posix shared memory

[yamt%mwd.biglobe.ne.jp@localhost (YAMAMOTO Takashi)]

i have userland implementation of posix shared memory.
(attached)

rmind proposed a kernel implementation while ago.
http://mail-index.netbsd.org/tech-kern/2009/07/30/msg005599.html
(thus cc:)

which way should we go?

IMO userland implementation is better because:
        - simpler
        - smaller
        - a bug would have less impact
        - "locking object in memory" functionality is better to be
          implemented with more generic api like fcntl if necessary

YAMAMOTO Takashi

/* $NetBSD$ */

/*-
 * Copyright (c)2013 YAMAMOTO Takashi,
 * All rights reserved.
 *
 * Redistribution and use in source and binary forms, with or without
 * modification, are permitted provided that the following conditions
 * are met:
 * 1. Redistributions of source code must retain the above copyright
 *    notice, this list of conditions and the following disclaimer.
 * 2. Redistributions in binary form must reproduce the above copyright
 *    notice, this list of conditions and the following disclaimer in the
 *    documentation and/or other materials provided with the distribution.
 *
 * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
 * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
 * SUCH DAMAGE.
 */

/*
 * POSIX shared memory
 *
 * this is a pure userland implementation.
 *
 * in this implementation, shared memory objects are regular files in
 * _PATH_SHM directory.
 * the directory should have mode 1333.  (or 1777.  this implmentation
 * doesn't need directory read access)
 *
 * for performance reasons, it might be desirable to mount a special
 * filesystem like tmpfs on the directory.
 * requirements for the filesystem:
 *  - it should support create, open, unlink, mmap of regular files.
 *  - it should support access controls with mode bits.
 *  - it should support directory sticky bit.
 *  - it doesn't need to support sub directories, symlinks, hardlinks,
 *    or special files.
 *  - it doesn't need to support persistency across a system reboot.
 *    POSIX allows either behaviours.
 *
 * why not implemented in kernel?
 * in addition to usual reasons to prefer userland implementations in
 * general, we can benefit from existing filesystem functionalities
 * for free.  eg. quota.
 */

#include <sys/cdefs.h>
__RCSID("$NetBSD$");

#include "namespace.h"

#include <sys/mman.h>

#include <errno.h>
#include <fcntl.h>
#include <limits.h>
#include <stdio.h>
#include <string.h>
#include <unistd.h>

#define _PATH_SHM       "/var/shm/"

#ifdef __weak_alias
__weak_alias(shm_open,_shm_open)
__weak_alias(shm_unlink,_shm_unlink)
#endif

static int
_mkname(const char *name, char buf[PATH_MAX])
{
        const char *p;
        int ret;

        /*
         * check the given name.
         *
         * SUSv3:
         * If name begins with the slash character, then
         * calling shm_open() with the same value of name refer to the same
         * shared memory object, as long as that name has not been removed.
         * If name does not begin with the slash character, the effect is
         * implementation-defined. The interpretation of slash characters
         * other than the leading slash character in name is implementation-
         * defined.
         */
        p = name + strspn(name, "/"); /* skip leading slashes */
        if (name[0] != '/' || strchr(p, '/')) {
                /*
                 * implementation-defined behaviour.
                 */
                errno = ENOTSUP;
                return -1;
        }

        /*
         * construct the absolute path of the backing file. 
         */
        ret = snprintf(buf, PATH_MAX, "%s%s", _PATH_SHM, p);
        if (ret == -1) {
                return -1;
        } else if ((size_t)ret >= PATH_MAX) {
                errno = ENAMETOOLONG;
                return -1;
        }
        return 0;
}

int
shm_open(const char *name, int oflags, mode_t mode)
{
        char real_name[PATH_MAX];

        if (_mkname(name, real_name)) {
                return -1;
        }
        return open(real_name, oflags, mode);
}

int
shm_unlink(const char *name)
{
        char real_name[PATH_MAX];

        if (_mkname(name, real_name)) {
                return -1;
        }
        return unlink(real_name);
}