* Copyright (c) 2003, 2010, Oracle and/or its affiliates. All rights reserved.
* Use is subject to license terms.
*
- * Copyright (c) 2011, 2012, Whamcloud, Inc.
+ * Copyright (c) 2011, 2013, Intel Corporation.
*/
/*
* This file is part of Lustre, http://www.lustre.org/
int it_open_error(int phase, struct lookup_intent *it)
{
+ if (it_disposition(it, DISP_OPEN_LEASE)) {
+ if (phase >= DISP_OPEN_LEASE)
+ return it->d.lustre.it_status;
+ else
+ return 0;
+ }
if (it_disposition(it, DISP_OPEN_OPEN)) {
if (phase >= DISP_OPEN_OPEN)
return it->d.lustre.it_status;
int mdc_set_lock_data(struct obd_export *exp, __u64 *lockh, void *data,
__u64 *bits)
{
- struct ldlm_lock *lock;
+ struct ldlm_lock *lock;
+ struct inode *new_inode = data;
ENTRY;
if(bits)
LASSERT(lock != NULL);
lock_res_and_lock(lock);
#ifdef __KERNEL__
- if (lock->l_ast_data && lock->l_ast_data != data) {
- struct inode *new_inode = data;
- struct inode *old_inode = lock->l_ast_data;
- LASSERTF(old_inode->i_state & I_FREEING,
- "Found existing inode %p/%lu/%u state %lu in lock: "
- "setting data to %p/%lu/%u\n", old_inode,
- old_inode->i_ino, old_inode->i_generation,
- old_inode->i_state,
- new_inode, new_inode->i_ino, new_inode->i_generation);
- }
+ if (lock->l_resource->lr_lvb_inode &&
+ lock->l_resource->lr_lvb_inode != data) {
+ struct inode *old_inode = lock->l_resource->lr_lvb_inode;
+ LASSERTF(old_inode->i_state & I_FREEING,
+ "Found existing inode %p/%lu/%u state %lu in lock: "
+ "setting data to %p/%lu/%u\n", old_inode,
+ old_inode->i_ino, old_inode->i_generation,
+ old_inode->i_state,
+ new_inode, new_inode->i_ino, new_inode->i_generation);
+ }
#endif
- lock->l_ast_data = data;
+ lock->l_resource->lr_lvb_inode = new_inode;
if (bits)
*bits = lock->l_policy_data.l_inodebits.bits;
RETURN(rc);
}
-int mdc_change_cbdata(struct obd_export *exp,
- const struct lu_fid *fid,
- ldlm_iterator_t it, void *data)
+int mdc_null_inode(struct obd_export *exp,
+ const struct lu_fid *fid)
{
- struct ldlm_res_id res_id;
- ENTRY;
+ struct ldlm_res_id res_id;
+ struct ldlm_resource *res;
+ struct ldlm_namespace *ns = class_exp2obd(exp)->obd_namespace;
+ ENTRY;
- fid_build_reg_res_name(fid, &res_id);
- ldlm_resource_iterate(class_exp2obd(exp)->obd_namespace,
- &res_id, it, data);
+ LASSERTF(ns != NULL, "no namespace passed\n");
- EXIT;
- return 0;
+ fid_build_reg_res_name(fid, &res_id);
+
+ res = ldlm_resource_get(ns, NULL, &res_id, 0, 0);
+ if(res == NULL)
+ RETURN(0);
+
+ lock_res(res);
+ res->lr_lvb_inode = NULL;
+ unlock_res(res);
+
+ ldlm_resource_putref(res);
+ RETURN(0);
}
/* find any ldlm lock of the inode in mdc
/* XXX: openlock is not cancelled for cross-refs. */
/* If inode is known, cancel conflicting OPEN locks. */
- if (fid_is_sane(&op_data->op_fid2)) {
- if (it->it_flags & (FMODE_WRITE|MDS_OPEN_TRUNC))
- mode = LCK_CW;
+ if (fid_is_sane(&op_data->op_fid2)) {
+ if (it->it_flags & MDS_OPEN_LEASE) { /* try to get lease */
+ if (it->it_flags & FMODE_WRITE)
+ mode = LCK_EX;
+ else
+ mode = LCK_PR;
+ } else {
+ if (it->it_flags & (FMODE_WRITE|MDS_OPEN_TRUNC))
+ mode = LCK_CW;
#ifdef FMODE_EXEC
- else if (it->it_flags & FMODE_EXEC)
- mode = LCK_PR;
+ else if (it->it_flags & FMODE_EXEC)
+ mode = LCK_PR;
#endif
- else
- mode = LCK_CR;
- count = mdc_resource_get_unused(exp, &op_data->op_fid2,
- &cancels, mode,
- MDS_INODELOCK_OPEN);
- }
+ else
+ mode = LCK_CR;
+ }
+ count = mdc_resource_get_unused(exp, &op_data->op_fid2,
+ &cancels, mode,
+ MDS_INODELOCK_OPEN);
+ }
/* If CREATE, cancel parent's UPDATE lock. */
if (it->it_op & IT_CREAT)
RETURN(req);
}
-static struct ptlrpc_request *ldlm_enqueue_pack(struct obd_export *exp)
+static struct ptlrpc_request *mdc_intent_layout_pack(struct obd_export *exp,
+ struct lookup_intent *it,
+ struct md_op_data *unused)
+{
+ struct obd_device *obd = class_exp2obd(exp);
+ struct ptlrpc_request *req;
+ struct ldlm_intent *lit;
+ struct layout_intent *layout;
+ int rc;
+ ENTRY;
+
+ req = ptlrpc_request_alloc(class_exp2cliimp(exp),
+ &RQF_LDLM_INTENT_LAYOUT);
+ if (req == NULL)
+ RETURN(ERR_PTR(-ENOMEM));
+
+ req_capsule_set_size(&req->rq_pill, &RMF_EADATA, RCL_CLIENT, 0);
+ rc = ldlm_prep_enqueue_req(exp, req, NULL, 0);
+ if (rc) {
+ ptlrpc_request_free(req);
+ RETURN(ERR_PTR(rc));
+ }
+
+ /* pack the intent */
+ lit = req_capsule_client_get(&req->rq_pill, &RMF_LDLM_INTENT);
+ lit->opc = (__u64)it->it_op;
+
+ /* pack the layout intent request */
+ layout = req_capsule_client_get(&req->rq_pill, &RMF_LAYOUT_INTENT);
+ /* LAYOUT_INTENT_ACCESS is generic, specific operation will be
+ * set for replication */
+ layout->li_opc = LAYOUT_INTENT_ACCESS;
+
+ req_capsule_set_size(&req->rq_pill, &RMF_DLM_LVB, RCL_SERVER,
+ obd->u.cli.cl_max_mds_easize);
+ ptlrpc_request_set_replen(req);
+ RETURN(req);
+}
+
+static struct ptlrpc_request *
+mdc_enqueue_pack(struct obd_export *exp, int lvb_len)
{
struct ptlrpc_request *req;
int rc;
RETURN(ERR_PTR(rc));
}
+ req_capsule_set_size(&req->rq_pill, &RMF_DLM_LVB, RCL_SERVER, lvb_len);
ptlrpc_request_set_replen(req);
RETURN(req);
}
struct ldlm_request *lockreq;
struct ldlm_reply *lockrep;
struct lustre_intent_data *intent = &it->d.lustre;
+ struct ldlm_lock *lock;
+ void *lvb_data = NULL;
+ int lvb_len = 0;
ENTRY;
LASSERT(rc >= 0);
memset(lockh, 0, sizeof(*lockh));
rc = 0;
} else { /* rc = 0 */
- struct ldlm_lock *lock = ldlm_handle2lock(lockh);
- LASSERT(lock);
+ lock = ldlm_handle2lock(lockh);
+ LASSERT(lock != NULL);
/* If the server gave us back a different lock mode, we should
* fix up our variables. */
mdc_set_open_replay_data(NULL, NULL, req);
}
- /* TODO: make sure LAYOUT lock must be granted along with EA */
-
if ((body->valid & (OBD_MD_FLDIREA | OBD_MD_FLEASIZE)) != 0) {
void *eadata;
- mdc_update_max_ea_from_body(exp, body);
+ mdc_update_max_ea_from_body(exp, body);
/*
* The eadata is opaque; just check that it is there.
if (eadata == NULL)
RETURN(-EPROTO);
+ /* save lvb data and length in case this is for layout
+ * lock */
+ lvb_data = eadata;
+ lvb_len = body->eadatasize;
+
/*
* We save the reply LOV EA in case we have to replay a
* create for recovery. If we didn't allocate a large
RETURN(-EPROTO);
}
} else if (it->it_op & IT_LAYOUT) {
- struct ldlm_lock *lock = ldlm_handle2lock(lockh);
-
- if (lock != NULL && lock->l_lvb_data == NULL) {
- int lvb_len;
-
- /* maybe the lock was granted right away and layout
- * is packed into RMF_DLM_LVB of req */
- lvb_len = req_capsule_get_size(pill, &RMF_DLM_LVB,
- RCL_SERVER);
- if (lvb_len > 0) {
- void *lvb;
- void *lmm;
-
- lvb = req_capsule_server_get(pill,
- &RMF_DLM_LVB);
- if (lvb == NULL) {
- LDLM_LOCK_PUT(lock);
- RETURN(-EPROTO);
- }
-
- OBD_ALLOC_LARGE(lmm, lvb_len);
- if (lmm == NULL) {
- LDLM_LOCK_PUT(lock);
- RETURN(-ENOMEM);
- }
- memcpy(lmm, lvb, lvb_len);
-
- /* install lvb_data */
- lock_res_and_lock(lock);
- LASSERT(lock->l_lvb_data == NULL);
- lock->l_lvb_data = lmm;
- lock->l_lvb_len = lvb_len;
- unlock_res_and_lock(lock);
- }
+ /* maybe the lock was granted right away and layout
+ * is packed into RMF_DLM_LVB of req */
+ lvb_len = req_capsule_get_size(pill, &RMF_DLM_LVB, RCL_SERVER);
+ if (lvb_len > 0) {
+ lvb_data = req_capsule_server_sized_get(pill,
+ &RMF_DLM_LVB, lvb_len);
+ if (lvb_data == NULL)
+ RETURN(-EPROTO);
}
- if (lock != NULL)
+ }
+
+ /* fill in stripe data for layout lock */
+ lock = ldlm_handle2lock(lockh);
+ if (lock != NULL && ldlm_has_layout(lock) && lvb_data != NULL) {
+ void *lmm;
+
+ LDLM_DEBUG(lock, "layout lock returned by: %s, lvb_len: %d\n",
+ ldlm_it2str(it->it_op), lvb_len);
+
+ OBD_ALLOC_LARGE(lmm, lvb_len);
+ if (lmm == NULL) {
LDLM_LOCK_PUT(lock);
+ RETURN(-ENOMEM);
+ }
+ memcpy(lmm, lvb_data, lvb_len);
+
+ /* install lvb_data */
+ lock_res_and_lock(lock);
+ if (lock->l_lvb_data == NULL) {
+ lock->l_lvb_data = lmm;
+ lock->l_lvb_len = lvb_len;
+ lmm = NULL;
+ }
+ unlock_res_and_lock(lock);
+ if (lmm != NULL)
+ OBD_FREE_LARGE(lmm, lvb_len);
}
+ if (lock != NULL)
+ LDLM_LOCK_PUT(lock);
RETURN(rc);
}
ldlm_policy_data_t const *policy = &lookup_policy;
int generation, resends = 0;
struct ldlm_reply *lockrep;
+ enum lvb_type lvb_type = 0;
ENTRY;
LASSERTF(!it || einfo->ei_type == LDLM_IBITS, "lock type %d\n",
policy = &update_policy;
einfo->ei_cbdata = NULL;
lmm = NULL;
- } else if (it->it_op & IT_UNLINK)
- req = mdc_intent_unlink_pack(exp, it, op_data);
- else if (it->it_op & (IT_GETATTR | IT_LOOKUP))
+ } else if (it->it_op & IT_UNLINK) {
+ req = mdc_intent_unlink_pack(exp, it, op_data);
+ } else if (it->it_op & (IT_GETATTR | IT_LOOKUP)) {
req = mdc_intent_getattr_pack(exp, it, op_data);
- else if (it->it_op & (IT_READDIR | IT_LAYOUT))
- req = ldlm_enqueue_pack(exp);
- else {
+ } else if (it->it_op & IT_READDIR) {
+ req = mdc_enqueue_pack(exp, 0);
+ } else if (it->it_op & IT_LAYOUT) {
+ if (!imp_connect_lvb_type(class_exp2cliimp(exp)))
+ RETURN(-EOPNOTSUPP);
+
+ req = mdc_intent_layout_pack(exp, it, op_data);
+ lvb_type = LVB_T_LAYOUT;
+ } else {
LBUG();
RETURN(-EINVAL);
}
}
rc = ldlm_cli_enqueue(exp, &req, einfo, &res_id, policy, &flags, NULL,
- 0, lockh, 0);
+ 0, lvb_type, lockh, 0);
if (!it) {
/* For flock requests we immediatelly return without further
delay and let caller deal with the rest, since rest of
this function metadata processing makes no sense for flock
- requests anyway */
+ requests anyway. But in case of problem during comms with
+ Server (ETIMEDOUT) or any signal/kill attempt (EINTR), we
+ can not rely on caller and this mainly for F_UNLCKs
+ (explicits or automatically generated by Kernel to clean
+ current FLocks upon exit) that can't be trashed */
+ if ((rc == -EINTR) || (rc == -ETIMEDOUT))
+ goto resend;
RETURN(rc);
}
lockrep = req_capsule_server_get(&req->rq_pill, &RMF_DLM_REP);
LASSERT(lockrep != NULL);
+ lockrep->lock_policy_res2 =
+ ptlrpc_status_ntoh(lockrep->lock_policy_res2);
+
/* Retry the create infinitely when we get -EINPROGRESS from
* server. This is required by the new quota design. */
if (it && it->it_op & IT_CREAT &&
}
}
- rc = mdc_finish_enqueue(exp, req, einfo, it, lockh, rc);
-
- RETURN(rc);
+ rc = mdc_finish_enqueue(exp, req, einfo, it, lockh, rc);
+ if (rc < 0) {
+ if (lustre_handle_is_used(lockh)) {
+ ldlm_lock_decref(lockh, einfo->ei_mode);
+ memset(lockh, 0, sizeof(*lockh));
+ }
+ ptlrpc_req_finished(req);
+ }
+ RETURN(rc);
}
static int mdc_finish_intent_lock(struct obd_export *exp,
struct mdt_body *mdt_body;
struct ldlm_lock *lock;
int rc;
-
+ ENTRY;
LASSERT(request != NULL);
LASSERT(request != LP_POISON);
ldlm_policy_data_t policy = lock->l_policy_data;
LDLM_DEBUG(lock, "matching against this");
- LASSERTF(fid_res_name_eq(&mdt_body->fid1,
- &lock->l_resource->lr_name),
- "Lock res_id: %lu/%lu/%lu, fid: %lu/%lu/%lu.\n",
- (unsigned long)lock->l_resource->lr_name.name[0],
- (unsigned long)lock->l_resource->lr_name.name[1],
- (unsigned long)lock->l_resource->lr_name.name[2],
- (unsigned long)fid_seq(&mdt_body->fid1),
- (unsigned long)fid_oid(&mdt_body->fid1),
- (unsigned long)fid_ver(&mdt_body->fid1));
- LDLM_LOCK_PUT(lock);
+ LASSERTF(fid_res_name_eq(&mdt_body->fid1,
+ &lock->l_resource->lr_name),
+ "Lock res_id: "DLDLMRES", fid: "DFID"\n",
+ PLDLMRES(lock->l_resource), PFID(&mdt_body->fid1));
+ LDLM_LOCK_PUT(lock);
memcpy(&old_lock, lockh, sizeof(*lockh));
if (ldlm_lock_match(NULL, LDLM_FL_BLOCK_GRANTED, NULL,
ldlm_blocking_callback cb_blocking,
__u64 extra_lock_flags)
{
- struct lustre_handle lockh;
- int rc = 0;
- ENTRY;
- LASSERT(it);
+ struct lustre_handle lockh;
+ int rc = 0;
+ ENTRY;
+ LASSERT(it);
- CDEBUG(D_DLMTRACE, "(name: %.*s,"DFID") in obj "DFID
- ", intent: %s flags %#o\n", op_data->op_namelen,
- op_data->op_name, PFID(&op_data->op_fid2),
- PFID(&op_data->op_fid1), ldlm_it2str(it->it_op),
- it->it_flags);
+ CDEBUG(D_DLMTRACE, "(name: %.*s,"DFID") in obj "DFID
+ ", intent: %s flags %#Lo\n", op_data->op_namelen,
+ op_data->op_name, PFID(&op_data->op_fid2),
+ PFID(&op_data->op_fid1), ldlm_it2str(it->it_op),
+ it->it_flags);
lockh.cookie = 0;
if (fid_is_sane(&op_data->op_fid2) &&
- (it->it_op & (IT_LOOKUP | IT_GETATTR | IT_LAYOUT))) {
+ (it->it_op & (IT_LOOKUP | IT_GETATTR))) {
/* We could just return 1 immediately, but since we should only
* be called in revalidate_it if we already have a lock, let's
* verify that. */
* this and use the request from revalidate. In this case, revalidate
* never dropped its reference, so the refcounts are all OK */
if (!it_disposition(it, DISP_ENQ_COMPLETE)) {
- struct ldlm_enqueue_info einfo =
- { LDLM_IBITS, it_to_lock_mode(it), cb_blocking,
- ldlm_completion_ast, NULL, NULL, NULL };
+ struct ldlm_enqueue_info einfo = {
+ .ei_type = LDLM_IBITS,
+ .ei_mode = it_to_lock_mode(it),
+ .ei_cb_bl = cb_blocking,
+ .ei_cb_cp = ldlm_completion_ast,
+ };
/* For case if upper layer did not alloc fid, do it now. */
if (!fid_is_sane(&op_data->op_fid2) && it->it_op & IT_CREAT) {
struct lookup_intent *it;
struct lustre_handle *lockh;
struct obd_device *obddev;
+ struct ldlm_reply *lockrep;
__u64 flags = LDLM_FL_HAS_INTENT;
ENTRY;
GOTO(out, rc);
}
+ lockrep = req_capsule_server_get(&req->rq_pill, &RMF_DLM_REP);
+ LASSERT(lockrep != NULL);
+
+ lockrep->lock_policy_res2 =
+ ptlrpc_status_ntoh(lockrep->lock_policy_res2);
+
rc = mdc_finish_enqueue(exp, req, einfo, it, lockh, rc);
if (rc)
GOTO(out, rc);
};
int rc = 0;
__u64 flags = LDLM_FL_HAS_INTENT;
- ENTRY;
+ ENTRY;
- CDEBUG(D_DLMTRACE,"name: %.*s in inode "DFID", intent: %s flags %#o\n",
- op_data->op_namelen, op_data->op_name, PFID(&op_data->op_fid1),
- ldlm_it2str(it->it_op), it->it_flags);
+ CDEBUG(D_DLMTRACE,"name: %.*s in inode "DFID", intent: %s flags %#Lo\n",
+ op_data->op_namelen, op_data->op_name, PFID(&op_data->op_fid1),
+ ldlm_it2str(it->it_op), it->it_flags);
fid_build_reg_res_name(&op_data->op_fid1, &res_id);
req = mdc_intent_getattr_pack(exp, it, op_data);
}
rc = ldlm_cli_enqueue(exp, &req, einfo, &res_id, &policy, &flags, NULL,
- 0, &minfo->mi_lockh, 1);
+ 0, LVB_T_NONE, &minfo->mi_lockh, 1);
if (rc < 0) {
mdc_exit_request(&obddev->u.cli);
ptlrpc_req_finished(req);