdeps/mach/hurd/sendmsg.c: Likewise. * sysdeps/mach/hurd/spawni.c: Likewise. * sysdeps/mach/powerpc/machine-sp.h: Likewise. * sysdeps/mach/powerpc/sysdep.h: Likewise. * sysdeps/mach/powerpc/thread_state.h: Likewise. * sysdeps/unix/bsd/bsd4.4/bits/socket.h: Likewise. * sysdeps/unix/sysv/linux/ia64/dl-iteratephdr-static.c: Likewise. * sysdeps/x86_64/gmp-mparam.h: Likewise.
2002-08-26 22:40:48 +00:00
Lesser General Public License for more details.
2002-08-26 Brian Youmans <3diff@gnu.org> * crypt/crypt.c: Changed copying permission notice to Lesser GPL from Library GPL. * crypt/crypt_util.c: Likewise. * crypt/ufc.c: Likewise. * elf/dl-conflict.c: Likewise. * elf/dl-iteratephdr.c: Likewise. * iconv/iconvconfig.h: Likewise. * linuxthreads/Examples/ex10.c: Likewise. * linuxthreads/Examples/ex11.c: Likewise. * linuxthreads/Examples/ex13.c: Likewise. * linuxthreads/Examples/ex8.c: Likewise. * linuxthreads/Examples/ex9.c: Likewise. * linuxthreads/barrier.c: Likewise. * linuxthreads/events.c: Likewise. * linuxthreads/lockfile.c: Likewise. * linuxthreads/no-tsd.c: Likewise. * linuxthreads/pt-machine.c: Likewise. * linuxthreads/ptclock_gettime.c: Likewise. * linuxthreads/ptclock_settime.c: Likewise. * linuxthreads/rwlock.c: Likewise. * linuxthreads/sysdeps/alpha/pspinlock.c: Likewise. * linuxthreads/sysdeps/alpha/pt-machine.h: Likewise. * linuxthreads/sysdeps/arm/pspinlock.c: Likewise. * linuxthreads/sysdeps/arm/pt-machine.h: Likewise. * linuxthreads/sysdeps/cris/pspinlock.c: Likewise. * linuxthreads/sysdeps/cris/pt-machine.h: Likewise. * linuxthreads/sysdeps/hppa/pspinlock.c: Likewise. * linuxthreads/sysdeps/hppa/pt-machine.h: Likewise. * linuxthreads/sysdeps/i386/i686/pt-machine.h: Likewise. * linuxthreads/sysdeps/i386/pspinlock.c: Likewise. * linuxthreads/sysdeps/i386/pt-machine.h: Likewise. * linuxthreads/sysdeps/i386/useldt.h: Likewise. * linuxthreads/sysdeps/ia64/pspinlock.c: Likewise. * linuxthreads/sysdeps/ia64/pt-machine.h: Likewise. * linuxthreads/sysdeps/m68k/pspinlock.c: Likewise. * linuxthreads/sysdeps/m68k/pt-machine.h: Likewise. * linuxthreads/sysdeps/mips/pspinlock.c: Likewise. * linuxthreads/sysdeps/mips/pt-machine.h: Likewise. * linuxthreads/sysdeps/powerpc/pspinlock.c: Likewise. * linuxthreads/sysdeps/powerpc/pt-machine.h: Likewise. * linuxthreads/sysdeps/pthread/bits/initspin.h: Likewise. * linuxthreads/sysdeps/pthread/bits/libc-lock.h: Likewise. * linuxthreads/sysdeps/pthread/bits/libc-tsd.h: Likewise. * linuxthreads/sysdeps/pthread/getcpuclockid.c: Likewise. * linuxthreads/sysdeps/pthread/posix-timer.h: Likewise. * linuxthreads/sysdeps/pthread/timer_create.c: Likewise. * linuxthreads/sysdeps/pthread/timer_delete.c: Likewise. * linuxthreads/sysdeps/pthread/timer_getoverr.c: Likewise. * linuxthreads/sysdeps/pthread/timer_gettime.c: Likewise. * linuxthreads/sysdeps/pthread/timer_routines.c: Likewise. * linuxthreads/sysdeps/pthread/timer_settime.c: Likewise. * linuxthreads/sysdeps/pthread/tst-timer.c: Likewise. * linuxthreads/sysdeps/s390/pspinlock.c: Likewise. * linuxthreads/sysdeps/s390/s390-32/pt-machine.h: Likewise. * linuxthreads/sysdeps/s390/s390-64/pt-machine.h: Likewise. * linuxthreads/sysdeps/sh/pspinlock.c: Likewise. * linuxthreads/sysdeps/sh/pt-machine.h: Likewise. * linuxthreads/sysdeps/sparc/sparc32/pspinlock.c: Likewise. * linuxthreads/sysdeps/sparc/sparc32/pt-machine.h: Likewise. * linuxthreads/sysdeps/sparc/sparc32/sparcv9/pspinlock.c: Likewise. * linuxthreads/sysdeps/sparc/sparc64/pspinlock.c: Likewise. * linuxthreads/sysdeps/sparc/sparc64/pt-machine.h: Likewise. * linuxthreads/sysdeps/unix/sysv/linux/bits/local_lim.h: Likewise. * linuxthreads/sysdeps/unix/sysv/linux/bits/posix_opt.h: Likewise. * linuxthreads/sysdeps/unix/sysv/linux/bits/sigthread.h: Likewise. * linuxthreads/sysdeps/unix/sysv/linux/hppa/bits/initspin.h: Likewise. * linuxthreads/sysdeps/unix/sysv/linux/i386/bits/posix_opt.h: Likewise. * linuxthreads/tststack.c: Likewise. * linuxthreads/unload.c: Likewise. * linuxthreads/weaks.c: Likewise. * linuxthreads/wrapsyscall.c: Likewise. * malloc/arena.c: Likewise. * malloc/hooks.c: Likewise. * malloc/malloc.c: Likewise. * posix/glob/Makefile.ami: Likewise. * posix/glob/Makefile.in: Likewise. * stdlib/gmp-impl.h: Likewise. * stdlib/gmp.h: Likewise. * sysdeps/generic/dl-iteratephdr-static.c: Likewise. * sysdeps/generic/strnlen.c: Likewise. * sysdeps/mach/hurd/powerpc/bits/sigcontext.h: Likewise. * sysdeps/mach/hurd/recvmsg.c: Likewise. * sysdeps/mach/hurd/sendmsg.c: Likewise. * sysdeps/mach/hurd/spawni.c: Likewise. * sysdeps/mach/powerpc/machine-sp.h: Likewise. * sysdeps/mach/powerpc/sysdep.h: Likewise. * sysdeps/mach/powerpc/thread_state.h: Likewise. * sysdeps/unix/bsd/bsd4.4/bits/socket.h: Likewise. * sysdeps/unix/sysv/linux/ia64/dl-iteratephdr-static.c: Likewise. * sysdeps/x86_64/gmp-mparam.h: Likewise.
2002-08-26 22:40:48 +00:00
You should have received a copy of the GNU Lesser General Public
License along with the GNU C Library; see the file COPYING.LIB. If
Prefer https to http for gnu.org and fsf.org URLs Also, change sources.redhat.com to sourceware.org. This patch was automatically generated by running the following shell script, which uses GNU sed, and which avoids modifying files imported from upstream: sed -ri ' s,(http|ftp)(://(.*\.)?(gnu|fsf|sourceware)\.org($|[^.]|\.[^a-z])),https\2,g s,(http|ftp)(://(.*\.)?)sources\.redhat\.com($|[^.]|\.[^a-z]),https\2sourceware.org\4,g ' \ $(find $(git ls-files) -prune -type f \ ! -name '*.po' \ ! -name 'ChangeLog*' \ ! -path COPYING ! -path COPYING.LIB \ ! -path manual/fdl-1.3.texi ! -path manual/lgpl-2.1.texi \ ! -path manual/texinfo.tex ! -path scripts/config.guess \ ! -path scripts/config.sub ! -path scripts/install-sh \ ! -path scripts/mkinstalldirs ! -path scripts/move-if-change \ ! -path INSTALL ! -path locale/programs/charmap-kw.h \ ! -path po/libc.pot ! -path sysdeps/gnu/errlist.c \ ! '(' -name configure \ -execdir test -f configure.ac -o -f configure.in ';' ')' \ ! '(' -name preconfigure \ -execdir test -f preconfigure.ac ';' ')' \ -print) and then by running 'make dist-prepare' to regenerate files built from the altered files, and then executing the following to cleanup: chmod a+x sysdeps/unix/sysv/linux/riscv/configure # Omit irrelevant whitespace and comment-only changes, # perhaps from a slightly-different Autoconf version. git checkout -f \ sysdeps/csky/configure \ sysdeps/hppa/configure \ sysdeps/riscv/configure \ sysdeps/unix/sysv/linux/csky/configure # Omit changes that caused a pre-commit check to fail like this: # remote: *** error: sysdeps/powerpc/powerpc64/ppc-mcount.S: trailing lines git checkout -f \ sysdeps/powerpc/powerpc64/ppc-mcount.S \ sysdeps/unix/sysv/linux/s390/s390-64/syscall.S # Omit change that caused a pre-commit check to fail like this: # remote: *** error: sysdeps/sparc/sparc64/multiarch/memcpy-ultra3.S: last line does not end in newline git checkout -f sysdeps/sparc/sparc64/multiarch/memcpy-ultra3.S
2019-09-07 05:40:42 +00:00
not, see <https://www.gnu.org/licenses/>. */
#include <stdbool.h>
Add framework for tunables The tunables framework allows us to uniformly manage and expose global variables inside glibc as switches to users. tunables/README has instructions for glibc developers to add new tunables. Tunables support can be enabled by passing the --enable-tunables configure flag to the configure script. This patch only adds a framework and does not pose any limitations on how tunable values are read from the user. It also adds environment variables used in malloc behaviour tweaking to the tunables framework as a PoC of the compatibility interface. * manual/install.texi: Add --enable-tunables option. * INSTALL: Regenerate. * README.tunables: New file. * Makeconfig (CPPFLAGS): Define TOP_NAMESPACE. (before-compile): Generate dl-tunable-list.h early. * config.h.in: Add HAVE_TUNABLES. * config.make.in: Add have-tunables. * configure.ac: Add --enable-tunables option. * configure: Regenerate. * csu/init-first.c (__libc_init_first): Move __libc_init_secure earlier... * csu/init-first.c (LIBC_START_MAIN):... to here. Include dl-tunables.h, libc-internal.h. (LIBC_START_MAIN) [!SHARED]: Initialize tunables for static binaries. * elf/Makefile (dl-routines): Add dl-tunables. * elf/Versions (ld): Add __tunable_set_val to GLIBC_PRIVATE namespace. * elf/dl-support (_dl_nondynamic_init): Unset MALLOC_CHECK_ only when !HAVE_TUNABLES. * elf/rtld.c (process_envvars): Likewise. * elf/dl-sysdep.c [HAVE_TUNABLES]: Include dl-tunables.h (_dl_sysdep_start): Call __tunables_init. * elf/dl-tunable-types.h: New file. * elf/dl-tunables.c: New file. * elf/dl-tunables.h: New file. * elf/dl-tunables.list: New file. * malloc/tst-malloc-usable-static.c: New test case. * malloc/Makefile (tests-static): Add it. * malloc/arena.c [HAVE_TUNABLES]: Include dl-tunables.h. Define TUNABLE_NAMESPACE. (DL_TUNABLE_CALLBACK (set_mallopt_check)): New function. (DL_TUNABLE_CALLBACK_FNDECL): New macro. Use it to define callback functions. (ptmalloc_init): Set tunable values. * scripts/gen-tunables.awk: New file. * sysdeps/mach/hurd/dl-sysdep.c: Include dl-tunables.h. (_dl_sysdep_start): Call __tunables_init.
2016-12-31 18:02:17 +00:00
#if HAVE_TUNABLES
# define TUNABLE_NAMESPACE malloc
#endif
#include <elf/dl-tunables.h>
/* Compile-time constants. */
2014-01-02 08:38:18 +00:00
#define HEAP_MIN_SIZE (32 * 1024)
#ifndef HEAP_MAX_SIZE
# ifdef DEFAULT_MMAP_THRESHOLD_MAX
# define HEAP_MAX_SIZE (2 * DEFAULT_MMAP_THRESHOLD_MAX)
# else
2014-01-02 08:38:18 +00:00
# define HEAP_MAX_SIZE (1024 * 1024) /* must be a power of two */
# endif
#endif
/* HEAP_MIN_SIZE and HEAP_MAX_SIZE limit the size of mmap()ed heaps
that are dynamically created for multi-threaded programs. The
maximum size must be a power of two, for fast determination of
which heap belongs to a chunk. It should be much larger than the
mmap threshold, so that requests with a size just below that
threshold can be fulfilled without creating too many heaps. */
/* When huge pages are used to create new arenas, the maximum and minumum
size are based on the runtime defined huge page size. */
static inline size_t
heap_min_size (void)
{
#if HAVE_TUNABLES
return mp_.hp_pagesize == 0 ? HEAP_MIN_SIZE : mp_.hp_pagesize;
#else
return HEAP_MIN_SIZE;
#endif
}
static inline size_t
heap_max_size (void)
{
#if HAVE_TUNABLES
return mp_.hp_pagesize == 0 ? HEAP_MAX_SIZE : mp_.hp_pagesize * 4;
#else
return HEAP_MAX_SIZE;
#endif
}
/***************************************************************************/
#define top(ar_ptr) ((ar_ptr)->top)
/* A heap is a single contiguous memory region holding (coalesceable)
malloc_chunks. It is allocated with mmap() and always starts at an
address aligned to HEAP_MAX_SIZE. */
2014-01-02 08:38:18 +00:00
typedef struct _heap_info
{
mstate ar_ptr; /* Arena for this heap. */
struct _heap_info *prev; /* Previous heap. */
size_t size; /* Current size in bytes. */
2014-01-02 08:38:18 +00:00
size_t mprotect_size; /* Size in bytes that has been mprotected
PROT_READ|PROT_WRITE. */
size_t pagesize; /* Page size used when allocating the arena. */
/* Make sure the following data is properly aligned, particularly
that sizeof (heap_info) + 2 * SIZE_SZ is a multiple of
MALLOC_ALIGNMENT. */
char pad[-3 * SIZE_SZ & MALLOC_ALIGN_MASK];
} heap_info;
/* Get a compile-time error if the heap_info padding is not correct
to make alignment work as expected in sYSMALLOc. */
extern int sanity_check_heap_info_alignment[(sizeof (heap_info)
2014-01-02 08:38:18 +00:00
+ 2 * SIZE_SZ) % MALLOC_ALIGNMENT
? -1 : 1];
/* Thread specific data. */
static __thread mstate thread_arena attribute_tls_model_ie;
/* Arena free list. free_list_lock synchronizes access to the
free_list variable below, and the next_free and attached_threads
members of struct malloc_state objects. No other locks must be
acquired after free_list_lock has been acquired. */
__libc_lock_define_initialized (static, free_list_lock);
#if IS_IN (libc)
static size_t narenas = 1;
#endif
static mstate free_list;
/* list_lock prevents concurrent writes to the next member of struct
malloc_state objects.
Read access to the next member is supposed to synchronize with the
atomic_write_barrier and the write to the next member in
_int_new_arena. This suffers from data races; see the FIXME
comments in _int_new_arena and reused_arena.
2015-12-23 16:23:33 +00:00
list_lock also prevents concurrent forks. At the time list_lock is
acquired, no arena lock must have been acquired, but it is
permitted to acquire arena locks subsequently, while list_lock is
acquired. */
__libc_lock_define_initialized (static, list_lock);
/* Already initialized? */
static bool __malloc_initialized = false;
/**************************************************************************/
/* arena_get() acquires an arena and locks the corresponding mutex.
First, try the one last locked successfully by this thread. (This
is the common case and handled with a macro for speed.) Then, loop
once over the circularly linked list of arenas. If no arena is
readily available, create a new one. In this latter case, `size'
is just a hint as to how much memory will be required immediately
in the new arena. */
#define arena_get(ptr, size) do { \
ptr = thread_arena; \
2014-01-02 08:38:18 +00:00
arena_lock (ptr, size); \
} while (0)
2014-01-02 08:38:18 +00:00
#define arena_lock(ptr, size) do { \
if (ptr) \
__libc_lock_lock (ptr->mutex); \
2014-01-02 08:38:18 +00:00
else \
ptr = arena_get2 ((size), NULL); \
2014-01-02 08:38:18 +00:00
} while (0)
/* find the heap and corresponding arena for a given ptr */
static inline heap_info *
heap_for_ptr (void *ptr)
{
size_t max_size = heap_max_size ();
return PTR_ALIGN_DOWN (ptr, max_size);
}
static inline struct malloc_state *
arena_for_chunk (mchunkptr ptr)
{
return chunk_main_arena (ptr) ? &main_arena : heap_for_ptr (ptr)->ar_ptr;
}
/**************************************************************************/
/* atfork support. */
/* The following three functions are called around fork from a
multi-threaded process. We do not use the general fork handler
mechanism to make sure that our handlers are the last ones being
called, so that other fork handlers can use the malloc
subsystem. */
void
__malloc_fork_lock_parent (void)
{
if (!__malloc_initialized)
return;
2014-01-02 08:38:18 +00:00
/* We do not acquire free_list_lock here because we completely
reconstruct free_list in __malloc_fork_unlock_child. */
__libc_lock_lock (list_lock);
for (mstate ar_ptr = &main_arena;; )
2014-01-02 08:38:18 +00:00
{
__libc_lock_lock (ar_ptr->mutex);
2014-01-02 08:38:18 +00:00
ar_ptr = ar_ptr->next;
if (ar_ptr == &main_arena)
break;
}
}
void
__malloc_fork_unlock_parent (void)
{
if (!__malloc_initialized)
return;
2014-01-02 08:38:18 +00:00
for (mstate ar_ptr = &main_arena;; )
2014-01-02 08:38:18 +00:00
{
__libc_lock_unlock (ar_ptr->mutex);
2014-01-02 08:38:18 +00:00
ar_ptr = ar_ptr->next;
if (ar_ptr == &main_arena)
break;
}
__libc_lock_unlock (list_lock);
}
void
__malloc_fork_unlock_child (void)
{
if (!__malloc_initialized)
return;
2014-01-02 08:38:18 +00:00
/* Push all arenas to the free list, except thread_arena, which is
attached to the current thread. */
__libc_lock_init (free_list_lock);
if (thread_arena != NULL)
thread_arena->attached_threads = 1;
free_list = NULL;
for (mstate ar_ptr = &main_arena;; )
2014-01-02 08:38:18 +00:00
{
__libc_lock_init (ar_ptr->mutex);
if (ar_ptr != thread_arena)
2014-01-02 08:38:18 +00:00
{
/* This arena is no longer attached to any thread. */
ar_ptr->attached_threads = 0;
2014-01-02 08:38:18 +00:00
ar_ptr->next_free = free_list;
free_list = ar_ptr;
}
ar_ptr = ar_ptr->next;
if (ar_ptr == &main_arena)
break;
}
__libc_lock_init (list_lock);
}
Add framework for tunables The tunables framework allows us to uniformly manage and expose global variables inside glibc as switches to users. tunables/README has instructions for glibc developers to add new tunables. Tunables support can be enabled by passing the --enable-tunables configure flag to the configure script. This patch only adds a framework and does not pose any limitations on how tunable values are read from the user. It also adds environment variables used in malloc behaviour tweaking to the tunables framework as a PoC of the compatibility interface. * manual/install.texi: Add --enable-tunables option. * INSTALL: Regenerate. * README.tunables: New file. * Makeconfig (CPPFLAGS): Define TOP_NAMESPACE. (before-compile): Generate dl-tunable-list.h early. * config.h.in: Add HAVE_TUNABLES. * config.make.in: Add have-tunables. * configure.ac: Add --enable-tunables option. * configure: Regenerate. * csu/init-first.c (__libc_init_first): Move __libc_init_secure earlier... * csu/init-first.c (LIBC_START_MAIN):... to here. Include dl-tunables.h, libc-internal.h. (LIBC_START_MAIN) [!SHARED]: Initialize tunables for static binaries. * elf/Makefile (dl-routines): Add dl-tunables. * elf/Versions (ld): Add __tunable_set_val to GLIBC_PRIVATE namespace. * elf/dl-support (_dl_nondynamic_init): Unset MALLOC_CHECK_ only when !HAVE_TUNABLES. * elf/rtld.c (process_envvars): Likewise. * elf/dl-sysdep.c [HAVE_TUNABLES]: Include dl-tunables.h (_dl_sysdep_start): Call __tunables_init. * elf/dl-tunable-types.h: New file. * elf/dl-tunables.c: New file. * elf/dl-tunables.h: New file. * elf/dl-tunables.list: New file. * malloc/tst-malloc-usable-static.c: New test case. * malloc/Makefile (tests-static): Add it. * malloc/arena.c [HAVE_TUNABLES]: Include dl-tunables.h. Define TUNABLE_NAMESPACE. (DL_TUNABLE_CALLBACK (set_mallopt_check)): New function. (DL_TUNABLE_CALLBACK_FNDECL): New macro. Use it to define callback functions. (ptmalloc_init): Set tunable values. * scripts/gen-tunables.awk: New file. * sysdeps/mach/hurd/dl-sysdep.c: Include dl-tunables.h. (_dl_sysdep_start): Call __tunables_init.
2016-12-31 18:02:17 +00:00
#if HAVE_TUNABLES
tunables: Clean up hooks to get and set tunables The TUNABLE_SET_VALUE and family of macros (and my later attempt to add a TUNABLE_GET) never quite went together very well because the overall interface was not clearly defined. This patch is an attempt to do just that. This patch consolidates the API to two simple sets of macros, TUNABLE_GET* and TUNABLE_SET*. If TUNABLE_NAMESPACE is defined, TUNABLE_GET takes just the tunable name, type and a (optionally NULL) callback function to get the value of the tunable. The callback function, if non-NULL, is called if the tunable was externally set (i.e. via GLIBC_TUNABLES or any future mechanism). For example: val = TUNABLE_GET (check, int32_t, check_callback) returns the value of the glibc.malloc.check tunable (assuming TUNABLE_NAMESPACE is set to malloc) as an int32_t into VAL after calling check_callback. Likewise, TUNABLE_SET can be used to set the value of the tunable, although this is currently possible only in the dynamic linker before it relocates itself. For example: TUNABLE_SET (check, int32_t, 2) will set glibc.malloc.check to 2. Of course, this is not possible since we set (or read) glibc.malloc.check long after it is relocated. To access or set a tunable outside of TUNABLE_NAMESPACE, use the TUNABLE_GET_FULL and TUNABLE_SET_FULL macros, which have the following prototype: TUNABLE_GET_FULL (glibc, tune, hwcap_mask, uint64_t, NULL) TUNABLE_SET_FULL (glibc, tune, hwcap_mask, uint64_t, 0xffff) In future the tunable list may get split into mutable and immutable tunables where mutable tunables can be modified by the library and userspace after relocation as well and TUNABLE_SET will be more useful than it currently is. However whenever we actually do that split, we will have to ensure that the mutable tunables are protected with locks. * elf/Versions (__tunable_set_val): Rename to __tunable_get_val. * elf/dl-tunables.c: Likewise. (do_tunable_update_val): New function. (__tunable_set_val): New function. (__tunable_get_val): Call CB only if the tunable was externally initialized. (tunables_strtoul): Replace strval with initialized. * elf/dl-tunables.h (strval): Replace with a bool initialized. (TUNABLE_ENUM_NAME, TUNABLE_ENUM_NAME1): Adjust names to prevent collision. (__tunable_set_val): New function. (TUNABLE_GET, TUNABLE_GET_FULL): New macros. (TUNABLE_SET, TUNABLE_SET_FULL): Likewise. (TUNABLE_SET_VAL): Remove. (TUNABLE_SET_VAL_WITH_CALLBACK): Likewise. * README.tunables: Document the new macros. * malloc/arena.c (ptmalloc_init): Adjust.
2017-06-01 14:54:46 +00:00
# define TUNABLE_CALLBACK_FNDECL(__name, __type) \
Add framework for tunables The tunables framework allows us to uniformly manage and expose global variables inside glibc as switches to users. tunables/README has instructions for glibc developers to add new tunables. Tunables support can be enabled by passing the --enable-tunables configure flag to the configure script. This patch only adds a framework and does not pose any limitations on how tunable values are read from the user. It also adds environment variables used in malloc behaviour tweaking to the tunables framework as a PoC of the compatibility interface. * manual/install.texi: Add --enable-tunables option. * INSTALL: Regenerate. * README.tunables: New file. * Makeconfig (CPPFLAGS): Define TOP_NAMESPACE. (before-compile): Generate dl-tunable-list.h early. * config.h.in: Add HAVE_TUNABLES. * config.make.in: Add have-tunables. * configure.ac: Add --enable-tunables option. * configure: Regenerate. * csu/init-first.c (__libc_init_first): Move __libc_init_secure earlier... * csu/init-first.c (LIBC_START_MAIN):... to here. Include dl-tunables.h, libc-internal.h. (LIBC_START_MAIN) [!SHARED]: Initialize tunables for static binaries. * elf/Makefile (dl-routines): Add dl-tunables. * elf/Versions (ld): Add __tunable_set_val to GLIBC_PRIVATE namespace. * elf/dl-support (_dl_nondynamic_init): Unset MALLOC_CHECK_ only when !HAVE_TUNABLES. * elf/rtld.c (process_envvars): Likewise. * elf/dl-sysdep.c [HAVE_TUNABLES]: Include dl-tunables.h (_dl_sysdep_start): Call __tunables_init. * elf/dl-tunable-types.h: New file. * elf/dl-tunables.c: New file. * elf/dl-tunables.h: New file. * elf/dl-tunables.list: New file. * malloc/tst-malloc-usable-static.c: New test case. * malloc/Makefile (tests-static): Add it. * malloc/arena.c [HAVE_TUNABLES]: Include dl-tunables.h. Define TUNABLE_NAMESPACE. (DL_TUNABLE_CALLBACK (set_mallopt_check)): New function. (DL_TUNABLE_CALLBACK_FNDECL): New macro. Use it to define callback functions. (ptmalloc_init): Set tunable values. * scripts/gen-tunables.awk: New file. * sysdeps/mach/hurd/dl-sysdep.c: Include dl-tunables.h. (_dl_sysdep_start): Call __tunables_init.
2016-12-31 18:02:17 +00:00
static inline int do_ ## __name (__type value); \
static void \
tunables: Clean up hooks to get and set tunables The TUNABLE_SET_VALUE and family of macros (and my later attempt to add a TUNABLE_GET) never quite went together very well because the overall interface was not clearly defined. This patch is an attempt to do just that. This patch consolidates the API to two simple sets of macros, TUNABLE_GET* and TUNABLE_SET*. If TUNABLE_NAMESPACE is defined, TUNABLE_GET takes just the tunable name, type and a (optionally NULL) callback function to get the value of the tunable. The callback function, if non-NULL, is called if the tunable was externally set (i.e. via GLIBC_TUNABLES or any future mechanism). For example: val = TUNABLE_GET (check, int32_t, check_callback) returns the value of the glibc.malloc.check tunable (assuming TUNABLE_NAMESPACE is set to malloc) as an int32_t into VAL after calling check_callback. Likewise, TUNABLE_SET can be used to set the value of the tunable, although this is currently possible only in the dynamic linker before it relocates itself. For example: TUNABLE_SET (check, int32_t, 2) will set glibc.malloc.check to 2. Of course, this is not possible since we set (or read) glibc.malloc.check long after it is relocated. To access or set a tunable outside of TUNABLE_NAMESPACE, use the TUNABLE_GET_FULL and TUNABLE_SET_FULL macros, which have the following prototype: TUNABLE_GET_FULL (glibc, tune, hwcap_mask, uint64_t, NULL) TUNABLE_SET_FULL (glibc, tune, hwcap_mask, uint64_t, 0xffff) In future the tunable list may get split into mutable and immutable tunables where mutable tunables can be modified by the library and userspace after relocation as well and TUNABLE_SET will be more useful than it currently is. However whenever we actually do that split, we will have to ensure that the mutable tunables are protected with locks. * elf/Versions (__tunable_set_val): Rename to __tunable_get_val. * elf/dl-tunables.c: Likewise. (do_tunable_update_val): New function. (__tunable_set_val): New function. (__tunable_get_val): Call CB only if the tunable was externally initialized. (tunables_strtoul): Replace strval with initialized. * elf/dl-tunables.h (strval): Replace with a bool initialized. (TUNABLE_ENUM_NAME, TUNABLE_ENUM_NAME1): Adjust names to prevent collision. (__tunable_set_val): New function. (TUNABLE_GET, TUNABLE_GET_FULL): New macros. (TUNABLE_SET, TUNABLE_SET_FULL): Likewise. (TUNABLE_SET_VAL): Remove. (TUNABLE_SET_VAL_WITH_CALLBACK): Likewise. * README.tunables: Document the new macros. * malloc/arena.c (ptmalloc_init): Adjust.
2017-06-01 14:54:46 +00:00
TUNABLE_CALLBACK (__name) (tunable_val_t *valp) \
Add framework for tunables The tunables framework allows us to uniformly manage and expose global variables inside glibc as switches to users. tunables/README has instructions for glibc developers to add new tunables. Tunables support can be enabled by passing the --enable-tunables configure flag to the configure script. This patch only adds a framework and does not pose any limitations on how tunable values are read from the user. It also adds environment variables used in malloc behaviour tweaking to the tunables framework as a PoC of the compatibility interface. * manual/install.texi: Add --enable-tunables option. * INSTALL: Regenerate. * README.tunables: New file. * Makeconfig (CPPFLAGS): Define TOP_NAMESPACE. (before-compile): Generate dl-tunable-list.h early. * config.h.in: Add HAVE_TUNABLES. * config.make.in: Add have-tunables. * configure.ac: Add --enable-tunables option. * configure: Regenerate. * csu/init-first.c (__libc_init_first): Move __libc_init_secure earlier... * csu/init-first.c (LIBC_START_MAIN):... to here. Include dl-tunables.h, libc-internal.h. (LIBC_START_MAIN) [!SHARED]: Initialize tunables for static binaries. * elf/Makefile (dl-routines): Add dl-tunables. * elf/Versions (ld): Add __tunable_set_val to GLIBC_PRIVATE namespace. * elf/dl-support (_dl_nondynamic_init): Unset MALLOC_CHECK_ only when !HAVE_TUNABLES. * elf/rtld.c (process_envvars): Likewise. * elf/dl-sysdep.c [HAVE_TUNABLES]: Include dl-tunables.h (_dl_sysdep_start): Call __tunables_init. * elf/dl-tunable-types.h: New file. * elf/dl-tunables.c: New file. * elf/dl-tunables.h: New file. * elf/dl-tunables.list: New file. * malloc/tst-malloc-usable-static.c: New test case. * malloc/Makefile (tests-static): Add it. * malloc/arena.c [HAVE_TUNABLES]: Include dl-tunables.h. Define TUNABLE_NAMESPACE. (DL_TUNABLE_CALLBACK (set_mallopt_check)): New function. (DL_TUNABLE_CALLBACK_FNDECL): New macro. Use it to define callback functions. (ptmalloc_init): Set tunable values. * scripts/gen-tunables.awk: New file. * sysdeps/mach/hurd/dl-sysdep.c: Include dl-tunables.h. (_dl_sysdep_start): Call __tunables_init.
2016-12-31 18:02:17 +00:00
{ \
__type value = (__type) (valp)->numval; \
Add framework for tunables The tunables framework allows us to uniformly manage and expose global variables inside glibc as switches to users. tunables/README has instructions for glibc developers to add new tunables. Tunables support can be enabled by passing the --enable-tunables configure flag to the configure script. This patch only adds a framework and does not pose any limitations on how tunable values are read from the user. It also adds environment variables used in malloc behaviour tweaking to the tunables framework as a PoC of the compatibility interface. * manual/install.texi: Add --enable-tunables option. * INSTALL: Regenerate. * README.tunables: New file. * Makeconfig (CPPFLAGS): Define TOP_NAMESPACE. (before-compile): Generate dl-tunable-list.h early. * config.h.in: Add HAVE_TUNABLES. * config.make.in: Add have-tunables. * configure.ac: Add --enable-tunables option. * configure: Regenerate. * csu/init-first.c (__libc_init_first): Move __libc_init_secure earlier... * csu/init-first.c (LIBC_START_MAIN):... to here. Include dl-tunables.h, libc-internal.h. (LIBC_START_MAIN) [!SHARED]: Initialize tunables for static binaries. * elf/Makefile (dl-routines): Add dl-tunables. * elf/Versions (ld): Add __tunable_set_val to GLIBC_PRIVATE namespace. * elf/dl-support (_dl_nondynamic_init): Unset MALLOC_CHECK_ only when !HAVE_TUNABLES. * elf/rtld.c (process_envvars): Likewise. * elf/dl-sysdep.c [HAVE_TUNABLES]: Include dl-tunables.h (_dl_sysdep_start): Call __tunables_init. * elf/dl-tunable-types.h: New file. * elf/dl-tunables.c: New file. * elf/dl-tunables.h: New file. * elf/dl-tunables.list: New file. * malloc/tst-malloc-usable-static.c: New test case. * malloc/Makefile (tests-static): Add it. * malloc/arena.c [HAVE_TUNABLES]: Include dl-tunables.h. Define TUNABLE_NAMESPACE. (DL_TUNABLE_CALLBACK (set_mallopt_check)): New function. (DL_TUNABLE_CALLBACK_FNDECL): New macro. Use it to define callback functions. (ptmalloc_init): Set tunable values. * scripts/gen-tunables.awk: New file. * sysdeps/mach/hurd/dl-sysdep.c: Include dl-tunables.h. (_dl_sysdep_start): Call __tunables_init.
2016-12-31 18:02:17 +00:00
do_ ## __name (value); \
}
tunables: Clean up hooks to get and set tunables The TUNABLE_SET_VALUE and family of macros (and my later attempt to add a TUNABLE_GET) never quite went together very well because the overall interface was not clearly defined. This patch is an attempt to do just that. This patch consolidates the API to two simple sets of macros, TUNABLE_GET* and TUNABLE_SET*. If TUNABLE_NAMESPACE is defined, TUNABLE_GET takes just the tunable name, type and a (optionally NULL) callback function to get the value of the tunable. The callback function, if non-NULL, is called if the tunable was externally set (i.e. via GLIBC_TUNABLES or any future mechanism). For example: val = TUNABLE_GET (check, int32_t, check_callback) returns the value of the glibc.malloc.check tunable (assuming TUNABLE_NAMESPACE is set to malloc) as an int32_t into VAL after calling check_callback. Likewise, TUNABLE_SET can be used to set the value of the tunable, although this is currently possible only in the dynamic linker before it relocates itself. For example: TUNABLE_SET (check, int32_t, 2) will set glibc.malloc.check to 2. Of course, this is not possible since we set (or read) glibc.malloc.check long after it is relocated. To access or set a tunable outside of TUNABLE_NAMESPACE, use the TUNABLE_GET_FULL and TUNABLE_SET_FULL macros, which have the following prototype: TUNABLE_GET_FULL (glibc, tune, hwcap_mask, uint64_t, NULL) TUNABLE_SET_FULL (glibc, tune, hwcap_mask, uint64_t, 0xffff) In future the tunable list may get split into mutable and immutable tunables where mutable tunables can be modified by the library and userspace after relocation as well and TUNABLE_SET will be more useful than it currently is. However whenever we actually do that split, we will have to ensure that the mutable tunables are protected with locks. * elf/Versions (__tunable_set_val): Rename to __tunable_get_val. * elf/dl-tunables.c: Likewise. (do_tunable_update_val): New function. (__tunable_set_val): New function. (__tunable_get_val): Call CB only if the tunable was externally initialized. (tunables_strtoul): Replace strval with initialized. * elf/dl-tunables.h (strval): Replace with a bool initialized. (TUNABLE_ENUM_NAME, TUNABLE_ENUM_NAME1): Adjust names to prevent collision. (__tunable_set_val): New function. (TUNABLE_GET, TUNABLE_GET_FULL): New macros. (TUNABLE_SET, TUNABLE_SET_FULL): Likewise. (TUNABLE_SET_VAL): Remove. (TUNABLE_SET_VAL_WITH_CALLBACK): Likewise. * README.tunables: Document the new macros. * malloc/arena.c (ptmalloc_init): Adjust.
2017-06-01 14:54:46 +00:00
TUNABLE_CALLBACK_FNDECL (set_mmap_threshold, size_t)
TUNABLE_CALLBACK_FNDECL (set_mmaps_max, int32_t)
TUNABLE_CALLBACK_FNDECL (set_top_pad, size_t)
TUNABLE_CALLBACK_FNDECL (set_perturb_byte, int32_t)
TUNABLE_CALLBACK_FNDECL (set_trim_threshold, size_t)
TUNABLE_CALLBACK_FNDECL (set_arena_max, size_t)
TUNABLE_CALLBACK_FNDECL (set_arena_test, size_t)
#if USE_TCACHE
TUNABLE_CALLBACK_FNDECL (set_tcache_max, size_t)
TUNABLE_CALLBACK_FNDECL (set_tcache_count, size_t)
TUNABLE_CALLBACK_FNDECL (set_tcache_unsorted_limit, size_t)
#endif
TUNABLE_CALLBACK_FNDECL (set_mxfast, size_t)
TUNABLE_CALLBACK_FNDECL (set_hugetlb, size_t)
Add framework for tunables The tunables framework allows us to uniformly manage and expose global variables inside glibc as switches to users. tunables/README has instructions for glibc developers to add new tunables. Tunables support can be enabled by passing the --enable-tunables configure flag to the configure script. This patch only adds a framework and does not pose any limitations on how tunable values are read from the user. It also adds environment variables used in malloc behaviour tweaking to the tunables framework as a PoC of the compatibility interface. * manual/install.texi: Add --enable-tunables option. * INSTALL: Regenerate. * README.tunables: New file. * Makeconfig (CPPFLAGS): Define TOP_NAMESPACE. (before-compile): Generate dl-tunable-list.h early. * config.h.in: Add HAVE_TUNABLES. * config.make.in: Add have-tunables. * configure.ac: Add --enable-tunables option. * configure: Regenerate. * csu/init-first.c (__libc_init_first): Move __libc_init_secure earlier... * csu/init-first.c (LIBC_START_MAIN):... to here. Include dl-tunables.h, libc-internal.h. (LIBC_START_MAIN) [!SHARED]: Initialize tunables for static binaries. * elf/Makefile (dl-routines): Add dl-tunables. * elf/Versions (ld): Add __tunable_set_val to GLIBC_PRIVATE namespace. * elf/dl-support (_dl_nondynamic_init): Unset MALLOC_CHECK_ only when !HAVE_TUNABLES. * elf/rtld.c (process_envvars): Likewise. * elf/dl-sysdep.c [HAVE_TUNABLES]: Include dl-tunables.h (_dl_sysdep_start): Call __tunables_init. * elf/dl-tunable-types.h: New file. * elf/dl-tunables.c: New file. * elf/dl-tunables.h: New file. * elf/dl-tunables.list: New file. * malloc/tst-malloc-usable-static.c: New test case. * malloc/Makefile (tests-static): Add it. * malloc/arena.c [HAVE_TUNABLES]: Include dl-tunables.h. Define TUNABLE_NAMESPACE. (DL_TUNABLE_CALLBACK (set_mallopt_check)): New function. (DL_TUNABLE_CALLBACK_FNDECL): New macro. Use it to define callback functions. (ptmalloc_init): Set tunable values. * scripts/gen-tunables.awk: New file. * sysdeps/mach/hurd/dl-sysdep.c: Include dl-tunables.h. (_dl_sysdep_start): Call __tunables_init.
2016-12-31 18:02:17 +00:00
#else
/* Initialization routine. */
#include <string.h>
extern char **_environ;
static char *
next_env_entry (char ***position)
{
char **current = *position;
char *result = NULL;
while (*current != NULL)
{
if (__builtin_expect ((*current)[0] == 'M', 0)
2014-01-02 08:38:18 +00:00
&& (*current)[1] == 'A'
&& (*current)[2] == 'L'
&& (*current)[3] == 'L'
&& (*current)[4] == 'O'
&& (*current)[5] == 'C'
&& (*current)[6] == '_')
{
result = &(*current)[7];
2014-01-02 08:38:18 +00:00
/* Save current position for next visit. */
*position = ++current;
2014-01-02 08:38:18 +00:00
break;
}
++current;
}
return result;
}
Add framework for tunables The tunables framework allows us to uniformly manage and expose global variables inside glibc as switches to users. tunables/README has instructions for glibc developers to add new tunables. Tunables support can be enabled by passing the --enable-tunables configure flag to the configure script. This patch only adds a framework and does not pose any limitations on how tunable values are read from the user. It also adds environment variables used in malloc behaviour tweaking to the tunables framework as a PoC of the compatibility interface. * manual/install.texi: Add --enable-tunables option. * INSTALL: Regenerate. * README.tunables: New file. * Makeconfig (CPPFLAGS): Define TOP_NAMESPACE. (before-compile): Generate dl-tunable-list.h early. * config.h.in: Add HAVE_TUNABLES. * config.make.in: Add have-tunables. * configure.ac: Add --enable-tunables option. * configure: Regenerate. * csu/init-first.c (__libc_init_first): Move __libc_init_secure earlier... * csu/init-first.c (LIBC_START_MAIN):... to here. Include dl-tunables.h, libc-internal.h. (LIBC_START_MAIN) [!SHARED]: Initialize tunables for static binaries. * elf/Makefile (dl-routines): Add dl-tunables. * elf/Versions (ld): Add __tunable_set_val to GLIBC_PRIVATE namespace. * elf/dl-support (_dl_nondynamic_init): Unset MALLOC_CHECK_ only when !HAVE_TUNABLES. * elf/rtld.c (process_envvars): Likewise. * elf/dl-sysdep.c [HAVE_TUNABLES]: Include dl-tunables.h (_dl_sysdep_start): Call __tunables_init. * elf/dl-tunable-types.h: New file. * elf/dl-tunables.c: New file. * elf/dl-tunables.h: New file. * elf/dl-tunables.list: New file. * malloc/tst-malloc-usable-static.c: New test case. * malloc/Makefile (tests-static): Add it. * malloc/arena.c [HAVE_TUNABLES]: Include dl-tunables.h. Define TUNABLE_NAMESPACE. (DL_TUNABLE_CALLBACK (set_mallopt_check)): New function. (DL_TUNABLE_CALLBACK_FNDECL): New macro. Use it to define callback functions. (ptmalloc_init): Set tunable values. * scripts/gen-tunables.awk: New file. * sysdeps/mach/hurd/dl-sysdep.c: Include dl-tunables.h. (_dl_sysdep_start): Call __tunables_init.
2016-12-31 18:02:17 +00:00
#endif
[BZ #77] Update. Add support for namespaces in the dynamic linker. * dlfcn/Makefile (libdl-routines): Add dlmopen. * dlfcn/Versions [libdl, GLIBC_2.3.4]: Add dlmopen. * dlfcn/dlfcn.h: Define Lmid_t, LM_ID_BASE, and LM_ID_NEWLM. Declare dlmopen. Document RTLD_DI_LMID. * dlfcn/dlinfo.c: Handle RTLD_DI_LMID. * dlfcn/dlmopen.c: New file. * dlfcn/dlopen.c: Pass new parameter to _dl_open. * dlfcn/dlopenold.c: Likewise. * elf/dl-addr.c: Adjust for removal of GL(dl_loaded). * elf/dl-caller.c: Likewise. * elf/dl-close.c: Likewise. * elf/dl-conflict.c: Likewise. * elf/dl-debug.c: Likewise. * elf/dl-lookup.c: Likewise. * elf/dl-sym.c: Likewise. * elf/dl-version.c: Likewise. * elf/do-lookup.h: Likewise. * elf/rtld.c: Likewise. * sysdeps/unix/sysv/linux/i386/dl-librecon.h: Likewise. * elf/dl-depsc: Likewise. Add new parameter to _dl_map_object. * elf/dl-fini.c: Call destructors in all namespaces. * elf/dl-iteratephdr.c: Compute total nloaded. Adjust for removal of GL(dl_loaded). * elf/dl-libc.c: Pass new parameter to _dl_open. Adjust for removal of GL(dl_loaded). * elf/dl-load.c (_dl_map_object_from_fd): Don't load ld.so a second time. Reuse the one from the main namespace in all others. Pass new parameter to _dl_new_object. Adjust for removal of GL(dl_loaded). * elf/dl-object.c: Take new parameter. Use it to initialize l_ns. Adjust for removal of GL(dl_loaded). * elf/dl-open.c (_dl_open): Take new parameter. Adjust for removal of GL(dl_loaded). * elf/dl-support.c: Replace global _dl_loaded etc variables with _dl_ns variable. * include/dlfcn.h: Adjust prototype of _dl_open. Define __LM_ID_CALLER. * include/link.h: Add l_real, l_ns, and l_direct_opencount elements. * sysdeps/generic/dl-tls.c: Bump TLS_STATIC_SURPLUS. Since libc is using TLS we need memory appropriate to the number of namespaces. * sysdeps/generic/ldsodefs.h (struct rtld_global): Replace _dl_loaded, _dl_nloaded, _dl_global_scope, _dl_main_searchlist, and _dl_global_scope_alloc with _dl_ns element. Define DL_NNS. Adjust prototypes of _dl_map_object and member in rtld_global_ro. * malloc/malloc.c: Include <dlfcn.h>. * malloc/arena.c (ptmalloc_init): If libc is not in primary namespace, never use brk. * elf/Makefile: Add rules to build and run tst-dlmopen1 and tst-dlmopen2. * elf/tst-dlmopen1.c: New file. * elf/tst-dlmopen1mod.c: New file. * elf/tst-dlmopen2.c: New file. * elf/dl-close.c: Improve reference counting by tracking direct loads. * elf/dl-lookup.c (add_dependency): Likewise. * elf/dl-open.c (dl_open_worker): Likewise. * elf/rtld.c (dl_main): Likewise. 2004-09-09 GOTO Masanori <gotom@debian.or.jp> [BZ #77] * elf/dl-close.c: Count down l_opencount to check not only for l_reldeps, but also l_initfini. 2004-10-13 Ulrich Drepper <drepper@redhat.com>
2004-10-14 02:08:23 +00:00
#if USE_TCACHE
static void tcache_key_initialize (void);
#endif
static void
ptmalloc_init (void)
{
if (__malloc_initialized)
2014-01-02 08:38:18 +00:00
return;
__malloc_initialized = true;
#if USE_TCACHE
tcache_key_initialize ();
#endif
#ifdef USE_MTAG
if ((TUNABLE_GET_FULL (glibc, mem, tagging, int32_t, NULL) & 1) != 0)
{
/* If the tunable says that we should be using tagged memory
and that morecore does not support tagged regions, then
disable it. */
if (__MTAG_SBRK_UNTAGGED)
__always_fail_morecore = true;
mtag_enabled = true;
mtag_mmap_flags = __MTAG_MMAP_FLAGS;
}
#endif
#if defined SHARED && IS_IN (libc)
/* In case this libc copy is in a non-default namespace, never use
brk. Likewise if dlopened from statically linked program. The
generic sbrk implementation also enforces this, but it is not
used on Hurd. */
if (!__libc_initial)
__always_fail_morecore = true;
[BZ #77] Update. Add support for namespaces in the dynamic linker. * dlfcn/Makefile (libdl-routines): Add dlmopen. * dlfcn/Versions [libdl, GLIBC_2.3.4]: Add dlmopen. * dlfcn/dlfcn.h: Define Lmid_t, LM_ID_BASE, and LM_ID_NEWLM. Declare dlmopen. Document RTLD_DI_LMID. * dlfcn/dlinfo.c: Handle RTLD_DI_LMID. * dlfcn/dlmopen.c: New file. * dlfcn/dlopen.c: Pass new parameter to _dl_open. * dlfcn/dlopenold.c: Likewise. * elf/dl-addr.c: Adjust for removal of GL(dl_loaded). * elf/dl-caller.c: Likewise. * elf/dl-close.c: Likewise. * elf/dl-conflict.c: Likewise. * elf/dl-debug.c: Likewise. * elf/dl-lookup.c: Likewise. * elf/dl-sym.c: Likewise. * elf/dl-version.c: Likewise. * elf/do-lookup.h: Likewise. * elf/rtld.c: Likewise. * sysdeps/unix/sysv/linux/i386/dl-librecon.h: Likewise. * elf/dl-depsc: Likewise. Add new parameter to _dl_map_object. * elf/dl-fini.c: Call destructors in all namespaces. * elf/dl-iteratephdr.c: Compute total nloaded. Adjust for removal of GL(dl_loaded). * elf/dl-libc.c: Pass new parameter to _dl_open. Adjust for removal of GL(dl_loaded). * elf/dl-load.c (_dl_map_object_from_fd): Don't load ld.so a second time. Reuse the one from the main namespace in all others. Pass new parameter to _dl_new_object. Adjust for removal of GL(dl_loaded). * elf/dl-object.c: Take new parameter. Use it to initialize l_ns. Adjust for removal of GL(dl_loaded). * elf/dl-open.c (_dl_open): Take new parameter. Adjust for removal of GL(dl_loaded). * elf/dl-support.c: Replace global _dl_loaded etc variables with _dl_ns variable. * include/dlfcn.h: Adjust prototype of _dl_open. Define __LM_ID_CALLER. * include/link.h: Add l_real, l_ns, and l_direct_opencount elements. * sysdeps/generic/dl-tls.c: Bump TLS_STATIC_SURPLUS. Since libc is using TLS we need memory appropriate to the number of namespaces. * sysdeps/generic/ldsodefs.h (struct rtld_global): Replace _dl_loaded, _dl_nloaded, _dl_global_scope, _dl_main_searchlist, and _dl_global_scope_alloc with _dl_ns element. Define DL_NNS. Adjust prototypes of _dl_map_object and member in rtld_global_ro. * malloc/malloc.c: Include <dlfcn.h>. * malloc/arena.c (ptmalloc_init): If libc is not in primary namespace, never use brk. * elf/Makefile: Add rules to build and run tst-dlmopen1 and tst-dlmopen2. * elf/tst-dlmopen1.c: New file. * elf/tst-dlmopen1mod.c: New file. * elf/tst-dlmopen2.c: New file. * elf/dl-close.c: Improve reference counting by tracking direct loads. * elf/dl-lookup.c (add_dependency): Likewise. * elf/dl-open.c (dl_open_worker): Likewise. * elf/rtld.c (dl_main): Likewise. 2004-09-09 GOTO Masanori <gotom@debian.or.jp> [BZ #77] * elf/dl-close.c: Count down l_opencount to check not only for l_reldeps, but also l_initfini. 2004-10-13 Ulrich Drepper <drepper@redhat.com>
2004-10-14 02:08:23 +00:00
#endif
thread_arena = &main_arena;
Add framework for tunables The tunables framework allows us to uniformly manage and expose global variables inside glibc as switches to users. tunables/README has instructions for glibc developers to add new tunables. Tunables support can be enabled by passing the --enable-tunables configure flag to the configure script. This patch only adds a framework and does not pose any limitations on how tunable values are read from the user. It also adds environment variables used in malloc behaviour tweaking to the tunables framework as a PoC of the compatibility interface. * manual/install.texi: Add --enable-tunables option. * INSTALL: Regenerate. * README.tunables: New file. * Makeconfig (CPPFLAGS): Define TOP_NAMESPACE. (before-compile): Generate dl-tunable-list.h early. * config.h.in: Add HAVE_TUNABLES. * config.make.in: Add have-tunables. * configure.ac: Add --enable-tunables option. * configure: Regenerate. * csu/init-first.c (__libc_init_first): Move __libc_init_secure earlier... * csu/init-first.c (LIBC_START_MAIN):... to here. Include dl-tunables.h, libc-internal.h. (LIBC_START_MAIN) [!SHARED]: Initialize tunables for static binaries. * elf/Makefile (dl-routines): Add dl-tunables. * elf/Versions (ld): Add __tunable_set_val to GLIBC_PRIVATE namespace. * elf/dl-support (_dl_nondynamic_init): Unset MALLOC_CHECK_ only when !HAVE_TUNABLES. * elf/rtld.c (process_envvars): Likewise. * elf/dl-sysdep.c [HAVE_TUNABLES]: Include dl-tunables.h (_dl_sysdep_start): Call __tunables_init. * elf/dl-tunable-types.h: New file. * elf/dl-tunables.c: New file. * elf/dl-tunables.h: New file. * elf/dl-tunables.list: New file. * malloc/tst-malloc-usable-static.c: New test case. * malloc/Makefile (tests-static): Add it. * malloc/arena.c [HAVE_TUNABLES]: Include dl-tunables.h. Define TUNABLE_NAMESPACE. (DL_TUNABLE_CALLBACK (set_mallopt_check)): New function. (DL_TUNABLE_CALLBACK_FNDECL): New macro. Use it to define callback functions. (ptmalloc_init): Set tunable values. * scripts/gen-tunables.awk: New file. * sysdeps/mach/hurd/dl-sysdep.c: Include dl-tunables.h. (_dl_sysdep_start): Call __tunables_init.
2016-12-31 18:02:17 +00:00
malloc_init_state (&main_arena);
Add framework for tunables The tunables framework allows us to uniformly manage and expose global variables inside glibc as switches to users. tunables/README has instructions for glibc developers to add new tunables. Tunables support can be enabled by passing the --enable-tunables configure flag to the configure script. This patch only adds a framework and does not pose any limitations on how tunable values are read from the user. It also adds environment variables used in malloc behaviour tweaking to the tunables framework as a PoC of the compatibility interface. * manual/install.texi: Add --enable-tunables option. * INSTALL: Regenerate. * README.tunables: New file. * Makeconfig (CPPFLAGS): Define TOP_NAMESPACE. (before-compile): Generate dl-tunable-list.h early. * config.h.in: Add HAVE_TUNABLES. * config.make.in: Add have-tunables. * configure.ac: Add --enable-tunables option. * configure: Regenerate. * csu/init-first.c (__libc_init_first): Move __libc_init_secure earlier... * csu/init-first.c (LIBC_START_MAIN):... to here. Include dl-tunables.h, libc-internal.h. (LIBC_START_MAIN) [!SHARED]: Initialize tunables for static binaries. * elf/Makefile (dl-routines): Add dl-tunables. * elf/Versions (ld): Add __tunable_set_val to GLIBC_PRIVATE namespace. * elf/dl-support (_dl_nondynamic_init): Unset MALLOC_CHECK_ only when !HAVE_TUNABLES. * elf/rtld.c (process_envvars): Likewise. * elf/dl-sysdep.c [HAVE_TUNABLES]: Include dl-tunables.h (_dl_sysdep_start): Call __tunables_init. * elf/dl-tunable-types.h: New file. * elf/dl-tunables.c: New file. * elf/dl-tunables.h: New file. * elf/dl-tunables.list: New file. * malloc/tst-malloc-usable-static.c: New test case. * malloc/Makefile (tests-static): Add it. * malloc/arena.c [HAVE_TUNABLES]: Include dl-tunables.h. Define TUNABLE_NAMESPACE. (DL_TUNABLE_CALLBACK (set_mallopt_check)): New function. (DL_TUNABLE_CALLBACK_FNDECL): New macro. Use it to define callback functions. (ptmalloc_init): Set tunable values. * scripts/gen-tunables.awk: New file. * sysdeps/mach/hurd/dl-sysdep.c: Include dl-tunables.h. (_dl_sysdep_start): Call __tunables_init.
2016-12-31 18:02:17 +00:00
#if HAVE_TUNABLES
tunables: Clean up hooks to get and set tunables The TUNABLE_SET_VALUE and family of macros (and my later attempt to add a TUNABLE_GET) never quite went together very well because the overall interface was not clearly defined. This patch is an attempt to do just that. This patch consolidates the API to two simple sets of macros, TUNABLE_GET* and TUNABLE_SET*. If TUNABLE_NAMESPACE is defined, TUNABLE_GET takes just the tunable name, type and a (optionally NULL) callback function to get the value of the tunable. The callback function, if non-NULL, is called if the tunable was externally set (i.e. via GLIBC_TUNABLES or any future mechanism). For example: val = TUNABLE_GET (check, int32_t, check_callback) returns the value of the glibc.malloc.check tunable (assuming TUNABLE_NAMESPACE is set to malloc) as an int32_t into VAL after calling check_callback. Likewise, TUNABLE_SET can be used to set the value of the tunable, although this is currently possible only in the dynamic linker before it relocates itself. For example: TUNABLE_SET (check, int32_t, 2) will set glibc.malloc.check to 2. Of course, this is not possible since we set (or read) glibc.malloc.check long after it is relocated. To access or set a tunable outside of TUNABLE_NAMESPACE, use the TUNABLE_GET_FULL and TUNABLE_SET_FULL macros, which have the following prototype: TUNABLE_GET_FULL (glibc, tune, hwcap_mask, uint64_t, NULL) TUNABLE_SET_FULL (glibc, tune, hwcap_mask, uint64_t, 0xffff) In future the tunable list may get split into mutable and immutable tunables where mutable tunables can be modified by the library and userspace after relocation as well and TUNABLE_SET will be more useful than it currently is. However whenever we actually do that split, we will have to ensure that the mutable tunables are protected with locks. * elf/Versions (__tunable_set_val): Rename to __tunable_get_val. * elf/dl-tunables.c: Likewise. (do_tunable_update_val): New function. (__tunable_set_val): New function. (__tunable_get_val): Call CB only if the tunable was externally initialized. (tunables_strtoul): Replace strval with initialized. * elf/dl-tunables.h (strval): Replace with a bool initialized. (TUNABLE_ENUM_NAME, TUNABLE_ENUM_NAME1): Adjust names to prevent collision. (__tunable_set_val): New function. (TUNABLE_GET, TUNABLE_GET_FULL): New macros. (TUNABLE_SET, TUNABLE_SET_FULL): Likewise. (TUNABLE_SET_VAL): Remove. (TUNABLE_SET_VAL_WITH_CALLBACK): Likewise. * README.tunables: Document the new macros. * malloc/arena.c (ptmalloc_init): Adjust.
2017-06-01 14:54:46 +00:00
TUNABLE_GET (top_pad, size_t, TUNABLE_CALLBACK (set_top_pad));
TUNABLE_GET (perturb, int32_t, TUNABLE_CALLBACK (set_perturb_byte));
TUNABLE_GET (mmap_threshold, size_t, TUNABLE_CALLBACK (set_mmap_threshold));
TUNABLE_GET (trim_threshold, size_t, TUNABLE_CALLBACK (set_trim_threshold));
TUNABLE_GET (mmap_max, int32_t, TUNABLE_CALLBACK (set_mmaps_max));
TUNABLE_GET (arena_max, size_t, TUNABLE_CALLBACK (set_arena_max));
TUNABLE_GET (arena_test, size_t, TUNABLE_CALLBACK (set_arena_test));
# if USE_TCACHE
TUNABLE_GET (tcache_max, size_t, TUNABLE_CALLBACK (set_tcache_max));
TUNABLE_GET (tcache_count, size_t, TUNABLE_CALLBACK (set_tcache_count));
TUNABLE_GET (tcache_unsorted_limit, size_t,
TUNABLE_CALLBACK (set_tcache_unsorted_limit));
# endif
TUNABLE_GET (mxfast, size_t, TUNABLE_CALLBACK (set_mxfast));
TUNABLE_GET (hugetlb, size_t, TUNABLE_CALLBACK (set_hugetlb));
if (mp_.hp_pagesize > 0)
/* Force mmap for main arena instead of sbrk, so hugepages are explicitly
used. */
__always_fail_morecore = true;
Add framework for tunables The tunables framework allows us to uniformly manage and expose global variables inside glibc as switches to users. tunables/README has instructions for glibc developers to add new tunables. Tunables support can be enabled by passing the --enable-tunables configure flag to the configure script. This patch only adds a framework and does not pose any limitations on how tunable values are read from the user. It also adds environment variables used in malloc behaviour tweaking to the tunables framework as a PoC of the compatibility interface. * manual/install.texi: Add --enable-tunables option. * INSTALL: Regenerate. * README.tunables: New file. * Makeconfig (CPPFLAGS): Define TOP_NAMESPACE. (before-compile): Generate dl-tunable-list.h early. * config.h.in: Add HAVE_TUNABLES. * config.make.in: Add have-tunables. * configure.ac: Add --enable-tunables option. * configure: Regenerate. * csu/init-first.c (__libc_init_first): Move __libc_init_secure earlier... * csu/init-first.c (LIBC_START_MAIN):... to here. Include dl-tunables.h, libc-internal.h. (LIBC_START_MAIN) [!SHARED]: Initialize tunables for static binaries. * elf/Makefile (dl-routines): Add dl-tunables. * elf/Versions (ld): Add __tunable_set_val to GLIBC_PRIVATE namespace. * elf/dl-support (_dl_nondynamic_init): Unset MALLOC_CHECK_ only when !HAVE_TUNABLES. * elf/rtld.c (process_envvars): Likewise. * elf/dl-sysdep.c [HAVE_TUNABLES]: Include dl-tunables.h (_dl_sysdep_start): Call __tunables_init. * elf/dl-tunable-types.h: New file. * elf/dl-tunables.c: New file. * elf/dl-tunables.h: New file. * elf/dl-tunables.list: New file. * malloc/tst-malloc-usable-static.c: New test case. * malloc/Makefile (tests-static): Add it. * malloc/arena.c [HAVE_TUNABLES]: Include dl-tunables.h. Define TUNABLE_NAMESPACE. (DL_TUNABLE_CALLBACK (set_mallopt_check)): New function. (DL_TUNABLE_CALLBACK_FNDECL): New macro. Use it to define callback functions. (ptmalloc_init): Set tunable values. * scripts/gen-tunables.awk: New file. * sysdeps/mach/hurd/dl-sysdep.c: Include dl-tunables.h. (_dl_sysdep_start): Call __tunables_init.
2016-12-31 18:02:17 +00:00
#else
if (__glibc_likely (_environ != NULL))
{
char **runp = _environ;
char *envline;
while (__builtin_expect ((envline = next_env_entry (&runp)) != NULL,
2014-01-02 08:38:18 +00:00
0))
{
size_t len = strcspn (envline, "=");
if (envline[len] != '=')
/* This is a "MALLOC_" variable at the end of the string
without a '=' character. Ignore it since otherwise we
will access invalid memory below. */
continue;
switch (len)
{
case 8:
if (!__builtin_expect (__libc_enable_secure, 0))
{
if (memcmp (envline, "TOP_PAD_", 8) == 0)
__libc_mallopt (M_TOP_PAD, strtol (&envline[9], NULL, 10));
2014-01-02 08:38:18 +00:00
else if (memcmp (envline, "PERTURB_", 8) == 0)
__libc_mallopt (M_PERTURB, strtol (&envline[9], NULL, 10));
2014-01-02 08:38:18 +00:00
}
break;
case 9:
if (!__builtin_expect (__libc_enable_secure, 0))
{
if (memcmp (envline, "MMAP_MAX_", 9) == 0)
__libc_mallopt (M_MMAP_MAX, strtol (&envline[10],
NULL, 10));
2014-01-02 08:38:18 +00:00
else if (memcmp (envline, "ARENA_MAX", 9) == 0)
__libc_mallopt (M_ARENA_MAX, strtol (&envline[10],
NULL, 10));
2014-01-02 08:38:18 +00:00
}
break;
case 10:
if (!__builtin_expect (__libc_enable_secure, 0))
{
if (memcmp (envline, "ARENA_TEST", 10) == 0)
__libc_mallopt (M_ARENA_TEST, strtol (&envline[11],
NULL, 10));
2014-01-02 08:38:18 +00:00
}
break;
case 15:
if (!__builtin_expect (__libc_enable_secure, 0))
{
if (memcmp (envline, "TRIM_THRESHOLD_", 15) == 0)
__libc_mallopt (M_TRIM_THRESHOLD, strtol (&envline[16],
NULL, 10));
2014-01-02 08:38:18 +00:00
else if (memcmp (envline, "MMAP_THRESHOLD_", 15) == 0)
__libc_mallopt (M_MMAP_THRESHOLD, strtol (&envline[16],
NULL, 10));
2014-01-02 08:38:18 +00:00
}
break;
default:
break;
}
}
}
Add framework for tunables The tunables framework allows us to uniformly manage and expose global variables inside glibc as switches to users. tunables/README has instructions for glibc developers to add new tunables. Tunables support can be enabled by passing the --enable-tunables configure flag to the configure script. This patch only adds a framework and does not pose any limitations on how tunable values are read from the user. It also adds environment variables used in malloc behaviour tweaking to the tunables framework as a PoC of the compatibility interface. * manual/install.texi: Add --enable-tunables option. * INSTALL: Regenerate. * README.tunables: New file. * Makeconfig (CPPFLAGS): Define TOP_NAMESPACE. (before-compile): Generate dl-tunable-list.h early. * config.h.in: Add HAVE_TUNABLES. * config.make.in: Add have-tunables. * configure.ac: Add --enable-tunables option. * configure: Regenerate. * csu/init-first.c (__libc_init_first): Move __libc_init_secure earlier... * csu/init-first.c (LIBC_START_MAIN):... to here. Include dl-tunables.h, libc-internal.h. (LIBC_START_MAIN) [!SHARED]: Initialize tunables for static binaries. * elf/Makefile (dl-routines): Add dl-tunables. * elf/Versions (ld): Add __tunable_set_val to GLIBC_PRIVATE namespace. * elf/dl-support (_dl_nondynamic_init): Unset MALLOC_CHECK_ only when !HAVE_TUNABLES. * elf/rtld.c (process_envvars): Likewise. * elf/dl-sysdep.c [HAVE_TUNABLES]: Include dl-tunables.h (_dl_sysdep_start): Call __tunables_init. * elf/dl-tunable-types.h: New file. * elf/dl-tunables.c: New file. * elf/dl-tunables.h: New file. * elf/dl-tunables.list: New file. * malloc/tst-malloc-usable-static.c: New test case. * malloc/Makefile (tests-static): Add it. * malloc/arena.c [HAVE_TUNABLES]: Include dl-tunables.h. Define TUNABLE_NAMESPACE. (DL_TUNABLE_CALLBACK (set_mallopt_check)): New function. (DL_TUNABLE_CALLBACK_FNDECL): New macro. Use it to define callback functions. (ptmalloc_init): Set tunable values. * scripts/gen-tunables.awk: New file. * sysdeps/mach/hurd/dl-sysdep.c: Include dl-tunables.h. (_dl_sysdep_start): Call __tunables_init.
2016-12-31 18:02:17 +00:00
#endif
}
/* Managing heaps and arenas (for concurrent threads) */
#if MALLOC_DEBUG > 1
/* Print the complete contents of a single heap to stderr. */
static void
2014-01-02 08:38:18 +00:00
dump_heap (heap_info *heap)
{
char *ptr;
mchunkptr p;
2014-01-02 08:38:18 +00:00
fprintf (stderr, "Heap %p, size %10lx:\n", heap, (long) heap->size);
ptr = (heap->ar_ptr != (mstate) (heap + 1)) ?
(char *) (heap + 1) : (char *) (heap + 1) + sizeof (struct malloc_state);
p = (mchunkptr) (((uintptr_t) ptr + MALLOC_ALIGN_MASK) &
2014-01-02 08:38:18 +00:00
~MALLOC_ALIGN_MASK);
for (;; )
{
fprintf (stderr, "chunk %p size %10lx", p, (long) chunksize_nomask(p));
2014-01-02 08:38:18 +00:00
if (p == top (heap->ar_ptr))
{
fprintf (stderr, " (top)\n");
break;
}
else if (chunksize_nomask(p) == (0 | PREV_INUSE))
2014-01-02 08:38:18 +00:00
{
fprintf (stderr, " (fence)\n");
break;
}
fprintf (stderr, "\n");
p = next_chunk (p);
}
}
#endif /* MALLOC_DEBUG > 1 */
/* If consecutive mmap (0, HEAP_MAX_SIZE << 1, ...) calls return decreasing
addresses as opposed to increasing, new_heap would badly fragment the
address space. In that case remember the second HEAP_MAX_SIZE part
aligned to HEAP_MAX_SIZE from last mmap (0, HEAP_MAX_SIZE << 1, ...)
call (if it is already aligned) and try to reuse it next time. We need
no locking for it, as kernel ensures the atomicity for us - worst case
we'll call mmap (addr, HEAP_MAX_SIZE, ...) for some value of addr in
multiple threads, but only one will succeed. */
static char *aligned_heap_area;
/* Create a new heap. size is automatically rounded up to a multiple
of the page size. */
static heap_info *
alloc_new_heap (size_t size, size_t top_pad, size_t pagesize,
int mmap_flags)
{
char *p1, *p2;
unsigned long ul;
heap_info *h;
size_t min_size = heap_min_size ();
size_t max_size = heap_max_size ();
if (size + top_pad < min_size)
size = min_size;
else if (size + top_pad <= max_size)
size += top_pad;
else if (size > max_size)
return 0;
else
size = max_size;
size = ALIGN_UP (size, pagesize);
/* A memory region aligned to a multiple of max_size is needed.
No swap space needs to be reserved for the following large
mapping (on Linux, this is the case for all non-writable mappings
anyway). */
p2 = MAP_FAILED;
2014-01-02 08:38:18 +00:00
if (aligned_heap_area)
{
p2 = (char *) MMAP (aligned_heap_area, max_size, PROT_NONE, mmap_flags);
2014-01-02 08:38:18 +00:00
aligned_heap_area = NULL;
if (p2 != MAP_FAILED && ((unsigned long) p2 & (max_size - 1)))
2014-01-02 08:38:18 +00:00
{
__munmap (p2, max_size);
2014-01-02 08:38:18 +00:00
p2 = MAP_FAILED;
}
}
2014-01-02 08:38:18 +00:00
if (p2 == MAP_FAILED)
{
p1 = (char *) MMAP (0, max_size << 1, PROT_NONE, mmap_flags);
2014-01-02 08:38:18 +00:00
if (p1 != MAP_FAILED)
{
p2 = (char *) (((uintptr_t) p1 + (max_size - 1))
& ~(max_size - 1));
2014-01-02 08:38:18 +00:00
ul = p2 - p1;
if (ul)
__munmap (p1, ul);
else
aligned_heap_area = p2 + max_size;
__munmap (p2 + max_size, max_size - ul);
2014-01-02 08:38:18 +00:00
}
else
2014-01-02 08:38:18 +00:00
{
/* Try to take the chance that an allocation of only max_size
2014-01-02 08:38:18 +00:00
is already aligned. */
p2 = (char *) MMAP (0, max_size, PROT_NONE, mmap_flags);
2014-01-02 08:38:18 +00:00
if (p2 == MAP_FAILED)
return 0;
if ((unsigned long) p2 & (max_size - 1))
2014-01-02 08:38:18 +00:00
{
__munmap (p2, max_size);
2014-01-02 08:38:18 +00:00
return 0;
}
}
}
if (__mprotect (p2, size, mtag_mmap_flags | PROT_READ | PROT_WRITE) != 0)
2014-01-02 08:38:18 +00:00
{
__munmap (p2, max_size);
2014-01-02 08:38:18 +00:00
return 0;
}
madvise_thp (p2, size);
2014-01-02 08:38:18 +00:00
h = (heap_info *) p2;
h->size = size;
h->mprotect_size = size;
h->pagesize = pagesize;
LIBC_PROBE (memory_heap_new, 2, h, h->size);
return h;
}
static heap_info *
new_heap (size_t size, size_t top_pad)
{
#if HAVE_TUNABLES
if (__glibc_unlikely (mp_.hp_pagesize != 0))
{
heap_info *h = alloc_new_heap (size, top_pad, mp_.hp_pagesize,
mp_.hp_flags);
if (h != NULL)
return h;
}
#endif
return alloc_new_heap (size, top_pad, GLRO (dl_pagesize), 0);
}
/* Grow a heap. size is automatically rounded up to a
multiple of the page size. */
static int
2014-01-02 08:38:18 +00:00
grow_heap (heap_info *h, long diff)
{
size_t pagesize = h->pagesize;
size_t max_size = heap_max_size ();
long new_size;
diff = ALIGN_UP (diff, pagesize);
2014-01-02 08:38:18 +00:00
new_size = (long) h->size + diff;
if ((unsigned long) new_size > (unsigned long) max_size)
return -1;
2014-01-02 08:38:18 +00:00
if ((unsigned long) new_size > h->mprotect_size)
{
if (__mprotect ((char *) h + h->mprotect_size,
(unsigned long) new_size - h->mprotect_size,
mtag_mmap_flags | PROT_READ | PROT_WRITE) != 0)
2014-01-02 08:38:18 +00:00
return -2;
h->mprotect_size = new_size;
}
h->size = new_size;
LIBC_PROBE (memory_heap_more, 2, h, h->size);
return 0;
}
/* Shrink a heap. */
static int
2014-01-02 08:38:18 +00:00
shrink_heap (heap_info *h, long diff)
{
long new_size;
2014-01-02 08:38:18 +00:00
new_size = (long) h->size - diff;
if (new_size < (long) sizeof (*h))
return -1;
2014-01-02 08:38:18 +00:00
/* Try to re-map the extra heap space freshly to save memory, and make it
inaccessible. See malloc-sysdep.h to know when this is true. */
if (__glibc_unlikely (check_may_shrink_heap ()))
{
2014-01-02 08:38:18 +00:00
if ((char *) MMAP ((char *) h + new_size, diff, PROT_NONE,
MAP_FIXED) == (char *) MAP_FAILED)
return -2;
h->mprotect_size = new_size;
}
else
2014-01-02 08:38:18 +00:00
__madvise ((char *) h + new_size, diff, MADV_DONTNEED);
/*fprintf(stderr, "shrink %p %08lx\n", h, new_size);*/
h->size = new_size;
LIBC_PROBE (memory_heap_less, 2, h, h->size);
return 0;
}
/* Delete a heap. */
static int
2014-01-02 08:38:18 +00:00
heap_trim (heap_info *heap, size_t pad)
{
mstate ar_ptr = heap->ar_ptr;
mchunkptr top_chunk = top (ar_ptr), p;
heap_info *prev_heap;
malloc: Consistently apply trim_threshold to all heaps [BZ #17195] Trimming heaps is a balance between saving memory and the system overhead required to update page tables and discard allocated pages. The malloc option M_TRIM_THRESHOLD is a tunable that users are meant to use to decide where this balance point is but it is only applied to the main arena. For scalability reasons, glibc malloc has per-thread heaps but these are shrunk with madvise() if there is one page free at the top of the heap. In some circumstances this can lead to high system overhead if a thread has a control flow like while (data_to_process) { buf = malloc(large_size); do_stuff(); free(buf); } For a large size, the free() will call madvise (pagetable teardown, page free and TLB flush) every time followed immediately by a malloc (fault, kernel page alloc, zeroing and charge accounting). The kernel overhead can dominate such a workload. This patch allows the user to tune when madvise gets called by applying the trim threshold to the per-thread heaps and using similar logic to the main arena when deciding whether to shrink. Alternatively if the dynamic brk/mmap threshold gets adjusted then the new values will be obeyed by the per-thread heaps. Bug 17195 was a test case motivated by a problem encountered in scientific applications written in python that performance badly due to high page fault overhead. The basic operation of such a program was posted by Julian Taylor https://sourceware.org/ml/libc-alpha/2015-02/msg00373.html With this patch applied, the overhead is eliminated. All numbers in this report are in seconds and were recorded by running Julian's program 30 times. pyarray glibc madvise 2.21 v2 System min 1.81 ( 0.00%) 0.00 (100.00%) System mean 1.93 ( 0.00%) 0.02 ( 99.20%) System stddev 0.06 ( 0.00%) 0.01 ( 88.99%) System max 2.06 ( 0.00%) 0.03 ( 98.54%) Elapsed min 3.26 ( 0.00%) 2.37 ( 27.30%) Elapsed mean 3.39 ( 0.00%) 2.41 ( 28.84%) Elapsed stddev 0.14 ( 0.00%) 0.02 ( 82.73%) Elapsed max 4.05 ( 0.00%) 2.47 ( 39.01%) glibc madvise 2.21 v2 User 141.86 142.28 System 57.94 0.60 Elapsed 102.02 72.66 Note that almost a minutes worth of system time is eliminted and the program completes 28% faster on average. To illustrate the problem without python this is a basic test-case for the worst case scenario where every free is a madvise followed by a an alloc /* gcc bench-free.c -lpthread -o bench-free */ static int num = 1024; void __attribute__((noinline,noclone)) dostuff (void *p) { } void *worker (void *data) { int i; for (i = num; i--;) { void *m = malloc (48*4096); dostuff (m); free (m); } return NULL; } int main() { int i; pthread_t t; void *ret; if (pthread_create (&t, NULL, worker, NULL)) exit (2); if (pthread_join (t, &ret)) exit (3); return 0; } Before the patch, this resulted in 1024 calls to madvise. With the patch applied, madvise is called twice because the default trim threshold is high enough to avoid this. This a more complex case where there is a mix of frees. It's simply a different worker function for the test case above void *worker (void *data) { int i; int j = 0; void *free_index[num]; for (i = num; i--;) { void *m = malloc ((i % 58) *4096); dostuff (m); if (i % 2 == 0) { free (m); } else { free_index[j++] = m; } } for (; j >= 0; j--) { free(free_index[j]); } return NULL; } glibc 2.21 calls malloc 90305 times but with the patch applied, it's called 13438. Increasing the trim threshold will decrease the number of times it's called with the option of eliminating the overhead. ebizzy is meant to generate a workload resembling common web application server workloads. It is threaded with a large working set that at its core has an allocation, do_stuff, free loop that also hits this case. The primary metric of the benchmark is records processed per second. This is running on my desktop which is a single socket machine with an I7-4770 and 8 cores. Each thread count was run for 30 seconds. It was only run once as the performance difference is so high that the variation is insignificant. glibc 2.21 patch threads 1 10230 44114 threads 2 19153 84925 threads 4 34295 134569 threads 8 51007 183387 Note that the saving happens to be a concidence as the size allocated by ebizzy was less than the default threshold. If a different number of chunks were specified then it may also be necessary to tune the threshold to compensate This is roughly quadrupling the performance of this benchmark. The difference in system CPU usage illustrates why. ebizzy running 1 thread with glibc 2.21 10230 records/s 306904 real 30.00 s user 7.47 s sys 22.49 s 22.49 seconds was spent in the kernel for a workload runinng 30 seconds. With the patch applied ebizzy running 1 thread with patch applied 44126 records/s 1323792 real 30.00 s user 29.97 s sys 0.00 s system CPU usage was zero with the patch applied. strace shows that glibc running this workload calls madvise approximately 9000 times a second. With the patch applied madvise was called twice during the workload (or 0.06 times per second). 2015-02-10 Mel Gorman <mgorman@suse.de> [BZ #17195] * malloc/arena.c (free): Apply trim threshold to per-thread heaps as well as the main arena.
2015-04-02 06:44:14 +00:00
long new_size, top_size, top_area, extra, prev_size, misalign;
size_t max_size = heap_max_size ();
/* Can this heap go away completely? */
2014-01-02 08:38:18 +00:00
while (top_chunk == chunk_at_offset (heap, sizeof (*heap)))
{
prev_heap = heap->prev;
prev_size = prev_heap->size - (MINSIZE - 2 * SIZE_SZ);
p = chunk_at_offset (prev_heap, prev_size);
/* fencepost must be properly aligned. */
misalign = ((long) p) & MALLOC_ALIGN_MASK;
p = chunk_at_offset (prev_heap, prev_size - misalign);
assert (chunksize_nomask (p) == (0 | PREV_INUSE)); /* must be fencepost */
2014-01-02 08:38:18 +00:00
p = prev_chunk (p);
new_size = chunksize (p) + (MINSIZE - 2 * SIZE_SZ) + misalign;
assert (new_size > 0 && new_size < (long) (2 * MINSIZE));
if (!prev_inuse (p))
new_size += prev_size (p);
assert (new_size > 0 && new_size < max_size);
if (new_size + (max_size - prev_heap->size) < pad + MINSIZE
+ heap->pagesize)
2014-01-02 08:38:18 +00:00
break;
ar_ptr->system_mem -= heap->size;
LIBC_PROBE (memory_heap_free, 2, heap, heap->size);
if ((char *) heap + max_size == aligned_heap_area)
aligned_heap_area = NULL;
__munmap (heap, max_size);
2014-01-02 08:38:18 +00:00
heap = prev_heap;
if (!prev_inuse (p)) /* consolidate backward */
{
p = prev_chunk (p);
unlink_chunk (ar_ptr, p);
2014-01-02 08:38:18 +00:00
}
assert (((unsigned long) ((char *) p + new_size) & (heap->pagesize - 1))
== 0);
2014-01-02 08:38:18 +00:00
assert (((char *) p + new_size) == ((char *) heap + heap->size));
top (ar_ptr) = top_chunk = p;
set_head (top_chunk, new_size | PREV_INUSE);
/*check_chunk(ar_ptr, top_chunk);*/
}
malloc: Consistently apply trim_threshold to all heaps [BZ #17195] Trimming heaps is a balance between saving memory and the system overhead required to update page tables and discard allocated pages. The malloc option M_TRIM_THRESHOLD is a tunable that users are meant to use to decide where this balance point is but it is only applied to the main arena. For scalability reasons, glibc malloc has per-thread heaps but these are shrunk with madvise() if there is one page free at the top of the heap. In some circumstances this can lead to high system overhead if a thread has a control flow like while (data_to_process) { buf = malloc(large_size); do_stuff(); free(buf); } For a large size, the free() will call madvise (pagetable teardown, page free and TLB flush) every time followed immediately by a malloc (fault, kernel page alloc, zeroing and charge accounting). The kernel overhead can dominate such a workload. This patch allows the user to tune when madvise gets called by applying the trim threshold to the per-thread heaps and using similar logic to the main arena when deciding whether to shrink. Alternatively if the dynamic brk/mmap threshold gets adjusted then the new values will be obeyed by the per-thread heaps. Bug 17195 was a test case motivated by a problem encountered in scientific applications written in python that performance badly due to high page fault overhead. The basic operation of such a program was posted by Julian Taylor https://sourceware.org/ml/libc-alpha/2015-02/msg00373.html With this patch applied, the overhead is eliminated. All numbers in this report are in seconds and were recorded by running Julian's program 30 times. pyarray glibc madvise 2.21 v2 System min 1.81 ( 0.00%) 0.00 (100.00%) System mean 1.93 ( 0.00%) 0.02 ( 99.20%) System stddev 0.06 ( 0.00%) 0.01 ( 88.99%) System max 2.06 ( 0.00%) 0.03 ( 98.54%) Elapsed min 3.26 ( 0.00%) 2.37 ( 27.30%) Elapsed mean 3.39 ( 0.00%) 2.41 ( 28.84%) Elapsed stddev 0.14 ( 0.00%) 0.02 ( 82.73%) Elapsed max 4.05 ( 0.00%) 2.47 ( 39.01%) glibc madvise 2.21 v2 User 141.86 142.28 System 57.94 0.60 Elapsed 102.02 72.66 Note that almost a minutes worth of system time is eliminted and the program completes 28% faster on average. To illustrate the problem without python this is a basic test-case for the worst case scenario where every free is a madvise followed by a an alloc /* gcc bench-free.c -lpthread -o bench-free */ static int num = 1024; void __attribute__((noinline,noclone)) dostuff (void *p) { } void *worker (void *data) { int i; for (i = num; i--;) { void *m = malloc (48*4096); dostuff (m); free (m); } return NULL; } int main() { int i; pthread_t t; void *ret; if (pthread_create (&t, NULL, worker, NULL)) exit (2); if (pthread_join (t, &ret)) exit (3); return 0; } Before the patch, this resulted in 1024 calls to madvise. With the patch applied, madvise is called twice because the default trim threshold is high enough to avoid this. This a more complex case where there is a mix of frees. It's simply a different worker function for the test case above void *worker (void *data) { int i; int j = 0; void *free_index[num]; for (i = num; i--;) { void *m = malloc ((i % 58) *4096); dostuff (m); if (i % 2 == 0) { free (m); } else { free_index[j++] = m; } } for (; j >= 0; j--) { free(free_index[j]); } return NULL; } glibc 2.21 calls malloc 90305 times but with the patch applied, it's called 13438. Increasing the trim threshold will decrease the number of times it's called with the option of eliminating the overhead. ebizzy is meant to generate a workload resembling common web application server workloads. It is threaded with a large working set that at its core has an allocation, do_stuff, free loop that also hits this case. The primary metric of the benchmark is records processed per second. This is running on my desktop which is a single socket machine with an I7-4770 and 8 cores. Each thread count was run for 30 seconds. It was only run once as the performance difference is so high that the variation is insignificant. glibc 2.21 patch threads 1 10230 44114 threads 2 19153 84925 threads 4 34295 134569 threads 8 51007 183387 Note that the saving happens to be a concidence as the size allocated by ebizzy was less than the default threshold. If a different number of chunks were specified then it may also be necessary to tune the threshold to compensate This is roughly quadrupling the performance of this benchmark. The difference in system CPU usage illustrates why. ebizzy running 1 thread with glibc 2.21 10230 records/s 306904 real 30.00 s user 7.47 s sys 22.49 s 22.49 seconds was spent in the kernel for a workload runinng 30 seconds. With the patch applied ebizzy running 1 thread with patch applied 44126 records/s 1323792 real 30.00 s user 29.97 s sys 0.00 s system CPU usage was zero with the patch applied. strace shows that glibc running this workload calls madvise approximately 9000 times a second. With the patch applied madvise was called twice during the workload (or 0.06 times per second). 2015-02-10 Mel Gorman <mgorman@suse.de> [BZ #17195] * malloc/arena.c (free): Apply trim threshold to per-thread heaps as well as the main arena.
2015-04-02 06:44:14 +00:00
/* Uses similar logic for per-thread arenas as the main arena with systrim
and _int_free by preserving the top pad and rounding down to the nearest
page. */
2014-01-02 08:38:18 +00:00
top_size = chunksize (top_chunk);
if ((unsigned long)(top_size) <
(unsigned long)(mp_.trim_threshold))
return 0;
malloc: Consistently apply trim_threshold to all heaps [BZ #17195] Trimming heaps is a balance between saving memory and the system overhead required to update page tables and discard allocated pages. The malloc option M_TRIM_THRESHOLD is a tunable that users are meant to use to decide where this balance point is but it is only applied to the main arena. For scalability reasons, glibc malloc has per-thread heaps but these are shrunk with madvise() if there is one page free at the top of the heap. In some circumstances this can lead to high system overhead if a thread has a control flow like while (data_to_process) { buf = malloc(large_size); do_stuff(); free(buf); } For a large size, the free() will call madvise (pagetable teardown, page free and TLB flush) every time followed immediately by a malloc (fault, kernel page alloc, zeroing and charge accounting). The kernel overhead can dominate such a workload. This patch allows the user to tune when madvise gets called by applying the trim threshold to the per-thread heaps and using similar logic to the main arena when deciding whether to shrink. Alternatively if the dynamic brk/mmap threshold gets adjusted then the new values will be obeyed by the per-thread heaps. Bug 17195 was a test case motivated by a problem encountered in scientific applications written in python that performance badly due to high page fault overhead. The basic operation of such a program was posted by Julian Taylor https://sourceware.org/ml/libc-alpha/2015-02/msg00373.html With this patch applied, the overhead is eliminated. All numbers in this report are in seconds and were recorded by running Julian's program 30 times. pyarray glibc madvise 2.21 v2 System min 1.81 ( 0.00%) 0.00 (100.00%) System mean 1.93 ( 0.00%) 0.02 ( 99.20%) System stddev 0.06 ( 0.00%) 0.01 ( 88.99%) System max 2.06 ( 0.00%) 0.03 ( 98.54%) Elapsed min 3.26 ( 0.00%) 2.37 ( 27.30%) Elapsed mean 3.39 ( 0.00%) 2.41 ( 28.84%) Elapsed stddev 0.14 ( 0.00%) 0.02 ( 82.73%) Elapsed max 4.05 ( 0.00%) 2.47 ( 39.01%) glibc madvise 2.21 v2 User 141.86 142.28 System 57.94 0.60 Elapsed 102.02 72.66 Note that almost a minutes worth of system time is eliminted and the program completes 28% faster on average. To illustrate the problem without python this is a basic test-case for the worst case scenario where every free is a madvise followed by a an alloc /* gcc bench-free.c -lpthread -o bench-free */ static int num = 1024; void __attribute__((noinline,noclone)) dostuff (void *p) { } void *worker (void *data) { int i; for (i = num; i--;) { void *m = malloc (48*4096); dostuff (m); free (m); } return NULL; } int main() { int i; pthread_t t; void *ret; if (pthread_create (&t, NULL, worker, NULL)) exit (2); if (pthread_join (t, &ret)) exit (3); return 0; } Before the patch, this resulted in 1024 calls to madvise. With the patch applied, madvise is called twice because the default trim threshold is high enough to avoid this. This a more complex case where there is a mix of frees. It's simply a different worker function for the test case above void *worker (void *data) { int i; int j = 0; void *free_index[num]; for (i = num; i--;) { void *m = malloc ((i % 58) *4096); dostuff (m); if (i % 2 == 0) { free (m); } else { free_index[j++] = m; } } for (; j >= 0; j--) { free(free_index[j]); } return NULL; } glibc 2.21 calls malloc 90305 times but with the patch applied, it's called 13438. Increasing the trim threshold will decrease the number of times it's called with the option of eliminating the overhead. ebizzy is meant to generate a workload resembling common web application server workloads. It is threaded with a large working set that at its core has an allocation, do_stuff, free loop that also hits this case. The primary metric of the benchmark is records processed per second. This is running on my desktop which is a single socket machine with an I7-4770 and 8 cores. Each thread count was run for 30 seconds. It was only run once as the performance difference is so high that the variation is insignificant. glibc 2.21 patch threads 1 10230 44114 threads 2 19153 84925 threads 4 34295 134569 threads 8 51007 183387 Note that the saving happens to be a concidence as the size allocated by ebizzy was less than the default threshold. If a different number of chunks were specified then it may also be necessary to tune the threshold to compensate This is roughly quadrupling the performance of this benchmark. The difference in system CPU usage illustrates why. ebizzy running 1 thread with glibc 2.21 10230 records/s 306904 real 30.00 s user 7.47 s sys 22.49 s 22.49 seconds was spent in the kernel for a workload runinng 30 seconds. With the patch applied ebizzy running 1 thread with patch applied 44126 records/s 1323792 real 30.00 s user 29.97 s sys 0.00 s system CPU usage was zero with the patch applied. strace shows that glibc running this workload calls madvise approximately 9000 times a second. With the patch applied madvise was called twice during the workload (or 0.06 times per second). 2015-02-10 Mel Gorman <mgorman@suse.de> [BZ #17195] * malloc/arena.c (free): Apply trim threshold to per-thread heaps as well as the main arena.
2015-04-02 06:44:14 +00:00
top_area = top_size - MINSIZE - 1;
if (top_area < 0 || (size_t) top_area <= pad)
malloc: Consistently apply trim_threshold to all heaps [BZ #17195] Trimming heaps is a balance between saving memory and the system overhead required to update page tables and discard allocated pages. The malloc option M_TRIM_THRESHOLD is a tunable that users are meant to use to decide where this balance point is but it is only applied to the main arena. For scalability reasons, glibc malloc has per-thread heaps but these are shrunk with madvise() if there is one page free at the top of the heap. In some circumstances this can lead to high system overhead if a thread has a control flow like while (data_to_process) { buf = malloc(large_size); do_stuff(); free(buf); } For a large size, the free() will call madvise (pagetable teardown, page free and TLB flush) every time followed immediately by a malloc (fault, kernel page alloc, zeroing and charge accounting). The kernel overhead can dominate such a workload. This patch allows the user to tune when madvise gets called by applying the trim threshold to the per-thread heaps and using similar logic to the main arena when deciding whether to shrink. Alternatively if the dynamic brk/mmap threshold gets adjusted then the new values will be obeyed by the per-thread heaps. Bug 17195 was a test case motivated by a problem encountered in scientific applications written in python that performance badly due to high page fault overhead. The basic operation of such a program was posted by Julian Taylor https://sourceware.org/ml/libc-alpha/2015-02/msg00373.html With this patch applied, the overhead is eliminated. All numbers in this report are in seconds and were recorded by running Julian's program 30 times. pyarray glibc madvise 2.21 v2 System min 1.81 ( 0.00%) 0.00 (100.00%) System mean 1.93 ( 0.00%) 0.02 ( 99.20%) System stddev 0.06 ( 0.00%) 0.01 ( 88.99%) System max 2.06 ( 0.00%) 0.03 ( 98.54%) Elapsed min 3.26 ( 0.00%) 2.37 ( 27.30%) Elapsed mean 3.39 ( 0.00%) 2.41 ( 28.84%) Elapsed stddev 0.14 ( 0.00%) 0.02 ( 82.73%) Elapsed max 4.05 ( 0.00%) 2.47 ( 39.01%) glibc madvise 2.21 v2 User 141.86 142.28 System 57.94 0.60 Elapsed 102.02 72.66 Note that almost a minutes worth of system time is eliminted and the program completes 28% faster on average. To illustrate the problem without python this is a basic test-case for the worst case scenario where every free is a madvise followed by a an alloc /* gcc bench-free.c -lpthread -o bench-free */ static int num = 1024; void __attribute__((noinline,noclone)) dostuff (void *p) { } void *worker (void *data) { int i; for (i = num; i--;) { void *m = malloc (48*4096); dostuff (m); free (m); } return NULL; } int main() { int i; pthread_t t; void *ret; if (pthread_create (&t, NULL, worker, NULL)) exit (2); if (pthread_join (t, &ret)) exit (3); return 0; } Before the patch, this resulted in 1024 calls to madvise. With the patch applied, madvise is called twice because the default trim threshold is high enough to avoid this. This a more complex case where there is a mix of frees. It's simply a different worker function for the test case above void *worker (void *data) { int i; int j = 0; void *free_index[num]; for (i = num; i--;) { void *m = malloc ((i % 58) *4096); dostuff (m); if (i % 2 == 0) { free (m); } else { free_index[j++] = m; } } for (; j >= 0; j--) { free(free_index[j]); } return NULL; } glibc 2.21 calls malloc 90305 times but with the patch applied, it's called 13438. Increasing the trim threshold will decrease the number of times it's called with the option of eliminating the overhead. ebizzy is meant to generate a workload resembling common web application server workloads. It is threaded with a large working set that at its core has an allocation, do_stuff, free loop that also hits this case. The primary metric of the benchmark is records processed per second. This is running on my desktop which is a single socket machine with an I7-4770 and 8 cores. Each thread count was run for 30 seconds. It was only run once as the performance difference is so high that the variation is insignificant. glibc 2.21 patch threads 1 10230 44114 threads 2 19153 84925 threads 4 34295 134569 threads 8 51007 183387 Note that the saving happens to be a concidence as the size allocated by ebizzy was less than the default threshold. If a different number of chunks were specified then it may also be necessary to tune the threshold to compensate This is roughly quadrupling the performance of this benchmark. The difference in system CPU usage illustrates why. ebizzy running 1 thread with glibc 2.21 10230 records/s 306904 real 30.00 s user 7.47 s sys 22.49 s 22.49 seconds was spent in the kernel for a workload runinng 30 seconds. With the patch applied ebizzy running 1 thread with patch applied 44126 records/s 1323792 real 30.00 s user 29.97 s sys 0.00 s system CPU usage was zero with the patch applied. strace shows that glibc running this workload calls madvise approximately 9000 times a second. With the patch applied madvise was called twice during the workload (or 0.06 times per second). 2015-02-10 Mel Gorman <mgorman@suse.de> [BZ #17195] * malloc/arena.c (free): Apply trim threshold to per-thread heaps as well as the main arena.
2015-04-02 06:44:14 +00:00
return 0;
/* Release in pagesize units and round down to the nearest page. */
extra = ALIGN_DOWN(top_area - pad, heap->pagesize);
if (extra == 0)
return 0;
2014-01-02 08:38:18 +00:00
/* Try to shrink. */
2014-01-02 08:38:18 +00:00
if (shrink_heap (heap, extra) != 0)
return 0;
2014-01-02 08:38:18 +00:00
ar_ptr->system_mem -= extra;
/* Success. Adjust top accordingly. */
2014-01-02 08:38:18 +00:00
set_head (top_chunk, (top_size - extra) | PREV_INUSE);
/*check_chunk(ar_ptr, top_chunk);*/
return 1;
}
/* Create a new arena with initial size "size". */
#if IS_IN (libc)
/* If REPLACED_ARENA is not NULL, detach it from this thread. Must be
called while free_list_lock is held. */
static void
detach_arena (mstate replaced_arena)
{
if (replaced_arena != NULL)
{
assert (replaced_arena->attached_threads > 0);
/* The current implementation only detaches from main_arena in
case of allocation failure. This means that it is likely not
beneficial to put the arena on free_list even if the
reference count reaches zero. */
--replaced_arena->attached_threads;
}
}
static mstate
2014-01-02 08:38:18 +00:00
_int_new_arena (size_t size)
{
mstate a;
heap_info *h;
char *ptr;
unsigned long misalign;
2014-01-02 08:38:18 +00:00
h = new_heap (size + (sizeof (*h) + sizeof (*a) + MALLOC_ALIGNMENT),
mp_.top_pad);
if (!h)
{
/* Maybe size is too large to fit in a single heap. So, just try
to create a minimally-sized arena and let _int_malloc() attempt
to deal with the large request via mmap_chunk(). */
h = new_heap (sizeof (*h) + sizeof (*a) + MALLOC_ALIGNMENT, mp_.top_pad);
if (!h)
return 0;
}
a = h->ar_ptr = (mstate) (h + 1);
malloc_init_state (a);
a->attached_threads = 1;
/*a->next = NULL;*/
a->system_mem = a->max_system_mem = h->size;
/* Set up the top chunk, with proper alignment. */
2014-01-02 08:38:18 +00:00
ptr = (char *) (a + 1);
misalign = (uintptr_t) chunk2mem (ptr) & MALLOC_ALIGN_MASK;
if (misalign > 0)
ptr += MALLOC_ALIGNMENT - misalign;
2014-01-02 08:38:18 +00:00
top (a) = (mchunkptr) ptr;
set_head (top (a), (((char *) h + h->size) - ptr) | PREV_INUSE);
LIBC_PROBE (memory_arena_new, 2, a, size);
mstate replaced_arena = thread_arena;
thread_arena = a;
__libc_lock_init (a->mutex);
__libc_lock_lock (list_lock);
/* Add the new arena to the global list. */
a->next = main_arena.next;
/* FIXME: The barrier is an attempt to synchronize with read access
in reused_arena, which does not acquire list_lock while
traversing the list. */
atomic_write_barrier ();
main_arena.next = a;
__libc_lock_unlock (list_lock);
__libc_lock_lock (free_list_lock);
detach_arena (replaced_arena);
__libc_lock_unlock (free_list_lock);
/* Lock this arena. NB: Another thread may have been attached to
this arena because the arena is now accessible from the
main_arena.next list and could have been picked by reused_arena.
This can only happen for the last arena created (before the arena
limit is reached). At this point, some arena has to be attached
to two threads. We could acquire the arena lock before list_lock
to make it less likely that reused_arena picks this new arena,
but this could result in a deadlock with
__malloc_fork_lock_parent. */
__libc_lock_lock (a->mutex);
return a;
}
/* Remove an arena from free_list. */
static mstate
get_free_list (void)
{
mstate replaced_arena = thread_arena;
mstate result = free_list;
if (result != NULL)
{
__libc_lock_lock (free_list_lock);
result = free_list;
if (result != NULL)
{
free_list = result->next_free;
/* The arena will be attached to this thread. */
assert (result->attached_threads == 0);
result->attached_threads = 1;
detach_arena (replaced_arena);
}
__libc_lock_unlock (free_list_lock);
if (result != NULL)
2014-01-02 08:38:18 +00:00
{
LIBC_PROBE (memory_arena_reuse_free_list, 1, result);
__libc_lock_lock (result->mutex);
thread_arena = result;
2014-01-02 08:38:18 +00:00
}
}
return result;
}
/* Remove the arena from the free list (if it is present).
free_list_lock must have been acquired by the caller. */
static void
remove_from_free_list (mstate arena)
{
mstate *previous = &free_list;
for (mstate p = free_list; p != NULL; p = p->next_free)
{
assert (p->attached_threads == 0);
if (p == arena)
{
/* Remove the requested arena from the list. */
*previous = p->next_free;
break;
}
else
previous = &p->next_free;
}
}
/* Lock and return an arena that can be reused for memory allocation.
Avoid AVOID_ARENA as we have already failed to allocate memory in
it and it is currently locked. */
static mstate
reused_arena (mstate avoid_arena)
{
mstate result;
/* FIXME: Access to next_to_use suffers from data races. */
static mstate next_to_use;
if (next_to_use == NULL)
next_to_use = &main_arena;
/* Iterate over all arenas (including those linked from
free_list). */
result = next_to_use;
do
{
if (!__libc_lock_trylock (result->mutex))
2014-01-02 08:38:18 +00:00
goto out;
/* FIXME: This is a data race, see _int_new_arena. */
result = result->next;
}
while (result != next_to_use);
/* Avoid AVOID_ARENA as we have already failed to allocate memory
in that arena and it is currently locked. */
if (result == avoid_arena)
result = result->next;
Avoid deadlock in malloc on backtrace (BZ #16159) When the malloc subsystem detects some kind of memory corruption, depending on the configuration it prints the error, a backtrace, a memory map and then aborts the process. In this process, the backtrace() call may result in a call to malloc, resulting in various kinds of problematic behavior. In one case, the malloc it calls may detect a corruption and call backtrace again, and a stack overflow may result due to the infinite recursion. In another case, the malloc it calls may deadlock on an arena lock with the malloc (or free, realloc, etc.) that detected the corruption. In yet another case, if the program is linked with pthreads, backtrace may do a pthread_once initialization, which deadlocks on itself. In all these cases, the program exit is not as intended. This is avoidable by marking the arena that malloc detected a corruption on, as unusable. The following patch does that. Features of this patch are as follows: - A flag is added to the mstate struct of the arena to indicate if the arena is corrupt. - The flag is checked whenever malloc functions try to get a lock on an arena. If the arena is unusable, a NULL is returned, causing the malloc to use mmap or try the next arena. - malloc_printerr sets the corrupt flag on the arena when it detects a corruption - free does not concern itself with the flag at all. It is not important since the backtrace workflow does not need free. A free in a parallel thread may cause another corruption, but that's not new - The flag check and set are not atomic and may race. This is fine since we don't care about contention during the flag check. We want to make sure that the malloc call in the backtrace does not trip on itself and all that action happens in the same thread and not across threads. I verified that the test case does not show any regressions due to this patch. I also ran the malloc benchmarks and found an insignificant difference in timings (< 2%). * malloc/Makefile (tests): New test case tst-malloc-backtrace. * malloc/arena.c (arena_lock): Check if arena is corrupt. (reused_arena): Find a non-corrupt arena. (heap_trim): Pass arena to unlink. * malloc/hooks.c (malloc_check_get_size): Pass arena to malloc_printerr. (top_check): Likewise. (free_check): Likewise. (realloc_check): Likewise. * malloc/malloc.c (malloc_printerr): Add arena argument. (unlink): Likewise. (munmap_chunk): Adjust. (ARENA_CORRUPTION_BIT): New macro. (arena_is_corrupt): Likewise. (set_arena_corrupt): Likewise. (sysmalloc): Use mmap if there are no usable arenas. (_int_malloc): Likewise. (__libc_malloc): Don't fail if arena_get returns NULL. (_mid_memalign): Likewise. (__libc_calloc): Likewise. (__libc_realloc): Adjust for additional argument to malloc_printerr. (_int_free): Likewise. (malloc_consolidate): Likewise. (_int_realloc): Likewise. (_int_memalign): Don't touch corrupt arenas. * malloc/tst-malloc-backtrace.c: New test case.
2015-05-19 01:10:37 +00:00
/* No arena available without contention. Wait for the next in line. */
LIBC_PROBE (memory_arena_reuse_wait, 3, &result->mutex, result, avoid_arena);
__libc_lock_lock (result->mutex);
2014-01-02 08:38:18 +00:00
out:
/* Attach the arena to the current thread. */
{
/* Update the arena thread attachment counters. */
mstate replaced_arena = thread_arena;
__libc_lock_lock (free_list_lock);
detach_arena (replaced_arena);
/* We may have picked up an arena on the free list. We need to
preserve the invariant that no arena on the free list has a
positive attached_threads counter (otherwise,
arena_thread_freeres cannot use the counter to determine if the
arena needs to be put on the free list). We unconditionally
remove the selected arena from the free list. The caller of
reused_arena checked the free list and observed it to be empty,
so the list is very short. */
remove_from_free_list (result);
++result->attached_threads;
__libc_lock_unlock (free_list_lock);
}
LIBC_PROBE (memory_arena_reuse, 2, result, avoid_arena);
thread_arena = result;
next_to_use = result->next;
return result;
}
static mstate
arena_get2 (size_t size, mstate avoid_arena)
{
mstate a;
2011-11-09 16:14:39 +00:00
static size_t narenas_limit;
a = get_free_list ();
if (a == NULL)
{
/* Nothing immediately available, so generate a new arena. */
if (narenas_limit == 0)
2014-01-02 08:38:18 +00:00
{
if (mp_.arena_max != 0)
narenas_limit = mp_.arena_max;
else if (narenas > mp_.arena_test)
{
int n = __get_nprocs_sched ();
2014-01-02 08:38:18 +00:00
if (n >= 1)
narenas_limit = NARENAS_FROM_NCORES (n);
else
/* We have no information about the system. Assume two
cores. */
narenas_limit = NARENAS_FROM_NCORES (2);
}
}
2011-11-09 16:14:39 +00:00
repeat:;
size_t n = narenas;
2012-01-31 19:42:34 +00:00
/* NB: the following depends on the fact that (size_t)0 - 1 is a
2014-01-02 08:38:18 +00:00
very large number and that the underflow is OK. If arena_max
is set the value of arena_test is irrelevant. If arena_test
is set but narenas is not yet larger or equal to arena_test
narenas_limit is 0. There is no possibility for narenas to
be too big for the test to always fail since there is not
enough address space to create that many arenas. */
if (__glibc_unlikely (n <= narenas_limit - 1))
2014-01-02 08:38:18 +00:00
{
if (catomic_compare_and_exchange_bool_acq (&narenas, n + 1, n))
goto repeat;
a = _int_new_arena (size);
if (__glibc_unlikely (a == NULL))
2014-01-02 08:38:18 +00:00
catomic_decrement (&narenas);
}
else
2014-01-02 08:38:18 +00:00
a = reused_arena (avoid_arena);
2011-11-09 16:14:39 +00:00
}
return a;
}
/* If we don't have the main arena, then maybe the failure is due to running
out of mmapped areas, so we can try allocating on the main arena.
Otherwise, it is likely that sbrk() has failed and there is still a chance
to mmap(), so try one of the other arenas. */
static mstate
arena_get_retry (mstate ar_ptr, size_t bytes)
{
LIBC_PROBE (memory_arena_retry, 2, bytes, ar_ptr);
2014-01-02 08:38:18 +00:00
if (ar_ptr != &main_arena)
{
__libc_lock_unlock (ar_ptr->mutex);
2014-01-02 08:38:18 +00:00
ar_ptr = &main_arena;
__libc_lock_lock (ar_ptr->mutex);
2014-01-02 08:38:18 +00:00
}
else
{
__libc_lock_unlock (ar_ptr->mutex);
ar_ptr = arena_get2 (bytes, ar_ptr);
2014-01-02 08:38:18 +00:00
}
return ar_ptr;
}
#endif
void
__malloc_arena_thread_freeres (void)
{
/* Shut down the thread cache first. This could deallocate data for
the thread arena, so do this before we put the arena on the free
list. */
tcache_thread_shutdown ();
mstate a = thread_arena;
thread_arena = NULL;
if (a != NULL)
{
__libc_lock_lock (free_list_lock);
/* If this was the last attached thread for this arena, put the
arena on the free list. */
assert (a->attached_threads > 0);
if (--a->attached_threads == 0)
{
a->next_free = free_list;
free_list = a;
}
__libc_lock_unlock (free_list_lock);
}
}
/*
* Local variables:
* c-basic-offset: 2
* End:
*/