From f65fd747b440ae2d8a7481ecc50e668c5e4d0cc9 Mon Sep 17 00:00:00 2001 From: Ulrich Drepper Date: Sun, 8 Dec 1996 08:01:13 +0000 Subject: update from main archive 961207 Sun Dec 8 06:56:49 1996 Ulrich Drepper * io/getwd.c: Use PATH_MAX not LOCAL_PATH_MAX. Fix typo in comment. * stdlib/canonicalize.c: Correct bugs in last change. Patch by HJ Lu. * libio/Makefile (routines): Remove ioprims. (aux): Remove cleanup. Add IO_DEBUG option for .o files. * libio/cleanups.c: Removed. * libio/ioprims.c: Removed. * libio/filedoalloc.c: More updates from libg++-2.8b5. * libio/fileops.c: Likewise. * libio/genops.c: Likewise. * libio/iolibio.h: Likewise. * libio/iopopen.c: Likewise. * libio/iovsprintf.c: Likewise. * libio/iovsscanf.c: Likewise. * libio/libio.h: Likewise. * libio/libioP.h: Likewise. * libio/memstream.c: Likewise. * libio/strfile.h: Likewise. * libio/vasprintf.c: Likewise. * libio/vsnprintf.c: Likewise. * libio/stdio.h: Define P_tmpdir only is __USE_SVID. * manual/arith.texi: Change references to ANSI C to ISO C. * manual/conf.texi: Likewise. * manual/creature.texi: Likewise. * manual/ctype.texi: Likewise. * manual/errno.texi: Likewise. * manual/filesys.texi: Likewise. * manual/intro.texi. Likewise. * manual/io.texi: Likewise. * manual/lang.texi: Likewise. * manual/libc.texinfo: Likewise. * manual/locale.texi: Likewise. * manual/maint.texi: Likewise. * manual/mbyte.texi: Likewise. * manual/memory.texi: Likewise. * manual/process.texi: Likewise. * manual/process.texi: Likewise. * manual/search.texi: Likewise. * manual/setjmp.texi: Likewise. * manual/signal.texi: Likewise. * manual/startup.texi: Likewise. * manual/stdio.texi: Likewise. * manual/string.texi: Likewise. * manual/time.texi: Likewise. * manual/locale.texi: Remove description of LC_RESPONSE and add LC_MESSAGES. * Makefile (subdirs): Change malloc in $(malloc). * config.make.in: Add variable malloc which is initialized from @malloc@. * configure.in: Add new option --enable-new-malloc to use new malloc. This is the default on Linux. * sysdeps/unix/sysv/linux/configure.in: Define malloc to new-malloc by default. * new-malloc/Makefile: New file. Improved malloc implementation. * new-malloc/malloc.c: Likewise. * new-malloc/malloc.h: Likewise. * new-malloc/mallocbug.c: Likewise. * new-malloc/obstack.c: Likewise. * new-malloc/obstack.h: Likewise. * new-malloc/thread-m.h: Likewise. * time/Makefile: Compile ap.c with NO_MCHECK flag for now. * time/ap.c: Don't call mcheck if NO_MCHECK is defined. * resolv/Makefile: Add rule to rebuiild libresolv.so when libc.so changed. * stdio/feof.c: Update copyright. * stdio/stdio.h: Add field for lock to FILE structure. Add cast to *MAGIC constants to prevent warnings. * stdio-common/bug7.c: Correct test. Stream must not be closed twice. * stdlib/Makefile (routines): Add secure-getenv. * stdlib/secure-getenv.c: New file. __secure_getenv function moved to here from sysdeps/generic/getenv.c. Otherwise an application cannot replace the getenv function in the libc. * sysdeps/generic/getenv.c: Remove __secure_getenv function. * sysdeps/stub/getenv.c: Remove __secure_getenv alias. * sysdeps/mach/libc-lock.h: Define__libc_mutex_lock to __mutex_lock. * sysdeps/posix/fdopen.c: Update copyright. Don't use EXFUN. * time/test-tz.c: Comment fifth test out. PROBLEM. * time/tzset.c: De-ANSI-declfy. (__tzset): Don't increment pointer tz when no DST information is given. Sat Dec 7 23:47:54 1996 Ulrich Drepper * sysdeps/mach/libc-lock.h [_LIBC]: Add definition of __libc_mutex_lock. Patch by Thomas Bushnell. * sysdeps/unix/sysv/linux/timebits.h: Load only if __USE_MISC. * sysdeps/unix/sysv/linux/Dist: Add llseek.c. Sat Dec 7 12:18:56 1996 Ulrich Drepper * time/strftime (%c format): Remove %Z from default string. Reported by Paul Eggert * io/getwd.c: Don't apply getcwd on user supplied buffer. --- malloc/malloc.c | 3443 +++++++++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 3443 insertions(+) create mode 100644 malloc/malloc.c (limited to 'malloc/malloc.c') diff --git a/malloc/malloc.c b/malloc/malloc.c new file mode 100644 index 0000000000..ed24d5d76d --- /dev/null +++ b/malloc/malloc.c @@ -0,0 +1,3443 @@ +/* Malloc implementation for multiple threads without lock contention. + Copyright (C) 1996 Free Software Foundation, Inc. + This file is part of the GNU C Library. + Contributed by Wolfram Gloger , 1996. + + The GNU C Library is free software; you can redistribute it and/or + modify it under the terms of the GNU Library General Public License as + published by the Free Software Foundation; either version 2 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 + Library General Public License for more details. + + You should have received a copy of the GNU Library General Public + License along with the GNU C Library; see the file COPYING.LIB. If not, + write to the Free Software Foundation, Inc., 59 Temple Place - Suite 330, + Boston, MA 02111-1307, USA. */ + +/* VERSION 2.6.4-pt Wed Dec 4 00:35:54 MET 1996 + + This work is mainly derived from malloc-2.6.4 by Doug Lea + , which is available from: + + ftp://g.oswego.edu/pub/misc/malloc.c + + Most of the original comments are reproduced in the code below. + +* Why use this malloc? + + This is not the fastest, most space-conserving, most portable, or + most tunable malloc ever written. However it is among the fastest + while also being among the most space-conserving, portable and tunable. + Consistent balance across these factors results in a good general-purpose + allocator. For a high-level description, see + http://g.oswego.edu/dl/html/malloc.html + + On many systems, the standard malloc implementation is by itself not + thread-safe, and therefore wrapped with a single global lock around + all malloc-related functions. In some applications, especially with + multiple available processors, this can lead to contention problems + and bad performance. This malloc version was designed with the goal + to avoid waiting for locks as much as possible. Statistics indicate + that this goal is achieved in many cases. + +* Synopsis of public routines + + (Much fuller descriptions are contained in the program documentation below.) + + ptmalloc_init(); + Initialize global configuration. When compiled for multiple threads, + this function must be called once before any other function in the + package. It is not required otherwise. It is called automatically + in the Linux/GNU C libray. + malloc(size_t n); + Return a pointer to a newly allocated chunk of at least n bytes, or null + if no space is available. + free(Void_t* p); + Release the chunk of memory pointed to by p, or no effect if p is null. + realloc(Void_t* p, size_t n); + Return a pointer to a chunk of size n that contains the same data + as does chunk p up to the minimum of (n, p's size) bytes, or null + if no space is available. The returned pointer may or may not be + the same as p. If p is null, equivalent to malloc. Unless the + #define REALLOC_ZERO_BYTES_FREES below is set, realloc with a + size argument of zero (re)allocates a minimum-sized chunk. + memalign(size_t alignment, size_t n); + Return a pointer to a newly allocated chunk of n bytes, aligned + in accord with the alignment argument, which must be a power of + two. + valloc(size_t n); + Equivalent to memalign(pagesize, n), where pagesize is the page + size of the system (or as near to this as can be figured out from + all the includes/defines below.) + pvalloc(size_t n); + Equivalent to valloc(minimum-page-that-holds(n)), that is, + round up n to nearest pagesize. + calloc(size_t unit, size_t quantity); + Returns a pointer to quantity * unit bytes, with all locations + set to zero. + cfree(Void_t* p); + Equivalent to free(p). + malloc_trim(size_t pad); + Release all but pad bytes of freed top-most memory back + to the system. Return 1 if successful, else 0. + malloc_usable_size(Void_t* p); + Report the number usable allocated bytes associated with allocated + chunk p. This may or may not report more bytes than were requested, + due to alignment and minimum size constraints. + malloc_stats(); + Prints brief summary statistics on stderr. + mallinfo() + Returns (by copy) a struct containing various summary statistics. + mallopt(int parameter_number, int parameter_value) + Changes one of the tunable parameters described below. Returns + 1 if successful in changing the parameter, else 0. + +* Vital statistics: + + Alignment: 8-byte + 8 byte alignment is currently hardwired into the design. This + seems to suffice for all current machines and C compilers. + + Assumed pointer representation: 4 or 8 bytes + Code for 8-byte pointers is untested by me but has worked + reliably by Wolfram Gloger, who contributed most of the + changes supporting this. + + Assumed size_t representation: 4 or 8 bytes + Note that size_t is allowed to be 4 bytes even if pointers are 8. + + Minimum overhead per allocated chunk: 4 or 8 bytes + Each malloced chunk has a hidden overhead of 4 bytes holding size + and status information. + + Minimum allocated size: 4-byte ptrs: 16 bytes (including 4 overhead) + 8-byte ptrs: 24/32 bytes (including, 4/8 overhead) + + When a chunk is freed, 12 (for 4byte ptrs) or 20 (for 8 byte + ptrs but 4 byte size) or 24 (for 8/8) additional bytes are + needed; 4 (8) for a trailing size field + and 8 (16) bytes for free list pointers. Thus, the minimum + allocatable size is 16/24/32 bytes. + + Even a request for zero bytes (i.e., malloc(0)) returns a + pointer to something of the minimum allocatable size. + + Maximum allocated size: 4-byte size_t: 2^31 - 8 bytes + 8-byte size_t: 2^63 - 16 bytes + + It is assumed that (possibly signed) size_t bit values suffice to + represent chunk sizes. `Possibly signed' is due to the fact + that `size_t' may be defined on a system as either a signed or + an unsigned type. To be conservative, values that would appear + as negative numbers are avoided. + Requests for sizes with a negative sign bit will return a + minimum-sized chunk. + + Maximum overhead wastage per allocated chunk: normally 15 bytes + + Alignnment demands, plus the minimum allocatable size restriction + make the normal worst-case wastage 15 bytes (i.e., up to 15 + more bytes will be allocated than were requested in malloc), with + two exceptions: + 1. Because requests for zero bytes allocate non-zero space, + the worst case wastage for a request of zero bytes is 24 bytes. + 2. For requests >= mmap_threshold that are serviced via + mmap(), the worst case wastage is 8 bytes plus the remainder + from a system page (the minimal mmap unit); typically 4096 bytes. + +* Limitations + + Here are some features that are NOT currently supported + + * No user-definable hooks for callbacks and the like. + * No automated mechanism for fully checking that all accesses + to malloced memory stay within their bounds. + * No support for compaction. + +* Synopsis of compile-time options: + + People have reported using previous versions of this malloc on all + versions of Unix, sometimes by tweaking some of the defines + below. It has been tested most extensively on Solaris and + Linux. People have also reported adapting this malloc for use in + stand-alone embedded systems. + + The implementation is in straight, hand-tuned ANSI C. Among other + consequences, it uses a lot of macros. Because of this, to be at + all usable, this code should be compiled using an optimizing compiler + (for example gcc -O2) that can simplify expressions and control + paths. + + __STD_C (default: derived from C compiler defines) + Nonzero if using ANSI-standard C compiler, a C++ compiler, or + a C compiler sufficiently close to ANSI to get away with it. + MALLOC_DEBUG (default: NOT defined) + Define to enable debugging. Adds fairly extensive assertion-based + checking to help track down memory errors, but noticeably slows down + execution. + REALLOC_ZERO_BYTES_FREES (default: NOT defined) + Define this if you think that realloc(p, 0) should be equivalent + to free(p). Otherwise, since malloc returns a unique pointer for + malloc(0), so does realloc(p, 0). + HAVE_MEMCPY (default: defined) + Define if you are not otherwise using ANSI STD C, but still + have memcpy and memset in your C library and want to use them. + Otherwise, simple internal versions are supplied. + USE_MEMCPY (default: 1 if HAVE_MEMCPY is defined, 0 otherwise) + Define as 1 if you want the C library versions of memset and + memcpy called in realloc and calloc (otherwise macro versions are used). + At least on some platforms, the simple macro versions usually + outperform libc versions. + HAVE_MMAP (default: defined as 1) + Define to non-zero to optionally make malloc() use mmap() to + allocate very large blocks. + HAVE_MREMAP (default: defined as 0 unless Linux libc set) + Define to non-zero to optionally make realloc() use mremap() to + reallocate very large blocks. + malloc_getpagesize (default: derived from system #includes) + Either a constant or routine call returning the system page size. + HAVE_USR_INCLUDE_MALLOC_H (default: NOT defined) + Optionally define if you are on a system with a /usr/include/malloc.h + that declares struct mallinfo. It is not at all necessary to + define this even if you do, but will ensure consistency. + INTERNAL_SIZE_T (default: size_t) + Define to a 32-bit type (probably `unsigned int') if you are on a + 64-bit machine, yet do not want or need to allow malloc requests of + greater than 2^31 to be handled. This saves space, especially for + very small chunks. + _LIBC (default: NOT defined) + Defined only when compiled as part of the Linux libc/glibc. + Also note that there is some odd internal name-mangling via defines + (for example, internally, `malloc' is named `mALLOc') needed + when compiling in this case. These look funny but don't otherwise + affect anything. + LACKS_UNISTD_H (default: undefined) + Define this if your system does not have a . + MORECORE (default: sbrk) + The name of the routine to call to obtain more memory from the system. + MORECORE_FAILURE (default: -1) + The value returned upon failure of MORECORE. + MORECORE_CLEARS (default 1) + True (1) if the routine mapped to MORECORE zeroes out memory (which + holds for sbrk). + DEFAULT_TRIM_THRESHOLD + DEFAULT_TOP_PAD + DEFAULT_MMAP_THRESHOLD + DEFAULT_MMAP_MAX + Default values of tunable parameters (described in detail below) + controlling interaction with host system routines (sbrk, mmap, etc). + These values may also be changed dynamically via mallopt(). The + preset defaults are those that give best performance for typical + programs/systems. + + +*/ + +/* + +* Compile-time options for multiple threads: + + USE_PTHREADS, USE_THR, USE_SPROC + Define one of these as 1 to select the thread interface: + POSIX threads, Solaris threads or SGI sproc's, respectively. + If none of these is defined as non-zero, you get a `normal' + malloc implementation which is not thread-safe. Support for + multiple threads requires HAVE_MMAP=1. As an exception, when + compiling for GNU libc, i.e. when _LIBC is defined, then none of + the USE_... symbols have to be defined. + + HEAP_MIN_SIZE + HEAP_MAX_SIZE + When thread support is enabled, additional `heap's are created + with mmap calls. These are limited in size; HEAP_MIN_SIZE should + be a multiple of the page size, while HEAP_MAX_SIZE must be a power + of two for alignment reasons. HEAP_MAX_SIZE should be at least + twice as large as the mmap threshold. + THREAD_STATS + When this is defined as non-zero, some statistics on mutex locking + are computed. + +*/ + + + + +/* Macros for handling mutexes and thread-specific data. This is + included first, because some thread-related header files (such as + pthread.h) should be included before any others. */ +#include "thread-m.h" + + +/* Preliminaries */ + +#ifndef __STD_C +#if defined (__STDC__) +#define __STD_C 1 +#else +#if __cplusplus +#define __STD_C 1 +#else +#define __STD_C 0 +#endif /*__cplusplus*/ +#endif /*__STDC__*/ +#endif /*__STD_C*/ + +#ifndef Void_t +#if __STD_C +#define Void_t void +#else +#define Void_t char +#endif +#endif /*Void_t*/ + +#if __STD_C +#include /* for size_t */ +#else +#include +#endif + +#ifdef __cplusplus +extern "C" { +#endif + +#include /* needed for malloc_stats */ + + +/* + Compile-time options +*/ + + +/* + Debugging: + + Because freed chunks may be overwritten with link fields, this + malloc will often die when freed memory is overwritten by user + programs. This can be very effective (albeit in an annoying way) + in helping track down dangling pointers. + + If you compile with -DMALLOC_DEBUG, a number of assertion checks are + enabled that will catch more memory errors. You probably won't be + able to make much sense of the actual assertion errors, but they + should help you locate incorrectly overwritten memory. The + checking is fairly extensive, and will slow down execution + noticeably. Calling malloc_stats or mallinfo with MALLOC_DEBUG set will + attempt to check every non-mmapped allocated and free chunk in the + course of computing the summmaries. (By nature, mmapped regions + cannot be checked very much automatically.) + + Setting MALLOC_DEBUG may also be helpful if you are trying to modify + this code. The assertions in the check routines spell out in more + detail the assumptions and invariants underlying the algorithms. + +*/ + +#if MALLOC_DEBUG +#include +#else +#define assert(x) ((void)0) +#endif + + +/* + INTERNAL_SIZE_T is the word-size used for internal bookkeeping + of chunk sizes. On a 64-bit machine, you can reduce malloc + overhead by defining INTERNAL_SIZE_T to be a 32 bit `unsigned int' + at the expense of not being able to handle requests greater than + 2^31. This limitation is hardly ever a concern; you are encouraged + to set this. However, the default version is the same as size_t. +*/ + +#ifndef INTERNAL_SIZE_T +#define INTERNAL_SIZE_T size_t +#endif + +/* + REALLOC_ZERO_BYTES_FREES should be set if a call to + realloc with zero bytes should be the same as a call to free. + Some people think it should. Otherwise, since this malloc + returns a unique pointer for malloc(0), so does realloc(p, 0). +*/ + + +/* #define REALLOC_ZERO_BYTES_FREES */ + + +/* + HAVE_MEMCPY should be defined if you are not otherwise using + ANSI STD C, but still have memcpy and memset in your C library + and want to use them in calloc and realloc. Otherwise simple + macro versions are defined here. + + USE_MEMCPY should be defined as 1 if you actually want to + have memset and memcpy called. People report that the macro + versions are often enough faster than libc versions on many + systems that it is better to use them. + +*/ + +#define HAVE_MEMCPY + +#ifndef USE_MEMCPY +#ifdef HAVE_MEMCPY +#define USE_MEMCPY 1 +#else +#define USE_MEMCPY 0 +#endif +#endif + +#if (__STD_C || defined(HAVE_MEMCPY)) + +#if __STD_C +void* memset(void*, int, size_t); +void* memcpy(void*, const void*, size_t); +#else +Void_t* memset(); +Void_t* memcpy(); +#endif +#endif + +#if USE_MEMCPY + +/* The following macros are only invoked with (2n+1)-multiples of + INTERNAL_SIZE_T units, with a positive integer n. This is exploited + for fast inline execution when n is small. */ + +#define MALLOC_ZERO(charp, nbytes) \ +do { \ + INTERNAL_SIZE_T mzsz = (nbytes); \ + if(mzsz <= 9*sizeof(mzsz)) { \ + INTERNAL_SIZE_T* mz = (INTERNAL_SIZE_T*) (charp); \ + if(mzsz >= 5*sizeof(mzsz)) { *mz++ = 0; \ + *mz++ = 0; \ + if(mzsz >= 7*sizeof(mzsz)) { *mz++ = 0; \ + *mz++ = 0; \ + if(mzsz >= 9*sizeof(mzsz)) { *mz++ = 0; \ + *mz++ = 0; }}} \ + *mz++ = 0; \ + *mz++ = 0; \ + *mz = 0; \ + } else memset((charp), 0, mzsz); \ +} while(0) + +#define MALLOC_COPY(dest,src,nbytes) \ +do { \ + INTERNAL_SIZE_T mcsz = (nbytes); \ + if(mcsz <= 9*sizeof(mcsz)) { \ + INTERNAL_SIZE_T* mcsrc = (INTERNAL_SIZE_T*) (src); \ + INTERNAL_SIZE_T* mcdst = (INTERNAL_SIZE_T*) (dest); \ + if(mcsz >= 5*sizeof(mcsz)) { *mcdst++ = *mcsrc++; \ + *mcdst++ = *mcsrc++; \ + if(mcsz >= 7*sizeof(mcsz)) { *mcdst++ = *mcsrc++; \ + *mcdst++ = *mcsrc++; \ + if(mcsz >= 9*sizeof(mcsz)) { *mcdst++ = *mcsrc++; \ + *mcdst++ = *mcsrc++; }}} \ + *mcdst++ = *mcsrc++; \ + *mcdst++ = *mcsrc++; \ + *mcdst = *mcsrc ; \ + } else memcpy(dest, src, mcsz); \ +} while(0) + +#else /* !USE_MEMCPY */ + +/* Use Duff's device for good zeroing/copying performance. */ + +#define MALLOC_ZERO(charp, nbytes) \ +do { \ + INTERNAL_SIZE_T* mzp = (INTERNAL_SIZE_T*)(charp); \ + long mctmp = (nbytes)/sizeof(INTERNAL_SIZE_T), mcn; \ + if (mctmp < 8) mcn = 0; else { mcn = (mctmp-1)/8; mctmp %= 8; } \ + switch (mctmp) { \ + case 0: for(;;) { *mzp++ = 0; \ + case 7: *mzp++ = 0; \ + case 6: *mzp++ = 0; \ + case 5: *mzp++ = 0; \ + case 4: *mzp++ = 0; \ + case 3: *mzp++ = 0; \ + case 2: *mzp++ = 0; \ + case 1: *mzp++ = 0; if(mcn <= 0) break; mcn--; } \ + } \ +} while(0) + +#define MALLOC_COPY(dest,src,nbytes) \ +do { \ + INTERNAL_SIZE_T* mcsrc = (INTERNAL_SIZE_T*) src; \ + INTERNAL_SIZE_T* mcdst = (INTERNAL_SIZE_T*) dest; \ + long mctmp = (nbytes)/sizeof(INTERNAL_SIZE_T), mcn; \ + if (mctmp < 8) mcn = 0; else { mcn = (mctmp-1)/8; mctmp %= 8; } \ + switch (mctmp) { \ + case 0: for(;;) { *mcdst++ = *mcsrc++; \ + case 7: *mcdst++ = *mcsrc++; \ + case 6: *mcdst++ = *mcsrc++; \ + case 5: *mcdst++ = *mcsrc++; \ + case 4: *mcdst++ = *mcsrc++; \ + case 3: *mcdst++ = *mcsrc++; \ + case 2: *mcdst++ = *mcsrc++; \ + case 1: *mcdst++ = *mcsrc++; if(mcn <= 0) break; mcn--; } \ + } \ +} while(0) + +#endif + + +/* + Define HAVE_MMAP to optionally make malloc() use mmap() to + allocate very large blocks. These will be returned to the + operating system immediately after a free(). +*/ + +#ifndef HAVE_MMAP +#define HAVE_MMAP 1 +#endif + +/* + Define HAVE_MREMAP to make realloc() use mremap() to re-allocate + large blocks. This is currently only possible on Linux with + kernel versions newer than 1.3.77. +*/ + +#ifndef HAVE_MREMAP +#define HAVE_MREMAP defined(__linux__) +#endif + +#if HAVE_MMAP + +#include +#include +#include + +#if !defined(MAP_ANONYMOUS) && defined(MAP_ANON) +#define MAP_ANONYMOUS MAP_ANON +#endif + +#endif /* HAVE_MMAP */ + +/* + Access to system page size. To the extent possible, this malloc + manages memory from the system in page-size units. + + The following mechanics for getpagesize were adapted from + bsd/gnu getpagesize.h +*/ + +#ifndef LACKS_UNISTD_H +# include +#endif + +#ifndef malloc_getpagesize +# ifdef _SC_PAGESIZE /* some SVR4 systems omit an underscore */ +# ifndef _SC_PAGE_SIZE +# define _SC_PAGE_SIZE _SC_PAGESIZE +# endif +# endif +# ifdef _SC_PAGE_SIZE +# define malloc_getpagesize sysconf(_SC_PAGE_SIZE) +# else +# if defined(BSD) || defined(DGUX) || defined(HAVE_GETPAGESIZE) + extern size_t getpagesize(); +# define malloc_getpagesize getpagesize() +# else +# include +# ifdef EXEC_PAGESIZE +# define malloc_getpagesize EXEC_PAGESIZE +# else +# ifdef NBPG +# ifndef CLSIZE +# define malloc_getpagesize NBPG +# else +# define malloc_getpagesize (NBPG * CLSIZE) +# endif +# else +# ifdef NBPC +# define malloc_getpagesize NBPC +# else +# ifdef PAGESIZE +# define malloc_getpagesize PAGESIZE +# else +# define malloc_getpagesize (4096) /* just guess */ +# endif +# endif +# endif +# endif +# endif +# endif +#endif + + + +/* + + This version of malloc supports the standard SVID/XPG mallinfo + routine that returns a struct containing the same kind of + information you can get from malloc_stats. It should work on + any SVID/XPG compliant system that has a /usr/include/malloc.h + defining struct mallinfo. (If you'd like to install such a thing + yourself, cut out the preliminary declarations as described above + and below and save them in a malloc.h file. But there's no + compelling reason to bother to do this.) + + The main declaration needed is the mallinfo struct that is returned + (by-copy) by mallinfo(). The SVID/XPG malloinfo struct contains a + bunch of fields, most of which are not even meaningful in this + version of malloc. Some of these fields are are instead filled by + mallinfo() with other numbers that might possibly be of interest. + + HAVE_USR_INCLUDE_MALLOC_H should be set if you have a + /usr/include/malloc.h file that includes a declaration of struct + mallinfo. If so, it is included; else an SVID2/XPG2 compliant + version is declared below. These must be precisely the same for + mallinfo() to work. + +*/ + +/* #define HAVE_USR_INCLUDE_MALLOC_H */ + +#if HAVE_USR_INCLUDE_MALLOC_H +#include "/usr/include/malloc.h" +#else +#include "malloc.h" +#endif + + + +#ifndef DEFAULT_TRIM_THRESHOLD +#define DEFAULT_TRIM_THRESHOLD (128 * 1024) +#endif + +/* + M_TRIM_THRESHOLD is the maximum amount of unused top-most memory + to keep before releasing via malloc_trim in free(). + + Automatic trimming is mainly useful in long-lived programs. + Because trimming via sbrk can be slow on some systems, and can + sometimes be wasteful (in cases where programs immediately + afterward allocate more large chunks) the value should be high + enough so that your overall system performance would improve by + releasing. + + The trim threshold and the mmap control parameters (see below) + can be traded off with one another. Trimming and mmapping are + two different ways of releasing unused memory back to the + system. Between these two, it is often possible to keep + system-level demands of a long-lived program down to a bare + minimum. For example, in one test suite of sessions measuring + the XF86 X server on Linux, using a trim threshold of 128K and a + mmap threshold of 192K led to near-minimal long term resource + consumption. + + If you are using this malloc in a long-lived program, it should + pay to experiment with these values. As a rough guide, you + might set to a value close to the average size of a process + (program) running on your system. Releasing this much memory + would allow such a process to run in memory. Generally, it's + worth it to tune for trimming rather tham memory mapping when a + program undergoes phases where several large chunks are + allocated and released in ways that can reuse each other's + storage, perhaps mixed with phases where there are no such + chunks at all. And in well-behaved long-lived programs, + controlling release of large blocks via trimming versus mapping + is usually faster. + + However, in most programs, these parameters serve mainly as + protection against the system-level effects of carrying around + massive amounts of unneeded memory. Since frequent calls to + sbrk, mmap, and munmap otherwise degrade performance, the default + parameters are set to relatively high values that serve only as + safeguards. + + The default trim value is high enough to cause trimming only in + fairly extreme (by current memory consumption standards) cases. + It must be greater than page size to have any useful effect. To + disable trimming completely, you can set to (unsigned long)(-1); + + +*/ + + +#ifndef DEFAULT_TOP_PAD +#define DEFAULT_TOP_PAD (0) +#endif + +/* + M_TOP_PAD is the amount of extra `padding' space to allocate or + retain whenever sbrk is called. It is used in two ways internally: + + * When sbrk is called to extend the top of the arena to satisfy + a new malloc request, this much padding is added to the sbrk + request. + + * When malloc_trim is called automatically from free(), + it is used as the `pad' argument. + + In both cases, the actual amount of padding is rounded + so that the end of the arena is always a system page boundary. + + The main reason for using padding is to avoid calling sbrk so + often. Having even a small pad greatly reduces the likelihood + that nearly every malloc request during program start-up (or + after trimming) will invoke sbrk, which needlessly wastes + time. + + Automatic rounding-up to page-size units is normally sufficient + to avoid measurable overhead, so the default is 0. However, in + systems where sbrk is relatively slow, it can pay to increase + this value, at the expense of carrying around more memory than + the program needs. + +*/ + + +#ifndef DEFAULT_MMAP_THRESHOLD +#define DEFAULT_MMAP_THRESHOLD (128 * 1024) +#endif + +/* + + M_MMAP_THRESHOLD is the request size threshold for using mmap() + to service a request. Requests of at least this size that cannot + be allocated using already-existing space will be serviced via mmap. + (If enough normal freed space already exists it is used instead.) + + Using mmap segregates relatively large chunks of memory so that + they can be individually obtained and released from the host + system. A request serviced through mmap is never reused by any + other request (at least not directly; the system may just so + happen to remap successive requests to the same locations). + + Segregating space in this way has the benefit that mmapped space + can ALWAYS be individually released back to the system, which + helps keep the system level memory demands of a long-lived + program low. Mapped memory can never become `locked' between + other chunks, as can happen with normally allocated chunks, which + menas that even trimming via malloc_trim would not release them. + + However, it has the disadvantages that: + + 1. The space cannot be reclaimed, consolidated, and then + used to service later requests, as happens with normal chunks. + 2. It can lead to more wastage because of mmap page alignment + requirements + 3. It causes malloc performance to be more dependent on host + system memory management support routines which may vary in + implementation quality and may impose arbitrary + limitations. Generally, servicing a request via normal + malloc steps is faster than going through a system's mmap. + + All together, these considerations should lead you to use mmap + only for relatively large requests. + + +*/ + + + +#ifndef DEFAULT_MMAP_MAX +#if HAVE_MMAP +#define DEFAULT_MMAP_MAX (1024) +#else +#define DEFAULT_MMAP_MAX (0) +#endif +#endif + +/* + M_MMAP_MAX is the maximum number of requests to simultaneously + service using mmap. This parameter exists because: + + 1. Some systems have a limited number of internal tables for + use by mmap. + 2. In most systems, overreliance on mmap can degrade overall + performance. + 3. If a program allocates many large regions, it is probably + better off using normal sbrk-based allocation routines that + can reclaim and reallocate normal heap memory. Using a + small value allows transition into this mode after the + first few allocations. + + Setting to 0 disables all use of mmap. If HAVE_MMAP is not set, + the default value is 0, and attempts to set it to non-zero values + in mallopt will fail. +*/ + + + +#define HEAP_MIN_SIZE (32*1024) +#define HEAP_MAX_SIZE (1024*1024) /* must be a power of two */ + +/* 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. +*/ + + + +#ifndef THREAD_STATS +#define THREAD_STATS 0 +#endif + +/* If THREAD_STATS is non-zero, some statistics on mutex locking are + computed. */ + + +/* + + Special defines for the Linux/GNU C library. + +*/ + + +#ifdef _LIBC + +#if __STD_C + +Void_t * __default_morecore (ptrdiff_t); +static Void_t *(*__morecore)(ptrdiff_t) = __default_morecore; + +#else + +Void_t * __default_morecore (); +static Void_t *(*__morecore)() = __default_morecore; + +#endif + +#define MORECORE (*__morecore) +#define MORECORE_FAILURE 0 +#define MORECORE_CLEARS 1 + +#else /* _LIBC */ + +#if __STD_C +extern Void_t* sbrk(ptrdiff_t); +#else +extern Void_t* sbrk(); +#endif + +#ifndef MORECORE +#define MORECORE sbrk +#endif + +#ifndef MORECORE_FAILURE +#define MORECORE_FAILURE -1 +#endif + +#ifndef MORECORE_CLEARS +#define MORECORE_CLEARS 1 +#endif + +#endif /* _LIBC */ + +#if 0 && defined(_LIBC) + +#define cALLOc __libc_calloc +#define fREe __libc_free +#define mALLOc __libc_malloc +#define mEMALIGn __libc_memalign +#define rEALLOc __libc_realloc +#define vALLOc __libc_valloc +#define pvALLOc __libc_pvalloc +#define mALLINFo __libc_mallinfo +#define mALLOPt __libc_mallopt + +#pragma weak calloc = __libc_calloc +#pragma weak free = __libc_free +#pragma weak cfree = __libc_free +#pragma weak malloc = __libc_malloc +#pragma weak memalign = __libc_memalign +#pragma weak realloc = __libc_realloc +#pragma weak valloc = __libc_valloc +#pragma weak pvalloc = __libc_pvalloc +#pragma weak mallinfo = __libc_mallinfo +#pragma weak mallopt = __libc_mallopt + +#else + +#define cALLOc calloc +#define fREe free +#define mALLOc malloc +#define mEMALIGn memalign +#define rEALLOc realloc +#define vALLOc valloc +#define pvALLOc pvalloc +#define mALLINFo mallinfo +#define mALLOPt mallopt + +#endif + +/* Public routines */ + +#if __STD_C + +#ifndef _LIBC +void ptmalloc_init(void); +#endif +Void_t* mALLOc(size_t); +void fREe(Void_t*); +Void_t* rEALLOc(Void_t*, size_t); +Void_t* mEMALIGn(size_t, size_t); +Void_t* vALLOc(size_t); +Void_t* pvALLOc(size_t); +Void_t* cALLOc(size_t, size_t); +void cfree(Void_t*); +int malloc_trim(size_t); +size_t malloc_usable_size(Void_t*); +void malloc_stats(void); +int mALLOPt(int, int); +struct mallinfo mALLINFo(void); +#else +#ifndef _LIBC +void ptmalloc_init(); +#endif +Void_t* mALLOc(); +void fREe(); +Void_t* rEALLOc(); +Void_t* mEMALIGn(); +Void_t* vALLOc(); +Void_t* pvALLOc(); +Void_t* cALLOc(); +void cfree(); +int malloc_trim(); +size_t malloc_usable_size(); +void malloc_stats(); +int mALLOPt(); +struct mallinfo mALLINFo(); +#endif + + +#ifdef __cplusplus +}; /* end of extern "C" */ +#endif + +#if !defined(NO_THREADS) && !HAVE_MMAP +"Can't have threads support without mmap" +#endif + + +/* + Type declarations +*/ + + +struct malloc_chunk +{ + INTERNAL_SIZE_T prev_size; /* Size of previous chunk (if free). */ + INTERNAL_SIZE_T size; /* Size in bytes, including overhead. */ + struct malloc_chunk* fd; /* double links -- used only if free. */ + struct malloc_chunk* bk; +}; + +typedef struct malloc_chunk* mchunkptr; + +/* + + malloc_chunk details: + + (The following includes lightly edited explanations by Colin Plumb.) + + Chunks of memory are maintained using a `boundary tag' method as + described in e.g., Knuth or Standish. (See the paper by Paul + Wilson ftp://ftp.cs.utexas.edu/pub/garbage/allocsrv.ps for a + survey of such techniques.) Sizes of free chunks are stored both + in the front of each chunk and at the end. This makes + consolidating fragmented chunks into bigger chunks very fast. The + size fields also hold bits representing whether chunks are free or + in use. + + An allocated chunk looks like this: + + + chunk-> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ + | Size of previous chunk, if allocated | | + +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ + | Size of chunk, in bytes |P| + mem-> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ + | User data starts here... . + . . + . (malloc_usable_space() bytes) . + . | +nextchunk-> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ + | Size of chunk | + +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ + + + Where "chunk" is the front of the chunk for the purpose of most of + the malloc code, but "mem" is the pointer that is returned to the + user. "Nextchunk" is the beginning of the next contiguous chunk. + + Chunks always begin on even word boundries, so the mem portion + (which is returned to the user) is also on an even word boundary, and + thus double-word aligned. + + Free chunks are stored in circular doubly-linked lists, and look like this: + + chunk-> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ + | Size of previous chunk | + +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ + `head:' | Size of chunk, in bytes |P| + mem-> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ + | Forward pointer to next chunk in list | + +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ + | Back pointer to previous chunk in list | + +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ + | Unused space (may be 0 bytes long) . + . . + . | +nextchunk-> +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ + `foot:' | Size of chunk, in bytes | + +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ + + The P (PREV_INUSE) bit, stored in the unused low-order bit of the + chunk size (which is always a multiple of two words), is an in-use + bit for the *previous* chunk. If that bit is *clear*, then the + word before the current chunk size contains the previous chunk + size, and can be used to find the front of the previous chunk. + (The very first chunk allocated always has this bit set, + preventing access to non-existent (or non-owned) memory.) + + Note that the `foot' of the current chunk is actually represented + as the prev_size of the NEXT chunk. (This makes it easier to + deal with alignments etc). + + The two exceptions to all this are + + 1. The special chunk `top', which doesn't bother using the + trailing size field since there is no + next contiguous chunk that would have to index off it. (After + initialization, `top' is forced to always exist. If it would + become less than MINSIZE bytes long, it is replenished via + malloc_extend_top.) + + 2. Chunks allocated via mmap, which have the second-lowest-order + bit (IS_MMAPPED) set in their size fields. Because they are + never merged or traversed from any other chunk, they have no + foot size or inuse information. + + Available chunks are kept in any of several places (all declared below): + + * `av': An array of chunks serving as bin headers for consolidated + chunks. Each bin is doubly linked. The bins are approximately + proportionally (log) spaced. There are a lot of these bins + (128). This may look excessive, but works very well in + practice. All procedures maintain the invariant that no + consolidated chunk physically borders another one. Chunks in + bins are kept in size order, with ties going to the + approximately least recently used chunk. + + The chunks in each bin are maintained in decreasing sorted order by + size. This is irrelevant for the small bins, which all contain + the same-sized chunks, but facilitates best-fit allocation for + larger chunks. (These lists are just sequential. Keeping them in + order almost never requires enough traversal to warrant using + fancier ordered data structures.) Chunks of the same size are + linked with the most recently freed at the front, and allocations + are taken from the back. This results in LRU or FIFO allocation + order, which tends to give each chunk an equal opportunity to be + consolidated with adjacent freed chunks, resulting in larger free + chunks and less fragmentation. + + * `top': The top-most available chunk (i.e., the one bordering the + end of available memory) is treated specially. It is never + included in any bin, is used only if no other chunk is + available, and is released back to the system if it is very + large (see M_TRIM_THRESHOLD). + + * `last_remainder': A bin holding only the remainder of the + most recently split (non-top) chunk. This bin is checked + before other non-fitting chunks, so as to provide better + locality for runs of sequentially allocated chunks. + + * Implicitly, through the host system's memory mapping tables. + If supported, requests greater than a threshold are usually + serviced via calls to mmap, and then later released via munmap. + +*/ + +/* + Bins + + The bins are an array of pairs of pointers serving as the + heads of (initially empty) doubly-linked lists of chunks, laid out + in a way so that each pair can be treated as if it were in a + malloc_chunk. (This way, the fd/bk offsets for linking bin heads + and chunks are the same). + + Bins for sizes < 512 bytes contain chunks of all the same size, spaced + 8 bytes apart. Larger bins are approximately logarithmically + spaced. (See the table below.) + + Bin layout: + + 64 bins of size 8 + 32 bins of size 64 + 16 bins of size 512 + 8 bins of size 4096 + 4 bins of size 32768 + 2 bins of size 262144 + 1 bin of size what's left + + There is actually a little bit of slop in the numbers in bin_index + for the sake of speed. This makes no difference elsewhere. + + The special chunks `top' and `last_remainder' get their own bins, + (this is implemented via yet more trickery with the av array), + although `top' is never properly linked to its bin since it is + always handled specially. + +*/ + +#define NAV 128 /* number of bins */ + +typedef struct malloc_chunk* mbinptr; + +/* An arena is a configuration of malloc_chunks together with an array + of bins. With multiple threads, it must be locked via a mutex + before changing its data structures. One or more `heaps' are + associated with each arena, except for the main_arena, which is + associated only with the `main heap', i.e. the conventional free + store obtained with calls to MORECORE() (usually sbrk). The `av' + array is never mentioned directly in the code, but instead used via + bin access macros. */ + +typedef struct _arena { + mbinptr av[2*NAV + 2]; + struct _arena *next; + mutex_t mutex; +} arena; + + +/* A heap is a single contiguous memory region holding (coalescable) + malloc_chunks. It is allocated with mmap() and always starts at an + address aligned to HEAP_MAX_SIZE. Not used unless compiling for + multiple threads. */ + +typedef struct _heap_info { + arena *ar_ptr; + size_t size; +} heap_info; + + +/* + Static functions (forward declarations) +*/ + +#if __STD_C +static void chunk_free(arena *ar_ptr, mchunkptr p); +static mchunkptr chunk_alloc(arena *ar_ptr, INTERNAL_SIZE_T size); +static int arena_trim(arena *ar_ptr, size_t pad); +#else +static void chunk_free(); +static mchunkptr chunk_alloc(); +static int arena_trim(); +#endif + + + +/* sizes, alignments */ + +#define SIZE_SZ (sizeof(INTERNAL_SIZE_T)) +#define MALLOC_ALIGNMENT (SIZE_SZ + SIZE_SZ) +#define MALLOC_ALIGN_MASK (MALLOC_ALIGNMENT - 1) +#define MINSIZE (sizeof(struct malloc_chunk)) + +/* conversion from malloc headers to user pointers, and back */ + +#define chunk2mem(p) ((Void_t*)((char*)(p) + 2*SIZE_SZ)) +#define mem2chunk(mem) ((mchunkptr)((char*)(mem) - 2*SIZE_SZ)) + +/* pad request bytes into a usable size */ + +#define request2size(req) \ + (((long)((req) + (SIZE_SZ + MALLOC_ALIGN_MASK)) < \ + (long)(MINSIZE + MALLOC_ALIGN_MASK)) ? MINSIZE : \ + (((req) + (SIZE_SZ + MALLOC_ALIGN_MASK)) & ~(MALLOC_ALIGN_MASK))) + +/* Check if m has acceptable alignment */ + +#define aligned_OK(m) (((unsigned long)((m)) & (MALLOC_ALIGN_MASK)) == 0) + + + + +/* + Physical chunk operations +*/ + + +/* size field is or'ed with PREV_INUSE when previous adjacent chunk in use */ + +#define PREV_INUSE 0x1 + +/* size field is or'ed with IS_MMAPPED if the chunk was obtained with mmap() */ + +#define IS_MMAPPED 0x2 + +/* Bits to mask off when extracting size */ + +#define SIZE_BITS (PREV_INUSE|IS_MMAPPED) + + +/* Ptr to next physical malloc_chunk. */ + +#define next_chunk(p) ((mchunkptr)( ((char*)(p)) + ((p)->size & ~PREV_INUSE) )) + +/* Ptr to previous physical malloc_chunk */ + +#define prev_chunk(p) ((mchunkptr)( ((char*)(p)) - ((p)->prev_size) )) + + +/* Treat space at ptr + offset as a chunk */ + +#define chunk_at_offset(p, s) ((mchunkptr)(((char*)(p)) + (s))) + + + + +/* + Dealing with use bits +*/ + +/* extract p's inuse bit */ + +#define inuse(p) \ + ((((mchunkptr)(((char*)(p))+((p)->size & ~PREV_INUSE)))->size) & PREV_INUSE) + +/* extract inuse bit of previous chunk */ + +#define prev_inuse(p) ((p)->size & PREV_INUSE) + +/* check for mmap()'ed chunk */ + +#define chunk_is_mmapped(p) ((p)->size & IS_MMAPPED) + +/* set/clear chunk as in use without otherwise disturbing */ + +#define set_inuse(p) \ + ((mchunkptr)(((char*)(p)) + ((p)->size & ~PREV_INUSE)))->size |= PREV_INUSE + +#define clear_inuse(p) \ + ((mchunkptr)(((char*)(p)) + ((p)->size & ~PREV_INUSE)))->size &= ~(PREV_INUSE) + +/* check/set/clear inuse bits in known places */ + +#define inuse_bit_at_offset(p, s)\ + (((mchunkptr)(((char*)(p)) + (s)))->size & PREV_INUSE) + +#define set_inuse_bit_at_offset(p, s)\ + (((mchunkptr)(((char*)(p)) + (s)))->size |= PREV_INUSE) + +#define clear_inuse_bit_at_offset(p, s)\ + (((mchunkptr)(((char*)(p)) + (s)))->size &= ~(PREV_INUSE)) + + + + +/* + Dealing with size fields +*/ + +/* Get size, ignoring use bits */ + +#define chunksize(p) ((p)->size & ~(SIZE_BITS)) + +/* Set size at head, without disturbing its use bit */ + +#define set_head_size(p, s) ((p)->size = (((p)->size & PREV_INUSE) | (s))) + +/* Set size/use ignoring previous bits in header */ + +#define set_head(p, s) ((p)->size = (s)) + +/* Set size at footer (only when chunk is not in use) */ + +#define set_foot(p, s) (((mchunkptr)((char*)(p) + (s)))->prev_size = (s)) + + + + + +/* access macros */ + +#define bin_at(a, i) ((mbinptr)((char*)&(((a)->av)[2*(i) + 2]) - 2*SIZE_SZ)) +#define init_bin(a, i) ((a)->av[2*i+2] = (a)->av[2*i+3] = bin_at((a), i)) +#define next_bin(b) ((mbinptr)((char*)(b) + 2 * sizeof(mbinptr))) +#define prev_bin(b) ((mbinptr)((char*)(b) - 2 * sizeof(mbinptr))) + +/* + The first 2 bins are never indexed. The corresponding av cells are instead + used for bookkeeping. This is not to save space, but to simplify + indexing, maintain locality, and avoid some initialization tests. +*/ + +#define binblocks(a) (bin_at(a,0)->size)/* bitvector of nonempty blocks */ +#define top(a) (bin_at(a,0)->fd) /* The topmost chunk */ +#define last_remainder(a) (bin_at(a,1)) /* remainder from last split */ + +/* + Because top initially points to its own bin with initial + zero size, thus forcing extension on the first malloc request, + we avoid having any special code in malloc to check whether + it even exists yet. But we still need to in malloc_extend_top. +*/ + +#define initial_top(a) ((mchunkptr)bin_at(a, 0)) + + + +/* field-extraction macros */ + +#define first(b) ((b)->fd) +#define last(b) ((b)->bk) + +/* + Indexing into bins +*/ + +#define bin_index(sz) \ +(((((unsigned long)(sz)) >> 9) == 0) ? (((unsigned long)(sz)) >> 3): \ + ((((unsigned long)(sz)) >> 9) <= 4) ? 56 + (((unsigned long)(sz)) >> 6): \ + ((((unsigned long)(sz)) >> 9) <= 20) ? 91 + (((unsigned long)(sz)) >> 9): \ + ((((unsigned long)(sz)) >> 9) <= 84) ? 110 + (((unsigned long)(sz)) >> 12): \ + ((((unsigned long)(sz)) >> 9) <= 340) ? 119 + (((unsigned long)(sz)) >> 15): \ + ((((unsigned long)(sz)) >> 9) <= 1364) ? 124 + (((unsigned long)(sz)) >> 18): \ + 126) +/* + bins for chunks < 512 are all spaced 8 bytes apart, and hold + identically sized chunks. This is exploited in malloc. +*/ + +#define MAX_SMALLBIN 63 +#define MAX_SMALLBIN_SIZE 512 +#define SMALLBIN_WIDTH 8 + +#define smallbin_index(sz) (((unsigned long)(sz)) >> 3) + +/* + Requests are `small' if both the corresponding and the next bin are small +*/ + +#define is_small_request(nb) ((nb) < MAX_SMALLBIN_SIZE - SMALLBIN_WIDTH) + + + +/* + To help compensate for the large number of bins, a one-level index + structure is used for bin-by-bin searching. `binblocks' is a + one-word bitvector recording whether groups of BINBLOCKWIDTH bins + have any (possibly) non-empty bins, so they can be skipped over + all at once during during traversals. The bits are NOT always + cleared as soon as all bins in a block are empty, but instead only + when all are noticed to be empty during traversal in malloc. +*/ + +#define BINBLOCKWIDTH 4 /* bins per block */ + +/* bin<->block macros */ + +#define idx2binblock(ix) ((unsigned)1 << ((ix) / BINBLOCKWIDTH)) +#define mark_binblock(a, ii) (binblocks(a) |= idx2binblock(ii)) +#define clear_binblock(a, ii) (binblocks(a) &= ~(idx2binblock(ii))) + + + + +/* Static bookkeeping data */ + +/* Helper macro to initialize bins */ +#define IAV(i) bin_at(&main_arena, i), bin_at(&main_arena, i) + +static arena main_arena = { + { + 0, 0, + IAV(0), IAV(1), IAV(2), IAV(3), IAV(4), IAV(5), IAV(6), IAV(7), + IAV(8), IAV(9), IAV(10), IAV(11), IAV(12), IAV(13), IAV(14), IAV(15), + IAV(16), IAV(17), IAV(18), IAV(19), IAV(20), IAV(21), IAV(22), IAV(23), + IAV(24), IAV(25), IAV(26), IAV(27), IAV(28), IAV(29), IAV(30), IAV(31), + IAV(32), IAV(33), IAV(34), IAV(35), IAV(36), IAV(37), IAV(38), IAV(39), + IAV(40), IAV(41), IAV(42), IAV(43), IAV(44), IAV(45), IAV(46), IAV(47), + IAV(48), IAV(49), IAV(50), IAV(51), IAV(52), IAV(53), IAV(54), IAV(55), + IAV(56), IAV(57), IAV(58), IAV(59), IAV(60), IAV(61), IAV(62), IAV(63), + IAV(64), IAV(65), IAV(66), IAV(67), IAV(68), IAV(69), IAV(70), IAV(71), + IAV(72), IAV(73), IAV(74), IAV(75), IAV(76), IAV(77), IAV(78), IAV(79), + IAV(80), IAV(81), IAV(82), IAV(83), IAV(84), IAV(85), IAV(86), IAV(87), + IAV(88), IAV(89), IAV(90), IAV(91), IAV(92), IAV(93), IAV(94), IAV(95), + IAV(96), IAV(97), IAV(98), IAV(99), IAV(100), IAV(101), IAV(102), IAV(103), + IAV(104), IAV(105), IAV(106), IAV(107), IAV(108), IAV(109), IAV(110), IAV(111), + IAV(112), IAV(113), IAV(114), IAV(115), IAV(116), IAV(117), IAV(118), IAV(119), + IAV(120), IAV(121), IAV(122), IAV(123), IAV(124), IAV(125), IAV(126), IAV(127) + }, + NULL, /* next */ + MUTEX_INITIALIZER /* mutex */ +}; + +#undef IAV + +/* Thread specific data */ + +static tsd_key_t arena_key; +static mutex_t list_lock = MUTEX_INITIALIZER; + +#if THREAD_STATS +static int stat_n_arenas = 0; +static int stat_n_heaps = 0; +static long stat_lock_direct = 0; +static long stat_lock_loop = 0; +#define THREAD_STAT(x) x +#else +#define THREAD_STAT(x) do ; while(0) +#endif + +/* variables holding tunable values */ + +static unsigned long trim_threshold = DEFAULT_TRIM_THRESHOLD; +static unsigned long top_pad = DEFAULT_TOP_PAD; +static unsigned int n_mmaps_max = DEFAULT_MMAP_MAX; +static unsigned long mmap_threshold = DEFAULT_MMAP_THRESHOLD; + +/* The first value returned from sbrk */ +static char* sbrk_base = (char*)(-1); + +/* The maximum memory obtained from system via sbrk */ +static unsigned long max_sbrked_mem = 0; + +/* The maximum via either sbrk or mmap */ +static unsigned long max_total_mem = 0; + +/* internal working copy of mallinfo */ +static struct mallinfo current_mallinfo = { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }; + +/* The total memory obtained from system via sbrk */ +#define sbrked_mem (current_mallinfo.arena) + +/* Tracking mmaps */ + +static unsigned int n_mmaps = 0; +static unsigned int max_n_mmaps = 0; +static unsigned long mmapped_mem = 0; +static unsigned long max_mmapped_mem = 0; + + + + + +/* Initialization routine. */ +#if defined(_LIBC) +static void ptmalloc_init __MALLOC_P ((void)) __attribute__ ((constructor)); + +static void +ptmalloc_init __MALLOC_P((void)) +#else +void +ptmalloc_init __MALLOC_P((void)) +#endif +{ + static int first = 1; + +#if defined(_LIBC) + /* Initialize the pthread. */ + if (__pthread_initialize != NULL) + __pthread_initialize (); +#endif + + if(first) { + first = 0; + mutex_init(&main_arena.mutex); + mutex_init(&list_lock); + tsd_key_create(&arena_key, NULL); + tsd_setspecific(arena_key, (Void_t *)&main_arena); + } +} + + + + + +/* Routines dealing with mmap(). */ + +#if HAVE_MMAP + +#ifndef MAP_ANONYMOUS + +static int dev_zero_fd = -1; /* Cached file descriptor for /dev/zero. */ + +#define MMAP(size, prot) ((dev_zero_fd < 0) ? \ + (dev_zero_fd = open("/dev/zero", O_RDWR), \ + mmap(0, (size), (prot), MAP_PRIVATE, dev_zero_fd, 0)) : \ + mmap(0, (size), (prot), MAP_PRIVATE, dev_zero_fd, 0)) + +#else + +#define MMAP(size, prot) \ + (mmap(0, (size), (prot), MAP_PRIVATE|MAP_ANONYMOUS, -1, 0)) + +#endif + +#if __STD_C +static mchunkptr mmap_chunk(size_t size) +#else +static mchunkptr mmap_chunk(size) size_t size; +#endif +{ + size_t page_mask = malloc_getpagesize - 1; + mchunkptr p; + + if(n_mmaps >= n_mmaps_max) return 0; /* too many regions */ + + /* For mmapped chunks, the overhead is one SIZE_SZ unit larger, because + * there is no following chunk whose prev_size field could be used. + */ + size = (size + SIZE_SZ + page_mask) & ~page_mask; + + p = (mchunkptr)MMAP(size, PROT_READ|PROT_WRITE); + if(p == (mchunkptr)-1) return 0; + + n_mmaps++; + if (n_mmaps > max_n_mmaps) max_n_mmaps = n_mmaps; + + /* We demand that eight bytes into a page must be 8-byte aligned. */ + assert(aligned_OK(chunk2mem(p))); + + /* The offset to the start of the mmapped region is stored + * in the prev_size field of the chunk; normally it is zero, + * but that can be changed in memalign(). + */ + p->prev_size = 0; + set_head(p, size|IS_MMAPPED); + + mmapped_mem += size; + if ((unsigned long)mmapped_mem > (unsigned long)max_mmapped_mem) + max_mmapped_mem = mmapped_mem; + if ((unsigned long)(mmapped_mem + sbrked_mem) > (unsigned long)max_total_mem) + max_total_mem = mmapped_mem + sbrked_mem; + return p; +} + +#if __STD_C +static void munmap_chunk(mchunkptr p) +#else +static void munmap_chunk(p) mchunkptr p; +#endif +{ + INTERNAL_SIZE_T size = chunksize(p); + int ret; + + assert (chunk_is_mmapped(p)); + assert(! ((char*)p >= sbrk_base && (char*)p < sbrk_base + sbrked_mem)); + assert((n_mmaps > 0)); + assert(((p->prev_size + size) & (malloc_getpagesize-1)) == 0); + + n_mmaps--; + mmapped_mem -= (size + p->prev_size); + + ret = munmap((char *)p - p->prev_size, size + p->prev_size); + + /* munmap returns non-zero on failure */ + assert(ret == 0); +} + +#if HAVE_MREMAP + +#if __STD_C +static mchunkptr mremap_chunk(mchunkptr p, size_t new_size) +#else +static mchunkptr mremap_chunk(p, new_size) mchunkptr p; size_t new_size; +#endif +{ + size_t page_mask = malloc_getpagesize - 1; + INTERNAL_SIZE_T offset = p->prev_size; + INTERNAL_SIZE_T size = chunksize(p); + char *cp; + + assert (chunk_is_mmapped(p)); + assert(! ((char*)p >= sbrk_base && (char*)p < sbrk_base + sbrked_mem)); + assert((n_mmaps > 0)); + assert(((size + offset) & (malloc_getpagesize-1)) == 0); + + /* Note the extra SIZE_SZ overhead as in mmap_chunk(). */ + new_size = (new_size + offset + SIZE_SZ + page_mask) & ~page_mask; + + cp = (char *)mremap((char *)p - offset, size + offset, new_size, + MREMAP_MAYMOVE); + + if (cp == (char *)-1) return 0; + + p = (mchunkptr)(cp + offset); + + assert(aligned_OK(chunk2mem(p))); + + assert((p->prev_size == offset)); + set_head(p, (new_size - offset)|IS_MMAPPED); + + mmapped_mem -= size + offset; + mmapped_mem += new_size; + if ((unsigned long)mmapped_mem > (unsigned long)max_mmapped_mem) + max_mmapped_mem = mmapped_mem; + if ((unsigned long)(mmapped_mem + sbrked_mem) > (unsigned long)max_total_mem) + max_total_mem = mmapped_mem + sbrked_mem; + return p; +} + +#endif /* HAVE_MREMAP */ + +#endif /* HAVE_MMAP */ + + + +/* Managing heaps and arenas (for concurrent threads) */ + +#ifndef NO_THREADS + +/* Create a new heap. size is automatically rounded up to a multiple + of the page size. */ + +static heap_info * +#if __STD_C +new_heap(size_t size) +#else +new_heap(size) size_t size; +#endif +{ + size_t page_mask = malloc_getpagesize - 1; + char *p1, *p2; + unsigned long ul; + heap_info *h; + + if(size < HEAP_MIN_SIZE) + size = HEAP_MIN_SIZE; + size = (size + page_mask) & ~page_mask; + if(size > HEAP_MAX_SIZE) + return 0; + p1 = (char *)MMAP(HEAP_MAX_SIZE<<1, PROT_NONE); + if(p1 == (char *)-1) + return 0; + p2 = (char *)(((unsigned long)p1 + HEAP_MAX_SIZE) & ~(HEAP_MAX_SIZE-1)); + ul = p2 - p1; + munmap(p1, ul); + munmap(p2 + HEAP_MAX_SIZE, HEAP_MAX_SIZE - ul); + if(mprotect(p2, size, PROT_READ|PROT_WRITE) != 0) { + munmap(p2, HEAP_MAX_SIZE); + return 0; + } + h = (heap_info *)p2; + h->size = size; + THREAD_STAT(stat_n_heaps++); + return h; +} + +/* Grow or shrink a heap. size is automatically rounded up to a + multiple of the page size. */ + +static int +#if __STD_C +grow_heap(heap_info *h, long diff) +#else +grow_heap(h, diff) heap_info *h; long diff; +#endif +{ + size_t page_mask = malloc_getpagesize - 1; + long new_size; + + if(diff >= 0) { + diff = (diff + page_mask) & ~page_mask; + new_size = (long)h->size + diff; + if(new_size > HEAP_MAX_SIZE) + return -1; + if(mprotect((char *)h + h->size, diff, PROT_READ|PROT_WRITE) != 0) + return -2; + } else { + new_size = (long)h->size + diff; + if(new_size < 0) + return -1; + if(mprotect((char *)h + new_size, -diff, PROT_NONE) != 0) + return -2; + } + h->size = new_size; + return 0; +} + +/* 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 + over the singly linked list of arenas. If no arena is readily + available, create a new one. */ + +#define arena_get(ptr, size) do { \ + Void_t *vptr = NULL; \ + ptr = (arena *)tsd_getspecific(arena_key, vptr); \ + if(ptr && !mutex_trylock(&ptr->mutex)) { \ + THREAD_STAT(stat_lock_direct++); \ + } else { \ + ptr = arena_get2(ptr, (size)); \ + } \ +} while(0) + +static arena * +#if __STD_C +arena_get2(arena *a_tsd, size_t size) +#else +arena_get2(a_tsd, size) arena *a_tsd; size_t size; +#endif +{ + arena *a; + heap_info *h; + char *ptr; + int i; + unsigned long misalign; + + /* Check the list for unlocked arenas. */ + if(a_tsd) { + for(a = a_tsd->next; a; a = a->next) { + if(!mutex_trylock(&a->mutex)) + goto done; + } + for(a = &main_arena; a != a_tsd; a = a->next) { + if(!mutex_tryloc