-/* -*- mode: c; c-basic-offset: 8; indent-tabs-mode: nil; -*-
- * vim:expandtab:shiftwidth=8:tabstop=8:
- *
+/*
* GPL HEADER START
*
* DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
* GPL HEADER END
*/
/*
- * Copyright 2008 Sun Microsystems, Inc. All rights reserved
+ * Copyright (c) 2007, 2010, Oracle and/or its affiliates. All rights reserved.
* Use is subject to license terms.
*/
/*
+ * Copyright (c) 2011, 2014, Intel Corporation.
+ */
+/*
* This file is part of Lustre, http://www.lustre.org/
* Lustre is a trademark of Sun Microsystems, Inc.
*
#ifndef _LUSTRE_MDC_H
#define _LUSTRE_MDC_H
-#ifdef __KERNEL__
-# include <linux/fs.h>
-# include <linux/dcache.h>
-# ifdef CONFIG_FS_POSIX_ACL
-# ifdef HAVE_XATTR_ACL
-# include <linux/xattr_acl.h>
-# endif /*HAVE_XATTR_ACL */
-# ifdef HAVE_LINUX_POSIX_ACL_XATTR_H
-# include <linux/posix_acl_xattr.h>
-# endif /* HAVE_LINUX_POSIX_ACL_XATTR_H */
-# endif /* CONFIG_FS_POSIX_ACL */
-# ifndef HAVE_VFS_INTENT_PATCHES
-# include <linux/lustre_intent.h>
-# endif /* HAVE_VFS_INTENT_PATCHES */
-#endif /* __KERNEL__ */
+/** \defgroup mdc mdc
+ *
+ * @{
+ */
+
+#include <linux/fs.h>
+#include <linux/dcache.h>
+#ifdef CONFIG_FS_POSIX_ACL
+# include <linux/posix_acl_xattr.h>
+#endif /* CONFIG_FS_POSIX_ACL */
#include <lustre_handles.h>
+#include <lustre_intent.h>
#include <libcfs/libcfs.h>
+#include <obd_class.h>
#include <lustre/lustre_idl.h>
#include <lustre_lib.h>
#include <lustre_dlm.h>
-#include <lustre_log.h>
#include <lustre_export.h>
struct ptlrpc_client;
struct ptlrpc_request;
struct obd_device;
+/**
+ * Serializes in-flight MDT-modifying RPC requests to preserve idempotency.
+ *
+ * This mutex is used to implement execute-once semantics on the MDT.
+ * The MDT stores the last transaction ID and result for every client in
+ * its last_rcvd file. If the client doesn't get a reply, it can safely
+ * resend the request and the MDT will reconstruct the reply being aware
+ * that the request has already been executed. Without this lock,
+ * execution status of concurrent in-flight requests would be
+ * overwritten.
+ *
+ * This design limits the extent to which we can keep a full pipeline of
+ * in-flight requests from a single client. This limitation could be
+ * overcome by allowing multiple slots per client in the last_rcvd file.
+ */
struct mdc_rpc_lock {
- struct semaphore rpcl_sem;
- struct lookup_intent *rpcl_it;
+ /** Lock protecting in-flight RPC concurrency. */
+ struct mutex rpcl_mutex;
+ /** Intent associated with currently executing request. */
+ struct lookup_intent *rpcl_it;
+ /** Used for MDS/RPC load testing purposes. */
+ int rpcl_fakes;
};
+#define MDC_FAKE_RPCL_IT ((void *)0x2c0012bfUL)
+
static inline void mdc_init_rpc_lock(struct mdc_rpc_lock *lck)
{
- sema_init(&lck->rpcl_sem, 1);
+ mutex_init(&lck->rpcl_mutex);
lck->rpcl_it = NULL;
}
static inline void mdc_get_rpc_lock(struct mdc_rpc_lock *lck,
- struct lookup_intent *it)
+ struct lookup_intent *it)
{
- ENTRY;
- if (!it || (it->it_op != IT_GETATTR && it->it_op != IT_LOOKUP)) {
- down(&lck->rpcl_sem);
- LASSERT(lck->rpcl_it == NULL);
- lck->rpcl_it = it;
- }
+ ENTRY;
+
+ if (it != NULL && (it->it_op == IT_GETATTR || it->it_op == IT_LOOKUP ||
+ it->it_op == IT_LAYOUT || it->it_op == IT_READDIR))
+ return;
+
+ /* This would normally block until the existing request finishes.
+ * If fail_loc is set it will block until the regular request is
+ * done, then set rpcl_it to MDC_FAKE_RPCL_IT. Once that is set
+ * it will only be cleared when all fake requests are finished.
+ * Only when all fake requests are finished can normal requests
+ * be sent, to ensure they are recoverable again. */
+ again:
+ mutex_lock(&lck->rpcl_mutex);
+
+ if (CFS_FAIL_CHECK_QUIET(OBD_FAIL_MDC_RPCS_SEM)) {
+ lck->rpcl_it = MDC_FAKE_RPCL_IT;
+ lck->rpcl_fakes++;
+ mutex_unlock(&lck->rpcl_mutex);
+ return;
+ }
+
+ /* This will only happen when the CFS_FAIL_CHECK() was
+ * just turned off but there are still requests in progress.
+ * Wait until they finish. It doesn't need to be efficient
+ * in this extremely rare case, just have low overhead in
+ * the common case when it isn't true. */
+ while (unlikely(lck->rpcl_it == MDC_FAKE_RPCL_IT)) {
+ mutex_unlock(&lck->rpcl_mutex);
+ schedule_timeout(cfs_time_seconds(1) / 4);
+ goto again;
+ }
+
+ LASSERT(lck->rpcl_it == NULL);
+ lck->rpcl_it = it;
}
static inline void mdc_put_rpc_lock(struct mdc_rpc_lock *lck,
- struct lookup_intent *it)
+ struct lookup_intent *it)
{
- if (!it || (it->it_op != IT_GETATTR && it->it_op != IT_LOOKUP)) {
- LASSERT(it == lck->rpcl_it);
- lck->rpcl_it = NULL;
- up(&lck->rpcl_sem);
- }
- EXIT;
+ if (it != NULL && (it->it_op == IT_GETATTR || it->it_op == IT_LOOKUP ||
+ it->it_op == IT_LAYOUT || it->it_op == IT_READDIR))
+ goto out;
+
+ if (lck->rpcl_it == MDC_FAKE_RPCL_IT) { /* OBD_FAIL_MDC_RPCS_SEM */
+ mutex_lock(&lck->rpcl_mutex);
+
+ LASSERTF(lck->rpcl_fakes > 0, "%d\n", lck->rpcl_fakes);
+ lck->rpcl_fakes--;
+
+ if (lck->rpcl_fakes == 0)
+ lck->rpcl_it = NULL;
+
+ } else {
+ LASSERTF(it == lck->rpcl_it, "%p != %p\n", it, lck->rpcl_it);
+ lck->rpcl_it = NULL;
+ }
+
+ mutex_unlock(&lck->rpcl_mutex);
+ out:
+ EXIT;
}
+/**
+ * Update the maximum possible easize.
+ *
+ * This value is learned from ptlrpc replies sent by the MDT. The
+ * default easize is initialized to the minimum value but allowed to
+ * grow up to a single page in size if required to handle the common
+ * case.
+ *
+ * \see client_obd::cl_default_mds_easize
+ *
+ * \param[in] exp export for MDC device
+ * \param[in] body body of ptlrpc reply from MDT
+ *
+ */
static inline void mdc_update_max_ea_from_body(struct obd_export *exp,
- struct mdt_body *body)
+ struct mdt_body *body)
{
- if (body->valid & OBD_MD_FLMODEASIZE) {
- if (exp->exp_obd->u.cli.cl_max_mds_easize < body->max_mdsize)
- exp->exp_obd->u.cli.cl_max_mds_easize =
- body->max_mdsize;
- if (exp->exp_obd->u.cli.cl_max_mds_cookiesize <
- body->max_cookiesize)
- exp->exp_obd->u.cli.cl_max_mds_cookiesize =
- body->max_cookiesize;
- }
-}
+ if (body->mbo_valid & OBD_MD_FLMODEASIZE) {
+ struct client_obd *cli = &exp->exp_obd->u.cli;
+ __u32 def_easize;
+ if (cli->cl_max_mds_easize < body->mbo_max_mdsize)
+ cli->cl_max_mds_easize = body->mbo_max_mdsize;
+
+ def_easize = min_t(__u32, body->mbo_max_mdsize,
+ OBD_MAX_DEFAULT_EA_SIZE);
+ cli->cl_default_mds_easize = def_easize;
+ }
+}
-struct mdc_cache_waiter {
- struct list_head mcw_entry;
- cfs_waitq_t mcw_waitq;
-};
/* mdc/mdc_locks.c */
-int it_disposition(struct lookup_intent *it, int flag);
-void it_clear_disposition(struct lookup_intent *it, int flag);
-void it_set_disposition(struct lookup_intent *it, int flag);
int it_open_error(int phase, struct lookup_intent *it);
-#ifdef HAVE_SPLIT_SUPPORT
-int mdc_sendpage(struct obd_export *exp, const struct lu_fid *fid,
- const struct page *page, int offset);
-#endif
+
+static inline bool cl_is_lov_delay_create(unsigned int flags)
+{
+ return (flags & O_LOV_DELAY_CREATE_1_8) != 0 ||
+ (flags & O_LOV_DELAY_CREATE_MASK) == O_LOV_DELAY_CREATE_MASK;
+}
+
+static inline void cl_lov_delay_create_clear(unsigned int *flags)
+{
+ if ((*flags & O_LOV_DELAY_CREATE_1_8) != 0)
+ *flags &= ~O_LOV_DELAY_CREATE_1_8;
+ if ((*flags & O_LOV_DELAY_CREATE_MASK) == O_LOV_DELAY_CREATE_MASK)
+ *flags &= ~O_LOV_DELAY_CREATE_MASK;
+}
+
+/** @} mdc */
+
#endif