linux-tcpip.c

非常经典的一个分布式系统

/* -*- mode: c; c-basic-offset: 8; indent-tabs-mode: nil; -*-
 * vim:expandtab:shiftwidth=8:tabstop=8:
 *
 * Copyright (C) 2005 Cluster File Systems, Inc.
 *
 *   This file is part of Lustre, http://www.lustre.org.
 *
 *   Lustre is free software; you can redistribute it and/or
 *   modify it under the terms of version 2 of the GNU General Public
 *   License as published by the Free Software Foundation.
 *
 *   Lustre 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 General Public License for more details.
 *
 *   You should have received a copy of the GNU General Public License
 *   along with Lustre; if not, write to the Free Software
 *   Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
 */
#define DEBUG_SUBSYSTEM S_LNET

#include <libcfs/kp30.h>
#include <libcfs/libcfs.h>

#include <linux/if.h>
#include <linux/in.h>
#include <linux/file.h>
/* For sys_open & sys_close */
#if LINUX_VERSION_CODE >= KERNEL_VERSION(2,5,0)
#include <linux/syscalls.h>
#else
#include <linux/fs.h>
#endif

int
libcfs_sock_ioctl(int cmd, unsigned long arg)
{
        mm_segment_t   oldmm = get_fs();
        struct socket  *sock;
        int             fd;
        int             rc;
        struct file     *sock_filp;

        rc = sock_create (PF_INET, SOCK_STREAM, 0, &sock);
        if (rc != 0) {
                CERROR ("Can't create socket: %d\n", rc);
                return rc;
        }

        fd = sock_map_fd(sock);
        if (fd < 0) {
                rc = fd;
                sock_release(sock);
                goto out;
        }

        sock_filp = fget(fd);
        if (!sock_filp) {
                rc = -ENOMEM;
                goto out_fd;
        }

        set_fs(KERNEL_DS);
#ifdef HAVE_UNLOCKED_IOCTL
        if (sock_filp->f_op->unlocked_ioctl)
                rc = sock_filp->f_op->unlocked_ioctl(sock_filp, cmd, arg);
        else
#endif
             {
                lock_kernel();
                rc =sock_filp->f_op->ioctl(sock_filp->f_dentry->d_inode,
                                           sock_filp, cmd, arg);
                unlock_kernel();
             }
        set_fs(oldmm);

        fput(sock_filp);

 out_fd:
        sys_close(fd);
 out:
        return rc;
}

int
libcfs_ipif_query (char *name, int *up, __u32 *ip, __u32 *mask)
{
        struct ifreq   ifr;
        int            nob;
        int            rc;
        __u32          val;

        nob = strnlen(name, IFNAMSIZ);
        if (nob == IFNAMSIZ) {
                CERROR("Interface name %s too long\n", name);
                return -EINVAL;
        }

        CLASSERT (sizeof(ifr.ifr_name) >= IFNAMSIZ);

        strcpy(ifr.ifr_name, name);
        rc = libcfs_sock_ioctl(SIOCGIFFLAGS, (unsigned long)&ifr);

        if (rc != 0) {
                CERROR("Can't get flags for interface %s\n", name);
                return rc;
        }

        if ((ifr.ifr_flags & IFF_UP) == 0) {
                CDEBUG(D_NET, "Interface %s down\n", name);
                *up = 0;
                *ip = *mask = 0;
                return 0;
        }

        *up = 1;

        strcpy(ifr.ifr_name, name);
        ifr.ifr_addr.sa_family = AF_INET;
        rc = libcfs_sock_ioctl(SIOCGIFADDR, (unsigned long)&ifr);

        if (rc != 0) {
                CERROR("Can't get IP address for interface %s\n", name);
                return rc;
        }

        val = ((struct sockaddr_in *)&ifr.ifr_addr)->sin_addr.s_addr;
        *ip = ntohl(val);

        strcpy(ifr.ifr_name, name);
        ifr.ifr_addr.sa_family = AF_INET;
        rc = libcfs_sock_ioctl(SIOCGIFNETMASK, (unsigned long)&ifr);

        if (rc != 0) {
                CERROR("Can't get netmask for interface %s\n", name);
                return rc;
        }

        val = ((struct sockaddr_in *)&ifr.ifr_netmask)->sin_addr.s_addr;
        *mask = ntohl(val);

        return 0;
}

EXPORT_SYMBOL(libcfs_ipif_query);

int
libcfs_ipif_enumerate (char ***namesp)
{
        /* Allocate and fill in 'names', returning # interfaces/error */
        char           **names;
        int             toobig;
        int             nalloc;
        int             nfound;
        struct ifreq   *ifr;
        struct ifconf   ifc;
        int             rc;
        int             nob;
        int             i;


        nalloc = 16;        /* first guess at max interfaces */
        toobig = 0;
        for (;;) {
                if (nalloc * sizeof(*ifr) > CFS_PAGE_SIZE) {
                        toobig = 1;
                        nalloc = CFS_PAGE_SIZE/sizeof(*ifr);
                        CWARN("Too many interfaces: only enumerating first %d\n",
                              nalloc);
                }

                LIBCFS_ALLOC(ifr, nalloc * sizeof(*ifr));
                if (ifr == NULL) {
                        CERROR ("ENOMEM enumerating up to %d interfaces\n", nalloc);
                        rc = -ENOMEM;
                        goto out0;
                }

                ifc.ifc_buf = (char *)ifr;
                ifc.ifc_len = nalloc * sizeof(*ifr);

                rc = libcfs_sock_ioctl(SIOCGIFCONF, (unsigned long)&ifc);

                if (rc < 0) {
                        CERROR ("Error %d enumerating interfaces\n", rc);
                        goto out1;
                }

                LASSERT (rc == 0);

                nfound = ifc.ifc_len/sizeof(*ifr);
                LASSERT (nfound <= nalloc);

                if (nfound < nalloc || toobig)
                        break;

                LIBCFS_FREE(ifr, nalloc * sizeof(*ifr));
                nalloc *= 2;
        }

        if (nfound == 0)
                goto out1;

        LIBCFS_ALLOC(names, nfound * sizeof(*names));
        if (names == NULL) {
                rc = -ENOMEM;
                goto out1;
        }
        /* NULL out all names[i] */
        memset (names, 0, nfound * sizeof(*names));

        for (i = 0; i < nfound; i++) {

                nob = strnlen (ifr[i].ifr_name, IFNAMSIZ);
                if (nob == IFNAMSIZ) {
                        /* no space for terminating NULL */
                        CERROR("interface name %.*s too long (%d max)\n",
                               nob, ifr[i].ifr_name, IFNAMSIZ);
                        rc = -ENAMETOOLONG;
                        goto out2;
                }

                LIBCFS_ALLOC(names[i], IFNAMSIZ);
                if (names[i] == NULL) {
                        rc = -ENOMEM;
                        goto out2;
                }

                memcpy(names[i], ifr[i].ifr_name, nob);
                names[i][nob] = 0;
        }

        *namesp = names;
        rc = nfound;

 out2:
        if (rc < 0)
                libcfs_ipif_free_enumeration(names, nfound);
 out1:
        LIBCFS_FREE(ifr, nalloc * sizeof(*ifr));
 out0:
        return rc;
}

EXPORT_SYMBOL(libcfs_ipif_enumerate);

void
libcfs_ipif_free_enumeration (char **names, int n)
{
        int      i;

        LASSERT (n > 0);

        for (i = 0; i < n && names[i] != NULL; i++)
                LIBCFS_FREE(names[i], IFNAMSIZ);

        LIBCFS_FREE(names, n * sizeof(*names));
}

EXPORT_SYMBOL(libcfs_ipif_free_enumeration);

int
libcfs_sock_write (struct socket *sock, void *buffer, int nob, int timeout)
{
        int            rc;
        mm_segment_t   oldmm = get_fs();
        long           ticks = timeout * HZ;
        unsigned long  then;
        struct timeval tv;

        LASSERT (nob > 0);
        /* Caller may pass a zero timeout if she thinks the socket buffer is
         * empty enough to take the whole message immediately */

        for (;;) {
                struct iovec  iov = {
                        .iov_base = buffer,
                        .iov_len  = nob
                };
                struct msghdr msg = {
                        .msg_name       = NULL,
                        .msg_namelen    = 0,
                        .msg_iov        = &iov,
                        .msg_iovlen     = 1,
                        .msg_control    = NULL,
                        .msg_controllen = 0,
                        .msg_flags      = (timeout == 0) ? MSG_DONTWAIT : 0
                };

                if (timeout != 0) {
                        /* Set send timeout to remaining time */
                        tv = (struct timeval) {
                                .tv_sec = ticks / HZ,
                                .tv_usec = ((ticks % HZ) * 1000000) / HZ
                        };
                        set_fs(KERNEL_DS);
                        rc = sock_setsockopt(sock, SOL_SOCKET, SO_SNDTIMEO,
                                             (char *)&tv, sizeof(tv));
                        set_fs(oldmm);
                        if (rc != 0) {
                                CERROR("Can't set socket send timeout "
                                       "%ld.%06d: %d\n",
                                       (long)tv.tv_sec, (int)tv.tv_usec, rc);
                                return rc;
                        }
                }

                set_fs (KERNEL_DS);
                then = jiffies;
                rc = sock_sendmsg (sock, &msg, iov.iov_len);
                ticks -= jiffies - then;
                set_fs (oldmm);

                if (rc == nob)
                        return 0;

                if (rc < 0)
                        return rc;

                if (rc == 0) {
                        CERROR ("Unexpected zero rc\n");
                        return (-ECONNABORTED);
                }

                if (ticks <= 0)
                        return -EAGAIN;

                buffer = ((char *)buffer) + rc;
                nob -= rc;
        }

        return (0);
}
EXPORT_SYMBOL(libcfs_sock_write);

int
libcfs_sock_read (struct socket *sock, void *buffer, int nob, int timeout)
{
        int            rc;
        mm_segment_t   oldmm = get_fs();
        long           ticks = timeout * HZ;