X-Git-Url: https://git.whamcloud.com/?p=fs%2Flustre-release.git;a=blobdiff_plain;f=lustre%2Fobdclass%2Fclass_obd.c;h=2747e10eff2681fa154a1649c1b548f55b1caf0a;hp=bb2b2c41e6b55a0a2113549ef84bac1f809efb48;hb=29e98f581ab68ecdfc1b923925eda2ec5b251f3a;hpb=2dc9c16e770415d56839e1996015fec5fab93f29 diff --git a/lustre/obdclass/class_obd.c b/lustre/obdclass/class_obd.c index bb2b2c4..2747e10 100644 --- a/lustre/obdclass/class_obd.c +++ b/lustre/obdclass/class_obd.c @@ -1,116 +1,195 @@ -/* -*- mode: c; c-basic-offset: 8; indent-tabs-mode: nil; -*- - * vim:expandtab:shiftwidth=8:tabstop=8: +/* + * GPL HEADER START * - * Object Devices Class Driver + * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER. * - * Copyright (C) 2001-2003 Cluster File Systems, Inc. + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License version 2 only, + * as published by the Free Software Foundation. * - * This file is part of Lustre, http://www.lustre.org. + * This program 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 version 2 for more details (a copy is included + * in the LICENSE file that accompanied this code). * - * 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. + * You should have received a copy of the GNU General Public License + * version 2 along with this program; If not, see + * http://www.sun.com/software/products/lustre/docs/GPLv2.pdf * - * 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. + * Please contact Sun Microsystems, Inc., 4150 Network Circle, Santa Clara, + * CA 95054 USA or visit www.sun.com if you need additional information or + * have any questions. * - * 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. + * GPL HEADER END + */ +/* + * Copyright (c) 1999, 2010, Oracle and/or its affiliates. All rights reserved. + * Use is subject to license terms. * - * These are the only exported functions, they provide some generic - * infrastructure for managing object devices + * Copyright (c) 2011, 2012, Intel Corporation. + */ +/* + * This file is part of Lustre, http://www.lustre.org/ + * Lustre is a trademark of Sun Microsystems, Inc. */ #define DEBUG_SUBSYSTEM S_CLASS -#ifndef EXPORT_SYMTAB -# define EXPORT_SYMTAB -#endif -#ifdef __KERNEL__ -#include /* for CONFIG_PROC_FS */ -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#include -#else +#ifndef __KERNEL__ # include +#else +# include #endif -#include -#include -#include -#include -#ifdef __KERNEL__ -#include -#include -#endif -#include +#include +#include +#include +#include +#include +#include +#include +#include +#ifdef HAVE_SERVER_SUPPORT +# include +# include +#endif /* HAVE_SERVER_SUPPORT */ +#include "llog_internal.h" #ifndef __KERNEL__ /* liblustre workaround */ -atomic_t portal_kmemory = {0}; +cfs_atomic_t libcfs_kmemory = {0}; #endif -struct semaphore obd_conf_sem; /* serialize configuration commands */ -struct obd_device obd_dev[MAX_OBD_DEVICES]; -struct list_head obd_types; +struct obd_device *obd_devs[MAX_OBD_DEVICES]; +EXPORT_SYMBOL(obd_devs); +cfs_list_t obd_types; +DEFINE_RWLOCK(obd_dev_lock); + +__u64 obd_max_pages = 0; +__u64 obd_max_alloc = 0; #ifndef __KERNEL__ -atomic_t obd_memory; -int obd_memmax; +__u64 obd_alloc; +__u64 obd_pages; #endif - -int proc_version; +DEFINE_SPINLOCK(obd_updatemax_lock); /* The following are visible and mutable through /proc/sys/lustre/. */ -unsigned int obd_fail_loc; -unsigned int obd_timeout = 100; -char obd_lustre_upcall[128] = "DEFAULT"; /* or NONE or /full/path/to/upcall */ -unsigned int obd_sync_filter; /* = 0, don't sync by default */ - -DECLARE_WAIT_QUEUE_HEAD(obd_race_waitq); - -#ifdef __KERNEL__ -/* opening /dev/obd */ -static int obd_class_open(struct inode * inode, struct file * file) +unsigned int obd_alloc_fail_rate = 0; +EXPORT_SYMBOL(obd_alloc_fail_rate); +unsigned int obd_debug_peer_on_timeout; +EXPORT_SYMBOL(obd_debug_peer_on_timeout); +unsigned int obd_dump_on_timeout; +EXPORT_SYMBOL(obd_dump_on_timeout); +unsigned int obd_dump_on_eviction; +EXPORT_SYMBOL(obd_dump_on_eviction); +unsigned int obd_max_dirty_pages = 256; +EXPORT_SYMBOL(obd_max_dirty_pages); +cfs_atomic_t obd_dirty_pages; +EXPORT_SYMBOL(obd_dirty_pages); +unsigned int obd_timeout = OBD_TIMEOUT_DEFAULT; /* seconds */ +EXPORT_SYMBOL(obd_timeout); +unsigned int ldlm_timeout = LDLM_TIMEOUT_DEFAULT; /* seconds */ +EXPORT_SYMBOL(ldlm_timeout); +unsigned int obd_timeout_set; +EXPORT_SYMBOL(obd_timeout_set); +unsigned int ldlm_timeout_set; +EXPORT_SYMBOL(ldlm_timeout_set); +/* Adaptive timeout defs here instead of ptlrpc module for /proc/sys/ access */ +unsigned int at_min = 0; +EXPORT_SYMBOL(at_min); +unsigned int at_max = 600; +EXPORT_SYMBOL(at_max); +unsigned int at_history = 600; +EXPORT_SYMBOL(at_history); +int at_early_margin = 5; +EXPORT_SYMBOL(at_early_margin); +int at_extra = 30; +EXPORT_SYMBOL(at_extra); + +cfs_atomic_t obd_dirty_transit_pages; +EXPORT_SYMBOL(obd_dirty_transit_pages); + +char obd_jobid_var[JOBSTATS_JOBID_VAR_MAX_LEN + 1] = JOBSTATS_DISABLE; +EXPORT_SYMBOL(obd_jobid_var); + +/* Get jobid of current process by reading the environment variable + * stored in between the "env_start" & "env_end" of task struct. + * + * TODO: + * It's better to cache the jobid for later use if there is any + * efficient way, the cl_env code probably could be reused for this + * purpose. + * + * If some job scheduler doesn't store jobid in the "env_start/end", + * then an upcall could be issued here to get the jobid by utilizing + * the userspace tools/api. Then, the jobid must be cached. + */ +int lustre_get_jobid(char *jobid) { - ENTRY; - - PORTAL_MODULE_USE; - RETURN(0); + int jobid_len = JOBSTATS_JOBID_SIZE; + int rc = 0; + ENTRY; + + memset(jobid, 0, JOBSTATS_JOBID_SIZE); + /* Jobstats isn't enabled */ + if (strcmp(obd_jobid_var, JOBSTATS_DISABLE) == 0) + RETURN(0); + + /* Use process name + fsuid as jobid */ + if (strcmp(obd_jobid_var, JOBSTATS_PROCNAME_UID) == 0) { + snprintf(jobid, JOBSTATS_JOBID_SIZE, "%s.%u", + cfs_curproc_comm(), cfs_curproc_fsuid()); + RETURN(0); + } + + rc = cfs_get_environ(obd_jobid_var, jobid, &jobid_len); + if (rc) { + if (rc == -EOVERFLOW) { + /* For the PBS_JOBID and LOADL_STEP_ID keys (which are + * variable length strings instead of just numbers), it + * might make sense to keep the unique parts for JobID, + * instead of just returning an error. That means a + * larger temp buffer for cfs_get_environ(), then + * truncating the string at some separator to fit into + * the specified jobid_len. Fix later if needed. */ + static bool printed; + if (unlikely(!printed)) { + LCONSOLE_ERROR_MSG(0x16b, "%s value too large " + "for JobID buffer (%d)\n", + obd_jobid_var, jobid_len); + printed = true; + } + } else { + CDEBUG((rc == -ENOENT || rc == -EINVAL || + rc == -EDEADLK) ? D_INFO : D_ERROR, + "Get jobid for (%s) failed: rc = %d\n", + obd_jobid_var, rc); + } + } + RETURN(rc); } +EXPORT_SYMBOL(lustre_get_jobid); -/* closing /dev/obd */ -static int obd_class_release(struct inode * inode, struct file * file) +int obd_alloc_fail(const void *ptr, const char *name, const char *type, + size_t size, const char *file, int line) { - ENTRY; - - PORTAL_MODULE_UNUSE; - RETURN(0); + if (ptr == NULL || + (cfs_rand() & OBD_ALLOC_FAIL_MASK) < obd_alloc_fail_rate) { + CERROR("%s%salloc of %s ("LPU64" bytes) failed at %s:%d\n", + ptr ? "force " :"", type, name, (__u64)size, file, + line); + CERROR(LPU64" total bytes and "LPU64" total pages " + "("LPU64" bytes) allocated by Lustre, " + "%d total bytes by LNET\n", + obd_memory_sum(), + obd_pages_sum() << CFS_PAGE_SHIFT, + obd_pages_sum(), + cfs_atomic_read(&libcfs_kmemory)); + return 1; + } + return 0; } -#endif +EXPORT_SYMBOL(obd_alloc_fail); static inline void obd_data2conn(struct lustre_handle *conn, struct obd_ioctl_data *data) @@ -125,11 +204,12 @@ static inline void obd_conn2data(struct obd_ioctl_data *data, data->ioc_cookie = conn->cookie; } -int class_resolve_dev_name(uint32_t len, char *name) +int class_resolve_dev_name(__u32 len, const char *name) { int rc; int dev; + ENTRY; if (!len || !name) { CERROR("No name passed,!\n"); GOTO(out, rc = -EINVAL); @@ -157,69 +237,45 @@ int class_handle_ioctl(unsigned int cmd, unsigned long arg) { char *buf = NULL; struct obd_ioctl_data *data; - struct portals_debug_ioctl_data *debug_data; + struct libcfs_debug_ioctl_data *debug_data; struct obd_device *obd = NULL; - int err = 0, len = 0, serialised = 0; + int err = 0, len = 0; ENTRY; - if (current->fsuid != 0) - RETURN(err = -EACCES); - - if ((cmd & 0xffffff00) == ((int)'T') << 8) /* ignore all tty ioctls */ - RETURN(err = -ENOTTY); - /* only for debugging */ - if (cmd == PTL_IOC_DEBUG_MASK) { - debug_data = (struct portals_debug_ioctl_data*)arg; - portal_subsystem_debug = debug_data->subs; - portal_debug = debug_data->debug; + if (cmd == LIBCFS_IOC_DEBUG_MASK) { + debug_data = (struct libcfs_debug_ioctl_data*)arg; + libcfs_subsystem_debug = debug_data->subs; + libcfs_debug = debug_data->debug; return 0; } - switch (cmd) { - case OBD_IOC_BRW_WRITE: - case OBD_IOC_BRW_READ: - case OBD_IOC_GETATTR: - case ECHO_IOC_ENQUEUE: - case ECHO_IOC_CANCEL: - case OBD_IOC_CLIENT_RECOVER: - case OBD_IOC_CATLOGLIST: - case OBD_IOC_LLOG_INFO: - case OBD_IOC_LLOG_PRINT: - case OBD_IOC_LLOG_CANCEL: - case OBD_IOC_LLOG_CHECK: - case OBD_IOC_LLOG_REMOVE: - break; - default: - down(&obd_conf_sem); - serialised = 1; - break; - } - - CDEBUG(D_IOCTL, "cmd = %x, obd = %p\n", cmd, obd); + CDEBUG(D_IOCTL, "cmd = %x\n", cmd); if (obd_ioctl_getdata(&buf, &len, (void *)arg)) { CERROR("OBD ioctl: data error\n"); - GOTO(out, err = -EINVAL); + RETURN(-EINVAL); } data = (struct obd_ioctl_data *)buf; switch (cmd) { case OBD_IOC_PROCESS_CFG: { - char *buf; struct lustre_cfg *lcfg; if (!data->ioc_plen1 || !data->ioc_pbuf1) { CERROR("No config buffer passed!\n"); GOTO(out, err = -EINVAL); } - err = lustre_cfg_getdata(&buf, data->ioc_plen1, - data->ioc_pbuf1, 0); - if (err) - GOTO(out, err); - lcfg = (struct lustre_cfg* ) buf; - - err = class_process_config(lcfg); - lustre_cfg_freedata(buf, data->ioc_plen1); + OBD_ALLOC(lcfg, data->ioc_plen1); + if (lcfg == NULL) + GOTO(out, err = -ENOMEM); + err = cfs_copy_from_user(lcfg, data->ioc_pbuf1, + data->ioc_plen1); + if (!err) + err = lustre_cfg_sanity_check(lcfg, data->ioc_plen1); + if (!err) + err = class_process_config(lcfg); + + OBD_FREE(lcfg, data->ioc_plen1); GOTO(out, err); } @@ -237,7 +293,7 @@ int class_handle_ioctl(unsigned int cmd, unsigned long arg) memcpy(data->ioc_bulk, BUILD_VERSION, strlen(BUILD_VERSION) + 1); - err = copy_to_user((void *)arg, data, len); + err = obd_ioctl_popdata((void *)arg, data, len); if (err) err = -EFAULT; GOTO(out, err); @@ -254,7 +310,7 @@ int class_handle_ioctl(unsigned int cmd, unsigned long arg) if (dev < 0) GOTO(out, err = -EINVAL); - err = copy_to_user((void *)arg, data, sizeof(*data)); + err = obd_ioctl_popdata((void *)arg, data, sizeof(*data)); if (err) err = -EFAULT; GOTO(out, err); @@ -288,30 +344,75 @@ int class_handle_ioctl(unsigned int cmd, unsigned long arg) CDEBUG(D_IOCTL, "device name %s, dev %d\n", data->ioc_inlbuf1, dev); - err = copy_to_user((void *)arg, data, sizeof(*data)); + err = obd_ioctl_popdata((void *)arg, data, sizeof(*data)); if (err) err = -EFAULT; GOTO(out, err); } - case OBD_IOC_CLOSE_UUID: { - ptl_nid_t peer_nid; - __u32 peer_nal; - CDEBUG(D_IOCTL, "closing all connections to uuid %s\n", + CDEBUG(D_IOCTL, "closing all connections to uuid %s (NOOP)\n", data->ioc_inlbuf1); - lustre_uuid_to_peer(data->ioc_inlbuf1, &peer_nal, &peer_nid); + GOTO(out, err = 0); + } + + case OBD_IOC_GETDEVICE: { + int index = data->ioc_count; + char *status, *str; + + if (!data->ioc_inlbuf1) { + CERROR("No buffer passed in ioctl\n"); + GOTO(out, err = -EINVAL); + } + if (data->ioc_inllen1 < 128) { + CERROR("ioctl buffer too small to hold version\n"); + GOTO(out, err = -EINVAL); + } + + obd = class_num2obd(index); + if (!obd) + GOTO(out, err = -ENOENT); + + if (obd->obd_stopping) + status = "ST"; + else if (obd->obd_set_up) + status = "UP"; + else if (obd->obd_attached) + status = "AT"; + else + status = "--"; + str = (char *)data->ioc_bulk; + snprintf(str, len - sizeof(*data), "%3d %s %s %s %s %d", + (int)index, status, obd->obd_type->typ_name, + obd->obd_name, obd->obd_uuid.uuid, + cfs_atomic_read(&obd->obd_refcount)); + err = obd_ioctl_popdata((void *)arg, data, len); + GOTO(out, err = 0); } } - if (data->ioc_dev >= MAX_OBD_DEVICES) { + if (data->ioc_dev == OBD_DEV_BY_DEVNAME) { + if (data->ioc_inllen4 <= 0 || data->ioc_inlbuf4 == NULL) + GOTO(out, err = -EINVAL); + if (strnlen(data->ioc_inlbuf4, MAX_OBD_NAME) >= MAX_OBD_NAME) + GOTO(out, err = -EINVAL); + obd = class_name2obd(data->ioc_inlbuf4); + } else if (data->ioc_dev < class_devno_max()) { + obd = class_num2obd(data->ioc_dev); + } else { CERROR("OBD ioctl: No device\n"); GOTO(out, err = -EINVAL); - } - obd = &obd_dev[data->ioc_dev]; - if (!(obd && obd->obd_set_up) || obd->obd_stopping) { + } + + if (obd == NULL) { + CERROR("OBD ioctl : No Device %d\n", data->ioc_dev); + GOTO(out, err = -EINVAL); + } + LASSERT(obd->obd_magic == OBD_DEVICE_MAGIC); + + if (!obd->obd_set_up || obd->obd_stopping) { CERROR("OBD ioctl: device not setup %d \n", data->ioc_dev); GOTO(out, err = -EINVAL); } @@ -322,9 +423,8 @@ int class_handle_ioctl(unsigned int cmd, unsigned long arg) CERROR("Device %d not attached\n", obd->obd_minor); GOTO(out, err = -ENODEV); } - CDEBUG(D_IOCTL, - "disabling committed-transno notifications on %d\n", - obd->obd_minor); + CDEBUG(D_HA, "%s: disabling committed-transno notification\n", + obd->obd_name); obd->obd_no_transno = 1; GOTO(out, err = 0); } @@ -334,7 +434,7 @@ int class_handle_ioctl(unsigned int cmd, unsigned long arg) if (err) GOTO(out, err); - err = copy_to_user((void *)arg, data, len); + err = obd_ioctl_popdata((void *)arg, data, len); if (err) err = -EFAULT; GOTO(out, err); @@ -344,332 +444,295 @@ int class_handle_ioctl(unsigned int cmd, unsigned long arg) out: if (buf) obd_ioctl_freedata(buf, len); - if (serialised) - up(&obd_conf_sem); RETURN(err); } /* class_handle_ioctl */ - - -#define OBD_MINOR 241 #ifdef __KERNEL__ -/* to control /dev/obd */ -static int obd_class_ioctl(struct inode *inode, struct file *filp, - unsigned int cmd, unsigned long arg) -{ - return class_handle_ioctl(cmd, arg); -} - -/* declare character device */ -static struct file_operations obd_psdev_fops = { - .owner = THIS_MODULE, - .ioctl = obd_class_ioctl, /* ioctl */ - .open = obd_class_open, /* open */ - .release = obd_class_release, /* release */ -}; - -/* modules setup */ -static struct miscdevice obd_psdev = { - .minor = OBD_MINOR, - .name = "obd_psdev", - .fops = &obd_psdev_fops, -}; +extern cfs_psdev_t obd_psdev; #else void *obd_psdev = NULL; #endif -EXPORT_SYMBOL(obd_dev); -EXPORT_SYMBOL(obdo_cachep); -EXPORT_SYMBOL(obd_fail_loc); -EXPORT_SYMBOL(obd_race_waitq); -EXPORT_SYMBOL(obd_timeout); -EXPORT_SYMBOL(obd_lustre_upcall); -EXPORT_SYMBOL(obd_sync_filter); -EXPORT_SYMBOL(ptlrpc_put_connection_superhack); -EXPORT_SYMBOL(ptlrpc_abort_inflight_superhack); -EXPORT_SYMBOL(proc_lustre_root); - -EXPORT_SYMBOL(class_register_type); -EXPORT_SYMBOL(class_unregister_type); -EXPORT_SYMBOL(class_get_type); -EXPORT_SYMBOL(class_put_type); -EXPORT_SYMBOL(class_name2dev); -EXPORT_SYMBOL(class_name2obd); -EXPORT_SYMBOL(class_uuid2dev); -EXPORT_SYMBOL(class_uuid2obd); -EXPORT_SYMBOL(class_find_client_obd); -EXPORT_SYMBOL(class_devices_in_group); -EXPORT_SYMBOL(__class_export_put); -EXPORT_SYMBOL(class_new_export); -EXPORT_SYMBOL(class_unlink_export); -EXPORT_SYMBOL(class_import_get); -EXPORT_SYMBOL(class_import_put); -EXPORT_SYMBOL(class_new_import); -EXPORT_SYMBOL(class_destroy_import); -EXPORT_SYMBOL(class_connect); -EXPORT_SYMBOL(class_conn2export); -EXPORT_SYMBOL(class_exp2obd); -EXPORT_SYMBOL(class_conn2obd); -EXPORT_SYMBOL(class_exp2cliimp); -EXPORT_SYMBOL(class_conn2cliimp); -EXPORT_SYMBOL(class_disconnect); -EXPORT_SYMBOL(class_disconnect_exports); - -EXPORT_SYMBOL(oig_init); -EXPORT_SYMBOL(oig_release); -EXPORT_SYMBOL(oig_add_one); -EXPORT_SYMBOL(oig_wait); -EXPORT_SYMBOL(oig_complete_one); - -/* uuid.c */ -EXPORT_SYMBOL(class_uuid_unparse); -EXPORT_SYMBOL(lustre_uuid_to_peer); - -EXPORT_SYMBOL(class_handle_hash); -EXPORT_SYMBOL(class_handle_unhash); -EXPORT_SYMBOL(class_handle2object); - -/* config.c */ -EXPORT_SYMBOL(class_get_profile); -EXPORT_SYMBOL(class_del_profile); -EXPORT_SYMBOL(class_process_config); -EXPORT_SYMBOL(class_config_parse_llog); -EXPORT_SYMBOL(class_config_dump_llog); -EXPORT_SYMBOL(class_attach); -EXPORT_SYMBOL(class_setup); -EXPORT_SYMBOL(class_cleanup); -EXPORT_SYMBOL(class_detach); - -/* mea.c */ -EXPORT_SYMBOL(mea_name2idx); -EXPORT_SYMBOL(raw_name2idx); - -#ifdef LPROCFS -int obd_proc_read_version(char *page, char **start, off_t off, int count, - int *eof, void *data) -{ - *eof = 1; - return snprintf(page, count, "%s\n", BUILD_VERSION); -} - -int obd_proc_read_kernel_version(char *page, char **start, off_t off, int count, - int *eof, void *data) -{ - *eof = 1; - return snprintf(page, count, "%u\n", LUSTRE_KERNEL_VERSION); -} - -int obd_proc_read_pinger(char *page, char **start, off_t off, int count, - int *eof, void *data) -{ - *eof = 1; - return snprintf(page, count, "%s\n", -#ifdef ENABLE_PINGER - "on" -#else - "off" -#endif - ); -} - -/* Root for /proc/fs/lustre */ -struct proc_dir_entry *proc_lustre_root = NULL; -struct lprocfs_vars lprocfs_base[] = { - { "version", obd_proc_read_version, NULL, NULL }, - { "kernel_version", obd_proc_read_kernel_version, NULL, NULL }, - { "pinger", obd_proc_read_pinger, NULL, NULL }, - { 0 } -}; - -static void *obd_device_list_seq_start(struct seq_file *p, loff_t*pos) +#define OBD_INIT_CHECK +#ifdef OBD_INIT_CHECK +int obd_init_checks(void) { - if (*pos >= MAX_OBD_DEVICES) - return NULL; - return &obd_dev[*pos]; -} + __u64 u64val, div64val; + char buf[64]; + int len, ret = 0; -static void obd_device_list_seq_stop(struct seq_file *p, void *v) -{ -} + CDEBUG(D_INFO, "LPU64=%s, LPD64=%s, LPX64=%s\n", LPU64, LPD64, LPX64); -static void *obd_device_list_seq_next(struct seq_file *p, void *v, loff_t *pos) -{ - ++*pos; - if (*pos >= MAX_OBD_DEVICES) - return NULL; - return &obd_dev[*pos]; -} + CDEBUG(D_INFO, "OBD_OBJECT_EOF = "LPX64"\n", (__u64)OBD_OBJECT_EOF); -static int obd_device_list_seq_show(struct seq_file *p, void *v) -{ - struct obd_device *obd = (struct obd_device *)v; - int index = obd - &obd_dev[0]; - char *status; - - if (!obd->obd_type) - return 0; - if (obd->obd_stopping) - status = "ST"; - else if (obd->obd_set_up) - status = "UP"; - else if (obd->obd_attached) - status = "AT"; - else - status = "--"; - - return seq_printf(p, "%3d %s %s %s %s %d\n", - (int)index, status, obd->obd_type->typ_name, - obd->obd_name, obd->obd_uuid.uuid, - atomic_read(&obd->obd_refcount)); -} - -struct seq_operations obd_device_list_sops = { - .start = obd_device_list_seq_start, - .stop = obd_device_list_seq_stop, - .next = obd_device_list_seq_next, - .show = obd_device_list_seq_show, -}; - -static int obd_device_list_open(struct inode *inode, struct file *file) -{ - struct proc_dir_entry *dp = PDE(inode); - struct seq_file *seq; - int rc = seq_open(file, &obd_device_list_sops); - - if (rc) - return rc; + u64val = OBD_OBJECT_EOF; + CDEBUG(D_INFO, "u64val OBD_OBJECT_EOF = "LPX64"\n", u64val); + if (u64val != OBD_OBJECT_EOF) { + CERROR("__u64 "LPX64"(%d) != 0xffffffffffffffff\n", + u64val, (int)sizeof(u64val)); + ret = -EINVAL; + } + len = snprintf(buf, sizeof(buf), LPX64, u64val); + if (len != 18) { + CWARN("LPX64 wrong length! strlen(%s)=%d != 18\n", buf, len); + ret = -EINVAL; + } - seq = file->private_data; - seq->private = dp->data; + div64val = OBD_OBJECT_EOF; + CDEBUG(D_INFO, "u64val OBD_OBJECT_EOF = "LPX64"\n", u64val); + if (u64val != OBD_OBJECT_EOF) { + CERROR("__u64 "LPX64"(%d) != 0xffffffffffffffff\n", + u64val, (int)sizeof(u64val)); + ret = -EOVERFLOW; + } + if (u64val >> 8 != OBD_OBJECT_EOF >> 8) { + CERROR("__u64 "LPX64"(%d) != 0xffffffffffffffff\n", + u64val, (int)sizeof(u64val)); + return -EOVERFLOW; + } + if (do_div(div64val, 256) != (u64val & 255)) { + CERROR("do_div("LPX64",256) != "LPU64"\n", u64val, u64val &255); + return -EOVERFLOW; + } + if (u64val >> 8 != div64val) { + CERROR("do_div("LPX64",256) "LPU64" != "LPU64"\n", + u64val, div64val, u64val >> 8); + return -EOVERFLOW; + } + len = snprintf(buf, sizeof(buf), LPX64, u64val); + if (len != 18) { + CWARN("LPX64 wrong length! strlen(%s)=%d != 18\n", buf, len); + ret = -EINVAL; + } + len = snprintf(buf, sizeof(buf), LPU64, u64val); + if (len != 20) { + CWARN("LPU64 wrong length! strlen(%s)=%d != 20\n", buf, len); + ret = -EINVAL; + } + len = snprintf(buf, sizeof(buf), LPD64, u64val); + if (len != 2) { + CWARN("LPD64 wrong length! strlen(%s)=%d != 2\n", buf, len); + ret = -EINVAL; + } + if ((u64val & ~CFS_PAGE_MASK) >= CFS_PAGE_SIZE) { + CWARN("mask failed: u64val "LPU64" >= "LPU64"\n", u64val, + (__u64)CFS_PAGE_SIZE); + ret = -EINVAL; + } - return 0; + return ret; } - -struct file_operations obd_device_list_fops = { - .owner = THIS_MODULE, - .open = obd_device_list_open, - .read = seq_read, - .llseek = seq_lseek, - .release = seq_release, -}; +#else +#define obd_init_checks() do {} while(0) #endif +extern spinlock_t obd_types_lock; +extern int class_procfs_init(void); +extern int class_procfs_clean(void); + #ifdef __KERNEL__ static int __init init_obdclass(void) #else int init_obdclass(void) #endif { - struct obd_device *obd; -#ifdef LPROCFS - struct proc_dir_entry *entry; + int i, err; +#ifdef __KERNEL__ + int lustre_register_fs(void); + + for (i = CAPA_SITE_CLIENT; i < CAPA_SITE_MAX; i++) + CFS_INIT_LIST_HEAD(&capa_list[i]); #endif - int err; - int i; - printk(KERN_INFO "Lustre: OBD class driver Build Version: " - BUILD_VERSION", info@clusterfs.com\n"); + LCONSOLE_INFO("Lustre: Build Version: "BUILD_VERSION"\n"); + + spin_lock_init(&obd_types_lock); + obd_zombie_impexp_init(); +#ifdef LPROCFS + obd_memory = lprocfs_alloc_stats(OBD_STATS_NUM, + LPROCFS_STATS_FLAG_NONE | + LPROCFS_STATS_FLAG_IRQ_SAFE); + if (obd_memory == NULL) { + CERROR("kmalloc of 'obd_memory' failed\n"); + RETURN(-ENOMEM); + } + + lprocfs_counter_init(obd_memory, OBD_MEMORY_STAT, + LPROCFS_CNTR_AVGMINMAX, + "memused", "bytes"); + lprocfs_counter_init(obd_memory, OBD_MEMORY_PAGES_STAT, + LPROCFS_CNTR_AVGMINMAX, + "pagesused", "pages"); +#endif + err = obd_init_checks(); + if (err == -EOVERFLOW) + return err; class_init_uuidlist(); err = class_handle_init(); if (err) return err; - sema_init(&obd_conf_sem, 1); - INIT_LIST_HEAD(&obd_types); + CFS_INIT_LIST_HEAD(&obd_types); - err = misc_register(&obd_psdev); + err = cfs_psdev_register(&obd_psdev); if (err) { - CERROR("cannot register %d err %d\n", OBD_MINOR, err); + CERROR("cannot register %d err %d\n", OBD_DEV_MINOR, err); return err; } - /* This struct is already zerod for us (static global) */ - for (i = 0, obd = obd_dev; i < MAX_OBD_DEVICES; i++, obd++) - obd->obd_minor = i; + /* This struct is already zeroed for us (static global) */ + for (i = 0; i < class_devno_max(); i++) + obd_devs[i] = NULL; + + /* Default the dirty page cache cap to 1/2 of system memory. + * For clients with less memory, a larger fraction is needed + * for other purposes (mostly for BGL). */ + if (cfs_num_physpages <= 512 << (20 - CFS_PAGE_SHIFT)) + obd_max_dirty_pages = cfs_num_physpages / 4; + else + obd_max_dirty_pages = cfs_num_physpages / 2; err = obd_init_caches(); if (err) return err; +#ifdef __KERNEL__ + err = class_procfs_init(); + if (err) + return err; +#endif + + err = lu_global_init(); + if (err) + return err; + + err = cl_global_init(); + if (err != 0) + return err; + +#if defined(__KERNEL__) && defined(HAVE_SERVER_SUPPORT) + err = dt_global_init(); + if (err != 0) + return err; + + err = lu_ucred_global_init(); + if (err != 0) + return err; +#endif + + err = llog_info_init(); + if (err) + return err; #ifdef __KERNEL__ - obd_sysctl_init(); + err = lustre_register_fs(); #endif + return err; +} + +void obd_update_maxusage(void) +{ + __u64 max1, max2; + + max1 = obd_pages_sum(); + max2 = obd_memory_sum(); + + spin_lock(&obd_updatemax_lock); + if (max1 > obd_max_pages) + obd_max_pages = max1; + if (max2 > obd_max_alloc) + obd_max_alloc = max2; + spin_unlock(&obd_updatemax_lock); +} +EXPORT_SYMBOL(obd_update_maxusage); + #ifdef LPROCFS - proc_lustre_root = proc_mkdir("lustre", proc_root_fs); - if (!proc_lustre_root) { - printk(KERN_ERR - "LustreError: error registering /proc/fs/lustre\n"); - RETURN(-ENOMEM); - } - proc_version = lprocfs_add_vars(proc_lustre_root, lprocfs_base, NULL); - entry = create_proc_entry("devices", 0444, proc_lustre_root); - if (entry == NULL) { - printk(KERN_ERR "LustreError: error registering " - "/proc/fs/lustre/devices\n"); - lprocfs_remove(proc_lustre_root); - RETURN(-ENOMEM); - } - entry->proc_fops = &obd_device_list_fops; -#endif - return 0; +__u64 obd_memory_max(void) +{ + __u64 ret; + + spin_lock(&obd_updatemax_lock); + ret = obd_max_alloc; + spin_unlock(&obd_updatemax_lock); + + return ret; +} +EXPORT_SYMBOL(obd_memory_max); + +__u64 obd_pages_max(void) +{ + __u64 ret; + + spin_lock(&obd_updatemax_lock); + ret = obd_max_pages; + spin_unlock(&obd_updatemax_lock); + + return ret; } +EXPORT_SYMBOL(obd_pages_max); +#endif /* liblustre doesn't call cleanup_obdclass, apparently. we carry on in this * ifdef to the end of the file to cover module and versioning goo.*/ #ifdef __KERNEL__ - static void cleanup_obdclass(void) { int i; - int leaked; + int lustre_unregister_fs(void); + __u64 memory_leaked, pages_leaked; + __u64 memory_max, pages_max; ENTRY; - misc_deregister(&obd_psdev); - for (i = 0; i < MAX_OBD_DEVICES; i++) { - struct obd_device *obd = &obd_dev[i]; - if (obd->obd_type && obd->obd_set_up && + lustre_unregister_fs(); + + cfs_psdev_deregister(&obd_psdev); + for (i = 0; i < class_devno_max(); i++) { + struct obd_device *obd = class_num2obd(i); + if (obd && obd->obd_set_up && OBT(obd) && OBP(obd, detach)) { /* XXX should this call generic detach otherwise? */ + LASSERT(obd->obd_magic == OBD_DEVICE_MAGIC); OBP(obd, detach)(obd); } } + llog_info_fini(); +#ifdef HAVE_SERVER_SUPPORT + lu_ucred_global_fini(); + dt_global_fini(); +#endif + cl_global_fini(); + lu_global_fini(); obd_cleanup_caches(); obd_sysctl_clean(); -#ifdef LPROCFS - if (proc_lustre_root) { - lprocfs_remove(proc_lustre_root); - proc_lustre_root = NULL; - } -#endif + + class_procfs_clean(); class_handle_cleanup(); class_exit_uuidlist(); + obd_zombie_impexp_stop(); - leaked = atomic_read(&obd_memory); - CDEBUG(leaked ? D_ERROR : D_INFO, - "obd mem max: %d leaked: %d\n", obd_memmax, leaked); + memory_leaked = obd_memory_sum(); + pages_leaked = obd_pages_sum(); + + memory_max = obd_memory_max(); + pages_max = obd_pages_max(); + + lprocfs_free_stats(&obd_memory); + CDEBUG((memory_leaked) ? D_ERROR : D_INFO, + "obd_memory max: "LPU64", leaked: "LPU64"\n", + memory_max, memory_leaked); + CDEBUG((pages_leaked) ? D_ERROR : D_INFO, + "obd_memory_pages max: "LPU64", leaked: "LPU64"\n", + pages_max, pages_leaked); EXIT; } -/* Check that we're building against the appropriate version of the Lustre - * kernel patch */ -#include -#define LUSTRE_MIN_VERSION 32 -#define LUSTRE_MAX_VERSION 36 -#if (LUSTRE_KERNEL_VERSION < LUSTRE_MIN_VERSION) -# error Cannot continue: Your Lustre kernel patch is older than the sources -#elif (LUSTRE_KERNEL_VERSION > LUSTRE_MAX_VERSION) -# error Cannot continue: Your Lustre sources are older than the kernel patch -#endif - -MODULE_AUTHOR("Cluster File Systems, Inc. "); +MODULE_AUTHOR("Sun Microsystems, Inc. "); MODULE_DESCRIPTION("Lustre Class Driver Build Version: " BUILD_VERSION); MODULE_LICENSE("GPL"); -module_init(init_obdclass); -module_exit(cleanup_obdclass); +cfs_module(obdclass, LUSTRE_VERSION_STRING, init_obdclass, cleanup_obdclass); #endif