Whamcloud - gitweb
- put GSS check after linux modules check to make sure that $LINUX is defined.
[fs/lustre-release.git] / lustre / liblustre / llite_lib.c
index 0b549c0..d1528e6 100644 (file)
@@ -3,7 +3,7 @@
  *
  * Lustre Light common routines
  *
- *  Copyright (c) 2002, 2003 Cluster File Systems, Inc.
+ *  Copyright (c) 2002-2004 Cluster File Systems, Inc.
  *
  *   This file is part of Lustre, http://www.lustre.org.
  *
 #include <sys/types.h>
 #include <sys/queue.h>
 
-#include <netinet/in.h>
-#include <sys/socket.h>
-#include <arpa/inet.h>
-
+#ifdef HAVE_XTIO_H
+#include <xtio.h>
+#endif
 #include <sysio.h>
 #include <fs.h>
 #include <mount.h>
 #include <inode.h>
+#ifdef HAVE_FILE_H
 #include <file.h>
+#endif
+
+/* env variables */
+#define ENV_LUSTRE_MNTPNT               "LIBLUSTRE_MOUNT_POINT"
+#define ENV_LUSTRE_MNTTGT               "LIBLUSTRE_MOUNT_TARGET"
+#define ENV_LUSTRE_TIMEOUT              "LIBLUSTRE_TIMEOUT"
+#define ENV_LUSTRE_DUMPFILE             "LIBLUSTRE_DUMPFILE"
+#define ENV_LUSTRE_DEBUG_MASK           "LIBLUSTRE_DEBUG_MASK"
+#define ENV_LUSTRE_DEBUG_SUBSYS         "LIBLUSTRE_DEBUG_SUBSYS"
+#define ENV_LUSTRE_NAL_NAME             "LIBLUSTRE_NAL_NAME"
+
+#ifdef REDSTORM
+#define CSTART_INIT
+#endif
 
 /* both sys/queue.h (libsysio require it) and portals/lists.h have definition
  * of 'LIST_HEAD'. undef it to suppress warnings
  */
 #undef LIST_HEAD
+#include <portals/ptlctl.h>
 
-#include <portals/ptlctl.h>    /* needed for parse_dump */
-#include <procbridge.h>
-
+#include "lutil.h"
 #include "llite_lib.h"
 
-unsigned int portal_subsystem_debug = ~0 - (S_PORTALS | S_QSWNAL | S_SOCKNAL |
-                                            S_GMNAL | S_IBNAL);
-
-ptl_handle_ni_t         tcpnal_ni;
-struct task_struct     *current;
-
-/* portals interfaces */
-
-struct ldlm_namespace;
-struct ldlm_res_id;
-struct obd_import;
-
-void *inter_module_get(char *arg)
+static int lllib_init(void)
 {
-        if (!strcmp(arg, "tcpnal_ni"))
-                return &tcpnal_ni;
-        else if (!strcmp(arg, "ldlm_cli_cancel_unused"))
-                return ldlm_cli_cancel_unused;
-        else if (!strcmp(arg, "ldlm_namespace_cleanup"))
-                return ldlm_namespace_cleanup;
-        else if (!strcmp(arg, "ldlm_replay_locks"))
-                return ldlm_replay_locks;
-        else
-                return NULL;
-}
-
-/* XXX move to proper place */
-char *portals_nid2str(int nal, ptl_nid_t nid, char *str)
-{
-        switch(nal){
-        case TCPNAL:
-                /* userspace NAL */
-        case SOCKNAL:
-                sprintf(str, "%u:%d.%d.%d.%d", (__u32)(nid >> 32),
-                        HIPQUAD(nid));
-                break;
-        case QSWNAL:
-        case GMNAL:
-        case IBNAL:
-        case SCIMACNAL:
-                sprintf(str, "%u:%u", (__u32)(nid >> 32), (__u32)nid);
-                break;
-        default:
-                return NULL;
-        }
-        return str;
-}
-
-void init_current(char *comm)
-{ 
-        current = malloc(sizeof(*current));
-        current->fs = malloc(sizeof(*current->fs));
-        current->fs->umask = umask(0777);
-        umask(current->fs->umask);
-        strncpy(current->comm, comm, sizeof(current->comm));
-        current->pid = getpid();
-        current->fsuid = 0;
-        current->fsgid = 0;
-        current->cap_effective = -1;
-        memset(&current->pending, 0, sizeof(current->pending));
-}
-
-/* FIXME */
-void generate_random_uuid(unsigned char uuid_out[16])
-{
-        int *arr = (int*)uuid_out;
-        int i;
-
-        for (i = 0; i < sizeof(uuid_out)/sizeof(int); i++)
-                arr[i] = rand();
-}
-
-ptl_nid_t tcpnal_mynid;
-
-int init_lib_portals()
-{
-        int max_interfaces;
-        int rc;
-        ENTRY;
+        liblustre_set_nal_nid();
 
-        rc = PtlInit(&max_interfaces);
-        if (rc != PTL_OK) {
-                CERROR("PtlInit failed: %d\n", rc);
-                RETURN (-ENXIO);
-        }
-        RETURN(0);
-}
-
-int
-libcfs_nal_cmd(struct portals_cfg *pcfg)
-{
-        /* handle portals command if we want */
-        return 0;
-}
-
-extern int class_handle_ioctl(unsigned int cmd, unsigned long arg);
-
-int lib_ioctl_nalcmd(int dev_id, int opc, void * ptr)
-{
-        struct portal_ioctl_data *ptldata;
-
-        if (opc == IOC_PORTAL_NAL_CMD) {
-                ptldata = (struct portal_ioctl_data *) ptr;
-
-                if (ptldata->ioc_nal_cmd == NAL_CMD_REGISTER_MYNID) {
-                        tcpnal_mynid = ptldata->ioc_nid;
-                        printf("mynid: %u.%u.%u.%u\n",
-                                (unsigned)(tcpnal_mynid>>24) & 0xFF,
-                                (unsigned)(tcpnal_mynid>>16) & 0xFF,
-                                (unsigned)(tcpnal_mynid>>8) & 0xFF,
-                                (unsigned)(tcpnal_mynid) & 0xFF);
-                }
-        }
-
-       return (0);
-}
-
-int lib_ioctl(int dev_id, int opc, void * ptr)
-{
-        int rc;
-
-       if (dev_id == OBD_DEV_ID) {
-                struct obd_ioctl_data *ioc = ptr;
-
-                //XXX hack!!!
-                ioc->ioc_plen1 = ioc->ioc_inllen1;
-                ioc->ioc_pbuf1 = ioc->ioc_bulk;
-                //XXX
-
-                rc = class_handle_ioctl(opc, (unsigned long)ptr);
-
-                printf ("proccssing ioctl cmd: %x, rc %d\n", opc,  rc);
-
-                if (rc)
-                        return rc;
-       }
-       return (0);
-}
-
-int lllib_init(char *dumpfile)
-{
-        if (!g_zconf) {
-                /* this parse only get my nid from config file
-                 * before initialize portals
-                 */
-                if (parse_dump(dumpfile, lib_ioctl_nalcmd))
-                        return -1;
-        } else {
-                /* XXX need setup mynid before tcpnal initialize */
-                tcpnal_mynid = ((uint64_t)getpid() << 32) | time(0);
-                printf("LibLustre: TCPNAL NID: %016llx\n", tcpnal_mynid);
-        }
-
-        init_current("dummy");
-        if (init_obdclass() ||
+        if (liblustre_init_current("dummy") ||
+            init_obdclass() ||
             init_lib_portals() ||
             ptlrpc_init() ||
             mdc_init() ||
@@ -210,22 +74,22 @@ int lllib_init(char *dumpfile)
             osc_init())
                 return -1;
 
-        if (!g_zconf && parse_dump(dumpfile, lib_ioctl))
-                return -1;
-
         return _sysio_fssw_register("llite", &llu_fssw_ops);
 }
  
-#if 0
-static void llu_check_request()
-{
-        liblustre_wait_event(0);
-}
+#ifndef CRAY_PORTALS
+#define LIBLUSTRE_NAL_NAME "tcp"
+#elif defined REDSTORM
+#define LIBLUSTRE_NAL_NAME "cray_qk_ernal"
+#else
+#define LIBLUSTRE_NAL_NAME "cray_pb_ernal"
 #endif
 
 int liblustre_process_log(struct config_llog_instance *cfg, int allow_recov)
 {
-        struct lustre_cfg lcfg;
+        struct lustre_cfg_bufs bufs;
+        struct lustre_cfg *lcfg;
+
         char  *peer = "MDS_PEER_UUID";
         struct obd_device *obd;
         struct lustre_handle mdc_conn = {0, };
@@ -236,6 +100,7 @@ int liblustre_process_log(struct config_llog_instance *cfg, int allow_recov)
         struct llog_ctxt *ctxt;
         ptl_nid_t nid = 0;
         int nal, err, rc = 0;
+        char *nal_name;
         ENTRY;
 
         generate_random_uuid(uuid);
@@ -246,38 +111,42 @@ int liblustre_process_log(struct config_llog_instance *cfg, int allow_recov)
                 RETURN(-EINVAL);
         }
 
-        nal = ptl_name2nal("tcp");
+        nal_name = getenv(ENV_LUSTRE_NAL_NAME);
+        if (!nal_name)
+                nal_name = "tcp";
+        nal = ptl_name2nal(nal_name);
         if (nal <= 0) {
-                CERROR("Can't parse NAL tcp\n");
+                CERROR("Can't parse NAL %s\n", nal_name);
                 RETURN(-EINVAL);
         }
-        LCFG_INIT(lcfg, LCFG_ADD_UUID, NULL);
-        lcfg.lcfg_nid = nid;
-        lcfg.lcfg_inllen1 = strlen(peer) + 1;
-        lcfg.lcfg_inlbuf1 = peer;
-        lcfg.lcfg_nal = nal;
-        err = class_process_config(&lcfg);
+        lustre_cfg_bufs_reset(&bufs, NULL);
+        lustre_cfg_bufs_set_string(&bufs, 1, peer);
+        lcfg = lustre_cfg_new(LCFG_ADD_UUID, &bufs);
+        lcfg->lcfg_nid = nid;
+        lcfg->lcfg_nal = nal;
+        err = class_process_config(lcfg);
+        lustre_cfg_free(lcfg);
         if (err < 0)
                 GOTO(out, err);
 
-        LCFG_INIT(lcfg, LCFG_ATTACH, name);
-        lcfg.lcfg_inlbuf1 = "mdc";
-        lcfg.lcfg_inllen1 = strlen(lcfg.lcfg_inlbuf1) + 1;
-        lcfg.lcfg_inlbuf2 = mdc_uuid.uuid;
-        lcfg.lcfg_inllen2 = strlen(lcfg.lcfg_inlbuf2) + 1;
-        err = class_process_config(&lcfg);
+        lustre_cfg_bufs_reset(&bufs, name);
+        lustre_cfg_bufs_set_string(&bufs, 1, OBD_MDC_DEVICENAME);
+        lustre_cfg_bufs_set_string(&bufs, 2, mdc_uuid.uuid);
+        lcfg = lustre_cfg_new(LCFG_ATTACH, &bufs);
+        err = class_process_config(lcfg);
+        lustre_cfg_free(lcfg);
         if (err < 0)
                 GOTO(out_del_uuid, err);
 
-        LCFG_INIT(lcfg, LCFG_SETUP, name);
-        lcfg.lcfg_inlbuf1 = g_zconf_mdsname;
-        lcfg.lcfg_inllen1 = strlen(lcfg.lcfg_inlbuf1) + 1;
-        lcfg.lcfg_inlbuf2 = peer;
-        lcfg.lcfg_inllen2 = strlen(lcfg.lcfg_inlbuf2) + 1;
-        err = class_process_config(&lcfg);
+        lustre_cfg_bufs_reset(&bufs, name);
+        lustre_cfg_bufs_set_string(&bufs, 1, g_zconf_mdsname);
+        lustre_cfg_bufs_set_string(&bufs, 2, peer);
+        lcfg = lustre_cfg_new(LCFG_SETUP, &bufs);
+        err = class_process_config(lcfg);
+        lustre_cfg_free(lcfg);
         if (err < 0)
                 GOTO(out_detach, err);
-        
+
         obd = class_name2obd(name);
         if (obd == NULL)
                 GOTO(out_cleanup, err = -EINVAL);
@@ -287,51 +156,49 @@ int liblustre_process_log(struct config_llog_instance *cfg, int allow_recov)
                            strlen("initial_recov"), "initial_recov",
                            sizeof(allow_recov), &allow_recov);
 
-        err = obd_connect(&mdc_conn, obd, &mdc_uuid);
+        err = obd_connect(&mdc_conn, obd, &mdc_uuid, NULL, 0);
         if (err) {
                 CERROR("cannot connect to %s: rc = %d\n",
                         g_zconf_mdsname, err);
                 GOTO(out_cleanup, err);
         }
-        
+
         exp = class_conn2export(&mdc_conn);
-        
+
         ctxt = exp->exp_obd->obd_llog_ctxt[LLOG_CONFIG_REPL_CTXT];
-        rc = class_config_parse_llog(ctxt, g_zconf_profile, cfg);
-        if (rc) {
-                CERROR("class_config_parse_llog failed: rc = %d\n", rc);
-        }
+        rc = class_config_process_llog(ctxt, g_zconf_profile, cfg);
+        if (rc)
+                CERROR("class_config_process_llog failed: rc = %d\n", rc);
 
         err = obd_disconnect(exp, 0);
 
 out_cleanup:
-        LCFG_INIT(lcfg, LCFG_CLEANUP, name);
-        err = class_process_config(&lcfg);
+        lustre_cfg_bufs_reset(&bufs, name);
+        lcfg = lustre_cfg_new(LCFG_CLEANUP, &bufs);
+        err = class_process_config(lcfg);
+        lustre_cfg_free(lcfg);
         if (err < 0)
                 GOTO(out, err);
 
 out_detach:
-        LCFG_INIT(lcfg, LCFG_DETACH, name);
-        err = class_process_config(&lcfg);
+        lustre_cfg_bufs_reset(&bufs, name);
+        lcfg = lustre_cfg_new(LCFG_DETACH, &bufs);
+        err = class_process_config(lcfg);
+        lustre_cfg_free(lcfg);
         if (err < 0)
                 GOTO(out, err);
 
 out_del_uuid:
-        LCFG_INIT(lcfg, LCFG_DEL_UUID, name);
-        lcfg.lcfg_inllen1 = strlen(peer) + 1;
-        lcfg.lcfg_inlbuf1 = peer;
-        err = class_process_config(&lcfg);
-
+        lustre_cfg_bufs_reset(&bufs, name);
+        lustre_cfg_bufs_set_string(&bufs, 1, peer);
+        lcfg = lustre_cfg_new(LCFG_DEL_UUID, &bufs);
+        err = class_process_config(lcfg);
+        lustre_cfg_free(lcfg);
 out:
         if (rc == 0)
                 rc = err;
-        
-        RETURN(rc);
-}
 
-static void sighandler_USR1(int signum)
-{
-        /* do nothing */
+        RETURN(rc);
 }
 
 /* parse host:/mdsname/profile string */
@@ -347,7 +214,7 @@ int ll_parse_mount_target(const char *target, char **mdsnid,
         if ((s = strchr(buf, ':'))) {
                 *mdsnid = buf;
                 *s = '\0';
-                                                                                                                        
+
                 while (*++s == '/')
                         ;
                 *mdsname = s;
@@ -361,18 +228,55 @@ int ll_parse_mount_target(const char *target, char **mdsnid,
         return -1;
 }
 
-/* env variables */
-#define ENV_LUSTRE_MNTPNT               "LIBLUSTRE_MOUNT_POINT"
-#define ENV_LUSTRE_MNTTGT               "LIBLUSTRE_MOUNT_TARGET"
-#define ENV_LUSTRE_TIMEOUT              "LIBLUSTRE_TIMEOUT"
-#define ENV_LUSTRE_DUMPFILE             "LIBLUSTRE_DUMPFILE"
+/*
+ * early liblustre init
+ * called from C startup in catamount apps, before main()
+ *
+ * The following is a skeleton sysio startup sequence,
+ * as implemented in C startup (skipping error handling).
+ * In this framework none of these calls need be made here
+ * or in the apps themselves.  The NAMESPACE_STRING specifying
+ * the initial set of fs ops (creates, mounts, etc.) is passed
+ * as an environment variable.
+ *
+ *      _sysio_init();
+ *      _sysio_incore_init();
+ *      _sysio_native_init();
+ *      _sysio_lustre_init();
+ *      _sysio_boot(NAMESPACE_STRING);
+ *
+ * the name _sysio_lustre_init() follows the naming convention
+ * established in other fs drivers from libsysio:
+ *  _sysio_incore_init(), _sysio_native_init()
+ *
+ * _sysio_lustre_init() must be called before _sysio_boot()
+ * to enable libsysio's processing of namespace init strings containing
+ * lustre filesystem operations
+ */
+int _sysio_lustre_init(void)
+{
+        int err;
 
-extern int _sysio_native_init();
+#if 0
+        portal_debug = -1;
+        portal_subsystem_debug = -1;
+#endif
+
+        liblustre_init_random();
+
+        err = lllib_init();
+        if (err) {
+                perror("init llite driver");
+        }
+        return err;
+}
 
+extern int _sysio_native_init();
 extern unsigned int obd_timeout;
 
+char *lustre_path = NULL;
+
 /* global variables */
-int     g_zconf = 0;            /* zeroconf or dumpfile */
 char   *g_zconf_mdsname = NULL; /* mdsname, for zeroconf */
 char   *g_zconf_mdsnid = NULL;  /* mdsnid, for zeroconf */
 char   *g_zconf_profile = NULL; /* profile, for zeroconf */
@@ -380,100 +284,111 @@ char   *g_zconf_profile = NULL; /* profile, for zeroconf */
 
 void __liblustre_setup_(void)
 {
-        char *lustre_path = NULL;
         char *target = NULL;
         char *timeout = NULL;
-        char *dumpfile = NULL;
+        char *debug_mask = NULL;
+        char *debug_subsys = NULL;
         char *root_driver = "native";
         char *lustre_driver = "llite";
         char *root_path = "/";
         unsigned mntflgs = 0;
-
        int err;
 
-        /* consider tha case of starting multiple liblustre instances
-         * at a same time on single node.
-         */
-        srand(time(NULL) + getpid());
-
-        signal(SIGUSR1, sighandler_USR1);
-
        lustre_path = getenv(ENV_LUSTRE_MNTPNT);
        if (!lustre_path) {
                 lustre_path = "/mnt/lustre";
        }
 
+        /* mount target */
         target = getenv(ENV_LUSTRE_MNTTGT);
         if (!target) {
-                dumpfile = getenv(ENV_LUSTRE_DUMPFILE);
-                if (!dumpfile) {
-                        CERROR("Neither mount target, nor dumpfile\n");
-                        exit(1);
-                }
-                g_zconf = 0;
-                printf("LibLustre: mount point %s, dumpfile %s\n",
-                        lustre_path, dumpfile);
-        } else {
-                if (ll_parse_mount_target(target,
-                                          &g_zconf_mdsnid,
-                                          &g_zconf_mdsname,
-                                          &g_zconf_profile)) {
-                        CERROR("mal-formed target %s \n", target);
-                        exit(1);
-                }
-                g_zconf = 1;
-                printf("LibLustre: mount point %s, target %s\n",
-                        lustre_path, target);
+                printf("LibLustre: no mount target specified\n");
+                exit(1);
         }
+        if (ll_parse_mount_target(target,
+                                  &g_zconf_mdsnid,
+                                  &g_zconf_mdsname,
+                                  &g_zconf_profile)) {
+                CERROR("mal-formed target %s \n", target);
+                exit(1);
+        }
+        if (!g_zconf_mdsnid || !g_zconf_mdsname || !g_zconf_profile) {
+                printf("Liblustre: invalid target %s\n", target);
+                exit(1);
+        }
+        printf("LibLustre: mount point %s, target %s\n",
+                lustre_path, target);
 
         timeout = getenv(ENV_LUSTRE_TIMEOUT);
         if (timeout) {
-                obd_timeout = (unsigned int) atoi(timeout);
+                obd_timeout = (unsigned int) strtol(timeout, NULL, 0);
                 printf("LibLustre: set obd timeout as %u seconds\n",
                         obd_timeout);
         }
 
-       if (_sysio_init() != 0) {
-               perror("init sysio");
-               exit(1);
-       }
+        /* debug masks */
+        debug_mask = getenv(ENV_LUSTRE_DEBUG_MASK);
+        if (debug_mask)
+                portal_debug = (unsigned int) strtol(debug_mask, NULL, 0);
+
+        debug_subsys = getenv(ENV_LUSTRE_DEBUG_SUBSYS);
+        if (debug_subsys)
+                portal_subsystem_debug =
+                                (unsigned int) strtol(debug_subsys, NULL, 0);
 
-        /* cygwin don't need native driver */
-#ifndef __CYGWIN__
+
+#ifdef INIT_SYSIO
+        /* initialize libsysio & mount rootfs */
+        if (_sysio_init()) {
+                perror("init sysio");
+                exit(1);
+        }
         _sysio_native_init();
-#endif
 
-       err = _sysio_mount_root(root_path, root_driver, mntflgs, NULL);
-       if (err) {
-               perror(root_driver);
-               exit(1);
-       }
+        err = _sysio_mount_root(root_path, root_driver, mntflgs, NULL);
+        if (err) {
+                perror(root_driver);
+                exit(1);
+        }
 
-#if 1
-       portal_debug = 0;
-       portal_subsystem_debug = 0;
-#endif
-       err = lllib_init(dumpfile);
-       if (err) {
-               perror("init llite driver");
-               exit(1);
-       }       
+        if (_sysio_lustre_init())
+                exit(1);
+#endif /* INIT_SYSIO */
 
         err = mount("/", lustre_path, lustre_driver, mntflgs, NULL);
-       if (err) {
-               errno = -err;
-               perror(lustre_driver);
-               exit(1);
-       }
-
-#if 0
-        __sysio_hook_sys_enter = llu_check_request;
-        __sysio_hook_sys_leave = NULL;
-#endif
+        if (err) {
+                errno = -err;
+                perror(lustre_driver);
+                exit(1);
+        }
 }
 
 void __liblustre_cleanup_(void)
 {
-       _sysio_shutdown();
+        /* user app might chdir to a lustre directory, and leave busy pnode
+         * during finaly libsysio cleanup. here we chdir back to "/".
+         * but it can't fix the situation that liblustre is mounted
+         * at "/".
+         */
+        chdir("/");
+#if 0
+        umount(lustre_path);
+#endif
+        /* we can't call umount here, because libsysio will not cleanup
+         * opening files for us. _sysio_shutdown() will cleanup fds at
+         * first but which will also close the sockets we need for umount
+         * liblutre. this delima lead to another hack in
+         * libsysio/src/file_hack.c FIXME
+         */
+#ifdef INIT_SYSIO
+        _sysio_shutdown();
+        cleanup_lib_portals();
         PtlFini();
+#else
+       /*
+        * don't do any libsysio or low level portals cleanups
+        * platform framework does it
+        */
+       cleanup_lib_portals();
+#endif
 }