* Copyright (c) 2008, 2010, Oracle and/or its affiliates. All rights reserved.
* Use is subject to license terms.
*
- * Copyright (c) 2011, 2016, Intel Corporation.
+ * Copyright (c) 2011, 2017, Intel Corporation.
*/
/*
* This file is part of Lustre, http://www.lustre.org/
- * Lustre is a trademark of Sun Microsystems, Inc.
*
* Client Lustre Object.
*
#include <linux/list.h>
#include <libcfs/libcfs.h>
-/* class_put_type() */
#include <obd_class.h>
#include <obd_support.h>
#include <lustre_fid.h>
-#include <libcfs/libcfs_hash.h> /* for cfs_hash stuff */
#include <cl_object.h>
#include <lu_object.h>
#include "cl_internal.h"
static struct kmem_cache *cl_env_kmem;
+struct kmem_cache *cl_dio_aio_kmem;
+struct kmem_cache *cl_sub_dio_kmem;
+struct kmem_cache *cl_page_kmem_array[16];
+unsigned short cl_page_kmem_size_array[16];
/** Lock class of cl_object_header::coh_attr_guard */
static struct lock_class_key cl_attr_guard_class;
* top-to-bottom to fill in parts of \a attr that this layer is responsible
* for.
*/
-int cl_object_attr_get(const struct lu_env *env, struct cl_object *obj,
+int cl_object_attr_get(const struct lu_env *env, struct cl_object *top,
struct cl_attr *attr)
{
- struct lu_object_header *top;
- int result;
+ struct cl_object *obj;
+ int result = 0;
- assert_spin_locked(cl_object_attr_guard(obj));
+ assert_spin_locked(cl_object_attr_guard(top));
ENTRY;
- top = obj->co_lu.lo_header;
- result = 0;
- list_for_each_entry(obj, &top->loh_layers, co_lu.lo_linkage) {
+ cl_object_for_each(obj, top) {
if (obj->co_ops->coo_attr_get != NULL) {
result = obj->co_ops->coo_attr_get(env, obj, attr);
if (result != 0) {
* updated. Calls cl_object_operations::coo_upd_attr() on every layer, bottom
* to top.
*/
-int cl_object_attr_update(const struct lu_env *env, struct cl_object *obj,
+int cl_object_attr_update(const struct lu_env *env, struct cl_object *top,
const struct cl_attr *attr, unsigned v)
{
- struct lu_object_header *top;
- int result;
+ struct cl_object *obj;
+ int result = 0;
- assert_spin_locked(cl_object_attr_guard(obj));
+ assert_spin_locked(cl_object_attr_guard(top));
ENTRY;
- top = obj->co_lu.lo_header;
- result = 0;
- list_for_each_entry_reverse(obj, &top->loh_layers, co_lu.lo_linkage) {
+ cl_object_for_each_reverse(obj, top) {
if (obj->co_ops->coo_attr_update != NULL) {
result = obj->co_ops->coo_attr_update(env, obj, attr,
v);
EXPORT_SYMBOL(cl_object_attr_update);
/**
+ * Mark the inode as dirty when the inode has uncommitted (unstable) pages.
+ * Thus when the system is under momory pressure, it will trigger writeback
+ * on background to commit and unpin the pages.
+ */
+void cl_object_dirty_for_sync(const struct lu_env *env, struct cl_object *top)
+{
+ struct cl_object *obj;
+
+ ENTRY;
+
+ cl_object_for_each(obj, top) {
+ if (obj->co_ops->coo_dirty_for_sync != NULL)
+ obj->co_ops->coo_dirty_for_sync(env, obj);
+ }
+ EXIT;
+}
+EXPORT_SYMBOL(cl_object_dirty_for_sync);
+
+/**
* Notifies layers (bottom-to-top) that glimpse AST was received.
*
* Layers have to fill \a lvb fields with information that will be shipped
*
* \see cl_lock_operations::clo_glimpse()
*/
-int cl_object_glimpse(const struct lu_env *env, struct cl_object *obj,
- struct ost_lvb *lvb)
+int cl_object_glimpse(const struct lu_env *env, struct cl_object *top,
+ struct ost_lvb *lvb)
{
- struct lu_object_header *top;
- int result;
+ struct cl_object *obj;
+ int result = 0;
- ENTRY;
- top = obj->co_lu.lo_header;
- result = 0;
- list_for_each_entry_reverse(obj, &top->loh_layers, co_lu.lo_linkage) {
- if (obj->co_ops->coo_glimpse != NULL) {
- result = obj->co_ops->coo_glimpse(env, obj, lvb);
- if (result != 0)
- break;
- }
- }
- LU_OBJECT_HEADER(D_DLMTRACE, env, lu_object_top(top),
+ ENTRY;
+ cl_object_for_each_reverse(obj, top) {
+ if (obj->co_ops->coo_glimpse != NULL) {
+ result = obj->co_ops->coo_glimpse(env, obj, lvb);
+ if (result != 0)
+ break;
+ }
+ }
+ LU_OBJECT_HEADER(D_DLMTRACE, env, lu_object_top(top->co_lu.lo_header),
"size: %llu mtime: %llu atime: %llu "
"ctime: %llu blocks: %llu\n",
- lvb->lvb_size, lvb->lvb_mtime, lvb->lvb_atime,
- lvb->lvb_ctime, lvb->lvb_blocks);
- RETURN(result);
+ lvb->lvb_size, lvb->lvb_mtime, lvb->lvb_atime,
+ lvb->lvb_ctime, lvb->lvb_blocks);
+ RETURN(result);
}
EXPORT_SYMBOL(cl_object_glimpse);
/**
* Updates a configuration of an object \a obj.
*/
-int cl_conf_set(const struct lu_env *env, struct cl_object *obj,
- const struct cl_object_conf *conf)
+int cl_conf_set(const struct lu_env *env, struct cl_object *top,
+ const struct cl_object_conf *conf)
{
- struct lu_object_header *top;
- int result;
+ struct cl_object *obj;
+ int result = 0;
- ENTRY;
- top = obj->co_lu.lo_header;
- result = 0;
- list_for_each_entry(obj, &top->loh_layers, co_lu.lo_linkage) {
- if (obj->co_ops->coo_conf_set != NULL) {
- result = obj->co_ops->coo_conf_set(env, obj, conf);
- if (result != 0)
- break;
- }
- }
- RETURN(result);
+ ENTRY;
+ cl_object_for_each(obj, top) {
+ if (obj->co_ops->coo_conf_set != NULL) {
+ result = obj->co_ops->coo_conf_set(env, obj, conf);
+ if (result)
+ break;
+ }
+ }
+ RETURN(result);
}
EXPORT_SYMBOL(cl_conf_set);
/**
* Prunes caches of pages and locks for this object.
*/
-int cl_object_prune(const struct lu_env *env, struct cl_object *obj)
+int cl_object_prune(const struct lu_env *env, struct cl_object *top)
{
- struct lu_object_header *top;
- struct cl_object *o;
- int result;
+ struct cl_object *obj;
+ int result = 0;
ENTRY;
- top = obj->co_lu.lo_header;
- result = 0;
- list_for_each_entry(o, &top->loh_layers, co_lu.lo_linkage) {
- if (o->co_ops->coo_prune != NULL) {
- result = o->co_ops->coo_prune(env, o);
- if (result != 0)
+ cl_object_for_each(obj, top) {
+ if (obj->co_ops->coo_prune != NULL) {
+ result = obj->co_ops->coo_prune(env, obj);
+ if (result)
break;
}
}
/**
* Get stripe information of this object.
*/
-int cl_object_getstripe(const struct lu_env *env, struct cl_object *obj,
- struct lov_user_md __user *uarg)
+int cl_object_getstripe(const struct lu_env *env, struct cl_object *top,
+ struct lov_user_md __user *uarg, size_t size)
{
- struct lu_object_header *top;
- int result = 0;
+ struct cl_object *obj;
+ int result = 0;
ENTRY;
- top = obj->co_lu.lo_header;
- list_for_each_entry(obj, &top->loh_layers, co_lu.lo_linkage) {
- if (obj->co_ops->coo_getstripe != NULL) {
- result = obj->co_ops->coo_getstripe(env, obj, uarg);
- if (result != 0)
+ cl_object_for_each(obj, top) {
+ if (obj->co_ops->coo_getstripe) {
+ result = obj->co_ops->coo_getstripe(env, obj, uarg,
+ size);
+ if (result)
break;
}
}
* \retval 0 success
* \retval < 0 error
*/
-int cl_object_fiemap(const struct lu_env *env, struct cl_object *obj,
+int cl_object_fiemap(const struct lu_env *env, struct cl_object *top,
struct ll_fiemap_info_key *key,
struct fiemap *fiemap, size_t *buflen)
{
- struct lu_object_header *top;
- int result = 0;
+ struct cl_object *obj;
+ int result = 0;
ENTRY;
- top = obj->co_lu.lo_header;
- list_for_each_entry(obj, &top->loh_layers, co_lu.lo_linkage) {
- if (obj->co_ops->coo_fiemap != NULL) {
+ cl_object_for_each(obj, top) {
+ if (obj->co_ops->coo_fiemap) {
result = obj->co_ops->coo_fiemap(env, obj, key, fiemap,
buflen);
- if (result != 0)
+ if (result)
break;
}
}
}
EXPORT_SYMBOL(cl_object_fiemap);
-int cl_object_layout_get(const struct lu_env *env, struct cl_object *obj,
+int cl_object_layout_get(const struct lu_env *env, struct cl_object *top,
struct cl_layout *cl)
{
- struct lu_object_header *top = obj->co_lu.lo_header;
+ struct cl_object *obj;
ENTRY;
- list_for_each_entry(obj, &top->loh_layers, co_lu.lo_linkage) {
- if (obj->co_ops->coo_layout_get != NULL)
+ cl_object_for_each(obj, top) {
+ if (obj->co_ops->coo_layout_get)
return obj->co_ops->coo_layout_get(env, obj, cl);
}
}
EXPORT_SYMBOL(cl_object_layout_get);
-loff_t cl_object_maxbytes(struct cl_object *obj)
+loff_t cl_object_maxbytes(struct cl_object *top)
{
- struct lu_object_header *top = obj->co_lu.lo_header;
+ struct cl_object *obj;
loff_t maxbytes = LLONG_MAX;
ENTRY;
- list_for_each_entry(obj, &top->loh_layers, co_lu.lo_linkage) {
- if (obj->co_ops->coo_maxbytes != NULL)
+ cl_object_for_each(obj, top) {
+ if (obj->co_ops->coo_maxbytes)
maxbytes = min_t(loff_t, obj->co_ops->coo_maxbytes(obj),
maxbytes);
}
}
EXPORT_SYMBOL(cl_object_maxbytes);
+int cl_object_flush(const struct lu_env *env, struct cl_object *top,
+ struct ldlm_lock *lock)
+{
+ struct cl_object *obj;
+ int rc = 0;
+ ENTRY;
+
+ cl_object_for_each(obj, top) {
+ if (obj->co_ops->coo_object_flush) {
+ rc = obj->co_ops->coo_object_flush(env, obj, lock);
+ if (rc)
+ break;
+ }
+ }
+ RETURN(rc);
+}
+EXPORT_SYMBOL(cl_object_flush);
+
+int cl_object_inode_ops(const struct lu_env *env, struct cl_object *top,
+ enum coo_inode_opc opc, void *data)
+{
+ struct cl_object *obj;
+ int rc = 0;
+
+ ENTRY;
+
+ cl_object_for_each(obj, top) {
+ if (obj->co_ops->coo_inode_ops) {
+ rc = obj->co_ops->coo_inode_ops(env, obj, opc, data);
+ if (rc)
+ break;
+ }
+ }
+ RETURN(rc);
+}
+EXPORT_SYMBOL(cl_object_inode_ops);
+
/**
* Helper function removing all object locks, and marking object for
* deletion. All object pages must have been deleted at this point.
*/
int cl_site_stats_print(const struct cl_site *site, struct seq_file *m)
{
- static const char *pstate[] = {
+ static const char *const pstate[] = {
[CPS_CACHED] = "c",
[CPS_OWNED] = "o",
[CPS_PAGEOUT] = "w",
*
*/
-/**
- * The most efficient way is to store cl_env pointer in task specific
- * structures. On Linux, it wont' be easy to use task_struct->journal_info
- * because Lustre code may call into other fs which has certain assumptions
- * about journal_info. Currently following fields in task_struct are identified
- * can be used for this purpose:
- * - cl_env: for liblustre.
- * - tux_info: ony on RedHat kernel.
- * - ...
- * \note As long as we use task_struct to store cl_env, we assume that once
- * called into Lustre, we'll never call into the other part of the kernel
- * which will use those fields in task_struct without explicitly exiting
- * Lustre.
- *
- * If there's no space in task_struct is available, hash will be used.
- * bz20044, bz22683.
- */
-
static unsigned cl_envs_cached_max = 32; /* XXX: prototype: arbitrary limit
* for now. */
static struct cl_env_cache {
void *ce_debug;
};
+static void cl_env_inc(enum cache_stats_item item)
+{
#ifdef CONFIG_DEBUG_PAGESTATE_TRACKING
-#define CL_ENV_INC(counter) atomic_inc(&cl_env_stats.cs_stats[CS_##counter])
-
-#define CL_ENV_DEC(counter) do { \
- LASSERT(atomic_read(&cl_env_stats.cs_stats[CS_##counter]) > 0); \
- atomic_dec(&cl_env_stats.cs_stats[CS_##counter]); \
-} while (0)
-#else
-#define CL_ENV_INC(counter)
-#define CL_ENV_DEC(counter)
+ atomic_inc(&cl_env_stats.cs_stats[item]);
#endif
+}
+
+static void cl_env_dec(enum cache_stats_item item)
+{
+#ifdef CONFIG_DEBUG_PAGESTATE_TRACKING
+ LASSERT(atomic_read(&cl_env_stats.cs_stats[item]) > 0);
+ atomic_dec(&cl_env_stats.cs_stats[item]);
+#endif
+}
static void cl_env_init0(struct cl_env *cle, void *debug)
{
cle->ce_ref = 1;
cle->ce_debug = debug;
- CL_ENV_INC(busy);
+ cl_env_inc(CS_busy);
}
static struct lu_env *cl_env_new(__u32 ctx_tags, __u32 ses_tags, void *debug)
OBD_SLAB_FREE_PTR(cle, cl_env_kmem);
env = ERR_PTR(rc);
} else {
- CL_ENV_INC(create);
- CL_ENV_INC(total);
+ cl_env_inc(CS_create);
+ cl_env_inc(CS_total);
}
} else
env = ERR_PTR(-ENOMEM);
static void cl_env_fini(struct cl_env *cle)
{
- CL_ENV_DEC(total);
- lu_context_fini(&cle->ce_lu.le_ctx);
- lu_context_fini(&cle->ce_ses);
- OBD_SLAB_FREE_PTR(cle, cl_env_kmem);
+ cl_env_dec(CS_total);
+ lu_context_fini(&cle->ce_lu.le_ctx);
+ lu_context_fini(&cle->ce_ses);
+ OBD_SLAB_FREE_PTR(cle, cl_env_kmem);
}
+/* Get a cl_env, either from the per-CPU cache for the current CPU, or by
+ * allocating a new one.
+ */
static struct lu_env *cl_env_obtain(void *debug)
{
struct cl_env *cle;
}
/**
- * Returns lu_env: if there already is an environment associated with the
- * current thread, it is returned, otherwise, new environment is allocated.
+ * Returns an lu_env.
+ *
+ * No link to thread, this returns an env from the cache or
+ * allocates a new one.
*
- * Allocations are amortized through the global cache of environments.
+ * If you need to get the specific environment you created for this thread,
+ * you must either pass the pointer directly or store it in the file/inode
+ * private data and retrieve it from there using ll_cl_add/ll_cl_find.
*
* \param refcheck pointer to a counter used to detect environment leaks. In
* the usual case cl_env_get() and cl_env_put() are called in the same lexical
* Release an environment.
*
* Decrement \a env reference counter. When counter drops to 0, nothing in
- * this thread is using environment and it is returned to the allocation
- * cache, or freed straight away, if cache is large enough.
+ * this thread is using environment and it is returned to the per-CPU cache or
+ * freed immediately if the cache is full.
*/
void cl_env_put(struct lu_env *env, __u16 *refcheck)
{
if (--cle->ce_ref == 0) {
int cpu = get_cpu();
- CL_ENV_DEC(busy);
- cle->ce_debug = NULL;
- cl_env_exit(cle);
- /*
- * Don't bother to take a lock here.
- *
- * Return environment to the cache only when it was allocated
- * with the standard tags.
- */
+ cl_env_dec(CS_busy);
+ cle->ce_debug = NULL;
+ cl_env_exit(cle);
+ /*
+ * Don't bother to take a lock here.
+ *
+ * Return environment to the cache only when it was allocated
+ * with the standard tags.
+ */
if (cl_envs[cpu].cec_count < cl_envs_cached_max &&
- (env->le_ctx.lc_tags & ~LCT_HAS_EXIT) == LCT_CL_THREAD &&
- (env->le_ses->lc_tags & ~LCT_HAS_EXIT) == LCT_SESSION) {
+ (env->le_ctx.lc_tags & ~LCT_HAS_EXIT) == lu_context_tags_default &&
+ (env->le_ses->lc_tags & ~LCT_HAS_EXIT) == lu_session_tags_default) {
read_lock(&cl_envs[cpu].cec_guard);
list_add(&cle->ce_linkage, &cl_envs[cpu].cec_envs);
cl_envs[cpu].cec_count++;
*/
void cl_attr2lvb(struct ost_lvb *lvb, const struct cl_attr *attr)
{
- ENTRY;
lvb->lvb_size = attr->cat_size;
lvb->lvb_mtime = attr->cat_mtime;
lvb->lvb_atime = attr->cat_atime;
lvb->lvb_ctime = attr->cat_ctime;
lvb->lvb_blocks = attr->cat_blocks;
- EXIT;
}
/**
*/
void cl_lvb2attr(struct cl_attr *attr, const struct ost_lvb *lvb)
{
- ENTRY;
attr->cat_size = lvb->lvb_size;
attr->cat_mtime = lvb->lvb_mtime;
attr->cat_atime = lvb->lvb_atime;
attr->cat_ctime = lvb->lvb_ctime;
attr->cat_blocks = lvb->lvb_blocks;
- EXIT;
}
EXPORT_SYMBOL(cl_lvb2attr);
static struct cl_env cl_env_percpu[NR_CPUS];
+static DEFINE_MUTEX(cl_env_percpu_mutex);
static int cl_env_percpu_init(void)
{
{
int i;
+ mutex_lock(&cl_env_percpu_mutex);
for_each_possible_cpu(i)
lu_env_refill(&cl_env_percpu[i].ce_lu);
+ mutex_unlock(&cl_env_percpu_mutex);
}
void cl_env_percpu_put(struct lu_env *env)
cle->ce_ref--;
LASSERT(cle->ce_ref == 0);
- CL_ENV_DEC(busy);
+ cl_env_dec(CS_busy);
cle->ce_debug = NULL;
put_cpu();
}
EXPORT_SYMBOL(cl_env_percpu_put);
-struct lu_env *cl_env_percpu_get()
+struct lu_env *cl_env_percpu_get(void)
{
struct cl_env *cle;
};
static struct lu_kmem_descr cl_object_caches[] = {
- {
- .ckd_cache = &cl_env_kmem,
- .ckd_name = "cl_env_kmem",
- .ckd_size = sizeof (struct cl_env)
- },
- {
- .ckd_cache = NULL
- }
+ {
+ .ckd_cache = &cl_env_kmem,
+ .ckd_name = "cl_env_kmem",
+ .ckd_size = sizeof(struct cl_env)
+ },
+ {
+ .ckd_cache = &cl_dio_aio_kmem,
+ .ckd_name = "cl_dio_aio_kmem",
+ .ckd_size = sizeof(struct cl_dio_aio)
+ },
+ {
+ .ckd_cache = &cl_sub_dio_kmem,
+ .ckd_name = "cl_sub_dio_kmem",
+ .ckd_size = sizeof(struct cl_sub_dio)
+ },
+ {
+ .ckd_cache = NULL
+ }
};
/**
{
int result;
- OBD_ALLOC(cl_envs, sizeof(*cl_envs) * num_possible_cpus());
+ OBD_ALLOC_PTR_ARRAY(cl_envs, num_possible_cpus());
if (cl_envs == NULL)
GOTO(out, result = -ENOMEM);
out_kmem:
lu_kmem_fini(cl_object_caches);
out_envs:
- OBD_FREE(cl_envs, sizeof(*cl_envs) * num_possible_cpus());
+ OBD_FREE_PTR_ARRAY(cl_envs, num_possible_cpus());
out:
return result;
}
*/
void cl_global_fini(void)
{
+ int i;
+
+ for (i = 0; i < ARRAY_SIZE(cl_page_kmem_array); i++) {
+ if (cl_page_kmem_array[i]) {
+ kmem_cache_destroy(cl_page_kmem_array[i]);
+ cl_page_kmem_array[i] = NULL;
+ }
+ }
cl_env_percpu_fini();
lu_context_key_degister(&cl_key);
lu_kmem_fini(cl_object_caches);
- OBD_FREE(cl_envs, sizeof(*cl_envs) * num_possible_cpus());
+ OBD_FREE_PTR_ARRAY(cl_envs, num_possible_cpus());
}