mm.3

Apache V2.0.15 Alpha For Linuxhttpd-2_0_15-alpha.tar.Z

.Vb 3
\& void    \fBmm_lib_error_set\fR(unsigned int, const char *str);
\& char   *\fBmm_lib_error_get\fR(void);
\& int     \fBmm_lib_version\fR(void);
.Ve
.SH "DESCRIPTION"
The \fBMM\fR library is a 2-layer abstraction library which simplifies the usage
of shared memory between forked (and this way strongly related) processes
under Unix platforms. On the first (lower) layer it hides all platform
dependent implementation details (allocation and locking) when dealing with
shared memory segments and on the second (higher) layer it provides a
high-level \fImalloc\fR\|(3)\-style API for a convenient and well known way to work
with data-structures inside those shared memory segments. 
.PP
The abbreviation \fBMM\fR is historically and originally comes from the phrase
``\fImemory mapped\fR'\*(R' as used by the POSIX.1 \fImmap\fR\|(2) function. Because this
facility is internally used by this library on most platforms to establish the
shared memory segments. 
.Sh "\s-1LIBRARY\s0 \s-1STRUCTURE\s0"
This library is structured into three main APIs which are internally based on
each other:
.Ip "\fBGlobal Malloc-Replacement \s-1API\s0\fR" 4
This is the most high-level \s-1API\s0 which directly can be used as replacement \s-1API\s0
for the \s-1POSIX\s0.1 memory allocation \s-1API\s0 (\fImalloc\fR\|(2) and friends). This is
useful when converting \fIheap\fR based data structures to \fIshared memory\fR
based data structures without the need to change the code dramatically.  All
which is needed is to prefix the \s-1POSIX\s0.1 memory allocation functions with
`\f(CWMM_\fR\*(R', i.e. `\f(CWmalloc\fR\*(R' becomes `\f(CWMM_malloc\fR\*(R', `\f(CWstrdup\fR\*(R' becomes
`\f(CWMM_strdup\fR\*(R', etc. This \s-1API\s0 internally uses just a global `\f(CWMM *\fR\*(R' pool for
calling the corresponding functions (those with prefix `\f(CWmm_\fR') of the
\fIStandard Malloc-Style \s-1API\s0\fR.
.Ip "\fBStandard Malloc-Style \s-1API\s0\fR" 4
This is the standard high-level memory allocation \s-1API\s0. Its interface is
similar to the \fIGlobal Malloc-Replacement \s-1API\s0\fR but it uses an explicit `\f(CWMM *\fR\*(R'
pool to operate on. That is why every function of this \s-1API\s0 has an argument of
type `\f(CWMM *\fR\*(R' as its first argument. This \s-1API\s0 provides a comfortable way to
work with small dynamically allocated shared memory chunks inside large
statically allocated shared memory segments. It is internally based on the
\fILow-Level Shared Memory \s-1API\s0\fR for creating the underlaying shared memory
segment.
.Ip "\fBLow-Level Shared Memory \s-1API\s0\fR" 4
This is the basis of the whole \fB\s-1MM\s0\fR library. It provides low-level functions
for creating shared memory segments with mutual exclusion (in short \fImutex\fR)
capabilities in a portable way. Internally the shared memory and mutex
facility is implemented in various platform-dependent ways. A list of
implementation variants follows under the next topic.
.Sh "\s-1SHARED\s0 \s-1MEMORY\s0 \s-1IMPLEMENTATION\s0"
Internally the shared memory facility is implemented in various
platform-dependent ways. Each way has its own advantages and disadvantages
(in addition to the fact that some variants aren't available at all on some
platforms). The \fB\s-1MM\s0\fR library's configuration procedure tries hard to make a
good decision. The implemented variants are now given for overview and
background reasons with their advantages and disadvantages and in an ascending
order, i.e. the \fB\s-1MM\s0\fR configuration mechanism chooses the last available one
in the list as the preferred variant.
.Ip "Classical mmap(2) on temporary file (\s-1MMFILE\s0)" 4
\fIAdvantage:\fR maximum portable.
\fIDisadvantage:\fR needs a temporary file on the filesystem.
.Ip "mmap(2) via \s-1POSIX\s0.1 shm_open(3) on temporary file (\s-1MMPOSX\s0)" 4
\fIAdvantage:\fR standardized by \s-1POSIX\s0.1 and theoretically portable.
\fIDisadvantage:\fR needs a temporary file on the filesystem and is
is usually not available on existing Unix platform.
.Ip "\s-1SVR4-\s0style mmap(2) on \f(CW/dev/zero\fR device (\s-1MMZERO\s0)" 4
\fIAdvantage:\fR widely available and mostly portable on \s-1SVR4\s0 platforms.
\fIDisadvantage:\fR needs the \f(CW/dev/zero/\fR device and a \fImmap\fR\|(2)
which supports memory mapping through this device.
.Ip "4.4BSD\-style mmap(2) via \f(CWMAP_ANON\fR facility (\s-1MMANON\s0)" 4
\fIAdvantage:\fR does not need a temporary file or external device.
\fIDisadvantage:\fR usually only available on \s-1BSD\s0 platforms and derivatives.
.Ip "SysV \s-1IPC\s0 shmget(2) (\s-1IPCSHM\s0)" 4
\fIAdvantage:\fR does not need a temporary file or external device.
\fIDisadvantage:\fR although available on mostly all modern Unix platforms, it has
strong restrictions like the maximum size of a single shared memory segment (can
be as small as 100KB, but depends on the platform).
.Sh "\s-1LOCKING\s0 \s-1IMPLEMENTATION\s0"
As for the shared memory facility, internally the locking facility is
implemented in various platform-dependent ways. A short overview of
implemented variants is given:
.Ip "4.2BSD\-style flock(2) on temporary file (\s-1FLOCK\s0)" 4
\fIAdvantage:\fR exists on a lot of platforms, especially on older Unix
derivates.  \fIDisadvantage:\fR needs a temporary file on the filesystem and has
to re-open file-descriptors to it in \fIeach\fR\|(!) \fIfork\fR\|(2)'ed child process.
.Ip "SysV \s-1IPC\s0 semget(2) (\s-1IPCSEM\s0)" 4
\fIAdvantage:\fR exists on a lot of platforms and does not need a temporary file.
\fIDisadvantage:\fR an unmeant termination of the application leads to a
semaphore leak because the facility does not allow a ``remove in advance'\*(R'
trick (as the \s-1IPC\s0 shared memory facility does) for safe cleanups.
.Ip "\s-1SVR4-\s0style fcntl(2) on temporary file (\s-1FCNTL\s0)" 4
\fIAdvantage:\fR exists on a lot of platforms and is also the most powerful
variant (although not always the fastest one). \fIDisadvantage:\fR needs a
temporary file.
.Sh "\s-1MEMORY\s0 \s-1ALLOCATION\s0 \s-1STRATEGY\s0"
The memory allocation strategy the \fIStandard Malloc-Style \s-1API\s0\fR functions use
internally is the following:
.Ip "\fBAllocation\fR" 4
If a chunk of memory has to be allocated, the internal list of free chunks
is searched for a minimal-size chunk which is larger or equal than the size of
the to be allocated chunk (a \fIbest fit\fR strategy>). 
.Sp
If a chunk is found which matches this best-fit criteria, but is still a lot
larger than the requested size, it is split into two chunks: One with exactly
the requested size (which is the resulting chunk given back) and one with the
remaining size (which is immediately re-inserted into the list of free
chunks). 
.Sp
If no fitting chunk is found at all in the list of free chunks, a new one is
created from the spare area of the shared memory segment until the segment is
full (in which case an \fIout of memory\fR error occurs).
.Ip "\fBDeallocation\fR" 4
If a chunk of memory has to be deallocated, it is inserted in sorted manner
into the internal list of free chunks. The insertion operation automatically
merges the chunk with a previous and/or a next free chunk if possible, i.e.
if the free chunks stay physically seamless (one after another) in memory, to
automatically form larger free chunks out of smaller ones. 
.Sp
This way the shared memory segment is automatically defragmented when memory
is deallocated.
.PP
This strategy reduces memory waste and fragmentation caused by small and
frequent allocations and deallocations to a minimum. 
.PP
The internal implementation of the list of free chunks is not specially
optimized (for instance by using binary search trees or even \fIsplay\fR trees,
etc), because it is assumed that the total amount of entries in the list of
free chunks is always small (caused both by the fact that shared memory
segments are usually a lot smaller than heaps and the fact that we always
defragment by merging the free chunks if possible).
.SH "API FUNCTIONS"
In the following all API functions are described in detail.
The order directly follows the one in the \fBSYNOPSIS\fR.
.Sh "Global Malloc-Replacement \s-1API\s0"
.Ip "int \fBMM_create\fR(size_t \fIsize\fR, const char *\fIfile\fR);" 4
This initializes the global shared memory pool with \fIsize\fR and \fIfile\fR and
has to be called \fIbefore\fR any \fIfork\fR\|(2) operations are performed by the
application.
.Ip "int \fBMM_permission\fR(mode_t \fImode\fR, uid_t \fIowner\fR, gid_t \fIgroup\fR);" 4
This sets the filesystem \fImode\fR, \fIowner\fR and \fIgroup\fR for the global shared
memory pool (has effects only if the underlaying shared memory segment
implementation is actually based on external auxiliary files).  The arguments
are directly passed through to \fIchmod\fR\|(2) and \fIchown\fR\|(2).
.Ip "void \fBMM_destroy\fR(void);" 4
This destroys the global shared memory pool and should be called \fIafter\fR all
child processes were killed.
.Ip "int \fBMM_lock\fR(mm_lock_mode \fImode\fR);" 4
This locks the global shared memory pool for the current process in order to
perform either shared/read-only (\fImode\fR is \f(CWMM_LOCK_RD\fR) or
exclusive/read-write (\fImode\fR is \f(CWMM_LOCK_RW\fR) critical operations inside the
global shared memory pool.
.Ip "int \fBMM_unlock\fR(void);" 4
This unlocks the global shared memory pool for the current process after the
critical operations were performed inside the global shared memory pool.
.Ip "void *\fBMM_malloc\fR(size_t \fIsize\fR);" 4
Identical to the \s-1POSIX\s0.1 \fImalloc\fR\|(3) function but instead of allocating
memory from the \fIheap\fR it allocates it from the global shared memory pool.
.Ip "void \fBMM_free\fR(void *\fIptr\fR);" 4
Identical to the \s-1POSIX\s0.1 \fIfree\fR\|(3) function but instead of deallocating
memory in the \fIheap\fR it deallocates it in the global shared memory pool.
.Ip "void *\fBMM_realloc\fR(void *\fIptr\fR, size_t \fIsize\fR);" 4
Identical to the \s-1POSIX\s0.1 \fIrealloc\fR\|(3) function but instead of reallocating
memory in the \fIheap\fR it reallocates it inside the global shared memory pool.
.Ip "void *\fBMM_calloc\fR(size_t \fInumber\fR, size_t \fIsize\fR);" 4
Identical to the \s-1POSIX\s0.1 \fIcalloc\fR\|(3) function but instead of allocating and
initializing memory from the \fIheap\fR it allocates and initializes it from the
global shared memory pool.
.Ip "char *\fBMM_strdup\fR(const char *\fIstr\fR);" 4
Identical to the \s-1POSIX\s0.1 \fIstrdup\fR\|(3) function but instead of creating the
string copy in the \fIheap\fR it creates it in the global shared memory pool.
.Ip "size_t \fBMM_sizeof\fR(const void *\fIptr\fR);" 4
This function returns the size in bytes of the chunk starting at \fIptr\fR when
\fIptr\fR was previously allocated with \fIMM_malloc\fR\|(3). The result is undefined
if \fIptr\fR was not previously allocated with \fIMM_malloc\fR\|(3).
.Ip "size_t \fBMM_maxsize\fR(void);" 4
This function returns the maximum size which is allowed
as the first argument to the \fIMM_create\fR\|(3) function.
.Ip "size_t \fBMM_available\fR(void);" 4
Returns the amount in bytes of still available (free) memory in the global
shared memory pool.
.Ip "char *\fBMM_error\fR(void);" 4
Returns the last error message which occurred inside the \fB\s-1MM\s0\fR library.
.Sh "Standard Malloc-Style \s-1API\s0"
.Ip "\s-1MM\s0 *\fBmm_create\fR(size_t \fIsize\fR, const char *\fIfile\fR);" 4
This creates a shared memory pool which has space for approximately a total of
\fIsize\fR bytes with the help of \fIfile\fR. Here \fIfile\fR is a filesystem path to a
file which need not to exist (and perhaps is never created because this
depends on the platform and chosen shared memory and mutex implementation).
The return value is a pointer to a \f(CWMM\fR structure which should be treated as
opaque by the application. It describes the internals of the created shared
memory pool. In case of an error \f(CWNULL\fR is returned.  A \fIsize\fR of 0 means to
allocate the maximum allowed size which is platform dependent and is between a
few \s-1KB\s0 and the soft limit of 64MB.
.Ip "int \fBmm_permission\fR(\s-1MM\s0 *\fImm\fR, mode_t \fImode\fR, uid_t \fIowner\fR, gid_t \fIgroup\fR);" 4
This sets the filesystem \fImode\fR, \fIowner\fR and \fIgroup\fR for the shared memory
pool \fImm\fR (has effects only when the underlaying shared memory segment
implementation is actually based on external auxiliary files).  The arguments
are directly passed through to \fIchmod\fR\|(2) and \fIchown\fR\|(2).
.Ip "void \fBmm_destroy\fR(\s-1MM\s0 *\fImm\fR);" 4
This destroys the complete shared memory pool \fImm\fR and with it all chunks
which were allocated in this pool. Additionally any created files on the
filesystem corresponding the to shared memory pool are unlinked.
.Ip "int \fBmm_lock\fR(\s-1MM\s0 *\fImm\fR, mm_lock_mode \fImode\fR);" 4
This locks the shared memory pool \fImm\fR for the current process in order to
perform either shared/read-only (\fImode\fR is \f(CWMM_LOCK_RD\fR) or
exclusive/read-write (\fImode\fR is \f(CWMM_LOCK_RW\fR) critical operations inside the
global shared memory pool.
.Ip "int \fBmm_unlock\fR(\s-1MM\s0 *\fImm\fR);" 4
This unlocks the shared memory pool \fImm\fR for the current process after
critical operations were performed inside the global shared memory pool.
.Ip "void *\fBmm_malloc\fR(\s-1MM\s0 *\fImm\fR, size_t \fIsize\fR);" 4
This function allocates \fIsize\fR bytes from the shared memory pool \fImm\fR and
returns either a (virtual memory word aligned) pointer to it or \f(CWNULL\fR in
case of an error (out of memory). It behaves like the \s-1POSIX\s0.1 \fImalloc\fR\|(3)
function but instead of allocating memory from the \fIheap\fR it allocates it
from the shared memory segment underlaying \fImm\fR.
.Ip "void \fBmm_free\fR(\s-1MM\s0 *\fImm\fR, void *\fIptr\fR);" 4
This deallocates the chunk starting at \fIptr\fR in the shared memory pool \fImm\fR.
It behaves like the \s-1POSIX\s0.1 \fIfree\fR\|(3) function but instead of deallocating
memory from the \fIheap\fR it deallocates it from the shared memory segment
underlaying \fImm\fR.
.Ip "void *\fBmm_realloc\fR(\s-1MM\s0 *\fImm\fR, void *\fIptr\fR, size_t \fIsize\fR);" 4
This function reallocates the chunk starting at \fIptr\fR inside the shared
memory pool \fImm\fR with the new size of \fIsize\fR bytes.  It behaves like the
\s-1POSIX\s0.1 \fIrealloc\fR\|(3) function but instead of reallocating memory in the
\fIheap\fR it reallocates it in the shared memory segment underlaying \fImm\fR.
.Ip "void *\fBmm_calloc\fR(\s-1MM\s0 *\fImm\fR, size_t \fInumber\fR, size_t \fIsize\fR);" 4
This is similar to \fImm_malloc\fR\|(3), but additionally clears the chunk. It behaves
like the \s-1POSIX\s0.1 \fIcalloc\fR\|(3) function.  It allocates space for \fInumber\fR
objects, each \fIsize\fR bytes in length from the shared memory pool \fImm\fR.  The
result is identical to calling \fImm_malloc\fR\|(3) with an argument of ``\fInumber\fR *
\fIsize\fR'\*(R', with the exception that the allocated memory is initialized to nul
bytes.
.Ip "char *\fBmm_strdup\fR(\s-1MM\s0 *\fImm\fR, const char *\fIstr\fR);" 4
This function behaves like the \s-1POSIX\s0.1 \fIstrdup\fR\|(3) function.  It allocates
sufficient memory inside the shared memory pool \fImm\fR for a copy of the string
\fIstr\fR, does the copy, and returns a pointer to it.  The pointer may
subsequently be used as an argument to the function \fImm_free\fR\|(3). If
insufficient shared memory is available, \f(CWNULL\fR is returned.
.Ip "size_t \fBmm_sizeof\fR(const void *\fIptr\fR);" 4
This function returns the size in bytes of the chunk starting at \fIptr\fR when
\fIptr\fR was previously allocated with \fImm_malloc\fR\|(3). The result is undefined
when \fIptr\fR was not previously allocated with \fImm_malloc\fR\|(3).
.Ip "size_t \fBmm_maxsize\fR(void);" 4
This function returns the maximum size which is allowed as the first argument
to the \fImm_create\fR\|(3) function.
.Ip "size_t \fBmm_available\fR(\s-1MM\s0 *\fImm\fR);" 4
Returns the amount in bytes of still available (free) memory in the 
shared memory pool \fImm\fR.
.Ip "char *\fBmm_error\fR(void);" 4
Returns the last error message which occurred inside the \fB\s-1MM\s0\fR library.
.Ip "void \fBmm_display_info\fR(\s-1MM\s0 *\fImm\fR);" 4
This is debugging function which displays a summary page for the shared memory
pool \fImm\fR describing various internal sizes and counters.
.Sh "Low-Level Shared Memory \s-1API\s0"
.Ip "void *\fBmm_core_create\fR(size_t \fIsize\fR, const char *\fIfile\fR);" 4
This creates a shared memory area which is at least \fIsize\fR bytes in size with