*
* 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
- *
- * 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.
+ * http://www.gnu.org/licenses/gpl-2.0.html
*
* GPL HEADER END
*/
* Copyright (c) 2007, 2010, Oracle and/or its affiliates. All rights reserved.
* Use is subject to license terms.
*
- * Copyright (c) 2011, 2013, Intel Corporation.
+ * Copyright (c) 2011, 2016, Intel Corporation.
*/
/*
* This file is part of Lustre, http://www.lustre.org/
#define DEBUG_SUBSYSTEM S_FLD
-#ifdef __KERNEL__
-# include <libcfs/libcfs.h>
-# include <linux/module.h>
-# include <asm/div64.h>
-#else /* __KERNEL__ */
-# include <liblustre.h>
-# include <libcfs/list.h>
-#endif
+#include <libcfs/libcfs.h>
+#include <linux/module.h>
+#include <linux/math64.h>
#include <obd.h>
#include <obd_class.h>
#include <lustre_mdc.h>
#include "fld_internal.h"
-/* TODO: these 3 functions are copies of flow-control code from mdc_lib.c
- * It should be common thing. The same about mdc RPC lock */
-static int fld_req_avail(struct client_obd *cli, struct mdc_cache_waiter *mcw)
-{
- int rc;
- ENTRY;
- client_obd_list_lock(&cli->cl_loi_list_lock);
- rc = cfs_list_empty(&mcw->mcw_entry);
- client_obd_list_unlock(&cli->cl_loi_list_lock);
- RETURN(rc);
-};
-
-static void fld_enter_request(struct client_obd *cli)
+static int fld_rrb_hash(struct lu_client_fld *fld, u64 seq)
{
- struct mdc_cache_waiter mcw;
- struct l_wait_info lwi = { 0 };
-
- client_obd_list_lock(&cli->cl_loi_list_lock);
- if (cli->cl_r_in_flight >= cli->cl_max_rpcs_in_flight) {
- cfs_list_add_tail(&mcw.mcw_entry, &cli->cl_cache_waiters);
- init_waitqueue_head(&mcw.mcw_waitq);
- client_obd_list_unlock(&cli->cl_loi_list_lock);
- l_wait_event(mcw.mcw_waitq, fld_req_avail(cli, &mcw), &lwi);
- } else {
- cli->cl_r_in_flight++;
- client_obd_list_unlock(&cli->cl_loi_list_lock);
- }
-}
-
-static void fld_exit_request(struct client_obd *cli)
-{
- cfs_list_t *l, *tmp;
- struct mdc_cache_waiter *mcw;
-
- client_obd_list_lock(&cli->cl_loi_list_lock);
- cli->cl_r_in_flight--;
- cfs_list_for_each_safe(l, tmp, &cli->cl_cache_waiters) {
-
- if (cli->cl_r_in_flight >= cli->cl_max_rpcs_in_flight) {
- /* No free request slots anymore */
- break;
- }
-
- mcw = cfs_list_entry(l, struct mdc_cache_waiter, mcw_entry);
- cfs_list_del_init(&mcw->mcw_entry);
- cli->cl_r_in_flight++;
- wake_up(&mcw->mcw_waitq);
- }
- client_obd_list_unlock(&cli->cl_loi_list_lock);
-}
-
-static int fld_rrb_hash(struct lu_client_fld *fld,
- seqno_t seq)
-{
- LASSERT(fld->lcf_count > 0);
- return do_div(seq, fld->lcf_count);
+ LASSERT(fld->lcf_count > 0);
+ return do_div(seq, fld->lcf_count);
}
static struct lu_fld_target *
-fld_rrb_scan(struct lu_client_fld *fld, seqno_t seq)
+fld_rrb_scan(struct lu_client_fld *fld, u64 seq)
{
struct lu_fld_target *target;
int hash;
else
hash = 0;
- cfs_list_for_each_entry(target, &fld->lcf_targets, ft_chain) {
+again:
+ list_for_each_entry(target, &fld->lcf_targets, ft_chain) {
if (target->ft_idx == hash)
RETURN(target);
}
- CERROR("%s: Can't find target by hash %d (seq "LPX64"). "
+ if (hash != 0) {
+ /* It is possible the remote target(MDT) are not connected to
+ * with client yet, so we will refer this to MDT0, which should
+ * be connected during mount */
+ hash = 0;
+ goto again;
+ }
+
+ CERROR("%s: Can't find target by hash %d (seq %#llx). "
"Targets (%d):\n", fld->lcf_name, hash, seq,
fld->lcf_count);
- cfs_list_for_each_entry(target, &fld->lcf_targets, ft_chain) {
+ list_for_each_entry(target, &fld->lcf_targets, ft_chain) {
const char *srv_name = target->ft_srv != NULL ?
target->ft_srv->lsf_name : "<null>";
const char *exp_name = target->ft_exp != NULL ?
(char *)target->ft_exp->exp_obd->obd_uuid.uuid :
"<null>";
- CERROR(" exp: 0x%p (%s), srv: 0x%p (%s), idx: "LPU64"\n",
+ CERROR(" exp: 0x%p (%s), srv: 0x%p (%s), idx: %llu\n",
target->ft_exp, exp_name, target->ft_srv,
srv_name, target->ft_idx);
}
.fh_scan_func = fld_rrb_scan
},
{
- 0,
+ NULL,
}
};
static struct lu_fld_target *
-fld_client_get_target(struct lu_client_fld *fld, seqno_t seq)
+fld_client_get_target(struct lu_client_fld *fld, u64 seq)
{
struct lu_fld_target *target;
ENTRY;
spin_unlock(&fld->lcf_lock);
if (target != NULL) {
- CDEBUG(D_INFO, "%s: Found target (idx "LPU64
- ") by seq "LPX64"\n", fld->lcf_name,
+ CDEBUG(D_INFO, "%s: Found target (idx %llu"
+ ") by seq %#llx\n", fld->lcf_name,
target->ft_idx, seq);
}
LASSERT(name != NULL);
LASSERT(tar->ft_srv != NULL || tar->ft_exp != NULL);
- if (fld->lcf_flags != LUSTRE_FLD_INIT) {
- CERROR("%s: Attempt to add target %s (idx "LPU64") "
- "on fly - skip it\n", fld->lcf_name, name,
- tar->ft_idx);
- RETURN(0);
- } else {
- CDEBUG(D_INFO, "%s: Adding target %s (idx "
- LPU64")\n", fld->lcf_name, name, tar->ft_idx);
- }
+ CDEBUG(D_INFO, "%s: Adding target %s (idx %llu)\n", fld->lcf_name,
+ name, tar->ft_idx);
OBD_ALLOC_PTR(target);
if (target == NULL)
RETURN(-ENOMEM);
spin_lock(&fld->lcf_lock);
- cfs_list_for_each_entry(tmp, &fld->lcf_targets, ft_chain) {
+ list_for_each_entry(tmp, &fld->lcf_targets, ft_chain) {
if (tmp->ft_idx == tar->ft_idx) {
spin_unlock(&fld->lcf_lock);
OBD_FREE_PTR(target);
- CERROR("Target %s exists in FLD and known as %s:#"LPU64"\n",
+ CERROR("Target %s exists in FLD and known as %s:#%llu\n",
name, fld_target_name(tmp), tmp->ft_idx);
RETURN(-EEXIST);
}
target->ft_srv = tar->ft_srv;
target->ft_idx = tar->ft_idx;
- cfs_list_add_tail(&target->ft_chain,
- &fld->lcf_targets);
+ list_add_tail(&target->ft_chain, &fld->lcf_targets);
fld->lcf_count++;
spin_unlock(&fld->lcf_lock);
ENTRY;
spin_lock(&fld->lcf_lock);
- cfs_list_for_each_entry_safe(target, tmp,
- &fld->lcf_targets, ft_chain) {
+ list_for_each_entry_safe(target, tmp, &fld->lcf_targets, ft_chain) {
if (target->ft_idx == idx) {
fld->lcf_count--;
- cfs_list_del(&target->ft_chain);
+ list_del(&target->ft_chain);
spin_unlock(&fld->lcf_lock);
if (target->ft_exp != NULL)
spin_unlock(&fld->lcf_lock);
RETURN(-ENOENT);
}
-EXPORT_SYMBOL(fld_client_del_target);
-#ifdef LPROCFS
+#ifdef CONFIG_PROC_FS
static int fld_client_proc_init(struct lu_client_fld *fld)
{
int rc;
ENTRY;
- fld->lcf_proc_dir = lprocfs_seq_register(fld->lcf_name,
- fld_type_proc_dir,
- NULL, NULL);
+ fld->lcf_proc_dir = lprocfs_register(fld->lcf_name, fld_type_proc_dir,
+ NULL, NULL);
if (IS_ERR(fld->lcf_proc_dir)) {
CERROR("%s: LProcFS failed in fld-init\n",
fld->lcf_name);
RETURN(rc);
}
- rc = lprocfs_seq_add_vars(fld->lcf_proc_dir,
- fld_client_proc_list, fld);
+ rc = lprocfs_add_vars(fld->lcf_proc_dir, fld_client_proc_list, fld);
if (rc) {
CERROR("%s: Can't init FLD proc, rc %d\n",
fld->lcf_name, rc);
}
EXIT;
}
-#else
+#else /* !CONFIG_PROC_FS */
static int fld_client_proc_init(struct lu_client_fld *fld)
{
return 0;
{
return;
}
-#endif
+#endif /* CONFIG_PROC_FS */
EXPORT_SYMBOL(fld_client_proc_fini);
RETURN(-EINVAL);
}
- fld->lcf_count = 0;
+ fld->lcf_count = 0;
spin_lock_init(&fld->lcf_lock);
- fld->lcf_hash = &fld_hash[hash];
- fld->lcf_flags = LUSTRE_FLD_INIT;
- CFS_INIT_LIST_HEAD(&fld->lcf_targets);
+ fld->lcf_hash = &fld_hash[hash];
+ INIT_LIST_HEAD(&fld->lcf_targets);
cache_size = FLD_CLIENT_CACHE_SIZE /
sizeof(struct fld_cache_entry);
ENTRY;
spin_lock(&fld->lcf_lock);
- cfs_list_for_each_entry_safe(target, tmp,
- &fld->lcf_targets, ft_chain) {
+ list_for_each_entry_safe(target, tmp, &fld->lcf_targets, ft_chain) {
fld->lcf_count--;
- cfs_list_del(&target->ft_chain);
+ list_del(&target->ft_chain);
if (target->ft_exp != NULL)
class_export_put(target->ft_exp);
OBD_FREE_PTR(target);
LASSERT(exp != NULL);
+again:
imp = class_exp2cliimp(exp);
switch (fld_op) {
case FLD_QUERY:
op = req_capsule_client_get(&req->rq_pill, &RMF_FLD_OPC);
*op = FLD_LOOKUP;
- if (imp->imp_connect_flags_orig & OBD_CONNECT_MDS_MDS)
+ /* For MDS_MDS seq lookup, it will always use LWP connection,
+ * but LWP will be evicted after restart, so cause the error.
+ * so we will set no_delay for seq lookup request, once the
+ * request fails because of the eviction. always retry here */
+ if (imp->imp_connect_flags_orig & OBD_CONNECT_MDS_MDS) {
req->rq_allow_replay = 1;
+ req->rq_no_delay = 1;
+ }
break;
case FLD_READ:
req = ptlrpc_request_alloc_pack(imp, &RQF_FLD_READ,
RETURN(-ENOMEM);
req_capsule_set_size(&req->rq_pill, &RMF_GENERIC_DATA,
- RCL_SERVER, PAGE_CACHE_SIZE);
+ RCL_SERVER, PAGE_SIZE);
break;
default:
rc = -EINVAL;
req->rq_reply_portal = MDC_REPLY_PORTAL;
ptlrpc_at_set_req_timeout(req);
- fld_enter_request(&exp->exp_obd->u.cli);
+ obd_get_request_slot(&exp->exp_obd->u.cli);
rc = ptlrpc_queue_wait(req);
- fld_exit_request(&exp->exp_obd->u.cli);
- if (rc)
+ obd_put_request_slot(&exp->exp_obd->u.cli);
+
+ if (rc == -ENOENT) {
+ /* Don't loop forever on non-existing FID sequences. */
+ GOTO(out_req, rc);
+ }
+
+ if (rc != 0) {
+ if (imp->imp_state != LUSTRE_IMP_CLOSED &&
+ !imp->imp_deactive &&
+ imp->imp_connect_flags_orig & OBD_CONNECT_MDS_MDS &&
+ OCD_HAS_FLAG(&imp->imp_connect_data, LIGHTWEIGHT) &&
+ rc != -ENOTSUPP) {
+ /* Since LWP is not replayable, so it will keep
+ * trying unless umount happens or the remote
+ * target does not support the operation, otherwise
+ * it would cause unecessary failure of the
+ * application. */
+ ptlrpc_req_finished(req);
+ rc = 0;
+ goto again;
+ }
GOTO(out_req, rc);
+ }
if (fld_op == FLD_QUERY) {
prange = req_capsule_server_get(&req->rq_pill,
return rc;
}
-int fld_client_lookup(struct lu_client_fld *fld, seqno_t seq, mdsno_t *mds,
+int fld_client_lookup(struct lu_client_fld *fld, u64 seq, u32 *mds,
__u32 flags, const struct lu_env *env)
{
struct lu_seq_range res = { 0 };
struct lu_fld_target *target;
+ struct lu_fld_target *origin;
int rc;
ENTRY;
- fld->lcf_flags |= LUSTRE_FLD_RUN;
-
- rc = fld_cache_lookup(fld->lcf_cache, seq, &res);
- if (rc == 0) {
- *mds = res.lsr_index;
- RETURN(0);
- }
+ rc = fld_cache_lookup(fld->lcf_cache, seq, &res);
+ if (rc == 0) {
+ *mds = res.lsr_index;
+ RETURN(0);
+ }
/* Can not find it in the cache */
target = fld_client_get_target(fld, seq);
LASSERT(target != NULL);
-
- CDEBUG(D_INFO, "%s: Lookup fld entry (seq: "LPX64") on "
- "target %s (idx "LPU64")\n", fld->lcf_name, seq,
+ origin = target;
+again:
+ CDEBUG(D_INFO, "%s: Lookup fld entry (seq: %#llx) on "
+ "target %s (idx %llu)\n", fld->lcf_name, seq,
fld_target_name(target), target->ft_idx);
res.lsr_start = seq;
fld_range_set_type(&res, flags);
-#if defined(__KERNEL__) && defined(HAVE_SERVER_SUPPORT)
+#ifdef HAVE_SERVER_SUPPORT
if (target->ft_srv != NULL) {
LASSERT(env != NULL);
rc = fld_server_lookup(env, target->ft_srv, seq, &res);
} else
-#endif
+#endif /* HAVE_SERVER_SUPPORT */
{
rc = fld_client_rpc(target->ft_exp, &res, FLD_QUERY, NULL);
}
+ if (rc == -ESHUTDOWN) {
+ /* If fld lookup failed because the target has been shutdown,
+ * then try next target in the list, until trying all targets
+ * or fld lookup succeeds */
+ spin_lock(&fld->lcf_lock);
+
+ /* If the next entry in the list is the head of the list,
+ * move to the next entry after the head and retrieve
+ * the target. Else retreive the next target entry. */
+
+ if (target->ft_chain.next == &fld->lcf_targets)
+ target = list_entry(target->ft_chain.next->next,
+ struct lu_fld_target, ft_chain);
+ else
+ target = list_entry(target->ft_chain.next,
+ struct lu_fld_target,
+ ft_chain);
+ spin_unlock(&fld->lcf_lock);
+ if (target != origin)
+ goto again;
+ }
if (rc == 0) {
*mds = res.lsr_index;
fld_cache_insert(fld->lcf_cache, &res);
{
fld_cache_flush(fld->lcf_cache);
}
-EXPORT_SYMBOL(fld_client_flush);
-#ifdef __KERNEL__
struct proc_dir_entry *fld_type_proc_dir;
-static int __init fld_mod_init(void)
+static int __init fld_init(void)
{
- fld_type_proc_dir = lprocfs_seq_register(LUSTRE_FLD_NAME,
- proc_lustre_root,
- NULL, NULL);
+ fld_type_proc_dir = lprocfs_register(LUSTRE_FLD_NAME,
+ proc_lustre_root,
+ NULL, NULL);
if (IS_ERR(fld_type_proc_dir))
return PTR_ERR(fld_type_proc_dir);
#ifdef HAVE_SERVER_SUPPORT
fld_server_mod_init();
-#endif
+#endif /* HAVE_SERVER_SUPPORT */
return 0;
}
-static void __exit fld_mod_exit(void)
+static void __exit fld_exit(void)
{
#ifdef HAVE_SERVER_SUPPORT
fld_server_mod_exit();
-#endif
+#endif /* HAVE_SERVER_SUPPORT */
if (fld_type_proc_dir != NULL && !IS_ERR(fld_type_proc_dir)) {
lprocfs_remove(&fld_type_proc_dir);
}
}
-MODULE_AUTHOR("Sun Microsystems, Inc. <http://www.lustre.org/>");
-MODULE_DESCRIPTION("Lustre FLD");
+MODULE_AUTHOR("OpenSFS, Inc. <http://www.lustre.org/>");
+MODULE_DESCRIPTION("Lustre FID Location Database");
+MODULE_VERSION(LUSTRE_VERSION_STRING);
MODULE_LICENSE("GPL");
-cfs_module(mdd, LUSTRE_VERSION_STRING, fld_mod_init, fld_mod_exit);
-#endif /* __KERNEL__ */
+module_init(fld_init);
+module_exit(fld_exit);