* [PATCH 01/48] staging/lustre: remove bogus ifndef EXPORT_SYMBOL
2013-07-22 16:06 [PATCH 00/48] staging/lustre: minor cleanup and Intel Lustre sync Peng Tao
@ 2013-07-22 16:06 ` Peng Tao
2013-07-22 16:06 ` [PATCH 02/48] staging/lustre/obdclass: restore lu_ref Peng Tao
` (47 subsequent siblings)
48 siblings, 0 replies; 62+ messages in thread
From: Peng Tao @ 2013-07-22 16:06 UTC (permalink / raw)
To: Greg Kroah-Hartman; +Cc: linux-kernel, Peng Tao, Peng Tao, Andreas Dilger
Signed-off-by: Peng Tao <tao.peng@emc.com>
Signed-off-by: Andreas Dilger <andreas.dilger@intel.com>
---
.../lustre/include/linux/libcfs/libcfs_prim.h | 4 ----
1 file changed, 4 deletions(-)
diff --git a/drivers/staging/lustre/include/linux/libcfs/libcfs_prim.h b/drivers/staging/lustre/include/linux/libcfs/libcfs_prim.h
index 9c40ed9..36def36 100644
--- a/drivers/staging/lustre/include/linux/libcfs/libcfs_prim.h
+++ b/drivers/staging/lustre/include/linux/libcfs/libcfs_prim.h
@@ -40,10 +40,6 @@
#ifndef __LIBCFS_PRIM_H__
#define __LIBCFS_PRIM_H__
-#ifndef EXPORT_SYMBOL
-# define EXPORT_SYMBOL(s)
-#endif
-
/*
* Schedule
*/
--
1.7.9.5
^ permalink raw reply related [flat|nested] 62+ messages in thread* [PATCH 02/48] staging/lustre/obdclass: restore lu_ref
2013-07-22 16:06 [PATCH 00/48] staging/lustre: minor cleanup and Intel Lustre sync Peng Tao
2013-07-22 16:06 ` [PATCH 01/48] staging/lustre: remove bogus ifndef EXPORT_SYMBOL Peng Tao
@ 2013-07-22 16:06 ` Peng Tao
2013-07-23 19:37 ` Greg Kroah-Hartman
2013-07-22 16:06 ` [PATCH 03/48] staging/lustre/llite: use READ, WRITE around ll_rw_stats_tally() Peng Tao
` (46 subsequent siblings)
48 siblings, 1 reply; 62+ messages in thread
From: Peng Tao @ 2013-07-22 16:06 UTC (permalink / raw)
To: Greg Kroah-Hartman; +Cc: linux-kernel, Peng Tao, Peng Tao, Andreas Dilger
lu_ref.[ch] were mistakenly removed by coan2. Add them back and guard
with CONFIG_LUSTRE_DEBUG_LU_REF_CHECK.
Signed-off-by: Peng Tao <tao.peng@emc.com>
Signed-off-by: Andreas Dilger <andreas.dilger@intel.com>
---
drivers/staging/lustre/lustre/Kconfig | 10 +
drivers/staging/lustre/lustre/include/lu_ref.h | 124 +++++--
drivers/staging/lustre/lustre/obdclass/Makefile | 5 +-
drivers/staging/lustre/lustre/obdclass/lu_ref.c | 403 ++++++++++++++++++++++-
4 files changed, 511 insertions(+), 31 deletions(-)
diff --git a/drivers/staging/lustre/lustre/Kconfig b/drivers/staging/lustre/lustre/Kconfig
index 893a3c6..0a8eb1ad 100644
--- a/drivers/staging/lustre/lustre/Kconfig
+++ b/drivers/staging/lustre/lustre/Kconfig
@@ -40,6 +40,16 @@ config LUSTRE_OBD_MAX_IOCTL_BUFFER
If unsure, use default.
+config LUSTRE_DEBUG_LU_REF_CHECK
+ bool "Enable Lustre DEBUG object refcount checks"
+ depends on LUSTRE_FS
+ default false
+ help
+ This option is mainly for debug purpose. It enables Lustre code to track
+ references between objects.
+
+ If unsure, say N.
+
config LUSTRE_DEBUG_EXPENSIVE_CHECK
bool "Enable Lustre DEBUG checks"
depends on LUSTRE_FS
diff --git a/drivers/staging/lustre/lustre/include/lu_ref.h b/drivers/staging/lustre/lustre/include/lu_ref.h
index 624c19b..adfbf51 100644
--- a/drivers/staging/lustre/lustre/include/lu_ref.h
+++ b/drivers/staging/lustre/lustre/include/lu_ref.h
@@ -36,11 +36,11 @@
* who acquired references to instance of struct foo, add lu_ref field to it:
*
* \code
- * struct foo {
- * atomic_t foo_refcount;
- * struct lu_ref foo_reference;
- * ...
- * };
+ * struct foo {
+ * atomic_t foo_refcount;
+ * struct lu_ref foo_reference;
+ * ...
+ * };
* \endcode
*
* foo::foo_reference has to be initialized by calling
@@ -54,28 +54,28 @@
* usages are:
*
* \code
- * struct bar *bar;
+ * struct bar *bar;
*
- * // bar owns a reference to foo.
- * bar->bar_foo = foo_get(foo);
- * lu_ref_add(&foo->foo_reference, "bar", bar);
+ * // bar owns a reference to foo.
+ * bar->bar_foo = foo_get(foo);
+ * lu_ref_add(&foo->foo_reference, "bar", bar);
*
- * ...
+ * ...
*
- * // reference from bar to foo is released.
- * lu_ref_del(&foo->foo_reference, "bar", bar);
- * foo_put(bar->bar_foo);
+ * // reference from bar to foo is released.
+ * lu_ref_del(&foo->foo_reference, "bar", bar);
+ * foo_put(bar->bar_foo);
*
*
- * // current thread acquired a temporary reference to foo.
- * foo_get(foo);
- * lu_ref_add(&foo->reference, __FUNCTION__, current);
+ * // current thread acquired a temporary reference to foo.
+ * foo_get(foo);
+ * lu_ref_add(&foo->reference, __FUNCTION__, current);
*
- * ...
+ * ...
*
- * // temporary reference is released.
- * lu_ref_del(&foo->reference, __FUNCTION__, current);
- * foo_put(foo);
+ * // temporary reference is released.
+ * lu_ref_del(&foo->reference, __FUNCTION__, current);
+ * foo_put(foo);
* \endcode
*
* \e Et \e cetera. Often it makes sense to include lu_ref_add() and
@@ -91,15 +91,15 @@
* lu_ref_del_at():
*
* \code
- * // There is a large number of bar's for a single foo.
- * bar->bar_foo = foo_get(foo);
- * bar->bar_foo_ref = lu_ref_add(&foo->foo_reference, "bar", bar);
+ * // There is a large number of bar's for a single foo.
+ * bar->bar_foo = foo_get(foo);
+ * bar->bar_foo_ref = lu_ref_add(&foo->foo_reference, "bar", bar);
*
- * ...
+ * ...
*
- * // reference from bar to foo is released.
- * lu_ref_del_at(&foo->foo_reference, bar->bar_foo_ref, "bar", bar);
- * foo_put(bar->bar_foo);
+ * // reference from bar to foo is released.
+ * lu_ref_del_at(&foo->foo_reference, bar->bar_foo_ref, "bar", bar);
+ * foo_put(bar->bar_foo);
* \endcode
*
* lu_ref interface degrades gracefully in case of memory shortages.
@@ -107,6 +107,75 @@
* @{
*/
+#ifdef CONFIG_LUSTRE_DEBUG_LU_REF_CHECK
+
+/* An incomplete type (defined locally in lu_ref.c) */
+struct lu_ref_link;
+
+/**
+ * Data-structure to keep track of references to a given object. This is used
+ * for debugging.
+ *
+ * lu_ref is embedded into an object which other entities (objects, threads,
+ * etc.) refer to.
+ */
+struct lu_ref {
+ /**
+ * Spin-lock protecting lu_ref::lf_list.
+ */
+ spinlock_t lf_guard;
+ /**
+ * List of all outstanding references (each represented by struct
+ * lu_ref_link), pointing to this object.
+ */
+ struct list_head lf_list;
+ /**
+ * # of links.
+ */
+ short lf_refs;
+ /**
+ * Flag set when lu_ref_add() failed to allocate lu_ref_link. It is
+ * used to mask spurious failure of the following lu_ref_del().
+ */
+ short lf_failed;
+ /**
+ * flags - attribute for the lu_ref, for pad and future use.
+ */
+ short lf_flags;
+ /**
+ * Where was I initialized?
+ */
+ short lf_line;
+ const char *lf_func;
+ /**
+ * Linkage into a global list of all lu_ref's (lu_ref_refs).
+ */
+ struct list_head lf_linkage;
+};
+
+int lu_ref_global_init(void);
+void lu_ref_global_fini(void);
+void lu_ref_init_loc(struct lu_ref *ref, const char *func, const int line);
+void lu_ref_fini (struct lu_ref *ref);
+#define lu_ref_init(ref) lu_ref_init_loc(ref, __FUNCTION__, __LINE__)
+
+struct lu_ref_link *lu_ref_add (struct lu_ref *ref, const char *scope,
+ const void *source);
+struct lu_ref_link *lu_ref_add_atomic(struct lu_ref *ref, const char *scope,
+ const void *source);
+void lu_ref_del (struct lu_ref *ref, const char *scope,
+ const void *source);
+void lu_ref_set_at (struct lu_ref *ref,
+ struct lu_ref_link *link,
+ const char *scope, const void *source0,
+ const void *source1);
+void lu_ref_del_at (struct lu_ref *ref,
+ struct lu_ref_link *link,
+ const char *scope, const void *source);
+void lu_ref_print (const struct lu_ref *ref);
+void lu_ref_print_all (void);
+
+#else /* !CONFIG_LUSTRE_DEBUG_LU_REF_CHECK */
struct lu_ref {};
@@ -164,6 +233,7 @@ static inline void lu_ref_print(const struct lu_ref *ref)
static inline void lu_ref_print_all(void)
{
}
+#endif /* CONFIG_LUSTRE_DEBUG_LU_REF_CHECK */
/** @} lu */
diff --git a/drivers/staging/lustre/lustre/obdclass/Makefile b/drivers/staging/lustre/lustre/obdclass/Makefile
index b80c13c..9930a9e 100644
--- a/drivers/staging/lustre/lustre/obdclass/Makefile
+++ b/drivers/staging/lustre/lustre/obdclass/Makefile
@@ -6,8 +6,9 @@ obdclass-y := linux/linux-module.o linux/linux-obdo.o linux/linux-sysctl.o \
lprocfs_jobstats.o lustre_handles.o lustre_peer.o llog_osd.o \
local_storage.o statfs_pack.o obdo.o obd_config.o obd_mount.o\
mea.o lu_object.o dt_object.o capa.o cl_object.o \
- cl_page.o cl_lock.o cl_io.o lu_ref.o acl.o idmap.o \
- lu_ucred.o
+ cl_page.o cl_lock.o cl_io.o acl.o idmap.o lu_ucred.o
+
+obdclass-$(CONFIG_LUSTRE_DEBUG_LU_REF_CHECK) += lu_ref.o
ccflags-y := -I$(src)/../include
diff --git a/drivers/staging/lustre/lustre/obdclass/lu_ref.c b/drivers/staging/lustre/lustre/obdclass/lu_ref.c
index 23a76f1..79d25c5 100644
--- a/drivers/staging/lustre/lustre/obdclass/lu_ref.c
+++ b/drivers/staging/lustre/lustre/obdclass/lu_ref.c
@@ -42,9 +42,408 @@
#define DEBUG_SUBSYSTEM S_CLASS
-# include <linux/libcfs/libcfs.h>
-
+#include <linux/gfp.h>
+#include <linux/libcfs.h>
#include <obd.h>
#include <obd_class.h>
#include <obd_support.h>
#include <lu_ref.h>
+
+/**
+ * Asserts a condition for a given lu_ref. Must be called with
+ * lu_ref::lf_guard held.
+ */
+#define REFASSERT(ref, expr) do { \
+ struct lu_ref *__tmp = (ref); \
+ \
+ if (unlikely(!(expr))) { \
+ lu_ref_print(__tmp); \
+ spin_unlock(&__tmp->lf_guard); \
+ lu_ref_print_all(); \
+ LASSERT(0); \
+ spin_lock(&__tmp->lf_guard); \
+ } \
+} while (0)
+
+struct lu_ref_link {
+ struct lu_ref *ll_ref;
+ struct list_head ll_linkage;
+ const char *ll_scope;
+ const void *ll_source;
+};
+
+static struct kmem_cache *lu_ref_link_kmem;
+
+static struct lu_kmem_descr lu_ref_caches[] = {
+ {
+ .ckd_cache = &lu_ref_link_kmem,
+ .ckd_name = "lu_ref_link_kmem",
+ .ckd_size = sizeof (struct lu_ref_link)
+ },
+ {
+ .ckd_cache = NULL
+ }
+};
+
+/**
+ * Global list of active (initialized, but not finalized) lu_ref's.
+ *
+ * Protected by lu_ref_refs_guard.
+ */
+static LIST_HEAD(lu_ref_refs);
+static spinlock_t lu_ref_refs_guard;
+static struct lu_ref lu_ref_marker = {
+ .lf_guard = __SPIN_LOCK_UNLOCKED(lu_ref_marker.lf_guard),
+ .lf_list = LIST_HEAD_INIT(lu_ref_marker.lf_list),
+ .lf_linkage = LIST_HEAD_INIT(lu_ref_marker.lf_linkage)
+};
+
+void lu_ref_print(const struct lu_ref *ref)
+{
+ struct lu_ref_link *link;
+
+ CERROR("lu_ref: %p %d %d %s:%d\n",
+ ref, ref->lf_refs, ref->lf_failed, ref->lf_func, ref->lf_line);
+ list_for_each_entry(link, &ref->lf_list, ll_linkage) {
+ CERROR(" link: %s %p\n", link->ll_scope, link->ll_source);
+ }
+}
+EXPORT_SYMBOL(lu_ref_print);
+
+static int lu_ref_is_marker(const struct lu_ref *ref)
+{
+ return (ref == &lu_ref_marker);
+}
+
+void lu_ref_print_all(void)
+{
+ struct lu_ref *ref;
+
+ spin_lock(&lu_ref_refs_guard);
+ list_for_each_entry(ref, &lu_ref_refs, lf_linkage) {
+ if (lu_ref_is_marker(ref))
+ continue;
+
+ spin_lock(&ref->lf_guard);
+ lu_ref_print(ref);
+ spin_unlock(&ref->lf_guard);
+ }
+ spin_unlock(&lu_ref_refs_guard);
+}
+EXPORT_SYMBOL(lu_ref_print_all);
+
+void lu_ref_init_loc(struct lu_ref *ref, const char *func, const int line)
+{
+ ref->lf_refs = 0;
+ ref->lf_func = func;
+ ref->lf_line = line;
+ spin_lock_init(&ref->lf_guard);
+ INIT_LIST_HEAD(&ref->lf_list);
+ spin_lock(&lu_ref_refs_guard);
+ list_add(&ref->lf_linkage, &lu_ref_refs);
+ spin_unlock(&lu_ref_refs_guard);
+}
+EXPORT_SYMBOL(lu_ref_init_loc);
+
+void lu_ref_fini(struct lu_ref *ref)
+{
+ REFASSERT(ref, list_empty(&ref->lf_list));
+ REFASSERT(ref, ref->lf_refs == 0);
+ spin_lock(&lu_ref_refs_guard);
+ list_del_init(&ref->lf_linkage);
+ spin_unlock(&lu_ref_refs_guard);
+}
+EXPORT_SYMBOL(lu_ref_fini);
+
+static struct lu_ref_link *lu_ref_add_context(struct lu_ref *ref,
+ gfp_t flags,
+ const char *scope,
+ const void *source)
+{
+ struct lu_ref_link *link;
+
+ link = NULL;
+ if (lu_ref_link_kmem != NULL) {
+ OBD_SLAB_ALLOC_PTR_GFP(link, lu_ref_link_kmem, flags);
+ if (link != NULL) {
+ link->ll_ref = ref;
+ link->ll_scope = scope;
+ link->ll_source = source;
+ spin_lock(&ref->lf_guard);
+ list_add_tail(&link->ll_linkage, &ref->lf_list);
+ ref->lf_refs++;
+ spin_unlock(&ref->lf_guard);
+ }
+ }
+
+ if (link == NULL) {
+ spin_lock(&ref->lf_guard);
+ ref->lf_failed++;
+ spin_unlock(&ref->lf_guard);
+ link = ERR_PTR(-ENOMEM);
+ }
+
+ return link;
+}
+
+struct lu_ref_link *lu_ref_add(struct lu_ref *ref, const char *scope,
+ const void *source)
+{
+ might_sleep();
+ return lu_ref_add_context(ref, GFP_IOFS, scope, source);
+}
+EXPORT_SYMBOL(lu_ref_add);
+
+/**
+ * Version of lu_ref_add() to be used in non-blockable contexts.
+ */
+struct lu_ref_link *lu_ref_add_atomic(struct lu_ref *ref, const char *scope,
+ const void *source)
+{
+ return lu_ref_add_context(ref, GFP_ATOMIC, scope, source);
+}
+EXPORT_SYMBOL(lu_ref_add_atomic);
+
+static inline int lu_ref_link_eq(const struct lu_ref_link *link,
+ const char *scope, const void *source)
+{
+ return link->ll_source == source && !strcmp(link->ll_scope, scope);
+}
+
+/**
+ * Maximal chain length seen so far.
+ */
+static unsigned lu_ref_chain_max_length = 127;
+
+/**
+ * Searches for a lu_ref_link with given [scope, source] within given lu_ref.
+ */
+static struct lu_ref_link *lu_ref_find(struct lu_ref *ref, const char *scope,
+ const void *source)
+{
+ struct lu_ref_link *link;
+ unsigned iterations;
+
+ iterations = 0;
+ list_for_each_entry(link, &ref->lf_list, ll_linkage) {
+ ++iterations;
+ if (lu_ref_link_eq(link, scope, source)) {
+ if (iterations > lu_ref_chain_max_length) {
+ CWARN("Long lu_ref chain %d \"%s\":%p\n",
+ iterations, scope, source);
+ lu_ref_chain_max_length = iterations * 3 / 2;
+ }
+ return link;
+ }
+ }
+ return NULL;
+}
+
+void lu_ref_del(struct lu_ref *ref, const char *scope, const void *source)
+{
+ struct lu_ref_link *link;
+
+ spin_lock(&ref->lf_guard);
+ link = lu_ref_find(ref, scope, source);
+ if (link != NULL) {
+ list_del(&link->ll_linkage);
+ ref->lf_refs--;
+ spin_unlock(&ref->lf_guard);
+ OBD_SLAB_FREE(link, lu_ref_link_kmem, sizeof(*link));
+ } else {
+ REFASSERT(ref, ref->lf_failed > 0);
+ ref->lf_failed--;
+ spin_unlock(&ref->lf_guard);
+ }
+}
+EXPORT_SYMBOL(lu_ref_del);
+
+void lu_ref_set_at(struct lu_ref *ref, struct lu_ref_link *link,
+ const char *scope,
+ const void *source0, const void *source1)
+{
+ spin_lock(&ref->lf_guard);
+ if (link != ERR_PTR(-ENOMEM)) {
+ REFASSERT(ref, link->ll_ref == ref);
+ REFASSERT(ref, lu_ref_link_eq(link, scope, source0));
+ link->ll_source = source1;
+ } else {
+ REFASSERT(ref, ref->lf_failed > 0);
+ }
+ spin_unlock(&ref->lf_guard);
+}
+EXPORT_SYMBOL(lu_ref_set_at);
+
+void lu_ref_del_at(struct lu_ref *ref, struct lu_ref_link *link,
+ const char *scope, const void *source)
+{
+ if (link != ERR_PTR(-ENOMEM)) {
+ spin_lock(&ref->lf_guard);
+ REFASSERT(ref, link->ll_ref == ref);
+ REFASSERT(ref, lu_ref_link_eq(link, scope, source));
+ list_del(&link->ll_linkage);
+ ref->lf_refs--;
+ spin_unlock(&ref->lf_guard);
+ OBD_SLAB_FREE(link, lu_ref_link_kmem, sizeof(*link));
+ } else {
+ spin_lock(&ref->lf_guard);
+ REFASSERT(ref, ref->lf_failed > 0);
+ ref->lf_failed--;
+ spin_unlock(&ref->lf_guard);
+ }
+}
+EXPORT_SYMBOL(lu_ref_del_at);
+
+#ifdef LPROCFS
+
+static void *lu_ref_seq_start(struct seq_file *seq, loff_t *pos)
+{
+ struct lu_ref *ref = seq->private;
+
+ spin_lock(&lu_ref_refs_guard);
+ if (list_empty(&ref->lf_linkage))
+ ref = NULL;
+ spin_unlock(&lu_ref_refs_guard);
+
+ return ref;
+}
+
+static void *lu_ref_seq_next(struct seq_file *seq, void *p, loff_t *pos)
+{
+ struct lu_ref *ref = p;
+ struct lu_ref *next;
+
+ LASSERT(seq->private == p);
+ LASSERT(!list_empty(&ref->lf_linkage));
+
+ spin_lock(&lu_ref_refs_guard);
+ next = list_entry(ref->lf_linkage.next, struct lu_ref, lf_linkage);
+ if (&next->lf_linkage == &lu_ref_refs) {
+ p = NULL;
+ } else {
+ (*pos)++;
+ list_move(&ref->lf_linkage, &next->lf_linkage);
+ }
+ spin_unlock(&lu_ref_refs_guard);
+ return p;
+}
+
+static void lu_ref_seq_stop(struct seq_file *seq, void *p)
+{
+ /* Nothing to do */
+}
+
+
+static int lu_ref_seq_show(struct seq_file *seq, void *p)
+{
+ struct lu_ref *ref = p;
+ struct lu_ref *next;
+
+ spin_lock(&lu_ref_refs_guard);
+ next = list_entry(ref->lf_linkage.next, struct lu_ref, lf_linkage);
+ if ((&next->lf_linkage == &lu_ref_refs) || lu_ref_is_marker(next)) {
+ spin_unlock(&lu_ref_refs_guard);
+ return 0;
+ }
+
+ /* print the entry */
+ spin_lock(&next->lf_guard);
+ seq_printf(seq, "lu_ref: %p %d %d %s:%d\n",
+ next, next->lf_refs, next->lf_failed,
+ next->lf_func, next->lf_line);
+ if (next->lf_refs > 64) {
+ seq_printf(seq, " too many references, skip\n");
+ } else {
+ struct lu_ref_link *link;
+ int i = 0;
+
+ list_for_each_entry(link, &next->lf_list, ll_linkage)
+ seq_printf(seq, " #%d link: %s %p\n",
+ i++, link->ll_scope, link->ll_source);
+ }
+ spin_unlock(&next->lf_guard);
+ spin_unlock(&lu_ref_refs_guard);
+
+ return 0;
+}
+
+static struct seq_operations lu_ref_seq_ops = {
+ .start = lu_ref_seq_start,
+ .stop = lu_ref_seq_stop,
+ .next = lu_ref_seq_next,
+ .show = lu_ref_seq_show
+};
+
+static int lu_ref_seq_open(struct inode *inode, struct file *file)
+{
+ struct lu_ref *marker = &lu_ref_marker;
+ int result = 0;
+
+ result = seq_open(file, &lu_ref_seq_ops);
+ if (result == 0) {
+ spin_lock(&lu_ref_refs_guard);
+ if (!list_empty(&marker->lf_linkage))
+ result = -EAGAIN;
+ else
+ list_add(&marker->lf_linkage, &lu_ref_refs);
+ spin_unlock(&lu_ref_refs_guard);
+
+ if (result == 0) {
+ struct seq_file *f = file->private_data;
+ f->private = marker;
+ } else {
+ seq_release(inode, file);
+ }
+ }
+
+ return result;
+}
+
+static int lu_ref_seq_release(struct inode *inode, struct file *file)
+{
+ struct lu_ref *ref = ((struct seq_file *)file->private_data)->private;
+
+ spin_lock(&lu_ref_refs_guard);
+ list_del_init(&ref->lf_linkage);
+ spin_unlock(&lu_ref_refs_guard);
+
+ return seq_release(inode, file);
+}
+
+static struct file_operations lu_ref_dump_fops = {
+ .owner = THIS_MODULE,
+ .open = lu_ref_seq_open,
+ .read = seq_read,
+ .llseek = seq_lseek,
+ .release = lu_ref_seq_release
+};
+
+int lu_ref_global_init(void)
+{
+ int result;
+
+ CDEBUG(D_CONSOLE,
+ "lu_ref tracking is enabled. Performance isn't.\n");
+
+ spin_lock_init(&lu_ref_refs_guard);
+ result = lu_kmem_init(lu_ref_caches);
+
+#ifdef LPROCFS
+ if (result == 0) {
+ result = lprocfs_seq_create(proc_lustre_root, "lu_refs",
+ 0444, &lu_ref_dump_fops, NULL);
+ if (result)
+ lu_kmem_fini(lu_ref_caches);
+ }
+#endif
+
+ return result;
+}
+
+void lu_ref_global_fini(void)
+{
+#ifdef LPROCFS
+ lprocfs_remove_proc_entry("lu_refs", proc_lustre_root);
+#endif
+ lu_kmem_fini(lu_ref_caches);
+}
--
1.7.9.5
^ permalink raw reply related [flat|nested] 62+ messages in thread* Re: [PATCH 02/48] staging/lustre/obdclass: restore lu_ref
2013-07-22 16:06 ` [PATCH 02/48] staging/lustre/obdclass: restore lu_ref Peng Tao
@ 2013-07-23 19:37 ` Greg Kroah-Hartman
2013-07-23 20:31 ` Dilger, Andreas
0 siblings, 1 reply; 62+ messages in thread
From: Greg Kroah-Hartman @ 2013-07-23 19:37 UTC (permalink / raw)
To: Peng Tao; +Cc: linux-kernel, Peng Tao, Andreas Dilger
On Tue, Jul 23, 2013 at 12:06:23AM +0800, Peng Tao wrote:
> lu_ref.[ch] were mistakenly removed by coan2. Add them back and guard
> with CONFIG_LUSTRE_DEBUG_LU_REF_CHECK.
Ick, why? Why not use the in-kernel infrastructure for this type of
thig, _iff_ you really need it.
Why do you need this at all, isn't the reference code fully debugged by
now, given that this has been running for years? I would only expect to
see something like this in new code that is still being developed.
thanks,
greg k-h
^ permalink raw reply [flat|nested] 62+ messages in thread
* Re: [PATCH 02/48] staging/lustre/obdclass: restore lu_ref
2013-07-23 19:37 ` Greg Kroah-Hartman
@ 2013-07-23 20:31 ` Dilger, Andreas
2013-07-23 20:52 ` Greg Kroah-Hartman
0 siblings, 1 reply; 62+ messages in thread
From: Dilger, Andreas @ 2013-07-23 20:31 UTC (permalink / raw)
To: Greg Kroah-Hartman, Peng Tao; +Cc: linux-kernel, Peng Tao
On 2013/07/23 1:37 PM, "Greg Kroah-Hartman" <gregkh@linuxfoundation.org>
wrote:
>On Tue, Jul 23, 2013 at 12:06:23AM +0800, Peng Tao wrote:
>> lu_ref.[ch] were mistakenly removed by coan2. Add them back and guard
>> with CONFIG_LUSTRE_DEBUG_LU_REF_CHECK.
>
>Ick, why? Why not use the in-kernel infrastructure for this type of
>thig, _iff_ you really need it.
>
>Why do you need this at all, isn't the reference code fully debugged by
>now, given that this has been running for years? I would only expect to
>see something like this in new code that is still being developed.
Lustre is still being developed, so it is useful to be able to enable this
debugging to catch any new problems that are introduced.
Cheers, Andreas
--
Andreas Dilger
Lustre Software Architect
Intel High Performance Data Division
^ permalink raw reply [flat|nested] 62+ messages in thread
* Re: [PATCH 02/48] staging/lustre/obdclass: restore lu_ref
2013-07-23 20:31 ` Dilger, Andreas
@ 2013-07-23 20:52 ` Greg Kroah-Hartman
2013-07-24 6:06 ` Peng Tao
0 siblings, 1 reply; 62+ messages in thread
From: Greg Kroah-Hartman @ 2013-07-23 20:52 UTC (permalink / raw)
To: Dilger, Andreas; +Cc: Peng Tao, linux-kernel, Peng Tao
On Tue, Jul 23, 2013 at 08:31:15PM +0000, Dilger, Andreas wrote:
> On 2013/07/23 1:37 PM, "Greg Kroah-Hartman" <gregkh@linuxfoundation.org>
> wrote:
>
> >On Tue, Jul 23, 2013 at 12:06:23AM +0800, Peng Tao wrote:
> >> lu_ref.[ch] were mistakenly removed by coan2. Add them back and guard
> >> with CONFIG_LUSTRE_DEBUG_LU_REF_CHECK.
> >
> >Ick, why? Why not use the in-kernel infrastructure for this type of
> >thig, _iff_ you really need it.
> >
> >Why do you need this at all, isn't the reference code fully debugged by
> >now, given that this has been running for years? I would only expect to
> >see something like this in new code that is still being developed.
>
> Lustre is still being developed, so it is useful to be able to enable this
> debugging to catch any new problems that are introduced.
Ok, then use the in-kernel infrastructure for this type of thing, don't
roll your own and add custom stuff that will later just have to be
ripped out and converted to use the in-kernel stuff.
thanks,
greg k-h
^ permalink raw reply [flat|nested] 62+ messages in thread
* Re: [PATCH 02/48] staging/lustre/obdclass: restore lu_ref
2013-07-23 20:52 ` Greg Kroah-Hartman
@ 2013-07-24 6:06 ` Peng Tao
0 siblings, 0 replies; 62+ messages in thread
From: Peng Tao @ 2013-07-24 6:06 UTC (permalink / raw)
To: Greg Kroah-Hartman; +Cc: Dilger, Andreas, linux-kernel
On Wed, Jul 24, 2013 at 4:52 AM, Greg Kroah-Hartman
<gregkh@linuxfoundation.org> wrote:
> On Tue, Jul 23, 2013 at 08:31:15PM +0000, Dilger, Andreas wrote:
>> On 2013/07/23 1:37 PM, "Greg Kroah-Hartman" <gregkh@linuxfoundation.org>
>> wrote:
>>
>> >On Tue, Jul 23, 2013 at 12:06:23AM +0800, Peng Tao wrote:
>> >> lu_ref.[ch] were mistakenly removed by coan2. Add them back and guard
>> >> with CONFIG_LUSTRE_DEBUG_LU_REF_CHECK.
>> >
>> >Ick, why? Why not use the in-kernel infrastructure for this type of
>> >thig, _iff_ you really need it.
>> >
>> >Why do you need this at all, isn't the reference code fully debugged by
>> >now, given that this has been running for years? I would only expect to
>> >see something like this in new code that is still being developed.
>>
>> Lustre is still being developed, so it is useful to be able to enable this
>> debugging to catch any new problems that are introduced.
>
> Ok, then use the in-kernel infrastructure for this type of thing, don't
> roll your own and add custom stuff that will later just have to be
> ripped out and converted to use the in-kernel stuff.
>
OK. Will drop this one.
Thanks,
Tao
^ permalink raw reply [flat|nested] 62+ messages in thread
* [PATCH 03/48] staging/lustre/llite: use READ, WRITE around ll_rw_stats_tally()
2013-07-22 16:06 [PATCH 00/48] staging/lustre: minor cleanup and Intel Lustre sync Peng Tao
2013-07-22 16:06 ` [PATCH 01/48] staging/lustre: remove bogus ifndef EXPORT_SYMBOL Peng Tao
2013-07-22 16:06 ` [PATCH 02/48] staging/lustre/obdclass: restore lu_ref Peng Tao
@ 2013-07-22 16:06 ` Peng Tao
2013-07-22 16:06 ` [PATCH 04/48] staging/lustre/llite: missing last bit in ll_have_md_lock Peng Tao
` (45 subsequent siblings)
48 siblings, 0 replies; 62+ messages in thread
From: Peng Tao @ 2013-07-22 16:06 UTC (permalink / raw)
To: Greg Kroah-Hartman
Cc: linux-kernel, John L. Hammond, Peng Tao, Andreas Dilger
From: "John L. Hammond" <john.hammond@intel.com>
In vvp_io_write_start() the stats function ll_rw_stats_tally() was
incorrectly called with a rw argument of 0. Correct this and use the
macros READ and WRITE in and around ll_rw_stats_tally() for clarity.
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-3384
Lustre-change: http://review.whamcloud.com/6447
Signed-off-by: John L. Hammond <john.hammond@intel.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Reviewed-by: Jinshan Xiong <jinshan.xiong@intel.com>
Signed-off-by: Peng Tao <tao.peng@emc.com>
Signed-off-by: Andreas Dilger <andreas.dilger@intel.com>
---
drivers/staging/lustre/lustre/llite/lproc_llite.c | 10 ++++++----
drivers/staging/lustre/lustre/llite/vvp_io.c | 4 ++--
2 files changed, 8 insertions(+), 6 deletions(-)
diff --git a/drivers/staging/lustre/lustre/llite/lproc_llite.c b/drivers/staging/lustre/lustre/llite/lproc_llite.c
index 5c7368c..a7b12fd 100644
--- a/drivers/staging/lustre/lustre/llite/lproc_llite.c
+++ b/drivers/staging/lustre/lustre/llite/lproc_llite.c
@@ -1303,8 +1303,9 @@ static int ll_rw_offset_stats_seq_show(struct seq_file *seq, void *v)
/* We stored the discontiguous offsets here; print them first */
for(i = 0; i < LL_OFFSET_HIST_MAX; i++) {
if (offset[i].rw_pid != 0)
- seq_printf(seq,"%3c %10d %14Lu %14Lu %17lu %17lu %14Lu",
- offset[i].rw_op ? 'W' : 'R',
+ seq_printf(seq,
+ "%3c %10d %14Lu %14Lu %17lu %17lu %14Lu",
+ offset[i].rw_op == READ ? 'R' : 'W',
offset[i].rw_pid,
offset[i].rw_range_start,
offset[i].rw_range_end,
@@ -1315,8 +1316,9 @@ static int ll_rw_offset_stats_seq_show(struct seq_file *seq, void *v)
/* Then print the current offsets for each process */
for(i = 0; i < LL_PROCESS_HIST_MAX; i++) {
if (process[i].rw_pid != 0)
- seq_printf(seq,"%3c %10d %14Lu %14Lu %17lu %17lu %14Lu",
- process[i].rw_op ? 'W' : 'R',
+ seq_printf(seq,
+ "%3c %10d %14Lu %14Lu %17lu %17lu %14Lu",
+ process[i].rw_op == READ ? 'R' : 'W',
process[i].rw_pid,
process[i].rw_range_start,
process[i].rw_last_file_pos,
diff --git a/drivers/staging/lustre/lustre/llite/vvp_io.c b/drivers/staging/lustre/lustre/llite/vvp_io.c
index eb964ac..c24dc268 100644
--- a/drivers/staging/lustre/lustre/llite/vvp_io.c
+++ b/drivers/staging/lustre/lustre/llite/vvp_io.c
@@ -525,7 +525,7 @@ out:
io->ci_continue = 0;
io->ci_nob += result;
ll_rw_stats_tally(ll_i2sbi(inode), current->pid,
- cio->cui_fd, pos, result, 0);
+ cio->cui_fd, pos, result, READ);
result = 0;
}
return result;
@@ -580,7 +580,7 @@ static int vvp_io_write_start(const struct lu_env *env,
io->ci_continue = 0;
io->ci_nob += result;
ll_rw_stats_tally(ll_i2sbi(inode), current->pid,
- cio->cui_fd, pos, result, 0);
+ cio->cui_fd, pos, result, WRITE);
result = 0;
}
RETURN(result);
--
1.7.9.5
^ permalink raw reply related [flat|nested] 62+ messages in thread* [PATCH 04/48] staging/lustre/llite: missing last bit in ll_have_md_lock
2013-07-22 16:06 [PATCH 00/48] staging/lustre: minor cleanup and Intel Lustre sync Peng Tao
` (2 preceding siblings ...)
2013-07-22 16:06 ` [PATCH 03/48] staging/lustre/llite: use READ, WRITE around ll_rw_stats_tally() Peng Tao
@ 2013-07-22 16:06 ` Peng Tao
2013-07-22 16:06 ` [PATCH 05/48] staging/lustre: fix 'program hangs' errors Peng Tao
` (44 subsequent siblings)
48 siblings, 0 replies; 62+ messages in thread
From: Peng Tao @ 2013-07-22 16:06 UTC (permalink / raw)
To: Greg Kroah-Hartman; +Cc: linux-kernel, wang di, Peng Tao, Andreas Dilger
From: wang di <di.wang@intel.com>
Missing the last bit during INODELOCK check in ll_have_md_lock.
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-3385
Lustre-change: http://review.whamcloud.com/6438
Signed-off-by: wang di <di.wang@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Signed-off-by: Peng Tao <tao.peng@emc.com>
Signed-off-by: Andreas Dilger <andreas.dilger@intel.com>
---
drivers/staging/lustre/lustre/llite/file.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/drivers/staging/lustre/lustre/llite/file.c b/drivers/staging/lustre/lustre/llite/file.c
index ed1e3f7..efe66d9 100644
--- a/drivers/staging/lustre/lustre/llite/file.c
+++ b/drivers/staging/lustre/lustre/llite/file.c
@@ -2448,7 +2448,7 @@ int ll_have_md_lock(struct inode *inode, __u64 *bits, ldlm_mode_t l_req_mode)
ldlm_lockname[mode]);
flags = LDLM_FL_BLOCK_GRANTED | LDLM_FL_CBPENDING | LDLM_FL_TEST_LOCK;
- for (i = 0; i < MDS_INODELOCK_MAXSHIFT && *bits != 0; i++) {
+ for (i = 0; i <= MDS_INODELOCK_MAXSHIFT && *bits != 0; i++) {
policy.l_inodebits.bits = *bits & (1 << i);
if (policy.l_inodebits.bits == 0)
continue;
--
1.7.9.5
^ permalink raw reply related [flat|nested] 62+ messages in thread* [PATCH 05/48] staging/lustre: fix 'program hangs' errors
2013-07-22 16:06 [PATCH 00/48] staging/lustre: minor cleanup and Intel Lustre sync Peng Tao
` (3 preceding siblings ...)
2013-07-22 16:06 ` [PATCH 04/48] staging/lustre/llite: missing last bit in ll_have_md_lock Peng Tao
@ 2013-07-22 16:06 ` Peng Tao
2013-07-22 16:06 ` [PATCH 06/48] staging/lustre/llite: check alloc in ll_file_data_get, ll_dir_ioctl Peng Tao
` (43 subsequent siblings)
48 siblings, 0 replies; 62+ messages in thread
From: Peng Tao @ 2013-07-22 16:06 UTC (permalink / raw)
To: Greg Kroah-Hartman
Cc: linux-kernel, Sebastien Buisson, Peng Tao, Andreas Dilger
From: Sebastien Buisson <sebastien.buisson@bull.net>
Fix 'program hangs' defects found by Coverity version 6.5.1:
Missing unlock (LOCK)
Returning without unlocking.
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-3054
Lustre-change: http://review.whamcloud.com/5870
Signed-off-by: Sebastien Buisson <sebastien.buisson@bull.net>
Reviewed-by: Dmitry Eremin <dmitry.eremin@intel.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Signed-off-by: Peng Tao <tao.peng@emc.com>
Signed-off-by: Andreas Dilger <andreas.dilger@intel.com>
---
drivers/staging/lustre/lnet/selftest/rpc.c | 4 +++-
1 file changed, 3 insertions(+), 1 deletion(-)
diff --git a/drivers/staging/lustre/lnet/selftest/rpc.c b/drivers/staging/lustre/lnet/selftest/rpc.c
index bc1f38b..0dfc76e 100644
--- a/drivers/staging/lustre/lnet/selftest/rpc.c
+++ b/drivers/staging/lustre/lnet/selftest/rpc.c
@@ -661,8 +661,10 @@ srpc_finish_service(struct srpc_service *sv)
cfs_percpt_for_each(scd, i, sv->sv_cpt_data) {
spin_lock(&scd->scd_lock);
- if (!swi_deschedule_workitem(&scd->scd_buf_wi))
+ if (!swi_deschedule_workitem(&scd->scd_buf_wi)) {
+ spin_unlock(&scd->scd_lock);
return 0;
+ }
if (scd->scd_buf_nposted > 0) {
CDEBUG(D_NET, "waiting for %d posted buffers to unlink",
--
1.7.9.5
^ permalink raw reply related [flat|nested] 62+ messages in thread* [PATCH 06/48] staging/lustre/llite: check alloc in ll_file_data_get, ll_dir_ioctl
2013-07-22 16:06 [PATCH 00/48] staging/lustre: minor cleanup and Intel Lustre sync Peng Tao
` (4 preceding siblings ...)
2013-07-22 16:06 ` [PATCH 05/48] staging/lustre: fix 'program hangs' errors Peng Tao
@ 2013-07-22 16:06 ` Peng Tao
2013-07-22 16:06 ` [PATCH 07/48] staging/lustre/llite: A not locked mutex can be unlocked Peng Tao
` (42 subsequent siblings)
48 siblings, 0 replies; 62+ messages in thread
From: Peng Tao @ 2013-07-22 16:06 UTC (permalink / raw)
To: Greg Kroah-Hartman
Cc: linux-kernel, John L. Hammond, Peng Tao, Andreas Dilger
From: "John L. Hammond" <john.hammond@intel.com>
In ll_file_data_get() and ll_dir_ioctl() return error on failed
allocations.
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-2753
Lustre-change: http://review.whamcloud.com/5845
Signed-off-by: John L. Hammond <john.hammond@intel.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Reviewed-by: Dmitry Eremin <dmitry.eremin@intel.com>
Reviewed-by: Sebastien Buisson <sebastien.buisson@bull.net>
Signed-off-by: Peng Tao <tao.peng@emc.com>
Signed-off-by: Andreas Dilger <andreas.dilger@intel.com>
---
drivers/staging/lustre/lustre/llite/dir.c | 2 ++
drivers/staging/lustre/lustre/llite/file.c | 2 ++
2 files changed, 4 insertions(+)
diff --git a/drivers/staging/lustre/lustre/llite/dir.c b/drivers/staging/lustre/lustre/llite/dir.c
index 1084999..bfc914e 100644
--- a/drivers/staging/lustre/lustre/llite/dir.c
+++ b/drivers/staging/lustre/lustre/llite/dir.c
@@ -1561,6 +1561,8 @@ out_rmdir:
RETURN(rc);
OBD_ALLOC_LARGE(lmm, lmmsize);
+ if (lmm == NULL)
+ RETURN(-ENOMEM);
if (copy_from_user(lmm, lum, lmmsize))
GOTO(free_lmm, rc = -EFAULT);
diff --git a/drivers/staging/lustre/lustre/llite/file.c b/drivers/staging/lustre/lustre/llite/file.c
index efe66d9..927b3a0 100644
--- a/drivers/staging/lustre/lustre/llite/file.c
+++ b/drivers/staging/lustre/lustre/llite/file.c
@@ -55,6 +55,8 @@ struct ll_file_data *ll_file_data_get(void)
struct ll_file_data *fd;
OBD_SLAB_ALLOC_PTR_GFP(fd, ll_file_data_slab, __GFP_IO);
+ if (fd == NULL)
+ return NULL;
fd->fd_write_failed = false;
return fd;
}
--
1.7.9.5
^ permalink raw reply related [flat|nested] 62+ messages in thread* [PATCH 07/48] staging/lustre/llite: A not locked mutex can be unlocked.
2013-07-22 16:06 [PATCH 00/48] staging/lustre: minor cleanup and Intel Lustre sync Peng Tao
` (5 preceding siblings ...)
2013-07-22 16:06 ` [PATCH 06/48] staging/lustre/llite: check alloc in ll_file_data_get, ll_dir_ioctl Peng Tao
@ 2013-07-22 16:06 ` Peng Tao
2013-07-22 16:06 ` [PATCH 08/48] staging/lustre/llite: check ll_prep_md_op_data() using IS_ERR() Peng Tao
` (41 subsequent siblings)
48 siblings, 0 replies; 62+ messages in thread
From: Peng Tao @ 2013-07-22 16:06 UTC (permalink / raw)
To: Greg Kroah-Hartman; +Cc: linux-kernel, Dmitry Eremin, Peng Tao, Andreas Dilger
From: Dmitry Eremin <dmitry.eremin@intel.com>
In case of memory pressure a not locked mutex can be unlocked
in function ll_file_open(). This is not allowed and subsequent
behavior is not defined.
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-3157
Lustre-change: http://review.whamcloud.com/6028
Signed-off-by: Dmitry Eremin <dmitry.eremin@intel.com>
Reviewed-by: John Hammond <johnlockwoodhammond@gmail.com>
Reviewed-by: Nikitas Angelinas <nikitas_angelinas@xyratex.com>
Reviewed-by: Sebastien Buisson <sebastien.buisson@bull.net>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Signed-off-by: Peng Tao <tao.peng@emc.com>
Signed-off-by: Andreas Dilger <andreas.dilger@intel.com>
---
drivers/staging/lustre/lustre/llite/file.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/drivers/staging/lustre/lustre/llite/file.c b/drivers/staging/lustre/lustre/llite/file.c
index 927b3a0..717682c 100644
--- a/drivers/staging/lustre/lustre/llite/file.c
+++ b/drivers/staging/lustre/lustre/llite/file.c
@@ -526,7 +526,7 @@ int ll_file_open(struct inode *inode, struct file *file)
fd = ll_file_data_get();
if (fd == NULL)
- GOTO(out_och_free, rc = -ENOMEM);
+ GOTO(out_openerr, rc = -ENOMEM);
fd->fd_file = file;
if (S_ISDIR(inode->i_mode)) {
--
1.7.9.5
^ permalink raw reply related [flat|nested] 62+ messages in thread* [PATCH 08/48] staging/lustre/llite: check ll_prep_md_op_data() using IS_ERR()
2013-07-22 16:06 [PATCH 00/48] staging/lustre: minor cleanup and Intel Lustre sync Peng Tao
` (6 preceding siblings ...)
2013-07-22 16:06 ` [PATCH 07/48] staging/lustre/llite: A not locked mutex can be unlocked Peng Tao
@ 2013-07-22 16:06 ` Peng Tao
2013-07-22 16:06 ` [PATCH 09/48] staging/lustre/ldlm: print FID in lvbo_init(), lvbo_update Peng Tao
` (40 subsequent siblings)
48 siblings, 0 replies; 62+ messages in thread
From: Peng Tao @ 2013-07-22 16:06 UTC (permalink / raw)
To: Greg Kroah-Hartman
Cc: linux-kernel, John L. Hammond, Peng Tao, Andreas Dilger
From: "John L. Hammond" <john.hammond@intel.com>
In ll_file_ioctl() and ll_swap_layouts() check the result of
ll_prep_md_op_data() using IS_ERR().
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-3283
Lustre-change: http://review.whamcloud.com/6275
Signed-off-by: John L. Hammond <john.hammond@intel.com>
Reviewed-by: Jinshan Xiong <jinshan.xiong@intel.com>
Reviewed-by: Fan Yong <fan.yong@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>
Signed-off-by: Peng Tao <tao.peng@emc.com>
Signed-off-by: Andreas Dilger <andreas.dilger@intel.com>
---
drivers/staging/lustre/lustre/llite/file.c | 24 ++++++++++++------------
1 file changed, 12 insertions(+), 12 deletions(-)
diff --git a/drivers/staging/lustre/lustre/llite/file.c b/drivers/staging/lustre/lustre/llite/file.c
index 717682c..50ef6a5 100644
--- a/drivers/staging/lustre/lustre/llite/file.c
+++ b/drivers/staging/lustre/lustre/llite/file.c
@@ -1832,12 +1832,12 @@ static int ll_swap_layouts(struct file *file1, struct file *file2,
rc = -ENOMEM;
op_data = ll_prep_md_op_data(NULL, llss->inode1, llss->inode2, NULL, 0,
0, LUSTRE_OPC_ANY, &msl);
- if (op_data != NULL) {
- rc = obd_iocontrol(LL_IOC_LOV_SWAP_LAYOUTS,
- ll_i2mdexp(llss->inode1),
- sizeof(*op_data), op_data, NULL);
- ll_finish_md_op_data(op_data);
- }
+ if (IS_ERR(op_data))
+ GOTO(free, rc = PTR_ERR(op_data));
+
+ rc = obd_iocontrol(LL_IOC_LOV_SWAP_LAYOUTS, ll_i2mdexp(llss->inode1),
+ sizeof(*op_data), op_data, NULL);
+ ll_finish_md_op_data(op_data);
putgl:
if (gid != 0) {
@@ -2031,9 +2031,9 @@ long ll_file_ioctl(struct file *file, unsigned int cmd, unsigned long arg)
op_data = ll_prep_md_op_data(NULL, inode, NULL, NULL, 0, 0,
LUSTRE_OPC_ANY, hus);
- if (op_data == NULL) {
+ if (IS_ERR(op_data)) {
OBD_FREE_PTR(hus);
- RETURN(-ENOMEM);
+ RETURN(PTR_ERR(op_data));
}
rc = obd_iocontrol(cmd, ll_i2mdexp(inode), sizeof(*op_data),
@@ -2069,9 +2069,9 @@ long ll_file_ioctl(struct file *file, unsigned int cmd, unsigned long arg)
op_data = ll_prep_md_op_data(NULL, inode, NULL, NULL, 0, 0,
LUSTRE_OPC_ANY, hss);
- if (op_data == NULL) {
+ if (IS_ERR(op_data)) {
OBD_FREE_PTR(hss);
- RETURN(-ENOMEM);
+ RETURN(PTR_ERR(op_data));
}
rc = obd_iocontrol(cmd, ll_i2mdexp(inode), sizeof(*op_data),
@@ -2093,9 +2093,9 @@ long ll_file_ioctl(struct file *file, unsigned int cmd, unsigned long arg)
op_data = ll_prep_md_op_data(NULL, inode, NULL, NULL, 0, 0,
LUSTRE_OPC_ANY, hca);
- if (op_data == NULL) {
+ if (IS_ERR(op_data)) {
OBD_FREE_PTR(hca);
- RETURN(-ENOMEM);
+ RETURN(PTR_ERR(op_data));
}
rc = obd_iocontrol(cmd, ll_i2mdexp(inode), sizeof(*op_data),
--
1.7.9.5
^ permalink raw reply related [flat|nested] 62+ messages in thread* [PATCH 09/48] staging/lustre/ldlm: print FID in lvbo_init(), lvbo_update
2013-07-22 16:06 [PATCH 00/48] staging/lustre: minor cleanup and Intel Lustre sync Peng Tao
` (7 preceding siblings ...)
2013-07-22 16:06 ` [PATCH 08/48] staging/lustre/llite: check ll_prep_md_op_data() using IS_ERR() Peng Tao
@ 2013-07-22 16:06 ` Peng Tao
2013-07-22 16:06 ` [PATCH 10/48] staging/lustre/ptlrpc: race in pinger (use-after-free situation) Peng Tao
` (39 subsequent siblings)
48 siblings, 0 replies; 62+ messages in thread
From: Peng Tao @ 2013-07-22 16:06 UTC (permalink / raw)
To: Greg Kroah-Hartman; +Cc: linux-kernel, Andreas Dilger, Peng Tao
From: Andreas Dilger <andreas.dilger@intel.com>
Print the namespace and OBD device name, as well as the first two
lock resource fields (typically the FID) if there is an error with
loading the object from disk. This will be more important with
FID-on-OST and also the MDS. Using fid_extract_from_res_name() isn't
possible in the LDLM code, since the lock resource may not be a FID.
Make fid_extract_quota_resid() argument order and name consistent
with other fid_*_res() functions, with FID first and resource second.
Fix a bug in ofd_lvbo_init() where NULL lvb is accessed on error.
Print FID in ofd_lvbo_update() CDEBUG() and CERROR() messages.
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-2193
Lustre-change: http://review.whamcloud.com/4501
Signed-off-by: Andreas Dilger <andreas.dilger@intel.com>
Reviewed-by: Jinshan Xiong <jinshan.xiong@intel.com>
Reviewed-by: Bobi Jam <bobijam@gmail.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>
Signed-off-by: Peng Tao <tao.peng@emc.com>
Signed-off-by: Andreas Dilger <andreas.dilger@intel.com>
---
.../lustre/lustre/include/lustre/lustre_idl.h | 4 +-
drivers/staging/lustre/lustre/include/lustre_fid.h | 89 ++++++++++----------
drivers/staging/lustre/lustre/ldlm/ldlm_resource.c | 5 +-
3 files changed, 50 insertions(+), 48 deletions(-)
diff --git a/drivers/staging/lustre/lustre/include/lustre/lustre_idl.h b/drivers/staging/lustre/lustre/include/lustre/lustre_idl.h
index 8825460..1cfebfe 100644
--- a/drivers/staging/lustre/lustre/include/lustre/lustre_idl.h
+++ b/drivers/staging/lustre/lustre/include/lustre/lustre_idl.h
@@ -911,7 +911,7 @@ static inline int lu_fid_cmp(const struct lu_fid *f0,
__diff_normalize(fid_ver(f0), fid_ver(f1));
}
-static inline void ostid_cpu_to_le(struct ost_id *src_oi,
+static inline void ostid_cpu_to_le(const struct ost_id *src_oi,
struct ost_id *dst_oi)
{
if (fid_seq_is_mdt0(ostid_seq(src_oi))) {
@@ -922,7 +922,7 @@ static inline void ostid_cpu_to_le(struct ost_id *src_oi,
}
}
-static inline void ostid_le_to_cpu(struct ost_id *src_oi,
+static inline void ostid_le_to_cpu(const struct ost_id *src_oi,
struct ost_id *dst_oi)
{
if (fid_seq_is_mdt0(ostid_seq(src_oi))) {
diff --git a/drivers/staging/lustre/lustre/include/lustre_fid.h b/drivers/staging/lustre/lustre/include/lustre_fid.h
index 7d20cba..7523b40 100644
--- a/drivers/staging/lustre/lustre/include/lustre_fid.h
+++ b/drivers/staging/lustre/lustre/include/lustre_fid.h
@@ -488,74 +488,75 @@ struct ldlm_namespace;
* renaming name[2,3] fields that need to be used for the quota identifier.
*/
static inline struct ldlm_res_id *
-fid_build_reg_res_name(const struct lu_fid *f,
- struct ldlm_res_id *name)
+fid_build_reg_res_name(const struct lu_fid *fid, struct ldlm_res_id *res)
{
- memset(name, 0, sizeof *name);
- name->name[LUSTRE_RES_ID_SEQ_OFF] = fid_seq(f);
- name->name[LUSTRE_RES_ID_VER_OID_OFF] = fid_ver_oid(f);
- return name;
+ memset(res, 0, sizeof(*res));
+ res->name[LUSTRE_RES_ID_SEQ_OFF] = fid_seq(fid);
+ res->name[LUSTRE_RES_ID_VER_OID_OFF] = fid_ver_oid(fid);
+
+ return res;
+}
+
+/*
+ * Return true if resource is for object identified by FID.
+ */
+static inline int fid_res_name_eq(const struct lu_fid *fid,
+ const struct ldlm_res_id *res)
+{
+ return res->name[LUSTRE_RES_ID_SEQ_OFF] == fid_seq(fid) &&
+ res->name[LUSTRE_RES_ID_VER_OID_OFF] == fid_ver_oid(fid);
+}
+
+/*
+ * Extract FID from LDLM resource. Reverse of fid_build_reg_res_name().
+ */
+static inline struct lu_fid *
+fid_extract_from_res_name(struct lu_fid *fid, const struct ldlm_res_id *res)
+{
+ fid->f_seq = res->name[LUSTRE_RES_ID_SEQ_OFF];
+ fid->f_oid = (__u32)(res->name[LUSTRE_RES_ID_VER_OID_OFF]);
+ fid->f_ver = (__u32)(res->name[LUSTRE_RES_ID_VER_OID_OFF] >> 32);
+ LASSERT(fid_res_name_eq(fid, res));
+
+ return fid;
}
/*
* Build (DLM) resource identifier from global quota FID and quota ID.
*/
static inline struct ldlm_res_id *
-fid_build_quota_resid(const struct lu_fid *glb_fid, union lquota_id *qid,
+fid_build_quota_res_name(const struct lu_fid *glb_fid, union lquota_id *qid,
struct ldlm_res_id *res)
{
fid_build_reg_res_name(glb_fid, res);
res->name[LUSTRE_RES_ID_QUOTA_SEQ_OFF] = fid_seq(&qid->qid_fid);
res->name[LUSTRE_RES_ID_QUOTA_VER_OID_OFF] = fid_ver_oid(&qid->qid_fid);
+
return res;
}
/*
* Extract global FID and quota ID from resource name
*/
-static inline void fid_extract_quota_resid(struct ldlm_res_id *res,
- struct lu_fid *glb_fid,
- union lquota_id *qid)
+static inline void fid_extract_from_quota_res(struct lu_fid *glb_fid,
+ union lquota_id *qid,
+ const struct ldlm_res_id *res)
{
- glb_fid->f_seq = res->name[LUSTRE_RES_ID_SEQ_OFF];
- glb_fid->f_oid = (__u32)res->name[LUSTRE_RES_ID_VER_OID_OFF];
- glb_fid->f_ver = (__u32)(res->name[LUSTRE_RES_ID_VER_OID_OFF] >> 32);
-
+ fid_extract_from_res_name(glb_fid, res);
qid->qid_fid.f_seq = res->name[LUSTRE_RES_ID_QUOTA_SEQ_OFF];
qid->qid_fid.f_oid = (__u32)res->name[LUSTRE_RES_ID_QUOTA_VER_OID_OFF];
qid->qid_fid.f_ver =
(__u32)(res->name[LUSTRE_RES_ID_QUOTA_VER_OID_OFF] >> 32);
}
-/*
- * Return true if resource is for object identified by fid.
- */
-static inline int fid_res_name_eq(const struct lu_fid *f,
- const struct ldlm_res_id *name)
-{
- return name->name[LUSTRE_RES_ID_SEQ_OFF] == fid_seq(f) &&
- name->name[LUSTRE_RES_ID_VER_OID_OFF] == fid_ver_oid(f);
-}
-
-/* reverse function of fid_build_reg_res_name() */
-static inline void fid_build_from_res_name(struct lu_fid *f,
- const struct ldlm_res_id *name)
-{
- fid_zero(f);
- f->f_seq = name->name[LUSTRE_RES_ID_SEQ_OFF];
- f->f_oid = name->name[LUSTRE_RES_ID_VER_OID_OFF] & 0xffffffff;
- f->f_ver = name->name[LUSTRE_RES_ID_VER_OID_OFF] >> 32;
- LASSERT(fid_res_name_eq(f, name));
-}
-
static inline struct ldlm_res_id *
-fid_build_pdo_res_name(const struct lu_fid *f,
- unsigned int hash,
- struct ldlm_res_id *name)
+fid_build_pdo_res_name(const struct lu_fid *fid, unsigned int hash,
+ struct ldlm_res_id *res)
{
- fid_build_reg_res_name(f, name);
- name->name[LUSTRE_RES_ID_HSH_OFF] = hash;
- return name;
+ fid_build_reg_res_name(fid, res);
+ res->name[LUSTRE_RES_ID_HSH_OFF] = hash;
+
+ return res;
}
/**
@@ -584,7 +585,7 @@ static inline void ostid_build_res_name(struct ost_id *oi,
name->name[LUSTRE_RES_ID_SEQ_OFF] = ostid_id(oi);
name->name[LUSTRE_RES_ID_VER_OID_OFF] = ostid_seq(oi);
} else {
- fid_build_reg_res_name((struct lu_fid *)oi, name);
+ fid_build_reg_res_name(&oi->oi_fid, name);
}
}
@@ -597,7 +598,7 @@ static inline void ostid_res_name_to_id(struct ost_id *oi,
ostid_set_id(oi, name->name[LUSTRE_RES_ID_SEQ_OFF]);
} else {
/* new resid */
- fid_build_from_res_name((struct lu_fid *)oi, name);
+ fid_extract_from_res_name(&oi->oi_fid, name);
}
}
@@ -644,7 +645,7 @@ static inline void ost_fid_from_resid(struct lu_fid *fid,
ostid_to_fid(fid, &oi, 0);
} else {
/* new resid */
- fid_build_from_res_name(fid, name);
+ fid_extract_from_res_name(fid, name);
}
}
diff --git a/drivers/staging/lustre/lustre/ldlm/ldlm_resource.c b/drivers/staging/lustre/lustre/ldlm/ldlm_resource.c
index 9052dc5..cb8659f 100644
--- a/drivers/staging/lustre/lustre/ldlm/ldlm_resource.c
+++ b/drivers/staging/lustre/lustre/ldlm/ldlm_resource.c
@@ -1128,8 +1128,9 @@ ldlm_resource_get(struct ldlm_namespace *ns, struct ldlm_resource *parent,
OBD_FAIL_TIMEOUT(OBD_FAIL_LDLM_CREATE_RESOURCE, 2);
rc = ns->ns_lvbo->lvbo_init(res);
if (rc < 0) {
- CERROR("lvbo_init failed for resource "
- LPU64": rc %d\n", name->name[0], rc);
+ CERROR("%s: lvbo_init failed for resource "LPX64":"
+ LPX64": rc = %d\n", ns->ns_obd->obd_name,
+ name->name[0], name->name[1], rc);
if (res->lr_lvb_data) {
OBD_FREE(res->lr_lvb_data, res->lr_lvb_len);
res->lr_lvb_data = NULL;
--
1.7.9.5
^ permalink raw reply related [flat|nested] 62+ messages in thread* [PATCH 10/48] staging/lustre/ptlrpc: race in pinger (use-after-free situation)
2013-07-22 16:06 [PATCH 00/48] staging/lustre: minor cleanup and Intel Lustre sync Peng Tao
` (8 preceding siblings ...)
2013-07-22 16:06 ` [PATCH 09/48] staging/lustre/ldlm: print FID in lvbo_init(), lvbo_update Peng Tao
@ 2013-07-22 16:06 ` Peng Tao
2013-07-22 16:06 ` [PATCH 11/48] staging/lustre/ptlrpc: Translate between host and network errnos Peng Tao
` (38 subsequent siblings)
48 siblings, 0 replies; 62+ messages in thread
From: Peng Tao @ 2013-07-22 16:06 UTC (permalink / raw)
To: Greg Kroah-Hartman
Cc: linux-kernel, Dmitry Eremin, Li Wei, Peng Tao, Andreas Dilger
From: Dmitry Eremin <dmitry.eremin@intel.com>
The race is result of use-after-free situation:
~ ptlrpc_stop_pinger() ~ ptlrpc_pinger_main()
---------------------------------------------------------------
thread_set_flags(SVC_STOPPING)
cfs_waitq_signal(pinger_thread) ...
... thread_set_flags(SVC_STOPPED)
l_wait_event(thread_is_stopped)
OBD_FREE_PTR(pinger_thread)
... cfs_waitq_signal(pinger_thread)
---------------------------------------------------------------
The memory used by pinger_thread might have been freed and
reallocated to something else, when ptlrpc_pinger_main()
used it in cvs_waitq_signal().
Signed-off-by: Li Wei <wei.g.li@intel.com>
Signed-off-by: Dmitry Eremin <dmitry.eremin@intel.com>
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-3032
Lustre-change: http://review.whamcloud.com/6040
Reviewed-by: Faccini Bruno <bruno.faccini@intel.com>
Reviewed-by: Mike Pershin <mike.pershin@intel.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Signed-off-by: Peng Tao <tao.peng@emc.com>
Signed-off-by: Andreas Dilger <andreas.dilger@intel.com>
---
drivers/staging/lustre/lustre/ptlrpc/pinger.c | 39 ++++++++++---------------
1 file changed, 16 insertions(+), 23 deletions(-)
diff --git a/drivers/staging/lustre/lustre/ptlrpc/pinger.c b/drivers/staging/lustre/lustre/ptlrpc/pinger.c
index ef5269a..f521251 100644
--- a/drivers/staging/lustre/lustre/ptlrpc/pinger.c
+++ b/drivers/staging/lustre/lustre/ptlrpc/pinger.c
@@ -369,7 +369,7 @@ static int ptlrpc_pinger_main(void *arg)
return 0;
}
-static struct ptlrpc_thread *pinger_thread = NULL;
+static struct ptlrpc_thread pinger_thread;
int ptlrpc_start_pinger(void)
{
@@ -377,29 +377,25 @@ int ptlrpc_start_pinger(void)
int rc;
ENTRY;
- if (pinger_thread != NULL)
+ if (!thread_is_init(&pinger_thread) &&
+ !thread_is_stopped(&pinger_thread))
RETURN(-EALREADY);
- OBD_ALLOC_PTR(pinger_thread);
- if (pinger_thread == NULL)
- RETURN(-ENOMEM);
- init_waitqueue_head(&pinger_thread->t_ctl_waitq);
+ init_waitqueue_head(&pinger_thread.t_ctl_waitq);
init_waitqueue_head(&suspend_timeouts_waitq);
- strcpy(pinger_thread->t_name, "ll_ping");
+ strcpy(pinger_thread.t_name, "ll_ping");
/* CLONE_VM and CLONE_FILES just avoid a needless copy, because we
* just drop the VM and FILES in cfs_daemonize_ctxt() right away. */
rc = PTR_ERR(kthread_run(ptlrpc_pinger_main,
- pinger_thread, pinger_thread->t_name));
+ &pinger_thread, pinger_thread.t_name));
if (IS_ERR_VALUE(rc)) {
CERROR("cannot start thread: %d\n", rc);
- OBD_FREE(pinger_thread, sizeof(*pinger_thread));
- pinger_thread = NULL;
RETURN(rc);
}
- l_wait_event(pinger_thread->t_ctl_waitq,
- thread_is_running(pinger_thread), &lwi);
+ l_wait_event(pinger_thread.t_ctl_waitq,
+ thread_is_running(&pinger_thread), &lwi);
if (suppress_pings)
CWARN("Pings will be suppressed at the request of the "
@@ -419,20 +415,17 @@ int ptlrpc_stop_pinger(void)
int rc = 0;
ENTRY;
- if (pinger_thread == NULL)
+ if (!thread_is_init(&pinger_thread) &&
+ !thread_is_stopped(&pinger_thread))
RETURN(-EALREADY);
ptlrpc_pinger_remove_timeouts();
- mutex_lock(&pinger_mutex);
- thread_set_flags(pinger_thread, SVC_STOPPING);
- wake_up(&pinger_thread->t_ctl_waitq);
- mutex_unlock(&pinger_mutex);
+ thread_set_flags(&pinger_thread, SVC_STOPPING);
+ wake_up(&pinger_thread.t_ctl_waitq);
- l_wait_event(pinger_thread->t_ctl_waitq,
- thread_is_stopped(pinger_thread), &lwi);
+ l_wait_event(pinger_thread.t_ctl_waitq,
+ thread_is_stopped(&pinger_thread), &lwi);
- OBD_FREE_PTR(pinger_thread);
- pinger_thread = NULL;
RETURN(rc);
}
@@ -617,8 +610,8 @@ int ptlrpc_pinger_remove_timeouts(void)
void ptlrpc_pinger_wake_up()
{
- thread_add_flags(pinger_thread, SVC_EVENT);
- wake_up(&pinger_thread->t_ctl_waitq);
+ thread_add_flags(&pinger_thread, SVC_EVENT);
+ wake_up(&pinger_thread.t_ctl_waitq);
}
/* Ping evictor thread */
--
1.7.9.5
^ permalink raw reply related [flat|nested] 62+ messages in thread* [PATCH 11/48] staging/lustre/ptlrpc: Translate between host and network errnos
2013-07-22 16:06 [PATCH 00/48] staging/lustre: minor cleanup and Intel Lustre sync Peng Tao
` (9 preceding siblings ...)
2013-07-22 16:06 ` [PATCH 10/48] staging/lustre/ptlrpc: race in pinger (use-after-free situation) Peng Tao
@ 2013-07-22 16:06 ` Peng Tao
2013-07-22 16:29 ` Paul Bolle
2013-07-22 16:06 ` [PATCH 12/48] staging/lustre/mdc: layout lock rpc must not take rpc_lock Peng Tao
` (37 subsequent siblings)
48 siblings, 1 reply; 62+ messages in thread
From: Peng Tao @ 2013-07-22 16:06 UTC (permalink / raw)
To: Greg Kroah-Hartman; +Cc: linux-kernel, Li Wei, Peng Tao, Andreas Dilger
From: Li Wei <wei.g.li@intel.com>
Lustre puts system errors (e.g., ENOTCONN) on wire as numbers
essentially specific to senders' architectures. While this is fine
for x86-only sites, where receivers share the same error number
definition with senders, problems will arise, however, for sites
involving multiple architectures with different error number
definitions. For instance, an ENOTCONN reply from a sparc server will
be put on wire as -57, which, for an x86 client, means EBADSLT
instead.
To solve the problem, this patch defines a set of network errors for
on-wire or on-disk uses. These errors correspond to a subset of the
x86 system errors and share the same number definition, maintaining
compatibility with existing x86 clients and servers.
Then, either error numbers could be translated at run time, or all
host errors going on wire could be replaced with network errors in the
code. This patch does the former by introducing both generic and
field-specific translation routines and calling them at proper places,
so that translations for existing fields are transparent.
(Personally, I tend to think the latter way might be worthwhile, as it
is more straightforward conceptually. Do we really need so many
different errors? Should errors returned by kernel routines really be
passed up and eventually put on wire? There could even be security
implications in that.)
Thank Fujitsu for the original idea and their contributions that make
this available upstream.
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-2743
Lustre-change: http://review.whamcloud.com/5577
Signed-off-by: Li Wei <wei.g.li@intel.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Reviewed-by: Hiroya Nozaki <nozaki.hiroya@jp.fujitsu.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>
Signed-off-by: Peng Tao <tao.peng@emc.com>
Signed-off-by: Andreas Dilger <andreas.dilger@intel.com>
---
drivers/staging/lustre/lustre/Kconfig | 5 +
.../lustre/lustre/include/lustre/lustre_errno.h | 215 +++++++++++
.../lustre/lustre/include/lustre/lustre_idl.h | 2 +
drivers/staging/lustre/lustre/include/lustre_net.h | 32 ++
drivers/staging/lustre/lustre/ldlm/ldlm_lock.c | 3 +-
drivers/staging/lustre/lustre/ldlm/ldlm_lockd.c | 2 +
drivers/staging/lustre/lustre/ldlm/ldlm_request.c | 6 +-
drivers/staging/lustre/lustre/mdc/mdc_locks.c | 10 +
drivers/staging/lustre/lustre/mdc/mdc_request.c | 1 +
drivers/staging/lustre/lustre/osc/osc_request.c | 2 +
drivers/staging/lustre/lustre/ptlrpc/Makefile | 1 +
drivers/staging/lustre/lustre/ptlrpc/errno.c | 382 ++++++++++++++++++++
drivers/staging/lustre/lustre/ptlrpc/niobuf.c | 3 +-
.../staging/lustre/lustre/ptlrpc/pack_generic.c | 3 +
14 files changed, 662 insertions(+), 5 deletions(-)
create mode 100644 drivers/staging/lustre/lustre/include/lustre/lustre_errno.h
create mode 100644 drivers/staging/lustre/lustre/ptlrpc/errno.c
diff --git a/drivers/staging/lustre/lustre/Kconfig b/drivers/staging/lustre/lustre/Kconfig
index 0a8eb1ad..19a1c33 100644
--- a/drivers/staging/lustre/lustre/Kconfig
+++ b/drivers/staging/lustre/lustre/Kconfig
@@ -59,3 +59,8 @@ config LUSTRE_DEBUG_EXPENSIVE_CHECK
expensive checks that may have a performance impact.
Use with caution. If unsure, say N.
+
+config LUSTRE_TRANSLATE_ERRNOS
+ bool
+ depends on LUSTRE_FS && !X86
+ default true
diff --git a/drivers/staging/lustre/lustre/include/lustre/lustre_errno.h b/drivers/staging/lustre/lustre/include/lustre/lustre_errno.h
new file mode 100644
index 0000000..2870487
--- /dev/null
+++ b/drivers/staging/lustre/lustre/include/lustre/lustre_errno.h
@@ -0,0 +1,215 @@
+/*
+ * GPL HEADER START
+ *
+ * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
+ *
+ * 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 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).
+ *
+ * You should have received a copy of the GNU General Public License
+ * version 2 along with this program; If not, see
+ * http://www.gnu.org/licenses/gpl-2.0.txt
+ *
+ * GPL HEADER END
+ */
+/*
+ * Copyright (C) 2011 FUJITSU LIMITED. All rights reserved.
+ *
+ * Copyright (c) 2013, Intel Corporation.
+ */
+
+#ifndef LUSTRE_ERRNO_H
+#define LUSTRE_ERRNO_H
+
+/*
+ * Only "network" errnos, which are defined below, are allowed on wire (or on
+ * disk). Generic routines exist to help translate between these and a subset
+ * of the "host" errnos. Some host errnos (e.g., EDEADLOCK) are intentionally
+ * left out. See also the comment on lustre_errno_hton_mapping[].
+ *
+ * To maintain compatibility with existing x86 clients and servers, each of
+ * these network errnos has the same numerical value as its corresponding host
+ * errno on x86.
+ */
+#define LUSTRE_EPERM 1 /* Operation not permitted */
+#define LUSTRE_ENOENT 2 /* No such file or directory */
+#define LUSTRE_ESRCH 3 /* No such process */
+#define LUSTRE_EINTR 4 /* Interrupted system call */
+#define LUSTRE_EIO 5 /* I/O error */
+#define LUSTRE_ENXIO 6 /* No such device or address */
+#define LUSTRE_E2BIG 7 /* Argument list too long */
+#define LUSTRE_ENOEXEC 8 /* Exec format error */
+#define LUSTRE_EBADF 9 /* Bad file number */
+#define LUSTRE_ECHILD 10 /* No child processes */
+#define LUSTRE_EAGAIN 11 /* Try again */
+#define LUSTRE_ENOMEM 12 /* Out of memory */
+#define LUSTRE_EACCES 13 /* Permission denied */
+#define LUSTRE_EFAULT 14 /* Bad address */
+#define LUSTRE_ENOTBLK 15 /* Block device required */
+#define LUSTRE_EBUSY 16 /* Device or resource busy */
+#define LUSTRE_EEXIST 17 /* File exists */
+#define LUSTRE_EXDEV 18 /* Cross-device link */
+#define LUSTRE_ENODEV 19 /* No such device */
+#define LUSTRE_ENOTDIR 20 /* Not a directory */
+#define LUSTRE_EISDIR 21 /* Is a directory */
+#define LUSTRE_EINVAL 22 /* Invalid argument */
+#define LUSTRE_ENFILE 23 /* File table overflow */
+#define LUSTRE_EMFILE 24 /* Too many open files */
+#define LUSTRE_ENOTTY 25 /* Not a typewriter */
+#define LUSTRE_ETXTBSY 26 /* Text file busy */
+#define LUSTRE_EFBIG 27 /* File too large */
+#define LUSTRE_ENOSPC 28 /* No space left on device */
+#define LUSTRE_ESPIPE 29 /* Illegal seek */
+#define LUSTRE_EROFS 30 /* Read-only file system */
+#define LUSTRE_EMLINK 31 /* Too many links */
+#define LUSTRE_EPIPE 32 /* Broken pipe */
+#define LUSTRE_EDOM 33 /* Math argument out of domain of
+ func */
+#define LUSTRE_ERANGE 34 /* Math result not representable */
+#define LUSTRE_EDEADLK 35 /* Resource deadlock would occur */
+#define LUSTRE_ENAMETOOLONG 36 /* File name too long */
+#define LUSTRE_ENOLCK 37 /* No record locks available */
+#define LUSTRE_ENOSYS 38 /* Function not implemented */
+#define LUSTRE_ENOTEMPTY 39 /* Directory not empty */
+#define LUSTRE_ELOOP 40 /* Too many symbolic links
+ encountered */
+#define LUSTRE_ENOMSG 42 /* No message of desired type */
+#define LUSTRE_EIDRM 43 /* Identifier removed */
+#define LUSTRE_ECHRNG 44 /* Channel number out of range */
+#define LUSTRE_EL2NSYNC 45 /* Level 2 not synchronized */
+#define LUSTRE_EL3HLT 46 /* Level 3 halted */
+#define LUSTRE_EL3RST 47 /* Level 3 reset */
+#define LUSTRE_ELNRNG 48 /* Link number out of range */
+#define LUSTRE_EUNATCH 49 /* Protocol driver not attached */
+#define LUSTRE_ENOCSI 50 /* No CSI structure available */
+#define LUSTRE_EL2HLT 51 /* Level 2 halted */
+#define LUSTRE_EBADE 52 /* Invalid exchange */
+#define LUSTRE_EBADR 53 /* Invalid request descriptor */
+#define LUSTRE_EXFULL 54 /* Exchange full */
+#define LUSTRE_ENOANO 55 /* No anode */
+#define LUSTRE_EBADRQC 56 /* Invalid request code */
+#define LUSTRE_EBADSLT 57 /* Invalid slot */
+#define LUSTRE_EBFONT 59 /* Bad font file format */
+#define LUSTRE_ENOSTR 60 /* Device not a stream */
+#define LUSTRE_ENODATA 61 /* No data available */
+#define LUSTRE_ETIME 62 /* Timer expired */
+#define LUSTRE_ENOSR 63 /* Out of streams resources */
+#define LUSTRE_ENONET 64 /* Machine is not on the network */
+#define LUSTRE_ENOPKG 65 /* Package not installed */
+#define LUSTRE_EREMOTE 66 /* Object is remote */
+#define LUSTRE_ENOLINK 67 /* Link has been severed */
+#define LUSTRE_EADV 68 /* Advertise error */
+#define LUSTRE_ESRMNT 69 /* Srmount error */
+#define LUSTRE_ECOMM 70 /* Communication error on send */
+#define LUSTRE_EPROTO 71 /* Protocol error */
+#define LUSTRE_EMULTIHOP 72 /* Multihop attempted */
+#define LUSTRE_EDOTDOT 73 /* RFS specific error */
+#define LUSTRE_EBADMSG 74 /* Not a data message */
+#define LUSTRE_EOVERFLOW 75 /* Value too large for defined data
+ type */
+#define LUSTRE_ENOTUNIQ 76 /* Name not unique on network */
+#define LUSTRE_EBADFD 77 /* File descriptor in bad state */
+#define LUSTRE_EREMCHG 78 /* Remote address changed */
+#define LUSTRE_ELIBACC 79 /* Can not access a needed shared
+ library */
+#define LUSTRE_ELIBBAD 80 /* Accessing a corrupted shared
+ library */
+#define LUSTRE_ELIBSCN 81 /* .lib section in a.out corrupted */
+#define LUSTRE_ELIBMAX 82 /* Attempting to link in too many shared
+ libraries */
+#define LUSTRE_ELIBEXEC 83 /* Cannot exec a shared library
+ directly */
+#define LUSTRE_EILSEQ 84 /* Illegal byte sequence */
+#define LUSTRE_ERESTART 85 /* Interrupted system call should be
+ restarted */
+#define LUSTRE_ESTRPIPE 86 /* Streams pipe error */
+#define LUSTRE_EUSERS 87 /* Too many users */
+#define LUSTRE_ENOTSOCK 88 /* Socket operation on non-socket */
+#define LUSTRE_EDESTADDRREQ 89 /* Destination address required */
+#define LUSTRE_EMSGSIZE 90 /* Message too long */
+#define LUSTRE_EPROTOTYPE 91 /* Protocol wrong type for socket */
+#define LUSTRE_ENOPROTOOPT 92 /* Protocol not available */
+#define LUSTRE_EPROTONOSUPPORT 93 /* Protocol not supported */
+#define LUSTRE_ESOCKTNOSUPPORT 94 /* Socket type not supported */
+#define LUSTRE_EOPNOTSUPP 95 /* Operation not supported on transport
+ endpoint */
+#define LUSTRE_EPFNOSUPPORT 96 /* Protocol family not supported */
+#define LUSTRE_EAFNOSUPPORT 97 /* Address family not supported by
+ protocol */
+#define LUSTRE_EADDRINUSE 98 /* Address already in use */
+#define LUSTRE_EADDRNOTAVAIL 99 /* Cannot assign requested address */
+#define LUSTRE_ENETDOWN 100 /* Network is down */
+#define LUSTRE_ENETUNREACH 101 /* Network is unreachable */
+#define LUSTRE_ENETRESET 102 /* Network dropped connection because of
+ reset */
+#define LUSTRE_ECONNABORTED 103 /* Software caused connection abort */
+#define LUSTRE_ECONNRESET 104 /* Connection reset by peer */
+#define LUSTRE_ENOBUFS 105 /* No buffer space available */
+#define LUSTRE_EISCONN 106 /* Transport endpoint is already
+ connected */
+#define LUSTRE_ENOTCONN 107 /* Transport endpoint is not
+ connected */
+#define LUSTRE_ESHUTDOWN 108 /* Cannot send after transport endpoint
+ shutdown */
+#define LUSTRE_ETOOMANYREFS 109 /* Too many references: cannot splice */
+#define LUSTRE_ETIMEDOUT 110 /* Connection timed out */
+#define LUSTRE_ECONNREFUSED 111 /* Connection refused */
+#define LUSTRE_EHOSTDOWN 112 /* Host is down */
+#define LUSTRE_EHOSTUNREACH 113 /* No route to host */
+#define LUSTRE_EALREADY 114 /* Operation already in progress */
+#define LUSTRE_EINPROGRESS 115 /* Operation now in progress */
+#define LUSTRE_ESTALE 116 /* Stale NFS file handle */
+#define LUSTRE_EUCLEAN 117 /* Structure needs cleaning */
+#define LUSTRE_ENOTNAM 118 /* Not a XENIX named type file */
+#define LUSTRE_ENAVAIL 119 /* No XENIX semaphores available */
+#define LUSTRE_EISNAM 120 /* Is a named type file */
+#define LUSTRE_EREMOTEIO 121 /* Remote I/O error */
+#define LUSTRE_EDQUOT 122 /* Quota exceeded */
+#define LUSTRE_ENOMEDIUM 123 /* No medium found */
+#define LUSTRE_EMEDIUMTYPE 124 /* Wrong medium type */
+#define LUSTRE_ECANCELED 125 /* Operation Canceled */
+#define LUSTRE_ENOKEY 126 /* Required key not available */
+#define LUSTRE_EKEYEXPIRED 127 /* Key has expired */
+#define LUSTRE_EKEYREVOKED 128 /* Key has been revoked */
+#define LUSTRE_EKEYREJECTED 129 /* Key was rejected by service */
+#define LUSTRE_EOWNERDEAD 130 /* Owner died */
+#define LUSTRE_ENOTRECOVERABLE 131 /* State not recoverable */
+#define LUSTRE_ERESTARTSYS 512
+#define LUSTRE_ERESTARTNOINTR 513
+#define LUSTRE_ERESTARTNOHAND 514 /* restart if no handler.. */
+#define LUSTRE_ENOIOCTLCMD 515 /* No ioctl command */
+#define LUSTRE_ERESTART_RESTARTBLOCK 516 /* restart by calling
+ sys_restart_syscall */
+#define LUSTRE_EBADHANDLE 521 /* Illegal NFS file handle */
+#define LUSTRE_ENOTSYNC 522 /* Update synchronization mismatch */
+#define LUSTRE_EBADCOOKIE 523 /* Cookie is stale */
+#define LUSTRE_ENOTSUPP 524 /* Operation is not supported */
+#define LUSTRE_ETOOSMALL 525 /* Buffer or request is too small */
+#define LUSTRE_ESERVERFAULT 526 /* An untranslatable error occurred */
+#define LUSTRE_EBADTYPE 527 /* Type not supported by server */
+#define LUSTRE_EJUKEBOX 528 /* Request initiated, but will not
+ complete before timeout */
+#define LUSTRE_EIOCBQUEUED 529 /* iocb queued, will get completion
+ event */
+#define LUSTRE_EIOCBRETRY 530 /* iocb queued, will trigger a retry */
+
+/*
+ * Translations are optimized away on x86. Host errnos that shouldn't be put
+ * on wire could leak through as a result. Do not count on this side effect.
+ */
+#ifdef CONFIG_LUSTRE_TRANSLATE_ERRNOS
+unsigned int lustre_errno_hton(unsigned int h);
+unsigned int lustre_errno_ntoh(unsigned int n);
+#else
+#define lustre_errno_hton(h) (h)
+#define lustre_errno_ntoh(n) (n)
+#endif
+
+#endif /* LUSTRE_ERRNO_H */
diff --git a/drivers/staging/lustre/lustre/include/lustre/lustre_idl.h b/drivers/staging/lustre/lustre/include/lustre/lustre_idl.h
index 1cfebfe..4829419 100644
--- a/drivers/staging/lustre/lustre/include/lustre/lustre_idl.h
+++ b/drivers/staging/lustre/lustre/include/lustre/lustre_idl.h
@@ -98,6 +98,8 @@
/* Defn's shared with user-space. */
#include <lustre/lustre_user.h>
+#include <lustre/lustre_errno.h>
+
/*
* GENERAL STUFF
*/
diff --git a/drivers/staging/lustre/lustre/include/lustre_net.h b/drivers/staging/lustre/lustre/include/lustre_net.h
index 1b99b5d..8d7cc7a 100644
--- a/drivers/staging/lustre/lustre/include/lustre_net.h
+++ b/drivers/staging/lustre/lustre/include/lustre_net.h
@@ -3166,6 +3166,38 @@ lustre_shrink_reply(struct ptlrpc_request *req, int segment,
req->rq_replen = lustre_shrink_msg(req->rq_repmsg, segment,
newlen, move_data);
}
+
+#ifdef CONFIG_LUSTRE_TRANSLATE_ERRNOS
+
+static inline int ptlrpc_status_hton(int h)
+{
+ /*
+ * Positive errnos must be network errnos, such as LUSTRE_EDEADLK,
+ * ELDLM_LOCK_ABORTED, etc.
+ */
+ if (h < 0)
+ return -lustre_errno_hton(-h);
+ else
+ return h;
+}
+
+static inline int ptlrpc_status_ntoh(int n)
+{
+ /*
+ * See the comment in ptlrpc_status_hton().
+ */
+ if (n < 0)
+ return -lustre_errno_ntoh(-n);
+ else
+ return n;
+}
+
+#else
+
+#define ptlrpc_status_hton(h) (h)
+#define ptlrpc_status_ntoh(n) (n)
+
+#endif
/** @} */
/** Change request phase of \a req to \a new_phase */
diff --git a/drivers/staging/lustre/lustre/ldlm/ldlm_lock.c b/drivers/staging/lustre/lustre/ldlm/ldlm_lock.c
index 93badf5..0b3ea88 100644
--- a/drivers/staging/lustre/lustre/ldlm/ldlm_lock.c
+++ b/drivers/staging/lustre/lustre/ldlm/ldlm_lock.c
@@ -2208,7 +2208,8 @@ struct ldlm_resource *ldlm_lock_convert(struct ldlm_lock *lock, int new_mode,
/* I can't check the type of lock here because the bitlock of lock
* is not held here, so do the allocation blindly. -jay */
OBD_SLAB_ALLOC_PTR_GFP(node, ldlm_interval_slab, __GFP_IO);
- if (node == NULL) /* Actually, this causes EDEADLOCK to be returned */
+ if (node == NULL)
+ /* Actually, this causes EDEADLOCK to be returned */
RETURN(NULL);
LASSERTF((new_mode == LCK_PW && lock->l_granted_mode == LCK_PR),
diff --git a/drivers/staging/lustre/lustre/ldlm/ldlm_lockd.c b/drivers/staging/lustre/lustre/ldlm/ldlm_lockd.c
index 324d5e4..f79b244 100644
--- a/drivers/staging/lustre/lustre/ldlm/ldlm_lockd.c
+++ b/drivers/staging/lustre/lustre/ldlm/ldlm_lockd.c
@@ -555,6 +555,8 @@ static int ldlm_handle_qc_callback(struct ptlrpc_request *req)
RETURN(-EPROTO);
}
+ oqctl->qc_stat = ptlrpc_status_ntoh(oqctl->qc_stat);
+
cli->cl_qchk_stat = oqctl->qc_stat;
return 0;
}
diff --git a/drivers/staging/lustre/lustre/ldlm/ldlm_request.c b/drivers/staging/lustre/lustre/ldlm/ldlm_request.c
index 1a690ed..ab41fea 100644
--- a/drivers/staging/lustre/lustre/ldlm/ldlm_request.c
+++ b/drivers/staging/lustre/lustre/ldlm/ldlm_request.c
@@ -1023,7 +1023,7 @@ static int ldlm_cli_convert_local(struct ldlm_lock *lock, int new_mode,
ldlm_reprocess_all(res);
rc = 0;
} else {
- rc = EDEADLOCK;
+ rc = LUSTRE_EDEADLK;
}
LDLM_DEBUG(lock, "client-side local convert handler END");
LDLM_LOCK_PUT(lock);
@@ -1095,7 +1095,7 @@ int ldlm_cli_convert(struct lustre_handle *lockh, int new_mode, __u32 *flags)
GOTO(out, rc);
}
} else {
- rc = EDEADLOCK;
+ rc = LUSTRE_EDEADLK;
}
EXIT;
out:
@@ -1248,7 +1248,7 @@ int ldlm_cli_cancel_req(struct obd_export *exp, struct list_head *cancels,
} else {
rc = ptlrpc_queue_wait(req);
}
- if (rc == ESTALE) {
+ if (rc == LUSTRE_ESTALE) {
CDEBUG(D_DLMTRACE, "client/server (nid %s) "
"out of sync -- not fatal\n",
libcfs_nid2str(req->rq_import->
diff --git a/drivers/staging/lustre/lustre/mdc/mdc_locks.c b/drivers/staging/lustre/lustre/mdc/mdc_locks.c
index 1cc90b6..bae2d67 100644
--- a/drivers/staging/lustre/lustre/mdc/mdc_locks.c
+++ b/drivers/staging/lustre/lustre/mdc/mdc_locks.c
@@ -840,6 +840,9 @@ resend:
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 &&
@@ -1139,6 +1142,7 @@ static int mdc_intent_getattr_async_interpret(const struct lu_env *env,
struct lookup_intent *it;
struct lustre_handle *lockh;
struct obd_device *obddev;
+ struct ldlm_reply *lockrep;
__u64 flags = LDLM_FL_HAS_INTENT;
ENTRY;
@@ -1159,6 +1163,12 @@ static int mdc_intent_getattr_async_interpret(const struct lu_env *env,
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);
diff --git a/drivers/staging/lustre/lustre/mdc/mdc_request.c b/drivers/staging/lustre/lustre/mdc/mdc_request.c
index 21e8a9d..b8bf2dc 100644
--- a/drivers/staging/lustre/lustre/mdc/mdc_request.c
+++ b/drivers/staging/lustre/lustre/mdc/mdc_request.c
@@ -1189,6 +1189,7 @@ static int mdc_ioc_hsm_progress(struct obd_export *exp,
GOTO(out, rc = -EPROTO);
*req_hpk = *hpk;
+ req_hpk->hpk_errval = lustre_errno_hton(hpk->hpk_errval);
ptlrpc_request_set_replen(req);
diff --git a/drivers/staging/lustre/lustre/osc/osc_request.c b/drivers/staging/lustre/lustre/osc/osc_request.c
index 53d6a35..0c0e07b 100644
--- a/drivers/staging/lustre/lustre/osc/osc_request.c
+++ b/drivers/staging/lustre/lustre/osc/osc_request.c
@@ -2347,6 +2347,8 @@ static int osc_enqueue_fini(struct ptlrpc_request *req, struct ost_lvb *lvb,
&RMF_DLM_REP);
LASSERT(rep != NULL);
+ rep->lock_policy_res1 =
+ ptlrpc_status_ntoh(rep->lock_policy_res1);
if (rep->lock_policy_res1)
rc = rep->lock_policy_res1;
}
diff --git a/drivers/staging/lustre/lustre/ptlrpc/Makefile b/drivers/staging/lustre/lustre/ptlrpc/Makefile
index 983eb66..a379558 100644
--- a/drivers/staging/lustre/lustre/ptlrpc/Makefile
+++ b/drivers/staging/lustre/lustre/ptlrpc/Makefile
@@ -16,6 +16,7 @@ ptlrpc_objs += sec.o sec_bulk.o sec_gc.o sec_config.o sec_lproc.o
ptlrpc_objs += sec_null.o sec_plain.o nrs.o nrs_fifo.o
ptlrpc-y := $(ldlm_objs) $(ptlrpc_objs)
+ptlrpc-$(CONFIG_LUSTRE_TRANSLATE_ERRNOS) += errno.c
obj-$(CONFIG_PTLRPC_GSS) += gss/
diff --git a/drivers/staging/lustre/lustre/ptlrpc/errno.c b/drivers/staging/lustre/lustre/ptlrpc/errno.c
new file mode 100644
index 0000000..46e259e
--- /dev/null
+++ b/drivers/staging/lustre/lustre/ptlrpc/errno.c
@@ -0,0 +1,382 @@
+/*
+ * GPL HEADER START
+ *
+ * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
+ *
+ * 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 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).
+ *
+ * You should have received a copy of the GNU General Public License
+ * version 2 along with this program; If not, see
+ * http://www.gnu.org/licenses/gpl-2.0.txt
+ *
+ * GPL HEADER END
+ */
+/*
+ * Copyright (C) 2011 FUJITSU LIMITED. All rights reserved.
+ *
+ * Copyright (c) 2013, Intel Corporation.
+ */
+
+#include <libcfs/libcfs.h>
+#include <lustre/lustre_errno.h>
+
+/*
+ * The two translation tables below must define a one-to-one mapping between
+ * host and network errnos.
+ *
+ * EWOULDBLOCK is equal to EAGAIN on all architectures except for parisc, which
+ * appears irrelevant. Thus, existing references to EWOULDBLOCK are fine.
+ *
+ * EDEADLOCK is equal to EDEADLK on x86 but not on sparc, at least. A sparc
+ * host has no context-free way to determine if a LUSTRE_EDEADLK represents an
+ * EDEADLK or an EDEADLOCK. Therefore, all existing references to EDEADLOCK
+ * that need to be transferred on wire have been replaced with EDEADLK.
+ */
+static int lustre_errno_hton_mapping[] = {
+ [EPERM] = LUSTRE_EPERM,
+ [ENOENT] = LUSTRE_ENOENT,
+ [ESRCH] = LUSTRE_ESRCH,
+ [EINTR] = LUSTRE_EINTR,
+ [EIO] = LUSTRE_EIO,
+ [ENXIO] = LUSTRE_ENXIO,
+ [E2BIG] = LUSTRE_E2BIG,
+ [ENOEXEC] = LUSTRE_ENOEXEC,
+ [EBADF] = LUSTRE_EBADF,
+ [ECHILD] = LUSTRE_ECHILD,
+ [EAGAIN] = LUSTRE_EAGAIN,
+ [ENOMEM] = LUSTRE_ENOMEM,
+ [EACCES] = LUSTRE_EACCES,
+ [EFAULT] = LUSTRE_EFAULT,
+ [ENOTBLK] = LUSTRE_ENOTBLK,
+ [EBUSY] = LUSTRE_EBUSY,
+ [EEXIST] = LUSTRE_EEXIST,
+ [EXDEV] = LUSTRE_EXDEV,
+ [ENODEV] = LUSTRE_ENODEV,
+ [ENOTDIR] = LUSTRE_ENOTDIR,
+ [EISDIR] = LUSTRE_EISDIR,
+ [EINVAL] = LUSTRE_EINVAL,
+ [ENFILE] = LUSTRE_ENFILE,
+ [EMFILE] = LUSTRE_EMFILE,
+ [ENOTTY] = LUSTRE_ENOTTY,
+ [ETXTBSY] = LUSTRE_ETXTBSY,
+ [EFBIG] = LUSTRE_EFBIG,
+ [ENOSPC] = LUSTRE_ENOSPC,
+ [ESPIPE] = LUSTRE_ESPIPE,
+ [EROFS] = LUSTRE_EROFS,
+ [EMLINK] = LUSTRE_EMLINK,
+ [EPIPE] = LUSTRE_EPIPE,
+ [EDOM] = LUSTRE_EDOM,
+ [ERANGE] = LUSTRE_ERANGE,
+ [EDEADLK] = LUSTRE_EDEADLK,
+ [ENAMETOOLONG] = LUSTRE_ENAMETOOLONG,
+ [ENOLCK] = LUSTRE_ENOLCK,
+ [ENOSYS] = LUSTRE_ENOSYS,
+ [ENOTEMPTY] = LUSTRE_ENOTEMPTY,
+ [ELOOP] = LUSTRE_ELOOP,
+ [ENOMSG] = LUSTRE_ENOMSG,
+ [EIDRM] = LUSTRE_EIDRM,
+ [ECHRNG] = LUSTRE_ECHRNG,
+ [EL2NSYNC] = LUSTRE_EL2NSYNC,
+ [EL3HLT] = LUSTRE_EL3HLT,
+ [EL3RST] = LUSTRE_EL3RST,
+ [ELNRNG] = LUSTRE_ELNRNG,
+ [EUNATCH] = LUSTRE_EUNATCH,
+ [ENOCSI] = LUSTRE_ENOCSI,
+ [EL2HLT] = LUSTRE_EL2HLT,
+ [EBADE] = LUSTRE_EBADE,
+ [EBADR] = LUSTRE_EBADR,
+ [EXFULL] = LUSTRE_EXFULL,
+ [ENOANO] = LUSTRE_ENOANO,
+ [EBADRQC] = LUSTRE_EBADRQC,
+ [EBADSLT] = LUSTRE_EBADSLT,
+ [EBFONT] = LUSTRE_EBFONT,
+ [ENOSTR] = LUSTRE_ENOSTR,
+ [ENODATA] = LUSTRE_ENODATA,
+ [ETIME] = LUSTRE_ETIME,
+ [ENOSR] = LUSTRE_ENOSR,
+ [ENONET] = LUSTRE_ENONET,
+ [ENOPKG] = LUSTRE_ENOPKG,
+ [EREMOTE] = LUSTRE_EREMOTE,
+ [ENOLINK] = LUSTRE_ENOLINK,
+ [EADV] = LUSTRE_EADV,
+ [ESRMNT] = LUSTRE_ESRMNT,
+ [ECOMM] = LUSTRE_ECOMM,
+ [EPROTO] = LUSTRE_EPROTO,
+ [EMULTIHOP] = LUSTRE_EMULTIHOP,
+ [EDOTDOT] = LUSTRE_EDOTDOT,
+ [EBADMSG] = LUSTRE_EBADMSG,
+ [EOVERFLOW] = LUSTRE_EOVERFLOW,
+ [ENOTUNIQ] = LUSTRE_ENOTUNIQ,
+ [EBADFD] = LUSTRE_EBADFD,
+ [EREMCHG] = LUSTRE_EREMCHG,
+ [ELIBACC] = LUSTRE_ELIBACC,
+ [ELIBBAD] = LUSTRE_ELIBBAD,
+ [ELIBSCN] = LUSTRE_ELIBSCN,
+ [ELIBMAX] = LUSTRE_ELIBMAX,
+ [ELIBEXEC] = LUSTRE_ELIBEXEC,
+ [EILSEQ] = LUSTRE_EILSEQ,
+ [ERESTART] = LUSTRE_ERESTART,
+ [ESTRPIPE] = LUSTRE_ESTRPIPE,
+ [EUSERS] = LUSTRE_EUSERS,
+ [ENOTSOCK] = LUSTRE_ENOTSOCK,
+ [EDESTADDRREQ] = LUSTRE_EDESTADDRREQ,
+ [EMSGSIZE] = LUSTRE_EMSGSIZE,
+ [EPROTOTYPE] = LUSTRE_EPROTOTYPE,
+ [ENOPROTOOPT] = LUSTRE_ENOPROTOOPT,
+ [EPROTONOSUPPORT] = LUSTRE_EPROTONOSUPPORT,
+ [ESOCKTNOSUPPORT] = LUSTRE_ESOCKTNOSUPPORT,
+ [EOPNOTSUPP] = LUSTRE_EOPNOTSUPP,
+ [EPFNOSUPPORT] = LUSTRE_EPFNOSUPPORT,
+ [EAFNOSUPPORT] = LUSTRE_EAFNOSUPPORT,
+ [EADDRINUSE] = LUSTRE_EADDRINUSE,
+ [EADDRNOTAVAIL] = LUSTRE_EADDRNOTAVAIL,
+ [ENETDOWN] = LUSTRE_ENETDOWN,
+ [ENETUNREACH] = LUSTRE_ENETUNREACH,
+ [ENETRESET] = LUSTRE_ENETRESET,
+ [ECONNABORTED] = LUSTRE_ECONNABORTED,
+ [ECONNRESET] = LUSTRE_ECONNRESET,
+ [ENOBUFS] = LUSTRE_ENOBUFS,
+ [EISCONN] = LUSTRE_EISCONN,
+ [ENOTCONN] = LUSTRE_ENOTCONN,
+ [ESHUTDOWN] = LUSTRE_ESHUTDOWN,
+ [ETOOMANYREFS] = LUSTRE_ETOOMANYREFS,
+ [ETIMEDOUT] = LUSTRE_ETIMEDOUT,
+ [ECONNREFUSED] = LUSTRE_ECONNREFUSED,
+ [EHOSTDOWN] = LUSTRE_EHOSTDOWN,
+ [EHOSTUNREACH] = LUSTRE_EHOSTUNREACH,
+ [EALREADY] = LUSTRE_EALREADY,
+ [EINPROGRESS] = LUSTRE_EINPROGRESS,
+ [ESTALE] = LUSTRE_ESTALE,
+ [EUCLEAN] = LUSTRE_EUCLEAN,
+ [ENOTNAM] = LUSTRE_ENOTNAM,
+ [ENAVAIL] = LUSTRE_ENAVAIL,
+ [EISNAM] = LUSTRE_EISNAM,
+ [EREMOTEIO] = LUSTRE_EREMOTEIO,
+ [EDQUOT] = LUSTRE_EDQUOT,
+ [ENOMEDIUM] = LUSTRE_ENOMEDIUM,
+ [EMEDIUMTYPE] = LUSTRE_EMEDIUMTYPE,
+ [ECANCELED] = LUSTRE_ECANCELED,
+ [ENOKEY] = LUSTRE_ENOKEY,
+ [EKEYEXPIRED] = LUSTRE_EKEYEXPIRED,
+ [EKEYREVOKED] = LUSTRE_EKEYREVOKED,
+ [EKEYREJECTED] = LUSTRE_EKEYREJECTED,
+ [EOWNERDEAD] = LUSTRE_EOWNERDEAD,
+ [ENOTRECOVERABLE] = LUSTRE_ENOTRECOVERABLE,
+ [ERESTARTSYS] = LUSTRE_ERESTARTSYS,
+ [ERESTARTNOINTR] = LUSTRE_ERESTARTNOINTR,
+ [ERESTARTNOHAND] = LUSTRE_ERESTARTNOHAND,
+ [ENOIOCTLCMD] = LUSTRE_ENOIOCTLCMD,
+ [ERESTART_RESTARTBLOCK] = LUSTRE_ERESTART_RESTARTBLOCK,
+ [EBADHANDLE] = LUSTRE_EBADHANDLE,
+ [ENOTSYNC] = LUSTRE_ENOTSYNC,
+ [EBADCOOKIE] = LUSTRE_EBADCOOKIE,
+ [ENOTSUPP] = LUSTRE_ENOTSUPP,
+ [ETOOSMALL] = LUSTRE_ETOOSMALL,
+ [ESERVERFAULT] = LUSTRE_ESERVERFAULT,
+ [EBADTYPE] = LUSTRE_EBADTYPE,
+ [EJUKEBOX] = LUSTRE_EJUKEBOX,
+ [EIOCBQUEUED] = LUSTRE_EIOCBQUEUED,
+ [EIOCBRETRY] = LUSTRE_EIOCBRETRY
+};
+
+static int lustre_errno_ntoh_mapping[] = {
+ [LUSTRE_EPERM] = EPERM,
+ [LUSTRE_ENOENT] = ENOENT,
+ [LUSTRE_ESRCH] = ESRCH,
+ [LUSTRE_EINTR] = EINTR,
+ [LUSTRE_EIO] = EIO,
+ [LUSTRE_ENXIO] = ENXIO,
+ [LUSTRE_E2BIG] = E2BIG,
+ [LUSTRE_ENOEXEC] = ENOEXEC,
+ [LUSTRE_EBADF] = EBADF,
+ [LUSTRE_ECHILD] = ECHILD,
+ [LUSTRE_EAGAIN] = EAGAIN,
+ [LUSTRE_ENOMEM] = ENOMEM,
+ [LUSTRE_EACCES] = EACCES,
+ [LUSTRE_EFAULT] = EFAULT,
+ [LUSTRE_ENOTBLK] = ENOTBLK,
+ [LUSTRE_EBUSY] = EBUSY,
+ [LUSTRE_EEXIST] = EEXIST,
+ [LUSTRE_EXDEV] = EXDEV,
+ [LUSTRE_ENODEV] = ENODEV,
+ [LUSTRE_ENOTDIR] = ENOTDIR,
+ [LUSTRE_EISDIR] = EISDIR,
+ [LUSTRE_EINVAL] = EINVAL,
+ [LUSTRE_ENFILE] = ENFILE,
+ [LUSTRE_EMFILE] = EMFILE,
+ [LUSTRE_ENOTTY] = ENOTTY,
+ [LUSTRE_ETXTBSY] = ETXTBSY,
+ [LUSTRE_EFBIG] = EFBIG,
+ [LUSTRE_ENOSPC] = ENOSPC,
+ [LUSTRE_ESPIPE] = ESPIPE,
+ [LUSTRE_EROFS] = EROFS,
+ [LUSTRE_EMLINK] = EMLINK,
+ [LUSTRE_EPIPE] = EPIPE,
+ [LUSTRE_EDOM] = EDOM,
+ [LUSTRE_ERANGE] = ERANGE,
+ [LUSTRE_EDEADLK] = EDEADLK,
+ [LUSTRE_ENAMETOOLONG] = ENAMETOOLONG,
+ [LUSTRE_ENOLCK] = ENOLCK,
+ [LUSTRE_ENOSYS] = ENOSYS,
+ [LUSTRE_ENOTEMPTY] = ENOTEMPTY,
+ [LUSTRE_ELOOP] = ELOOP,
+ [LUSTRE_ENOMSG] = ENOMSG,
+ [LUSTRE_EIDRM] = EIDRM,
+ [LUSTRE_ECHRNG] = ECHRNG,
+ [LUSTRE_EL2NSYNC] = EL2NSYNC,
+ [LUSTRE_EL3HLT] = EL3HLT,
+ [LUSTRE_EL3RST] = EL3RST,
+ [LUSTRE_ELNRNG] = ELNRNG,
+ [LUSTRE_EUNATCH] = EUNATCH,
+ [LUSTRE_ENOCSI] = ENOCSI,
+ [LUSTRE_EL2HLT] = EL2HLT,
+ [LUSTRE_EBADE] = EBADE,
+ [LUSTRE_EBADR] = EBADR,
+ [LUSTRE_EXFULL] = EXFULL,
+ [LUSTRE_ENOANO] = ENOANO,
+ [LUSTRE_EBADRQC] = EBADRQC,
+ [LUSTRE_EBADSLT] = EBADSLT,
+ [LUSTRE_EBFONT] = EBFONT,
+ [LUSTRE_ENOSTR] = ENOSTR,
+ [LUSTRE_ENODATA] = ENODATA,
+ [LUSTRE_ETIME] = ETIME,
+ [LUSTRE_ENOSR] = ENOSR,
+ [LUSTRE_ENONET] = ENONET,
+ [LUSTRE_ENOPKG] = ENOPKG,
+ [LUSTRE_EREMOTE] = EREMOTE,
+ [LUSTRE_ENOLINK] = ENOLINK,
+ [LUSTRE_EADV] = EADV,
+ [LUSTRE_ESRMNT] = ESRMNT,
+ [LUSTRE_ECOMM] = ECOMM,
+ [LUSTRE_EPROTO] = EPROTO,
+ [LUSTRE_EMULTIHOP] = EMULTIHOP,
+ [LUSTRE_EDOTDOT] = EDOTDOT,
+ [LUSTRE_EBADMSG] = EBADMSG,
+ [LUSTRE_EOVERFLOW] = EOVERFLOW,
+ [LUSTRE_ENOTUNIQ] = ENOTUNIQ,
+ [LUSTRE_EBADFD] = EBADFD,
+ [LUSTRE_EREMCHG] = EREMCHG,
+ [LUSTRE_ELIBACC] = ELIBACC,
+ [LUSTRE_ELIBBAD] = ELIBBAD,
+ [LUSTRE_ELIBSCN] = ELIBSCN,
+ [LUSTRE_ELIBMAX] = ELIBMAX,
+ [LUSTRE_ELIBEXEC] = ELIBEXEC,
+ [LUSTRE_EILSEQ] = EILSEQ,
+ [LUSTRE_ERESTART] = ERESTART,
+ [LUSTRE_ESTRPIPE] = ESTRPIPE,
+ [LUSTRE_EUSERS] = EUSERS,
+ [LUSTRE_ENOTSOCK] = ENOTSOCK,
+ [LUSTRE_EDESTADDRREQ] = EDESTADDRREQ,
+ [LUSTRE_EMSGSIZE] = EMSGSIZE,
+ [LUSTRE_EPROTOTYPE] = EPROTOTYPE,
+ [LUSTRE_ENOPROTOOPT] = ENOPROTOOPT,
+ [LUSTRE_EPROTONOSUPPORT] = EPROTONOSUPPORT,
+ [LUSTRE_ESOCKTNOSUPPORT] = ESOCKTNOSUPPORT,
+ [LUSTRE_EOPNOTSUPP] = EOPNOTSUPP,
+ [LUSTRE_EPFNOSUPPORT] = EPFNOSUPPORT,
+ [LUSTRE_EAFNOSUPPORT] = EAFNOSUPPORT,
+ [LUSTRE_EADDRINUSE] = EADDRINUSE,
+ [LUSTRE_EADDRNOTAVAIL] = EADDRNOTAVAIL,
+ [LUSTRE_ENETDOWN] = ENETDOWN,
+ [LUSTRE_ENETUNREACH] = ENETUNREACH,
+ [LUSTRE_ENETRESET] = ENETRESET,
+ [LUSTRE_ECONNABORTED] = ECONNABORTED,
+ [LUSTRE_ECONNRESET] = ECONNRESET,
+ [LUSTRE_ENOBUFS] = ENOBUFS,
+ [LUSTRE_EISCONN] = EISCONN,
+ [LUSTRE_ENOTCONN] = ENOTCONN,
+ [LUSTRE_ESHUTDOWN] = ESHUTDOWN,
+ [LUSTRE_ETOOMANYREFS] = ETOOMANYREFS,
+ [LUSTRE_ETIMEDOUT] = ETIMEDOUT,
+ [LUSTRE_ECONNREFUSED] = ECONNREFUSED,
+ [LUSTRE_EHOSTDOWN] = EHOSTDOWN,
+ [LUSTRE_EHOSTUNREACH] = EHOSTUNREACH,
+ [LUSTRE_EALREADY] = EALREADY,
+ [LUSTRE_EINPROGRESS] = EINPROGRESS,
+ [LUSTRE_ESTALE] = ESTALE,
+ [LUSTRE_EUCLEAN] = EUCLEAN,
+ [LUSTRE_ENOTNAM] = ENOTNAM,
+ [LUSTRE_ENAVAIL] = ENAVAIL,
+ [LUSTRE_EISNAM] = EISNAM,
+ [LUSTRE_EREMOTEIO] = EREMOTEIO,
+ [LUSTRE_EDQUOT] = EDQUOT,
+ [LUSTRE_ENOMEDIUM] = ENOMEDIUM,
+ [LUSTRE_EMEDIUMTYPE] = EMEDIUMTYPE,
+ [LUSTRE_ECANCELED] = ECANCELED,
+ [LUSTRE_ENOKEY] = ENOKEY,
+ [LUSTRE_EKEYEXPIRED] = EKEYEXPIRED,
+ [LUSTRE_EKEYREVOKED] = EKEYREVOKED,
+ [LUSTRE_EKEYREJECTED] = EKEYREJECTED,
+ [LUSTRE_EOWNERDEAD] = EOWNERDEAD,
+ [LUSTRE_ENOTRECOVERABLE] = ENOTRECOVERABLE,
+ [LUSTRE_ERESTARTSYS] = ERESTARTSYS,
+ [LUSTRE_ERESTARTNOINTR] = ERESTARTNOINTR,
+ [LUSTRE_ERESTARTNOHAND] = ERESTARTNOHAND,
+ [LUSTRE_ENOIOCTLCMD] = ENOIOCTLCMD,
+ [LUSTRE_ERESTART_RESTARTBLOCK] = ERESTART_RESTARTBLOCK,
+ [LUSTRE_EBADHANDLE] = EBADHANDLE,
+ [LUSTRE_ENOTSYNC] = ENOTSYNC,
+ [LUSTRE_EBADCOOKIE] = EBADCOOKIE,
+ [LUSTRE_ENOTSUPP] = ENOTSUPP,
+ [LUSTRE_ETOOSMALL] = ETOOSMALL,
+ [LUSTRE_ESERVERFAULT] = ESERVERFAULT,
+ [LUSTRE_EBADTYPE] = EBADTYPE,
+ [LUSTRE_EJUKEBOX] = EJUKEBOX,
+ [LUSTRE_EIOCBQUEUED] = EIOCBQUEUED,
+ [LUSTRE_EIOCBRETRY] = EIOCBRETRY
+};
+
+unsigned int lustre_errno_hton(unsigned int h)
+{
+ unsigned int n;
+
+ if (h == 0) {
+ n = 0;
+ } else if (h < ARRAY_SIZE(lustre_errno_hton_mapping)) {
+ n = lustre_errno_hton_mapping[h];
+ if (n == 0)
+ goto generic;
+ } else {
+generic:
+ /*
+ * A generic errno is better than the unknown one that could
+ * mean anything to a different host.
+ */
+ n = LUSTRE_EIO;
+ }
+
+ return n;
+}
+EXPORT_SYMBOL(lustre_errno_hton);
+
+unsigned int lustre_errno_ntoh(unsigned int n)
+{
+ unsigned int h;
+
+ if (n == 0) {
+ h = 0;
+ } else if (n < ARRAY_SIZE(lustre_errno_ntoh_mapping)) {
+ h = lustre_errno_ntoh_mapping[n];
+ if (h == 0)
+ goto generic;
+ } else {
+generic:
+ /*
+ * Similar to the situation in lustre_errno_hton(), an unknown
+ * network errno could coincide with anything. Hence, it is
+ * better to return a generic errno.
+ */
+ h = EIO;
+ }
+
+ return h;
+}
+EXPORT_SYMBOL(lustre_errno_ntoh);
diff --git a/drivers/staging/lustre/lustre/ptlrpc/niobuf.c b/drivers/staging/lustre/lustre/ptlrpc/niobuf.c
index de3f0db..5f1b889 100644
--- a/drivers/staging/lustre/lustre/ptlrpc/niobuf.c
+++ b/drivers/staging/lustre/lustre/ptlrpc/niobuf.c
@@ -400,7 +400,8 @@ int ptlrpc_send_reply(struct ptlrpc_request *req, int flags)
req->rq_type = PTL_RPC_MSG_REPLY;
lustre_msg_set_type(req->rq_repmsg, req->rq_type);
- lustre_msg_set_status(req->rq_repmsg, req->rq_status);
+ lustre_msg_set_status(req->rq_repmsg,
+ ptlrpc_status_hton(req->rq_status));
lustre_msg_set_opc(req->rq_repmsg,
req->rq_reqmsg ? lustre_msg_get_opc(req->rq_reqmsg) : 0);
diff --git a/drivers/staging/lustre/lustre/ptlrpc/pack_generic.c b/drivers/staging/lustre/lustre/ptlrpc/pack_generic.c
index 1437636..be246d4 100644
--- a/drivers/staging/lustre/lustre/ptlrpc/pack_generic.c
+++ b/drivers/staging/lustre/lustre/ptlrpc/pack_generic.c
@@ -642,6 +642,9 @@ static inline int lustre_unpack_ptlrpc_body_v2(struct ptlrpc_request *req,
return -EINVAL;
}
+ if (!inout)
+ pb->pb_status = ptlrpc_status_ntoh(pb->pb_status);
+
return 0;
}
--
1.7.9.5
^ permalink raw reply related [flat|nested] 62+ messages in thread* Re: [PATCH 11/48] staging/lustre/ptlrpc: Translate between host and network errnos
2013-07-22 16:06 ` [PATCH 11/48] staging/lustre/ptlrpc: Translate between host and network errnos Peng Tao
@ 2013-07-22 16:29 ` Paul Bolle
2013-07-22 16:36 ` Peng Tao
0 siblings, 1 reply; 62+ messages in thread
From: Paul Bolle @ 2013-07-22 16:29 UTC (permalink / raw)
To: Peng Tao
Cc: Greg Kroah-Hartman, linux-kernel, Li Wei, Peng Tao, Andreas Dilger
On Tue, 2013-07-23 at 00:06 +0800, Peng Tao wrote:
> From: Li Wei <wei.g.li@intel.com>
>
> Lustre puts system errors (e.g., ENOTCONN) on wire as numbers
> essentially specific to senders' architectures. While this is fine
> for x86-only sites, where receivers share the same error number
> definition with senders, problems will arise, however, for sites
> involving multiple architectures with different error number
> definitions. For instance, an ENOTCONN reply from a sparc server will
> be put on wire as -57, which, for an x86 client, means EBADSLT
> instead.
>
> To solve the problem, this patch defines a set of network errors for
> on-wire or on-disk uses. These errors correspond to a subset of the
> x86 system errors and share the same number definition, maintaining
> compatibility with existing x86 clients and servers.
>
> Then, either error numbers could be translated at run time, or all
> host errors going on wire could be replaced with network errors in the
> code. This patch does the former by introducing both generic and
> field-specific translation routines and calling them at proper places,
> so that translations for existing fields are transparent.
> (Personally, I tend to think the latter way might be worthwhile, as it
> is more straightforward conceptually. Do we really need so many
> different errors? Should errors returned by kernel routines really be
> passed up and eventually put on wire? There could even be security
> implications in that.)
>
> Thank Fujitsu for the original idea and their contributions that make
> this available upstream.
>
> Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-2743
> Lustre-change: http://review.whamcloud.com/5577
> Signed-off-by: Li Wei <wei.g.li@intel.com>
> Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
> Reviewed-by: Hiroya Nozaki <nozaki.hiroya@jp.fujitsu.com>
> Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>
> Signed-off-by: Peng Tao <tao.peng@emc.com>
> Signed-off-by: Andreas Dilger <andreas.dilger@intel.com>
> ---
> drivers/staging/lustre/lustre/Kconfig | 5 +
> .../lustre/lustre/include/lustre/lustre_errno.h | 215 +++++++++++
> .../lustre/lustre/include/lustre/lustre_idl.h | 2 +
> drivers/staging/lustre/lustre/include/lustre_net.h | 32 ++
> drivers/staging/lustre/lustre/ldlm/ldlm_lock.c | 3 +-
> drivers/staging/lustre/lustre/ldlm/ldlm_lockd.c | 2 +
> drivers/staging/lustre/lustre/ldlm/ldlm_request.c | 6 +-
> drivers/staging/lustre/lustre/mdc/mdc_locks.c | 10 +
> drivers/staging/lustre/lustre/mdc/mdc_request.c | 1 +
> drivers/staging/lustre/lustre/osc/osc_request.c | 2 +
> drivers/staging/lustre/lustre/ptlrpc/Makefile | 1 +
> drivers/staging/lustre/lustre/ptlrpc/errno.c | 382 ++++++++++++++++++++
> drivers/staging/lustre/lustre/ptlrpc/niobuf.c | 3 +-
> .../staging/lustre/lustre/ptlrpc/pack_generic.c | 3 +
> 14 files changed, 662 insertions(+), 5 deletions(-)
> create mode 100644 drivers/staging/lustre/lustre/include/lustre/lustre_errno.h
> create mode 100644 drivers/staging/lustre/lustre/ptlrpc/errno.c
>
> diff --git a/drivers/staging/lustre/lustre/Kconfig b/drivers/staging/lustre/lustre/Kconfig
> index 0a8eb1ad..19a1c33 100644
> --- a/drivers/staging/lustre/lustre/Kconfig
> +++ b/drivers/staging/lustre/lustre/Kconfig
> @@ -59,3 +59,8 @@ config LUSTRE_DEBUG_EXPENSIVE_CHECK
> expensive checks that may have a performance impact.
>
> Use with caution. If unsure, say N.
> +
> +config LUSTRE_TRANSLATE_ERRNOS
> + bool
> + depends on LUSTRE_FS && !X86
> + default true
Did you mean "default y"?
Paul Bolle
^ permalink raw reply [flat|nested] 62+ messages in thread
* Re: [PATCH 11/48] staging/lustre/ptlrpc: Translate between host and network errnos
2013-07-22 16:29 ` Paul Bolle
@ 2013-07-22 16:36 ` Peng Tao
0 siblings, 0 replies; 62+ messages in thread
From: Peng Tao @ 2013-07-22 16:36 UTC (permalink / raw)
To: Paul Bolle
Cc: Greg Kroah-Hartman, Linux Kernel Mailing List, Li Wei, Andreas Dilger
On Tue, Jul 23, 2013 at 12:29 AM, Paul Bolle <pebolle@tiscali.nl> wrote:
> On Tue, 2013-07-23 at 00:06 +0800, Peng Tao wrote:
>> From: Li Wei <wei.g.li@intel.com>
>>
>> Lustre puts system errors (e.g., ENOTCONN) on wire as numbers
>> essentially specific to senders' architectures. While this is fine
>> for x86-only sites, where receivers share the same error number
>> definition with senders, problems will arise, however, for sites
>> involving multiple architectures with different error number
>> definitions. For instance, an ENOTCONN reply from a sparc server will
>> be put on wire as -57, which, for an x86 client, means EBADSLT
>> instead.
>>
>> To solve the problem, this patch defines a set of network errors for
>> on-wire or on-disk uses. These errors correspond to a subset of the
>> x86 system errors and share the same number definition, maintaining
>> compatibility with existing x86 clients and servers.
>>
>> Then, either error numbers could be translated at run time, or all
>> host errors going on wire could be replaced with network errors in the
>> code. This patch does the former by introducing both generic and
>> field-specific translation routines and calling them at proper places,
>> so that translations for existing fields are transparent.
>> (Personally, I tend to think the latter way might be worthwhile, as it
>> is more straightforward conceptually. Do we really need so many
>> different errors? Should errors returned by kernel routines really be
>> passed up and eventually put on wire? There could even be security
>> implications in that.)
>>
>> Thank Fujitsu for the original idea and their contributions that make
>> this available upstream.
>>
>> Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-2743
>> Lustre-change: http://review.whamcloud.com/5577
>> Signed-off-by: Li Wei <wei.g.li@intel.com>
>> Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
>> Reviewed-by: Hiroya Nozaki <nozaki.hiroya@jp.fujitsu.com>
>> Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>
>> Signed-off-by: Peng Tao <tao.peng@emc.com>
>> Signed-off-by: Andreas Dilger <andreas.dilger@intel.com>
>> ---
>> drivers/staging/lustre/lustre/Kconfig | 5 +
>> .../lustre/lustre/include/lustre/lustre_errno.h | 215 +++++++++++
>> .../lustre/lustre/include/lustre/lustre_idl.h | 2 +
>> drivers/staging/lustre/lustre/include/lustre_net.h | 32 ++
>> drivers/staging/lustre/lustre/ldlm/ldlm_lock.c | 3 +-
>> drivers/staging/lustre/lustre/ldlm/ldlm_lockd.c | 2 +
>> drivers/staging/lustre/lustre/ldlm/ldlm_request.c | 6 +-
>> drivers/staging/lustre/lustre/mdc/mdc_locks.c | 10 +
>> drivers/staging/lustre/lustre/mdc/mdc_request.c | 1 +
>> drivers/staging/lustre/lustre/osc/osc_request.c | 2 +
>> drivers/staging/lustre/lustre/ptlrpc/Makefile | 1 +
>> drivers/staging/lustre/lustre/ptlrpc/errno.c | 382 ++++++++++++++++++++
>> drivers/staging/lustre/lustre/ptlrpc/niobuf.c | 3 +-
>> .../staging/lustre/lustre/ptlrpc/pack_generic.c | 3 +
>> 14 files changed, 662 insertions(+), 5 deletions(-)
>> create mode 100644 drivers/staging/lustre/lustre/include/lustre/lustre_errno.h
>> create mode 100644 drivers/staging/lustre/lustre/ptlrpc/errno.c
>>
>> diff --git a/drivers/staging/lustre/lustre/Kconfig b/drivers/staging/lustre/lustre/Kconfig
>> index 0a8eb1ad..19a1c33 100644
>> --- a/drivers/staging/lustre/lustre/Kconfig
>> +++ b/drivers/staging/lustre/lustre/Kconfig
>> @@ -59,3 +59,8 @@ config LUSTRE_DEBUG_EXPENSIVE_CHECK
>> expensive checks that may have a performance impact.
>>
>> Use with caution. If unsure, say N.
>> +
>> +config LUSTRE_TRANSLATE_ERRNOS
>> + bool
>> + depends on LUSTRE_FS && !X86
>> + default true
>
> Did you mean "default y"?
>
You are right. I wasn't aware of the mistake and have all the bool
default to either "true" or "false"...
Thanks for pointing it out. I'll send patch to fix all of them.
Thanks,
Tao
^ permalink raw reply [flat|nested] 62+ messages in thread
* [PATCH 12/48] staging/lustre/mdc: layout lock rpc must not take rpc_lock
2013-07-22 16:06 [PATCH 00/48] staging/lustre: minor cleanup and Intel Lustre sync Peng Tao
` (10 preceding siblings ...)
2013-07-22 16:06 ` [PATCH 11/48] staging/lustre/ptlrpc: Translate between host and network errnos Peng Tao
@ 2013-07-22 16:06 ` Peng Tao
2013-07-22 16:06 ` [PATCH 13/48] staging/lustre/ldlm: split client namespaces into active and inactive Peng Tao
` (36 subsequent siblings)
48 siblings, 0 replies; 62+ messages in thread
From: Peng Tao @ 2013-07-22 16:06 UTC (permalink / raw)
To: Greg Kroah-Hartman; +Cc: linux-kernel, jcl, Peng Tao, Andreas Dilger
From: jcl <jacques-charles.lafoucriere@cea.fr>
When a client issue an RPC to get a layout lock, it
must not hold rpc_lock because in case of a restore
the rpc can be blocking for a long time
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-3200
Lustre-change: http://review.whamcloud.com/6115
Signed-off-by: JC Lafoucriere <jacques-charles.lafoucriere@cea.fr>
Reviewed-by: Jinshan Xiong <jinshan.xiong@intel.com>
Reviewed-by: Johann Lombardi <johann.lombardi@intel.com>
Signed-off-by: Peng Tao <tao.peng@emc.com>
Signed-off-by: Andreas Dilger <andreas.dilger@intel.com>
---
drivers/staging/lustre/lustre/include/lustre_mdc.h | 6 ++++--
drivers/staging/lustre/lustre/mdc/mdc_locks.c | 2 +-
2 files changed, 5 insertions(+), 3 deletions(-)
diff --git a/drivers/staging/lustre/lustre/include/lustre_mdc.h b/drivers/staging/lustre/lustre/include/lustre_mdc.h
index fb1561a..82d1f2f 100644
--- a/drivers/staging/lustre/lustre/include/lustre_mdc.h
+++ b/drivers/staging/lustre/lustre/include/lustre_mdc.h
@@ -86,7 +86,8 @@ static inline void mdc_get_rpc_lock(struct mdc_rpc_lock *lck,
{
ENTRY;
- if (it != NULL && (it->it_op == IT_GETATTR || it->it_op == IT_LOOKUP))
+ if (it != NULL && (it->it_op == IT_GETATTR || it->it_op == IT_LOOKUP ||
+ it->it_op == IT_LAYOUT))
return;
/* This would normally block until the existing request finishes.
@@ -123,7 +124,8 @@ static inline void mdc_get_rpc_lock(struct mdc_rpc_lock *lck,
static inline void mdc_put_rpc_lock(struct mdc_rpc_lock *lck,
struct lookup_intent *it)
{
- if (it != NULL && (it->it_op == IT_GETATTR || it->it_op == IT_LOOKUP))
+ if (it != NULL && (it->it_op == IT_GETATTR || it->it_op == IT_LOOKUP ||
+ it->it_op == IT_LAYOUT))
goto out;
if (lck->rpcl_it == MDC_FAKE_RPCL_IT) { /* OBD_FAIL_MDC_RPCS_SEM */
diff --git a/drivers/staging/lustre/lustre/mdc/mdc_locks.c b/drivers/staging/lustre/lustre/mdc/mdc_locks.c
index bae2d67..9532134 100644
--- a/drivers/staging/lustre/lustre/mdc/mdc_locks.c
+++ b/drivers/staging/lustre/lustre/mdc/mdc_locks.c
@@ -884,7 +884,7 @@ 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);
--
1.7.9.5
^ permalink raw reply related [flat|nested] 62+ messages in thread* [PATCH 13/48] staging/lustre/ldlm: split client namespaces into active and inactive
2013-07-22 16:06 [PATCH 00/48] staging/lustre: minor cleanup and Intel Lustre sync Peng Tao
` (11 preceding siblings ...)
2013-07-22 16:06 ` [PATCH 12/48] staging/lustre/mdc: layout lock rpc must not take rpc_lock Peng Tao
@ 2013-07-22 16:06 ` Peng Tao
2013-07-22 16:06 ` [PATCH 14/48] staging/lustre: Only wake up ldlm_poold as frequently as the check interval Peng Tao
` (35 subsequent siblings)
48 siblings, 0 replies; 62+ messages in thread
From: Peng Tao @ 2013-07-22 16:06 UTC (permalink / raw)
To: Greg Kroah-Hartman
Cc: linux-kernel, Oleg Drokin, Oleg Drokin, Peng Tao, Andreas Dilger
From: Oleg Drokin <green@whamcloud.com>
The main reason behind this is ldlm_poold walks all namespaces currently
no matter if there are any locks or not. On large systems this could take
quite a bit of time, esp. since ldlm_poold is currently woken up once per
second.
Now every time a client namespace loses it's last resource it is placed
into an inactive list that is not touched by ldlm_poold as pointless.
On creation of a first resource in a namespace it is placed back into
the active list.
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-2924
Lustre-change: http://review.whamcloud.com/5624
Signed-off-by: Oleg Drokin <oleg.drokin@intel.com>
Reviewed-by: Hiroya Nozaki <nozaki.hiroya@jp.fujitsu.com>
Reviewed-by: Niu Yawei <yawei.niu@intel.com>
Signed-off-by: Peng Tao <tao.peng@emc.com>
Signed-off-by: Andreas Dilger <andreas.dilger@intel.com>
---
drivers/staging/lustre/lustre/include/lustre_dlm.h | 2 -
drivers/staging/lustre/lustre/ldlm/ldlm_internal.h | 46 +++++++++++++--
drivers/staging/lustre/lustre/ldlm/ldlm_pool.c | 59 ++++++++++++++++----
drivers/staging/lustre/lustre/ldlm/ldlm_resource.c | 55 +++++++++++++++---
4 files changed, 135 insertions(+), 27 deletions(-)
diff --git a/drivers/staging/lustre/lustre/include/lustre_dlm.h b/drivers/staging/lustre/lustre/include/lustre_dlm.h
index 317f928..48cb6f8 100644
--- a/drivers/staging/lustre/lustre/include/lustre_dlm.h
+++ b/drivers/staging/lustre/lustre/include/lustre_dlm.h
@@ -1471,8 +1471,6 @@ void ldlm_namespace_free(struct ldlm_namespace *ns,
struct obd_import *imp, int force);
void ldlm_namespace_register(struct ldlm_namespace *ns, ldlm_side_t client);
void ldlm_namespace_unregister(struct ldlm_namespace *ns, ldlm_side_t client);
-void ldlm_namespace_move_locked(struct ldlm_namespace *ns, ldlm_side_t client);
-struct ldlm_namespace *ldlm_namespace_first_locked(ldlm_side_t client);
void ldlm_namespace_get(struct ldlm_namespace *ns);
void ldlm_namespace_put(struct ldlm_namespace *ns);
int ldlm_proc_setup(void);
diff --git a/drivers/staging/lustre/lustre/ldlm/ldlm_internal.h b/drivers/staging/lustre/lustre/ldlm/ldlm_internal.h
index 141a957..785c1a1 100644
--- a/drivers/staging/lustre/lustre/ldlm/ldlm_internal.h
+++ b/drivers/staging/lustre/lustre/ldlm/ldlm_internal.h
@@ -36,23 +36,46 @@
#define MAX_STRING_SIZE 128
-extern atomic_t ldlm_srv_namespace_nr;
-extern atomic_t ldlm_cli_namespace_nr;
+extern int ldlm_srv_namespace_nr;
+extern int ldlm_cli_namespace_nr;
extern struct mutex ldlm_srv_namespace_lock;
extern struct list_head ldlm_srv_namespace_list;
extern struct mutex ldlm_cli_namespace_lock;
-extern struct list_head ldlm_cli_namespace_list;
+extern struct list_head ldlm_cli_active_namespace_list;
+extern struct list_head ldlm_cli_inactive_namespace_list;
-static inline atomic_t *ldlm_namespace_nr(ldlm_side_t client)
+static inline int ldlm_namespace_nr_read(ldlm_side_t client)
{
return client == LDLM_NAMESPACE_SERVER ?
- &ldlm_srv_namespace_nr : &ldlm_cli_namespace_nr;
+ ldlm_srv_namespace_nr : ldlm_cli_namespace_nr;
+}
+
+static inline void ldlm_namespace_nr_inc(ldlm_side_t client)
+{
+ if (client == LDLM_NAMESPACE_SERVER)
+ ldlm_srv_namespace_nr++;
+ else
+ ldlm_cli_namespace_nr++;
+}
+
+static inline void ldlm_namespace_nr_dec(ldlm_side_t client)
+{
+ if (client == LDLM_NAMESPACE_SERVER)
+ ldlm_srv_namespace_nr--;
+ else
+ ldlm_cli_namespace_nr--;
}
static inline struct list_head *ldlm_namespace_list(ldlm_side_t client)
{
return client == LDLM_NAMESPACE_SERVER ?
- &ldlm_srv_namespace_list : &ldlm_cli_namespace_list;
+ &ldlm_srv_namespace_list : &ldlm_cli_active_namespace_list;
+}
+
+static inline struct list_head *ldlm_namespace_inactive_list(ldlm_side_t client)
+{
+ return client == LDLM_NAMESPACE_SERVER ?
+ &ldlm_srv_namespace_list : &ldlm_cli_inactive_namespace_list;
}
static inline struct mutex *ldlm_namespace_lock(ldlm_side_t client)
@@ -61,6 +84,17 @@ static inline struct mutex *ldlm_namespace_lock(ldlm_side_t client)
&ldlm_srv_namespace_lock : &ldlm_cli_namespace_lock;
}
+/* ns_bref is the number of resources in this namespace with the notable
+ * exception of quota namespaces which have their empty refcount at 1 */
+static inline int ldlm_ns_empty(struct ldlm_namespace *ns)
+{
+ return atomic_read(&ns->ns_bref) == 0;
+}
+
+void ldlm_namespace_move_to_active_locked(struct ldlm_namespace *, ldlm_side_t);
+void ldlm_namespace_move_to_inactive_locked(struct ldlm_namespace *, ldlm_side_t);
+struct ldlm_namespace *ldlm_namespace_first_locked(ldlm_side_t);
+
/* ldlm_request.c */
/* Cancel lru flag, it indicates we cancel aged locks. */
enum {
diff --git a/drivers/staging/lustre/lustre/ldlm/ldlm_pool.c b/drivers/staging/lustre/lustre/ldlm/ldlm_pool.c
index b3b6028..1e6802f 100644
--- a/drivers/staging/lustre/lustre/ldlm/ldlm_pool.c
+++ b/drivers/staging/lustre/lustre/ldlm/ldlm_pool.c
@@ -1039,6 +1039,7 @@ static int ldlm_pools_shrink(ldlm_side_t client, int nr,
{
int total = 0, cached = 0, nr_ns;
struct ldlm_namespace *ns;
+ struct ldlm_namespace *ns_old = NULL; /* loop detection */
void *cookie;
if (client == LDLM_NAMESPACE_CLIENT && nr != 0 &&
@@ -1053,7 +1054,7 @@ static int ldlm_pools_shrink(ldlm_side_t client, int nr,
/*
* Find out how many resources we may release.
*/
- for (nr_ns = atomic_read(ldlm_namespace_nr(client));
+ for (nr_ns = ldlm_namespace_nr_read(client);
nr_ns > 0; nr_ns--)
{
mutex_lock(ldlm_namespace_lock(client));
@@ -1063,8 +1064,23 @@ static int ldlm_pools_shrink(ldlm_side_t client, int nr,
return 0;
}
ns = ldlm_namespace_first_locked(client);
+
+ if (ns == ns_old) {
+ mutex_unlock(ldlm_namespace_lock(client));
+ break;
+ }
+
+ if (ldlm_ns_empty(ns)) {
+ ldlm_namespace_move_to_inactive_locked(ns, client);
+ mutex_unlock(ldlm_namespace_lock(client));
+ continue;
+ }
+
+ if (ns_old == NULL)
+ ns_old = ns;
+
ldlm_namespace_get(ns);
- ldlm_namespace_move_locked(ns, client);
+ ldlm_namespace_move_to_active_locked(ns, client);
mutex_unlock(ldlm_namespace_lock(client));
total += ldlm_pool_shrink(&ns->ns_pool, 0, gfp_mask);
ldlm_namespace_put(ns);
@@ -1078,7 +1094,7 @@ static int ldlm_pools_shrink(ldlm_side_t client, int nr,
/*
* Shrink at least ldlm_namespace_nr(client) namespaces.
*/
- for (nr_ns = atomic_read(ldlm_namespace_nr(client));
+ for (nr_ns = ldlm_namespace_nr_read(client) - nr_ns;
nr_ns > 0; nr_ns--)
{
int cancel, nr_locks;
@@ -1099,7 +1115,7 @@ static int ldlm_pools_shrink(ldlm_side_t client, int nr,
}
ns = ldlm_namespace_first_locked(client);
ldlm_namespace_get(ns);
- ldlm_namespace_move_locked(ns, client);
+ ldlm_namespace_move_to_active_locked(ns, client);
mutex_unlock(ldlm_namespace_lock(client));
nr_locks = ldlm_pool_granted(&ns->ns_pool);
@@ -1132,6 +1148,7 @@ void ldlm_pools_recalc(ldlm_side_t client)
{
__u32 nr_l = 0, nr_p = 0, l;
struct ldlm_namespace *ns;
+ struct ldlm_namespace *ns_old = NULL;
int nr, equal = 0;
/*
@@ -1190,16 +1207,14 @@ void ldlm_pools_recalc(ldlm_side_t client)
* for _all_ pools.
*/
l = LDLM_POOL_HOST_L /
- atomic_read(
- ldlm_namespace_nr(client));
+ ldlm_namespace_nr_read(client);
} else {
/*
* All the rest of greedy pools will have
* all locks in equal parts.
*/
l = (LDLM_POOL_HOST_L - nr_l) /
- (atomic_read(
- ldlm_namespace_nr(client)) -
+ (ldlm_namespace_nr_read(client) -
nr_p);
}
ldlm_pool_setup(&ns->ns_pool, l);
@@ -1210,7 +1225,7 @@ void ldlm_pools_recalc(ldlm_side_t client)
/*
* Recalc at least ldlm_namespace_nr(client) namespaces.
*/
- for (nr = atomic_read(ldlm_namespace_nr(client)); nr > 0; nr--) {
+ for (nr = ldlm_namespace_nr_read(client); nr > 0; nr--) {
int skip;
/*
* Lock the list, get first @ns in the list, getref, move it
@@ -1226,6 +1241,30 @@ void ldlm_pools_recalc(ldlm_side_t client)
}
ns = ldlm_namespace_first_locked(client);
+ if (ns_old == ns) { /* Full pass complete */
+ mutex_unlock(ldlm_namespace_lock(client));
+ break;
+ }
+
+ /* We got an empty namespace, need to move it back to inactive
+ * list.
+ * The race with parallel resource creation is fine:
+ * - If they do namespace_get before our check, we fail the
+ * check and they move this item to the end of the list anyway
+ * - If we do the check and then they do namespace_get, then
+ * we move the namespace to inactive and they will move
+ * it back to active (synchronised by the lock, so no clash
+ * there).
+ */
+ if (ldlm_ns_empty(ns)) {
+ ldlm_namespace_move_to_inactive_locked(ns, client);
+ mutex_unlock(ldlm_namespace_lock(client));
+ continue;
+ }
+
+ if (ns_old == NULL)
+ ns_old = ns;
+
spin_lock(&ns->ns_lock);
/*
* skip ns which is being freed, and we don't want to increase
@@ -1239,7 +1278,7 @@ void ldlm_pools_recalc(ldlm_side_t client)
}
spin_unlock(&ns->ns_lock);
- ldlm_namespace_move_locked(ns, client);
+ ldlm_namespace_move_to_active_locked(ns, client);
mutex_unlock(ldlm_namespace_lock(client));
/*
diff --git a/drivers/staging/lustre/lustre/ldlm/ldlm_resource.c b/drivers/staging/lustre/lustre/ldlm/ldlm_resource.c
index cb8659f..7162838 100644
--- a/drivers/staging/lustre/lustre/ldlm/ldlm_resource.c
+++ b/drivers/staging/lustre/lustre/ldlm/ldlm_resource.c
@@ -48,14 +48,19 @@
struct kmem_cache *ldlm_resource_slab, *ldlm_lock_slab;
-atomic_t ldlm_srv_namespace_nr = ATOMIC_INIT(0);
-atomic_t ldlm_cli_namespace_nr = ATOMIC_INIT(0);
+int ldlm_srv_namespace_nr = 0;
+int ldlm_cli_namespace_nr = 0;
struct mutex ldlm_srv_namespace_lock;
LIST_HEAD(ldlm_srv_namespace_list);
struct mutex ldlm_cli_namespace_lock;
-LIST_HEAD(ldlm_cli_namespace_list);
+/* Client Namespaces that have active resources in them.
+ * Once all resources go away, ldlm_poold moves such namespaces to the
+ * inactive list */
+LIST_HEAD(ldlm_cli_active_namespace_list);
+/* Client namespaces that don't have any locks in them */
+LIST_HEAD(ldlm_cli_inactive_namespace_list);
proc_dir_entry_t *ldlm_type_proc_dir = NULL;
proc_dir_entry_t *ldlm_ns_proc_dir = NULL;
@@ -636,7 +641,7 @@ struct ldlm_namespace *ldlm_namespace_new(struct obd_device *obd, char *name,
GOTO(out_hash, rc);
}
- idx = atomic_read(ldlm_namespace_nr(client));
+ idx = ldlm_namespace_nr_read(client);
rc = ldlm_pool_init(&ns->ns_pool, ns, idx, client);
if (rc) {
CERROR("Can't initialize lock pool, rc %d\n", rc);
@@ -953,6 +958,12 @@ void ldlm_namespace_get(struct ldlm_namespace *ns)
}
EXPORT_SYMBOL(ldlm_namespace_get);
+/* This is only for callers that care about refcount */
+int ldlm_namespace_get_return(struct ldlm_namespace *ns)
+{
+ return atomic_inc_return(&ns->ns_bref);
+}
+
void ldlm_namespace_put(struct ldlm_namespace *ns)
{
if (atomic_dec_and_lock(&ns->ns_bref, &ns->ns_lock)) {
@@ -967,8 +978,8 @@ void ldlm_namespace_register(struct ldlm_namespace *ns, ldlm_side_t client)
{
mutex_lock(ldlm_namespace_lock(client));
LASSERT(list_empty(&ns->ns_list_chain));
- list_add(&ns->ns_list_chain, ldlm_namespace_list(client));
- atomic_inc(ldlm_namespace_nr(client));
+ list_add(&ns->ns_list_chain, ldlm_namespace_inactive_list(client));
+ ldlm_namespace_nr_inc(client);
mutex_unlock(ldlm_namespace_lock(client));
}
@@ -981,12 +992,13 @@ void ldlm_namespace_unregister(struct ldlm_namespace *ns, ldlm_side_t client)
* using list_empty(&ns->ns_list_chain). This is why it is
* important to use list_del_init() here. */
list_del_init(&ns->ns_list_chain);
- atomic_dec(ldlm_namespace_nr(client));
+ ldlm_namespace_nr_dec(client);
mutex_unlock(ldlm_namespace_lock(client));
}
/** Should be called with ldlm_namespace_lock(client) taken. */
-void ldlm_namespace_move_locked(struct ldlm_namespace *ns, ldlm_side_t client)
+void ldlm_namespace_move_to_active_locked(struct ldlm_namespace *ns,
+ ldlm_side_t client)
{
LASSERT(!list_empty(&ns->ns_list_chain));
LASSERT(mutex_is_locked(ldlm_namespace_lock(client)));
@@ -994,6 +1006,16 @@ void ldlm_namespace_move_locked(struct ldlm_namespace *ns, ldlm_side_t client)
}
/** Should be called with ldlm_namespace_lock(client) taken. */
+void ldlm_namespace_move_to_inactive_locked(struct ldlm_namespace *ns,
+ ldlm_side_t client)
+{
+ LASSERT(!list_empty(&ns->ns_list_chain));
+ LASSERT(mutex_is_locked(ldlm_namespace_lock(client)));
+ list_move_tail(&ns->ns_list_chain,
+ ldlm_namespace_inactive_list(client));
+}
+
+/** Should be called with ldlm_namespace_lock(client) taken. */
struct ldlm_namespace *ldlm_namespace_first_locked(ldlm_side_t client)
{
LASSERT(mutex_is_locked(ldlm_namespace_lock(client)));
@@ -1049,6 +1071,7 @@ ldlm_resource_get(struct ldlm_namespace *ns, struct ldlm_resource *parent,
struct ldlm_resource *res;
cfs_hash_bd_t bd;
__u64 version;
+ int ns_refcount = 0;
LASSERT(ns != NULL);
LASSERT(parent == NULL);
@@ -1119,7 +1142,7 @@ ldlm_resource_get(struct ldlm_namespace *ns, struct ldlm_resource *parent,
/* We won! Let's add the resource. */
cfs_hash_bd_add_locked(ns->ns_rs_hash, &bd, &res->lr_hash);
if (cfs_hash_bd_count_get(&bd) == 1)
- ldlm_namespace_get(ns);
+ ns_refcount = ldlm_namespace_get_return(ns);
cfs_hash_bd_unlock(ns->ns_rs_hash, &bd, 1);
if (ns->ns_lvbo && ns->ns_lvbo->lvbo_init) {
@@ -1145,6 +1168,20 @@ ldlm_resource_get(struct ldlm_namespace *ns, struct ldlm_resource *parent,
/* We create resource with locked lr_lvb_mutex. */
mutex_unlock(&res->lr_lvb_mutex);
+ /* Let's see if we happened to be the very first resource in this
+ * namespace. If so, and this is a client namespace, we need to move
+ * the namespace into the active namespaces list to be patrolled by
+ * the ldlm_poold.
+ * A notable exception, for quota namespaces qsd_lib.c already took a
+ * namespace reference, so it won't be participating in all of this,
+ * but I guess that's ok since we have no business cancelling quota
+ * locks anyway */
+ if (ns_is_client(ns) && ns_refcount == 1) {
+ mutex_lock(ldlm_namespace_lock(LDLM_NAMESPACE_CLIENT));
+ ldlm_namespace_move_to_active_locked(ns, LDLM_NAMESPACE_CLIENT);
+ mutex_unlock(ldlm_namespace_lock(LDLM_NAMESPACE_CLIENT));
+ }
+
return res;
}
EXPORT_SYMBOL(ldlm_resource_get);
--
1.7.9.5
^ permalink raw reply related [flat|nested] 62+ messages in thread* [PATCH 14/48] staging/lustre: Only wake up ldlm_poold as frequently as the check interval
2013-07-22 16:06 [PATCH 00/48] staging/lustre: minor cleanup and Intel Lustre sync Peng Tao
` (12 preceding siblings ...)
2013-07-22 16:06 ` [PATCH 13/48] staging/lustre/ldlm: split client namespaces into active and inactive Peng Tao
@ 2013-07-22 16:06 ` Peng Tao
2013-07-22 16:06 ` [PATCH 15/48] staging/lustre: Make quota namespace refcounting consistent Peng Tao
` (34 subsequent siblings)
48 siblings, 0 replies; 62+ messages in thread
From: Peng Tao @ 2013-07-22 16:06 UTC (permalink / raw)
To: Greg Kroah-Hartman
Cc: linux-kernel, Oleg Drokin, Oleg Drokin, Peng Tao, Andreas Dilger
From: Oleg Drokin <green@whamcloud.com>
We used to wake up ldlm poold every second, but that's overkill,
we should just see how much time is left until next closest recalc
interval hits and sleep this much.
This will make "per-second" client grant statistic not actually
per-second, but I don't think we need any precision in that code
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-2924
Lustre-change: http://review.whamcloud.com/5793
Signed-off-by: Oleg Drokin <oleg.drokin@intel.com>
Reviewed-by: Vitaly Fertman <vitaly_fertman@xyratex.com>
Reviewed-by: Hiroya Nozaki <nozaki.hiroya@jp.fujitsu.com>
Reviewed-by: Niu Yawei <yawei.niu@intel.com>
Signed-off-by: Peng Tao <tao.peng@emc.com>
Signed-off-by: Andreas Dilger <andreas.dilger@intel.com>
---
drivers/staging/lustre/lustre/include/lustre_dlm.h | 2 +-
drivers/staging/lustre/lustre/ldlm/ldlm_pool.c | 24 ++++++++++++--------
2 files changed, 16 insertions(+), 10 deletions(-)
diff --git a/drivers/staging/lustre/lustre/include/lustre_dlm.h b/drivers/staging/lustre/lustre/include/lustre_dlm.h
index 48cb6f8..25437b7 100644
--- a/drivers/staging/lustre/lustre/include/lustre_dlm.h
+++ b/drivers/staging/lustre/lustre/include/lustre_dlm.h
@@ -1643,7 +1643,7 @@ void unlock_res_and_lock(struct ldlm_lock *lock);
* There are not used outside of ldlm.
* @{
*/
-void ldlm_pools_recalc(ldlm_side_t client);
+int ldlm_pools_recalc(ldlm_side_t client);
int ldlm_pools_init(void);
void ldlm_pools_fini(void);
diff --git a/drivers/staging/lustre/lustre/ldlm/ldlm_pool.c b/drivers/staging/lustre/lustre/ldlm/ldlm_pool.c
index 1e6802f..3277a0d 100644
--- a/drivers/staging/lustre/lustre/ldlm/ldlm_pool.c
+++ b/drivers/staging/lustre/lustre/ldlm/ldlm_pool.c
@@ -578,7 +578,6 @@ int ldlm_pool_recalc(struct ldlm_pool *pl)
goto recalc;
spin_lock(&pl->pl_lock);
- recalc_interval_sec = cfs_time_current_sec() - pl->pl_recalc_time;
if (recalc_interval_sec > 0) {
/*
* Update pool statistics every 1s.
@@ -598,12 +597,12 @@ int ldlm_pool_recalc(struct ldlm_pool *pl)
count = pl->pl_ops->po_recalc(pl);
lprocfs_counter_add(pl->pl_stats, LDLM_POOL_RECALC_STAT,
count);
- return count;
}
+ recalc_interval_sec = pl->pl_recalc_time - cfs_time_current_sec() +
+ pl->pl_recalc_period;
- return 0;
+ return recalc_interval_sec;
}
-EXPORT_SYMBOL(ldlm_pool_recalc);
/**
* Pool shrink wrapper. Will call either client or server pool recalc callback
@@ -1144,12 +1143,13 @@ static int ldlm_pools_cli_shrink(SHRINKER_ARGS(sc, nr_to_scan, gfp_mask))
shrink_param(sc, gfp_mask));
}
-void ldlm_pools_recalc(ldlm_side_t client)
+int ldlm_pools_recalc(ldlm_side_t client)
{
__u32 nr_l = 0, nr_p = 0, l;
struct ldlm_namespace *ns;
struct ldlm_namespace *ns_old = NULL;
int nr, equal = 0;
+ int time = 50; /* seconds of sleep if no active namespaces */
/*
* No need to setup pool limit for client pools.
@@ -1285,16 +1285,22 @@ void ldlm_pools_recalc(ldlm_side_t client)
* After setup is done - recalc the pool.
*/
if (!skip) {
- ldlm_pool_recalc(&ns->ns_pool);
+ int ttime = ldlm_pool_recalc(&ns->ns_pool);
+
+ if (ttime < time)
+ time = ttime;
+
ldlm_namespace_put(ns);
}
}
+ return time;
}
EXPORT_SYMBOL(ldlm_pools_recalc);
static int ldlm_pools_thread_main(void *arg)
{
struct ptlrpc_thread *thread = (struct ptlrpc_thread *)arg;
+ int s_time, c_time;
ENTRY;
thread_set_flags(thread, SVC_RUNNING);
@@ -1309,14 +1315,14 @@ static int ldlm_pools_thread_main(void *arg)
/*
* Recal all pools on this tick.
*/
- ldlm_pools_recalc(LDLM_NAMESPACE_SERVER);
- ldlm_pools_recalc(LDLM_NAMESPACE_CLIENT);
+ s_time = ldlm_pools_recalc(LDLM_NAMESPACE_SERVER);
+ c_time = ldlm_pools_recalc(LDLM_NAMESPACE_CLIENT);
/*
* Wait until the next check time, or until we're
* stopped.
*/
- lwi = LWI_TIMEOUT(cfs_time_seconds(LDLM_POOLS_THREAD_PERIOD),
+ lwi = LWI_TIMEOUT(cfs_time_seconds(min(s_time, c_time)),
NULL, NULL);
l_wait_event(thread->t_ctl_waitq,
thread_is_stopping(thread) ||
--
1.7.9.5
^ permalink raw reply related [flat|nested] 62+ messages in thread* [PATCH 15/48] staging/lustre: Make quota namespace refcounting consistent
2013-07-22 16:06 [PATCH 00/48] staging/lustre: minor cleanup and Intel Lustre sync Peng Tao
` (13 preceding siblings ...)
2013-07-22 16:06 ` [PATCH 14/48] staging/lustre: Only wake up ldlm_poold as frequently as the check interval Peng Tao
@ 2013-07-22 16:06 ` Peng Tao
2013-07-22 16:06 ` [PATCH 16/48] staging/lustre/dlmlock: compress out unused space Peng Tao
` (33 subsequent siblings)
48 siblings, 0 replies; 62+ messages in thread
From: Peng Tao @ 2013-07-22 16:06 UTC (permalink / raw)
To: Greg Kroah-Hartman
Cc: linux-kernel, Oleg Drokin, Oleg Drokin, Peng Tao, Andreas Dilger
From: Oleg Drokin <green@whamcloud.com>
It seems quota namespace is needlessly referenced on connect,
but that's not necessary as it could not go away until entire
obd goes away.
On the other hand this extra reference disturbs other logic
depending on empty namespace having zero refcount, so this patch
drops such extra referencing.
This picks client side change of the original patch.
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-2924
Lustre-change: http://review.whamcloud.com/6234
Signed-off-by: Oleg Drokin <oleg.drokin@intel.com>
Reviewed-by: Niu Yawei <yawei.niu@intel.com>
Reviewed-by: Johann Lombardi <johann.lombardi@intel.com>
Signed-off-by: Peng Tao <tao.peng@emc.com>
Signed-off-by: Andreas Dilger <andreas.dilger@intel.com>
---
drivers/staging/lustre/lustre/ldlm/ldlm_internal.h | 3 +--
drivers/staging/lustre/lustre/ldlm/ldlm_resource.c | 6 +-----
2 files changed, 2 insertions(+), 7 deletions(-)
diff --git a/drivers/staging/lustre/lustre/ldlm/ldlm_internal.h b/drivers/staging/lustre/lustre/ldlm/ldlm_internal.h
index 785c1a1..bf6f003 100644
--- a/drivers/staging/lustre/lustre/ldlm/ldlm_internal.h
+++ b/drivers/staging/lustre/lustre/ldlm/ldlm_internal.h
@@ -84,8 +84,7 @@ static inline struct mutex *ldlm_namespace_lock(ldlm_side_t client)
&ldlm_srv_namespace_lock : &ldlm_cli_namespace_lock;
}
-/* ns_bref is the number of resources in this namespace with the notable
- * exception of quota namespaces which have their empty refcount at 1 */
+/* ns_bref is the number of resources in this namespace */
static inline int ldlm_ns_empty(struct ldlm_namespace *ns)
{
return atomic_read(&ns->ns_bref) == 0;
diff --git a/drivers/staging/lustre/lustre/ldlm/ldlm_resource.c b/drivers/staging/lustre/lustre/ldlm/ldlm_resource.c
index 7162838..17eab79 100644
--- a/drivers/staging/lustre/lustre/ldlm/ldlm_resource.c
+++ b/drivers/staging/lustre/lustre/ldlm/ldlm_resource.c
@@ -1171,11 +1171,7 @@ ldlm_resource_get(struct ldlm_namespace *ns, struct ldlm_resource *parent,
/* Let's see if we happened to be the very first resource in this
* namespace. If so, and this is a client namespace, we need to move
* the namespace into the active namespaces list to be patrolled by
- * the ldlm_poold.
- * A notable exception, for quota namespaces qsd_lib.c already took a
- * namespace reference, so it won't be participating in all of this,
- * but I guess that's ok since we have no business cancelling quota
- * locks anyway */
+ * the ldlm_poold. */
if (ns_is_client(ns) && ns_refcount == 1) {
mutex_lock(ldlm_namespace_lock(LDLM_NAMESPACE_CLIENT));
ldlm_namespace_move_to_active_locked(ns, LDLM_NAMESPACE_CLIENT);
--
1.7.9.5
^ permalink raw reply related [flat|nested] 62+ messages in thread* [PATCH 16/48] staging/lustre/dlmlock: compress out unused space
2013-07-22 16:06 [PATCH 00/48] staging/lustre: minor cleanup and Intel Lustre sync Peng Tao
` (14 preceding siblings ...)
2013-07-22 16:06 ` [PATCH 15/48] staging/lustre: Make quota namespace refcounting consistent Peng Tao
@ 2013-07-22 16:06 ` Peng Tao
2013-07-22 16:06 ` [PATCH 17/48] staging/lustre/md: fix lu_ucred.c boilerplate Peng Tao
` (32 subsequent siblings)
48 siblings, 0 replies; 62+ messages in thread
From: Peng Tao @ 2013-07-22 16:06 UTC (permalink / raw)
To: Greg Kroah-Hartman; +Cc: linux-kernel, Bruce Korb, Peng Tao, Andreas Dilger
From: Bruce Korb <bruce_korb@xyratex.com>
* lustre/include/lustre_dlm.h: Remove all bit fields and the unused
weighing callback procedure. respell LDLM_AST_DISCARD_DATA as
LDLM_FL_AST_DISCARD_DATA to match other flags.
* .gitignore: ignore emacs temporary files
* autogen.sh: rebuild the lock bits, if autogen is available.
* contrib/bit-masks/lustre_dlm_flags.def: define the ldlm_lock flags
* contrib/bit-masks/lustre_dlm_flags.tpl: template for emitting text
* contrib/bit-masks/Makefile: construct the .c and .h files
The .c file is for constructing a crash extension and is not
preserved.
* contrib/bit-masks/.gitignore: ignore built products
* lustre/contrib/wireshark/packet-lustre.c: use built files instead
of local versions of the defines.
In the rest of the modified sources, replace flag field references
with bit mask references.
* lustre/osc/osc_lock.c: removed osc_lock_weigh, too
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-2771
Lustre-change: http://review.whamcloud.com/5312
Signed-off-by: Bruce Korb <bruce_korb@xyratex.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>
Reviewed-by: Keith Mannthey <Keith.Mannthey@intel.com>
Reviewed-by: Keith Mannthey <keith.mannthey@intel.com>
Reviewed-by: <bruce.korb@gmail.com>
Signed-off-by: Peng Tao <tao.peng@emc.com>
Signed-off-by: Andreas Dilger <andreas.dilger@intel.com>
---
drivers/staging/lustre/lustre/include/lustre_dlm.h | 198 +--------
.../lustre/lustre/include/lustre_dlm_flags.h | 460 ++++++++++++++++++++
drivers/staging/lustre/lustre/ldlm/l_lock.c | 8 +-
drivers/staging/lustre/lustre/ldlm/ldlm_flock.c | 2 +-
drivers/staging/lustre/lustre/ldlm/ldlm_lock.c | 48 +-
drivers/staging/lustre/lustre/ldlm/ldlm_lockd.c | 4 +-
drivers/staging/lustre/lustre/ldlm/ldlm_request.c | 7 +-
drivers/staging/lustre/lustre/ldlm/ldlm_resource.c | 4 +-
drivers/staging/lustre/lustre/llite/dir.c | 15 +-
drivers/staging/lustre/lustre/llite/file.c | 19 +-
drivers/staging/lustre/lustre/mdc/mdc_locks.c | 9 +-
drivers/staging/lustre/lustre/mgc/mgc_request.c | 8 +-
drivers/staging/lustre/lustre/osc/osc_lock.c | 124 ++----
13 files changed, 573 insertions(+), 333 deletions(-)
create mode 100644 drivers/staging/lustre/lustre/include/lustre_dlm_flags.h
diff --git a/drivers/staging/lustre/lustre/include/lustre_dlm.h b/drivers/staging/lustre/lustre/include/lustre_dlm.h
index 25437b7..66db562 100644
--- a/drivers/staging/lustre/lustre/include/lustre_dlm.h
+++ b/drivers/staging/lustre/lustre/include/lustre_dlm.h
@@ -57,6 +57,8 @@
#include <interval_tree.h> /* for interval_node{}, ldlm_extent */
#include <lu_ref.h>
+#include "lustre_dlm_flags.h"
+
struct obd_ops;
struct obd_device;
@@ -96,161 +98,6 @@ typedef enum {
} ldlm_side_t;
/**
- * Declaration of flags sent through the wire.
- **/
-#define LDLM_FL_LOCK_CHANGED 0x000001 /* extent, mode, or resource changed */
-
-/**
- * If the server returns one of these flags, then the lock was put on that list.
- * If the client sends one of these flags (during recovery ONLY!), it wants the
- * lock added to the specified list, no questions asked.
- */
-#define LDLM_FL_BLOCK_GRANTED 0x000002
-#define LDLM_FL_BLOCK_CONV 0x000004
-#define LDLM_FL_BLOCK_WAIT 0x000008
-
-/* Used to be LDLM_FL_CBPENDING 0x000010 moved to non-wire flags */
-
-#define LDLM_FL_AST_SENT 0x000020 /* blocking or cancel packet was
- * queued for sending. */
-/* Used to be LDLM_FL_WAIT_NOREPROC 0x000040 moved to non-wire flags */
-/* Used to be LDLM_FL_CANCEL 0x000080 moved to non-wire flags */
-
-/**
- * Lock is being replayed. This could probably be implied by the fact that one
- * of BLOCK_{GRANTED,CONV,WAIT} is set, but that is pretty dangerous.
- */
-#define LDLM_FL_REPLAY 0x000100
-
-#define LDLM_FL_INTENT_ONLY 0x000200 /* Don't grant lock, just do intent. */
-
-/* Used to be LDLM_FL_LOCAL_ONLY 0x000400 moved to non-wire flags */
-/* Used to be LDLM_FL_FAILED 0x000800 moved to non-wire flags */
-
-#define LDLM_FL_HAS_INTENT 0x001000 /* lock request has intent */
-
-/* Used to be LDLM_FL_CANCELING 0x002000 moved to non-wire flags */
-/* Used to be LDLM_FL_LOCAL 0x004000 moved to non-wire flags */
-
-#define LDLM_FL_DISCARD_DATA 0x010000 /* discard (no writeback) on cancel */
-
-#define LDLM_FL_NO_TIMEOUT 0x020000 /* Blocked by group lock - wait
- * indefinitely */
-
-/** file & record locking */
-#define LDLM_FL_BLOCK_NOWAIT 0x040000 /* Server told not to wait if blocked.
- * For AGL, OST will not send glimpse
- * callback. */
-#define LDLM_FL_TEST_LOCK 0x080000 // return blocking lock
-
-/* Used to be LDLM_FL_LVB_READY 0x100000 moved to non-wire flags */
-/* Used to be LDLM_FL_KMS_IGNORE 0x200000 moved to non-wire flags */
-/* Used to be LDLM_FL_NO_LRU 0x400000 moved to non-wire flags */
-
-/* Immediatelly cancel such locks when they block some other locks. Send
- * cancel notification to original lock holder, but expect no reply. This is
- * for clients (like liblustre) that cannot be expected to reliably response
- * to blocking AST. */
-#define LDLM_FL_CANCEL_ON_BLOCK 0x800000
-
-/* Flags flags inherited from parent lock when doing intents. */
-#define LDLM_INHERIT_FLAGS (LDLM_FL_CANCEL_ON_BLOCK)
-
-/* Used to be LDLM_FL_CP_REQD 0x1000000 moved to non-wire flags */
-/* Used to be LDLM_FL_CLEANED 0x2000000 moved to non-wire flags */
-/* Used to be LDLM_FL_ATOMIC_CB 0x4000000 moved to non-wire flags */
-/* Used to be LDLM_FL_BL_AST 0x10000000 moved to non-wire flags */
-/* Used to be LDLM_FL_BL_DONE 0x20000000 moved to non-wire flags */
-
-/* measure lock contention and return -EUSERS if locking contention is high */
-#define LDLM_FL_DENY_ON_CONTENTION 0x40000000
-
-/* These are flags that are mapped into the flags and ASTs of blocking locks */
-#define LDLM_AST_DISCARD_DATA 0x80000000 /* Add FL_DISCARD to blocking ASTs */
-
-/* Flags sent in AST lock_flags to be mapped into the receiving lock. */
-#define LDLM_AST_FLAGS (LDLM_FL_DISCARD_DATA)
-
-/*
- * --------------------------------------------------------------------------
- * NOTE! Starting from this point, that is, LDLM_FL_* flags with values above
- * 0x80000000 will not be sent over the wire.
- * --------------------------------------------------------------------------
- */
-
-/**
- * Declaration of flags not sent through the wire.
- **/
-
-/**
- * Used for marking lock as a target for -EINTR while cp_ast sleep
- * emulation + race with upcoming bl_ast.
- */
-#define LDLM_FL_FAIL_LOC 0x100000000ULL
-
-/**
- * Used while processing the unused list to know that we have already
- * handled this lock and decided to skip it.
- */
-#define LDLM_FL_SKIPPED 0x200000000ULL
-/* this lock is being destroyed */
-#define LDLM_FL_CBPENDING 0x400000000ULL
-/* not a real flag, not saved in lock */
-#define LDLM_FL_WAIT_NOREPROC 0x800000000ULL
-/* cancellation callback already run */
-#define LDLM_FL_CANCEL 0x1000000000ULL
-#define LDLM_FL_LOCAL_ONLY 0x2000000000ULL
-/* don't run the cancel callback under ldlm_cli_cancel_unused */
-#define LDLM_FL_FAILED 0x4000000000ULL
-/* lock cancel has already been sent */
-#define LDLM_FL_CANCELING 0x8000000000ULL
-/* local lock (ie, no srv/cli split) */
-#define LDLM_FL_LOCAL 0x10000000000ULL
-/* XXX FIXME: This is being added to b_size as a low-risk fix to the fact that
- * the LVB filling happens _after_ the lock has been granted, so another thread
- * can match it before the LVB has been updated. As a dirty hack, we set
- * LDLM_FL_LVB_READY only after we've done the LVB poop.
- * this is only needed on LOV/OSC now, where LVB is actually used and callers
- * must set it in input flags.
- *
- * The proper fix is to do the granting inside of the completion AST, which can
- * be replaced with a LVB-aware wrapping function for OSC locks. That change is
- * pretty high-risk, though, and would need a lot more testing. */
-#define LDLM_FL_LVB_READY 0x20000000000ULL
-/* A lock contributes to the known minimum size (KMS) calculation until it has
- * finished the part of its cancelation that performs write back on its dirty
- * pages. It can remain on the granted list during this whole time. Threads
- * racing to update the KMS after performing their writeback need to know to
- * exclude each other's locks from the calculation as they walk the granted
- * list. */
-#define LDLM_FL_KMS_IGNORE 0x40000000000ULL
-/* completion AST to be executed */
-#define LDLM_FL_CP_REQD 0x80000000000ULL
-/* cleanup_resource has already handled the lock */
-#define LDLM_FL_CLEANED 0x100000000000ULL
-/* optimization hint: LDLM can run blocking callback from current context
- * w/o involving separate thread. in order to decrease cs rate */
-#define LDLM_FL_ATOMIC_CB 0x200000000000ULL
-
-/* It may happen that a client initiates two operations, e.g. unlink and
- * mkdir, such that the server sends a blocking AST for conflicting
- * locks to this client for the first operation, whereas the second
- * operation has canceled this lock and is waiting for rpc_lock which is
- * taken by the first operation. LDLM_FL_BL_AST is set by
- * ldlm_callback_handler() in the lock to prevent the Early Lock Cancel
- * (ELC) code from cancelling it.
- *
- * LDLM_FL_BL_DONE is to be set by ldlm_cancel_callback() when lock
- * cache is dropped to let ldlm_callback_handler() return EINVAL to the
- * server. It is used when ELC RPC is already prepared and is waiting
- * for rpc_lock, too late to send a separate CANCEL RPC. */
-#define LDLM_FL_BL_AST 0x400000000000ULL
-#define LDLM_FL_BL_DONE 0x800000000000ULL
-/* Don't put lock into the LRU list, so that it is not canceled due to aging.
- * Used by MGC locks, they are cancelled only at unmount or by callback. */
-#define LDLM_FL_NO_LRU 0x1000000000000ULL
-
-/**
* The blocking callback is overloaded to perform two functions. These flags
* indicate which operation should be performed.
*/
@@ -720,8 +567,6 @@ typedef int (*ldlm_completion_callback)(struct ldlm_lock *lock, __u64 flags,
void *data);
/** Type for glimpse callback function of a lock. */
typedef int (*ldlm_glimpse_callback)(struct ldlm_lock *lock, void *data);
-/** Type for weight callback function of a lock. */
-typedef unsigned long (*ldlm_weigh_callback)(struct ldlm_lock *lock);
/** Work list for sending GL ASTs to multiple locks. */
struct ldlm_glimpse_work {
@@ -890,9 +735,6 @@ struct ldlm_lock {
*/
ldlm_glimpse_callback l_glimpse_ast;
- /** XXX apparently unused "weight" handler. To be removed? */
- ldlm_weigh_callback l_weigh_ast;
-
/**
* Lock export.
* This is a pointer to actual client export for locks that were granted
@@ -919,11 +761,11 @@ struct ldlm_lock {
ldlm_policy_data_t l_policy_data;
/**
- * Lock state flags.
- * Like whenever we receive any blocking requests for this lock, etc.
- * Protected by lr_lock.
+ * Lock state flags. Protected by lr_lock.
+ * \see lustre_dlm_flags.h where the bits are defined.
*/
__u64 l_flags;
+
/**
* Lock r/w usage counters.
* Protected by lr_lock.
@@ -952,34 +794,6 @@ struct ldlm_lock {
/** Originally requested extent for the extent lock. */
struct ldlm_extent l_req_extent;
- unsigned int l_failed:1,
- /**
- * Set for locks that were removed from class hash table and will be
- * destroyed when last reference to them is released. Set by
- * ldlm_lock_destroy_internal().
- *
- * Protected by lock and resource locks.
- */
- l_destroyed:1,
- /*
- * it's set in lock_res_and_lock() and unset in unlock_res_and_lock().
- *
- * NB: compared with check_res_locked(), checking this bit is cheaper.
- * Also, spin_is_locked() is deprecated for kernel code; one reason is
- * because it works only for SMP so user needs to add extra macros like
- * LASSERT_SPIN_LOCKED for uniprocessor kernels.
- */
- l_res_locked:1,
- /*
- * It's set once we call ldlm_add_waiting_lock_res_locked()
- * to start the lock-timeout timer and it will never be reset.
- *
- * Protected by lock_res_and_lock().
- */
- l_waited:1,
- /** Flag whether this is a server namespace lock. */
- l_ns_srv:1;
-
/*
* Client-side-only members.
*/
@@ -1230,7 +1044,6 @@ struct ldlm_enqueue_info {
void *ei_cb_bl; /** blocking lock callback */
void *ei_cb_cp; /** lock completion callback */
void *ei_cb_gl; /** lock glimpse callback */
- void *ei_cb_wg; /** lock weigh callback */
void *ei_cbdata; /** Data to be passed into callbacks. */
};
@@ -1328,7 +1141,6 @@ struct ldlm_callback_suite {
ldlm_completion_callback lcs_completion;
ldlm_blocking_callback lcs_blocking;
ldlm_glimpse_callback lcs_glimpse;
- ldlm_weigh_callback lcs_weigh;
};
/* ldlm_lockd.c */
diff --git a/drivers/staging/lustre/lustre/include/lustre_dlm_flags.h b/drivers/staging/lustre/lustre/include/lustre_dlm_flags.h
new file mode 100644
index 0000000..8c34d9d
--- /dev/null
+++ b/drivers/staging/lustre/lustre/include/lustre_dlm_flags.h
@@ -0,0 +1,460 @@
+/* -*- buffer-read-only: t -*- vi: set ro:
+ *
+ * DO NOT EDIT THIS FILE (lustre_dlm_flags.h)
+ *
+ * It has been AutoGen-ed
+ * From the definitions lustre_dlm_flags.def
+ * and the template file lustre_dlm_flags.tpl
+ *
+ * lustre is free software: you can redistribute it and/or modify it
+ * under the terms of the GNU General Public License as published by the
+ * Free Software Foundation, either version 3 of the License, or
+ * (at your option) any later version.
+ *
+ * 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.
+ *
+ * You should have received a copy of the GNU General Public License along
+ * with this program. If not, see <http://www.gnu.org/licenses/>.
+ */
+/**
+ * \file lustre_dlm_flags.h
+ * The flags and collections of flags (masks) for \see struct ldlm_lock.
+ * This file is derived from flag definitions in lustre_dlm_flags.def.
+ * The format is defined in the lustre_dlm_flags.tpl template file.
+ *
+ * \addtogroup LDLM Lustre Distributed Lock Manager
+ * @{
+ *
+ * \name flags
+ * The flags and collections of flags (masks) for \see struct ldlm_lock.
+ * @{
+ */
+#ifndef LDLM_ALL_FLAGS_MASK
+
+/** l_flags bits marked as "all_flags" bits */
+#define LDLM_FL_ALL_FLAGS_MASK 0x007FFFFFC08F132FULL
+
+/** l_flags bits marked as "ast" bits */
+#define LDLM_FL_AST_MASK 0x0000000080000000ULL
+
+/** l_flags bits marked as "blocked" bits */
+#define LDLM_FL_BLOCKED_MASK 0x000000000000000EULL
+
+/** l_flags bits marked as "gone" bits */
+#define LDLM_FL_GONE_MASK 0x0006004000000000ULL
+
+/** l_flags bits marked as "hide_lock" bits */
+#define LDLM_FL_HIDE_LOCK_MASK 0x0000206400000000ULL
+
+/** l_flags bits marked as "inherit" bits */
+#define LDLM_FL_INHERIT_MASK 0x0000000000800000ULL
+
+/** l_flags bits marked as "local_only" bits */
+#define LDLM_FL_LOCAL_ONLY_MASK 0x007FFFFF00000000ULL
+
+/** l_flags bits marked as "on_wire" bits */
+#define LDLM_FL_ON_WIRE_MASK 0x00000000C08F132FULL
+
+/** extent, mode, or resource changed */
+#define LDLM_FL_LOCK_CHANGED 0x0000000000000001ULL // bit 0
+#define ldlm_is_lock_changed(_l) LDLM_TEST_FLAG(( _l), 1ULL << 0)
+#define ldlm_set_lock_changed(_l) LDLM_SET_FLAG(( _l), 1ULL << 0)
+#define ldlm_clear_lock_changed(_l) LDLM_CLEAR_FLAG((_l), 1ULL << 0)
+
+/**
+ * Server placed lock on granted list, or a recovering client wants the
+ * lock added to the granted list, no questions asked. */
+#define LDLM_FL_BLOCK_GRANTED 0x0000000000000002ULL // bit 1
+#define ldlm_is_block_granted(_l) LDLM_TEST_FLAG(( _l), 1ULL << 1)
+#define ldlm_set_block_granted(_l) LDLM_SET_FLAG(( _l), 1ULL << 1)
+#define ldlm_clear_block_granted(_l) LDLM_CLEAR_FLAG((_l), 1ULL << 1)
+
+/**
+ * Server placed lock on conv list, or a recovering client wants the lock
+ * added to the conv list, no questions asked. */
+#define LDLM_FL_BLOCK_CONV 0x0000000000000004ULL // bit 2
+#define ldlm_is_block_conv(_l) LDLM_TEST_FLAG(( _l), 1ULL << 2)
+#define ldlm_set_block_conv(_l) LDLM_SET_FLAG(( _l), 1ULL << 2)
+#define ldlm_clear_block_conv(_l) LDLM_CLEAR_FLAG((_l), 1ULL << 2)
+
+/**
+ * Server placed lock on wait list, or a recovering client wants the lock
+ * added to the wait list, no questions asked. */
+#define LDLM_FL_BLOCK_WAIT 0x0000000000000008ULL // bit 3
+#define ldlm_is_block_wait(_l) LDLM_TEST_FLAG(( _l), 1ULL << 3)
+#define ldlm_set_block_wait(_l) LDLM_SET_FLAG(( _l), 1ULL << 3)
+#define ldlm_clear_block_wait(_l) LDLM_CLEAR_FLAG((_l), 1ULL << 3)
+
+/** blocking or cancel packet was queued for sending. */
+#define LDLM_FL_AST_SENT 0x0000000000000020ULL // bit 5
+#define ldlm_is_ast_sent(_l) LDLM_TEST_FLAG(( _l), 1ULL << 5)
+#define ldlm_set_ast_sent(_l) LDLM_SET_FLAG(( _l), 1ULL << 5)
+#define ldlm_clear_ast_sent(_l) LDLM_CLEAR_FLAG((_l), 1ULL << 5)
+
+/**
+ * Lock is being replayed. This could probably be implied by the fact that
+ * one of BLOCK_{GRANTED,CONV,WAIT} is set, but that is pretty dangerous. */
+#define LDLM_FL_REPLAY 0x0000000000000100ULL // bit 8
+#define ldlm_is_replay(_l) LDLM_TEST_FLAG(( _l), 1ULL << 8)
+#define ldlm_set_replay(_l) LDLM_SET_FLAG(( _l), 1ULL << 8)
+#define ldlm_clear_replay(_l) LDLM_CLEAR_FLAG((_l), 1ULL << 8)
+
+/** Don't grant lock, just do intent. */
+#define LDLM_FL_INTENT_ONLY 0x0000000000000200ULL // bit 9
+#define ldlm_is_intent_only(_l) LDLM_TEST_FLAG(( _l), 1ULL << 9)
+#define ldlm_set_intent_only(_l) LDLM_SET_FLAG(( _l), 1ULL << 9)
+#define ldlm_clear_intent_only(_l) LDLM_CLEAR_FLAG((_l), 1ULL << 9)
+
+/** lock request has intent */
+#define LDLM_FL_HAS_INTENT 0x0000000000001000ULL // bit 12
+#define ldlm_is_has_intent(_l) LDLM_TEST_FLAG(( _l), 1ULL << 12)
+#define ldlm_set_has_intent(_l) LDLM_SET_FLAG(( _l), 1ULL << 12)
+#define ldlm_clear_has_intent(_l) LDLM_CLEAR_FLAG((_l), 1ULL << 12)
+
+/** discard (no writeback) on cancel */
+#define LDLM_FL_DISCARD_DATA 0x0000000000010000ULL // bit 16
+#define ldlm_is_discard_data(_l) LDLM_TEST_FLAG(( _l), 1ULL << 16)
+#define ldlm_set_discard_data(_l) LDLM_SET_FLAG(( _l), 1ULL << 16)
+#define ldlm_clear_discard_data(_l) LDLM_CLEAR_FLAG((_l), 1ULL << 16)
+
+/** Blocked by group lock - wait indefinitely */
+#define LDLM_FL_NO_TIMEOUT 0x0000000000020000ULL // bit 17
+#define ldlm_is_no_timeout(_l) LDLM_TEST_FLAG(( _l), 1ULL << 17)
+#define ldlm_set_no_timeout(_l) LDLM_SET_FLAG(( _l), 1ULL << 17)
+#define ldlm_clear_no_timeout(_l) LDLM_CLEAR_FLAG((_l), 1ULL << 17)
+
+/**
+ * Server told not to wait if blocked. For AGL, OST will not send glimpse
+ * callback. */
+#define LDLM_FL_BLOCK_NOWAIT 0x0000000000040000ULL // bit 18
+#define ldlm_is_block_nowait(_l) LDLM_TEST_FLAG(( _l), 1ULL << 18)
+#define ldlm_set_block_nowait(_l) LDLM_SET_FLAG(( _l), 1ULL << 18)
+#define ldlm_clear_block_nowait(_l) LDLM_CLEAR_FLAG((_l), 1ULL << 18)
+
+/** return blocking lock */
+#define LDLM_FL_TEST_LOCK 0x0000000000080000ULL // bit 19
+#define ldlm_is_test_lock(_l) LDLM_TEST_FLAG(( _l), 1ULL << 19)
+#define ldlm_set_test_lock(_l) LDLM_SET_FLAG(( _l), 1ULL << 19)
+#define ldlm_clear_test_lock(_l) LDLM_CLEAR_FLAG((_l), 1ULL << 19)
+
+/**
+ * Immediatelly cancel such locks when they block some other locks. Send
+ * cancel notification to original lock holder, but expect no reply. This
+ * is for clients (like liblustre) that cannot be expected to reliably
+ * response to blocking AST. */
+#define LDLM_FL_CANCEL_ON_BLOCK 0x0000000000800000ULL // bit 23
+#define ldlm_is_cancel_on_block(_l) LDLM_TEST_FLAG(( _l), 1ULL << 23)
+#define ldlm_set_cancel_on_block(_l) LDLM_SET_FLAG(( _l), 1ULL << 23)
+#define ldlm_clear_cancel_on_block(_l) LDLM_CLEAR_FLAG((_l), 1ULL << 23)
+
+/**
+ * measure lock contention and return -EUSERS if locking contention is high */
+#define LDLM_FL_DENY_ON_CONTENTION 0x0000000040000000ULL // bit 30
+#define ldlm_is_deny_on_contention(_l) LDLM_TEST_FLAG(( _l), 1ULL << 30)
+#define ldlm_set_deny_on_contention(_l) LDLM_SET_FLAG(( _l), 1ULL << 30)
+#define ldlm_clear_deny_on_contention(_l) LDLM_CLEAR_FLAG((_l), 1ULL << 30)
+
+/**
+ * These are flags that are mapped into the flags and ASTs of blocking
+ * locks Add FL_DISCARD to blocking ASTs */
+#define LDLM_FL_AST_DISCARD_DATA 0x0000000080000000ULL // bit 31
+#define ldlm_is_ast_discard_data(_l) LDLM_TEST_FLAG(( _l), 1ULL << 31)
+#define ldlm_set_ast_discard_data(_l) LDLM_SET_FLAG(( _l), 1ULL << 31)
+#define ldlm_clear_ast_discard_data(_l) LDLM_CLEAR_FLAG((_l), 1ULL << 31)
+
+/**
+ * Used for marking lock as a target for -EINTR while cp_ast sleep emulation
+ * + race with upcoming bl_ast. */
+#define LDLM_FL_FAIL_LOC 0x0000000100000000ULL // bit 32
+#define ldlm_is_fail_loc(_l) LDLM_TEST_FLAG(( _l), 1ULL << 32)
+#define ldlm_set_fail_loc(_l) LDLM_SET_FLAG(( _l), 1ULL << 32)
+#define ldlm_clear_fail_loc(_l) LDLM_CLEAR_FLAG((_l), 1ULL << 32)
+
+/**
+ * Used while processing the unused list to know that we have already
+ * handled this lock and decided to skip it. */
+#define LDLM_FL_SKIPPED 0x0000000200000000ULL // bit 33
+#define ldlm_is_skipped(_l) LDLM_TEST_FLAG(( _l), 1ULL << 33)
+#define ldlm_set_skipped(_l) LDLM_SET_FLAG(( _l), 1ULL << 33)
+#define ldlm_clear_skipped(_l) LDLM_CLEAR_FLAG((_l), 1ULL << 33)
+
+/** this lock is being destroyed */
+#define LDLM_FL_CBPENDING 0x0000000400000000ULL // bit 34
+#define ldlm_is_cbpending(_l) LDLM_TEST_FLAG(( _l), 1ULL << 34)
+#define ldlm_set_cbpending(_l) LDLM_SET_FLAG(( _l), 1ULL << 34)
+#define ldlm_clear_cbpending(_l) LDLM_CLEAR_FLAG((_l), 1ULL << 34)
+
+/** not a real flag, not saved in lock */
+#define LDLM_FL_WAIT_NOREPROC 0x0000000800000000ULL // bit 35
+#define ldlm_is_wait_noreproc(_l) LDLM_TEST_FLAG(( _l), 1ULL << 35)
+#define ldlm_set_wait_noreproc(_l) LDLM_SET_FLAG(( _l), 1ULL << 35)
+#define ldlm_clear_wait_noreproc(_l) LDLM_CLEAR_FLAG((_l), 1ULL << 35)
+
+/** cancellation callback already run */
+#define LDLM_FL_CANCEL 0x0000001000000000ULL // bit 36
+#define ldlm_is_cancel(_l) LDLM_TEST_FLAG(( _l), 1ULL << 36)
+#define ldlm_set_cancel(_l) LDLM_SET_FLAG(( _l), 1ULL << 36)
+#define ldlm_clear_cancel(_l) LDLM_CLEAR_FLAG((_l), 1ULL << 36)
+
+/** whatever it might mean */
+#define LDLM_FL_LOCAL_ONLY 0x0000002000000000ULL // bit 37
+#define ldlm_is_local_only(_l) LDLM_TEST_FLAG(( _l), 1ULL << 37)
+#define ldlm_set_local_only(_l) LDLM_SET_FLAG(( _l), 1ULL << 37)
+#define ldlm_clear_local_only(_l) LDLM_CLEAR_FLAG((_l), 1ULL << 37)
+
+/** don't run the cancel callback under ldlm_cli_cancel_unused */
+#define LDLM_FL_FAILED 0x0000004000000000ULL // bit 38
+#define ldlm_is_failed(_l) LDLM_TEST_FLAG(( _l), 1ULL << 38)
+#define ldlm_set_failed(_l) LDLM_SET_FLAG(( _l), 1ULL << 38)
+#define ldlm_clear_failed(_l) LDLM_CLEAR_FLAG((_l), 1ULL << 38)
+
+/** lock cancel has already been sent */
+#define LDLM_FL_CANCELING 0x0000008000000000ULL // bit 39
+#define ldlm_is_canceling(_l) LDLM_TEST_FLAG(( _l), 1ULL << 39)
+#define ldlm_set_canceling(_l) LDLM_SET_FLAG(( _l), 1ULL << 39)
+#define ldlm_clear_canceling(_l) LDLM_CLEAR_FLAG((_l), 1ULL << 39)
+
+/** local lock (ie, no srv/cli split) */
+#define LDLM_FL_LOCAL 0x0000010000000000ULL // bit 40
+#define ldlm_is_local(_l) LDLM_TEST_FLAG(( _l), 1ULL << 40)
+#define ldlm_set_local(_l) LDLM_SET_FLAG(( _l), 1ULL << 40)
+#define ldlm_clear_local(_l) LDLM_CLEAR_FLAG((_l), 1ULL << 40)
+
+/**
+ * XXX FIXME: This is being added to b_size as a low-risk fix to the
+ * fact that the LVB filling happens _after_ the lock has been granted,
+ * so another thread can match it before the LVB has been updated. As a
+ * dirty hack, we set LDLM_FL_LVB_READY only after we've done the LVB poop.
+ * this is only needed on LOV/OSC now, where LVB is actually used and
+ * callers must set it in input flags.
+ *
+ * The proper fix is to do the granting inside of the completion AST,
+ * which can be replaced with a LVB-aware wrapping function for OSC locks.
+ * That change is pretty high-risk, though, and would need a lot more
+ * testing. */
+#define LDLM_FL_LVB_READY 0x0000020000000000ULL // bit 41
+#define ldlm_is_lvb_ready(_l) LDLM_TEST_FLAG(( _l), 1ULL << 41)
+#define ldlm_set_lvb_ready(_l) LDLM_SET_FLAG(( _l), 1ULL << 41)
+#define ldlm_clear_lvb_ready(_l) LDLM_CLEAR_FLAG((_l), 1ULL << 41)
+
+/**
+ * A lock contributes to the known minimum size (KMS) calculation until it
+ * has finished the part of its cancelation that performs write back on its
+ * dirty pages. It can remain on the granted list during this whole time.
+ * Threads racing to update the KMS after performing their writeback need
+ * to know to exclude each other's locks from the calculation as they walk
+ * the granted list. */
+#define LDLM_FL_KMS_IGNORE 0x0000040000000000ULL // bit 42
+#define ldlm_is_kms_ignore(_l) LDLM_TEST_FLAG(( _l), 1ULL << 42)
+#define ldlm_set_kms_ignore(_l) LDLM_SET_FLAG(( _l), 1ULL << 42)
+#define ldlm_clear_kms_ignore(_l) LDLM_CLEAR_FLAG((_l), 1ULL << 42)
+
+/** completion AST to be executed */
+#define LDLM_FL_CP_REQD 0x0000080000000000ULL // bit 43
+#define ldlm_is_cp_reqd(_l) LDLM_TEST_FLAG(( _l), 1ULL << 43)
+#define ldlm_set_cp_reqd(_l) LDLM_SET_FLAG(( _l), 1ULL << 43)
+#define ldlm_clear_cp_reqd(_l) LDLM_CLEAR_FLAG((_l), 1ULL << 43)
+
+/** cleanup_resource has already handled the lock */
+#define LDLM_FL_CLEANED 0x0000100000000000ULL // bit 44
+#define ldlm_is_cleaned(_l) LDLM_TEST_FLAG(( _l), 1ULL << 44)
+#define ldlm_set_cleaned(_l) LDLM_SET_FLAG(( _l), 1ULL << 44)
+#define ldlm_clear_cleaned(_l) LDLM_CLEAR_FLAG((_l), 1ULL << 44)
+
+/**
+ * optimization hint: LDLM can run blocking callback from current context
+ * w/o involving separate thread. in order to decrease cs rate */
+#define LDLM_FL_ATOMIC_CB 0x0000200000000000ULL // bit 45
+#define ldlm_is_atomic_cb(_l) LDLM_TEST_FLAG(( _l), 1ULL << 45)
+#define ldlm_set_atomic_cb(_l) LDLM_SET_FLAG(( _l), 1ULL << 45)
+#define ldlm_clear_atomic_cb(_l) LDLM_CLEAR_FLAG((_l), 1ULL << 45)
+
+/**
+ * It may happen that a client initiates two operations, e.g. unlink and
+ * mkdir, such that the server sends a blocking AST for conflicting locks
+ * to this client for the first operation, whereas the second operation
+ * has canceled this lock and is waiting for rpc_lock which is taken by
+ * the first operation. LDLM_FL_BL_AST is set by ldlm_callback_handler() in
+ * the lock to prevent the Early Lock Cancel (ELC) code from cancelling it.
+ *
+ * LDLM_FL_BL_DONE is to be set by ldlm_cancel_callback() when lock cache is
+ * dropped to let ldlm_callback_handler() return EINVAL to the server. It
+ * is used when ELC RPC is already prepared and is waiting for rpc_lock,
+ * too late to send a separate CANCEL RPC. */
+#define LDLM_FL_BL_AST 0x0000400000000000ULL // bit 46
+#define ldlm_is_bl_ast(_l) LDLM_TEST_FLAG(( _l), 1ULL << 46)
+#define ldlm_set_bl_ast(_l) LDLM_SET_FLAG(( _l), 1ULL << 46)
+#define ldlm_clear_bl_ast(_l) LDLM_CLEAR_FLAG((_l), 1ULL << 46)
+
+/** whatever it might mean */
+#define LDLM_FL_BL_DONE 0x0000800000000000ULL // bit 47
+#define ldlm_is_bl_done(_l) LDLM_TEST_FLAG(( _l), 1ULL << 47)
+#define ldlm_set_bl_done(_l) LDLM_SET_FLAG(( _l), 1ULL << 47)
+#define ldlm_clear_bl_done(_l) LDLM_CLEAR_FLAG((_l), 1ULL << 47)
+
+/**
+ * Don't put lock into the LRU list, so that it is not canceled due
+ * to aging. Used by MGC locks, they are cancelled only at unmount or
+ * by callback. */
+#define LDLM_FL_NO_LRU 0x0001000000000000ULL // bit 48
+#define ldlm_is_no_lru(_l) LDLM_TEST_FLAG(( _l), 1ULL << 48)
+#define ldlm_set_no_lru(_l) LDLM_SET_FLAG(( _l), 1ULL << 48)
+#define ldlm_clear_no_lru(_l) LDLM_CLEAR_FLAG((_l), 1ULL << 48)
+
+/**
+ * Set for locks that failed and where the server has been notified.
+ *
+ * Protected by lock and resource locks. */
+#define LDLM_FL_FAIL_NOTIFIED 0x0002000000000000ULL // bit 49
+#define ldlm_is_fail_notified(_l) LDLM_TEST_FLAG(( _l), 1ULL << 49)
+#define ldlm_set_fail_notified(_l) LDLM_SET_FLAG(( _l), 1ULL << 49)
+#define ldlm_clear_fail_notified(_l) LDLM_CLEAR_FLAG((_l), 1ULL << 49)
+
+/**
+ * Set for locks that were removed from class hash table and will
+ * be destroyed when last reference to them is released. Set by
+ * ldlm_lock_destroy_internal().
+ *
+ * Protected by lock and resource locks. */
+#define LDLM_FL_DESTROYED 0x0004000000000000ULL // bit 50
+#define ldlm_is_destroyed(_l) LDLM_TEST_FLAG(( _l), 1ULL << 50)
+#define ldlm_set_destroyed(_l) LDLM_SET_FLAG(( _l), 1ULL << 50)
+#define ldlm_clear_destroyed(_l) LDLM_CLEAR_FLAG((_l), 1ULL << 50)
+
+/** flag whether this is a server namespace lock */
+#define LDLM_FL_SERVER_LOCK 0x0008000000000000ULL // bit 51
+#define ldlm_is_server_lock(_l) LDLM_TEST_FLAG(( _l), 1ULL << 51)
+#define ldlm_set_server_lock(_l) LDLM_SET_FLAG(( _l), 1ULL << 51)
+#define ldlm_clear_server_lock(_l) LDLM_CLEAR_FLAG((_l), 1ULL << 51)
+
+/**
+ * It's set in lock_res_and_lock() and unset in unlock_res_and_lock().
+ *
+ * NB: compared with check_res_locked(), checking this bit is cheaper.
+ * Also, spin_is_locked() is deprecated for kernel code; one reason is
+ * because it works only for SMP so user needs to add extra macros like
+ * LASSERT_SPIN_LOCKED for uniprocessor kernels. */
+#define LDLM_FL_RES_LOCKED 0x0010000000000000ULL // bit 52
+#define ldlm_is_res_locked(_l) LDLM_TEST_FLAG(( _l), 1ULL << 52)
+#define ldlm_set_res_locked(_l) LDLM_SET_FLAG(( _l), 1ULL << 52)
+#define ldlm_clear_res_locked(_l) LDLM_CLEAR_FLAG((_l), 1ULL << 52)
+
+/**
+ * It's set once we call ldlm_add_waiting_lock_res_locked() to start the
+ * lock-timeout timer and it will never be reset.
+ *
+ * Protected by lock and resource locks. */
+#define LDLM_FL_WAITED 0x0020000000000000ULL // bit 53
+#define ldlm_is_waited(_l) LDLM_TEST_FLAG(( _l), 1ULL << 53)
+#define ldlm_set_waited(_l) LDLM_SET_FLAG(( _l), 1ULL << 53)
+#define ldlm_clear_waited(_l) LDLM_CLEAR_FLAG((_l), 1ULL << 53)
+
+/** Flag whether this is a server namespace lock. */
+#define LDLM_FL_NS_SRV 0x0040000000000000ULL // bit 54
+#define ldlm_is_ns_srv(_l) LDLM_TEST_FLAG(( _l), 1ULL << 54)
+#define ldlm_set_ns_srv(_l) LDLM_SET_FLAG(( _l), 1ULL << 54)
+#define ldlm_clear_ns_srv(_l) LDLM_CLEAR_FLAG((_l), 1ULL << 54)
+
+/** test for ldlm_lock flag bit set */
+#define LDLM_TEST_FLAG(_l, _b) (((_l)->l_flags & (_b)) != 0)
+
+/** set a ldlm_lock flag bit */
+#define LDLM_SET_FLAG(_l, _b) (((_l)->l_flags |= (_b))
+
+/** clear a ldlm_lock flag bit */
+#define LDLM_CLEAR_FLAG(_l, _b) (((_l)->l_flags &= ~(_b))
+
+/** Mask of flags inherited from parent lock when doing intents. */
+#define LDLM_INHERIT_FLAGS LDLM_FL_INHERIT_MASK
+
+/** Mask of Flags sent in AST lock_flags to map into the receiving lock. */
+#define LDLM_AST_FLAGS LDLM_FL_AST_MASK
+
+/** @} subgroup */
+/** @} group */
+#ifdef WIRESHARK_COMPILE
+static int hf_lustre_ldlm_fl_lock_changed = -1;
+static int hf_lustre_ldlm_fl_block_granted = -1;
+static int hf_lustre_ldlm_fl_block_conv = -1;
+static int hf_lustre_ldlm_fl_block_wait = -1;
+static int hf_lustre_ldlm_fl_ast_sent = -1;
+static int hf_lustre_ldlm_fl_replay = -1;
+static int hf_lustre_ldlm_fl_intent_only = -1;
+static int hf_lustre_ldlm_fl_has_intent = -1;
+static int hf_lustre_ldlm_fl_discard_data = -1;
+static int hf_lustre_ldlm_fl_no_timeout = -1;
+static int hf_lustre_ldlm_fl_block_nowait = -1;
+static int hf_lustre_ldlm_fl_test_lock = -1;
+static int hf_lustre_ldlm_fl_cancel_on_block = -1;
+static int hf_lustre_ldlm_fl_deny_on_contention = -1;
+static int hf_lustre_ldlm_fl_ast_discard_data = -1;
+static int hf_lustre_ldlm_fl_fail_loc = -1;
+static int hf_lustre_ldlm_fl_skipped = -1;
+static int hf_lustre_ldlm_fl_cbpending = -1;
+static int hf_lustre_ldlm_fl_wait_noreproc = -1;
+static int hf_lustre_ldlm_fl_cancel = -1;
+static int hf_lustre_ldlm_fl_local_only = -1;
+static int hf_lustre_ldlm_fl_failed = -1;
+static int hf_lustre_ldlm_fl_canceling = -1;
+static int hf_lustre_ldlm_fl_local = -1;
+static int hf_lustre_ldlm_fl_lvb_ready = -1;
+static int hf_lustre_ldlm_fl_kms_ignore = -1;
+static int hf_lustre_ldlm_fl_cp_reqd = -1;
+static int hf_lustre_ldlm_fl_cleaned = -1;
+static int hf_lustre_ldlm_fl_atomic_cb = -1;
+static int hf_lustre_ldlm_fl_bl_ast = -1;
+static int hf_lustre_ldlm_fl_bl_done = -1;
+static int hf_lustre_ldlm_fl_no_lru = -1;
+static int hf_lustre_ldlm_fl_fail_notified = -1;
+static int hf_lustre_ldlm_fl_destroyed = -1;
+static int hf_lustre_ldlm_fl_server_lock = -1;
+static int hf_lustre_ldlm_fl_res_locked = -1;
+static int hf_lustre_ldlm_fl_waited = -1;
+static int hf_lustre_ldlm_fl_ns_srv = -1;
+
+const value_string lustre_ldlm_flags_vals[] = {
+ {LDLM_FL_LOCK_CHANGED, "LDLM_FL_LOCK_CHANGED"},
+ {LDLM_FL_BLOCK_GRANTED, "LDLM_FL_BLOCK_GRANTED"},
+ {LDLM_FL_BLOCK_CONV, "LDLM_FL_BLOCK_CONV"},
+ {LDLM_FL_BLOCK_WAIT, "LDLM_FL_BLOCK_WAIT"},
+ {LDLM_FL_AST_SENT, "LDLM_FL_AST_SENT"},
+ {LDLM_FL_REPLAY, "LDLM_FL_REPLAY"},
+ {LDLM_FL_INTENT_ONLY, "LDLM_FL_INTENT_ONLY"},
+ {LDLM_FL_HAS_INTENT, "LDLM_FL_HAS_INTENT"},
+ {LDLM_FL_DISCARD_DATA, "LDLM_FL_DISCARD_DATA"},
+ {LDLM_FL_NO_TIMEOUT, "LDLM_FL_NO_TIMEOUT"},
+ {LDLM_FL_BLOCK_NOWAIT, "LDLM_FL_BLOCK_NOWAIT"},
+ {LDLM_FL_TEST_LOCK, "LDLM_FL_TEST_LOCK"},
+ {LDLM_FL_CANCEL_ON_BLOCK, "LDLM_FL_CANCEL_ON_BLOCK"},
+ {LDLM_FL_DENY_ON_CONTENTION, "LDLM_FL_DENY_ON_CONTENTION"},
+ {LDLM_FL_AST_DISCARD_DATA, "LDLM_FL_AST_DISCARD_DATA"},
+ {LDLM_FL_FAIL_LOC, "LDLM_FL_FAIL_LOC"},
+ {LDLM_FL_SKIPPED, "LDLM_FL_SKIPPED"},
+ {LDLM_FL_CBPENDING, "LDLM_FL_CBPENDING"},
+ {LDLM_FL_WAIT_NOREPROC, "LDLM_FL_WAIT_NOREPROC"},
+ {LDLM_FL_CANCEL, "LDLM_FL_CANCEL"},
+ {LDLM_FL_LOCAL_ONLY, "LDLM_FL_LOCAL_ONLY"},
+ {LDLM_FL_FAILED, "LDLM_FL_FAILED"},
+ {LDLM_FL_CANCELING, "LDLM_FL_CANCELING"},
+ {LDLM_FL_LOCAL, "LDLM_FL_LOCAL"},
+ {LDLM_FL_LVB_READY, "LDLM_FL_LVB_READY"},
+ {LDLM_FL_KMS_IGNORE, "LDLM_FL_KMS_IGNORE"},
+ {LDLM_FL_CP_REQD, "LDLM_FL_CP_REQD"},
+ {LDLM_FL_CLEANED, "LDLM_FL_CLEANED"},
+ {LDLM_FL_ATOMIC_CB, "LDLM_FL_ATOMIC_CB"},
+ {LDLM_FL_BL_AST, "LDLM_FL_BL_AST"},
+ {LDLM_FL_BL_DONE, "LDLM_FL_BL_DONE"},
+ {LDLM_FL_NO_LRU, "LDLM_FL_NO_LRU"},
+ {LDLM_FL_FAIL_NOTIFIED, "LDLM_FL_FAIL_NOTIFIED"},
+ {LDLM_FL_DESTROYED, "LDLM_FL_DESTROYED"},
+ {LDLM_FL_SERVER_LOCK, "LDLM_FL_SERVER_LOCK"},
+ {LDLM_FL_RES_LOCKED, "LDLM_FL_RES_LOCKED"},
+ {LDLM_FL_WAITED, "LDLM_FL_WAITED"},
+ {LDLM_FL_NS_SRV, "LDLM_FL_NS_SRV"},
+ { 0, NULL }
+};
+#endif /* WIRESHARK_COMPILE */
+#endif /* LDLM_ALL_FLAGS_MASK */
diff --git a/drivers/staging/lustre/lustre/ldlm/l_lock.c b/drivers/staging/lustre/lustre/ldlm/l_lock.c
index 853409a..32f4d52 100644
--- a/drivers/staging/lustre/lustre/ldlm/l_lock.c
+++ b/drivers/staging/lustre/lustre/ldlm/l_lock.c
@@ -51,12 +51,12 @@
struct ldlm_resource *lock_res_and_lock(struct ldlm_lock *lock)
{
/* on server-side resource of lock doesn't change */
- if (!lock->l_ns_srv)
+ if ((lock->l_flags & LDLM_FL_NS_SRV) == 0)
spin_lock(&lock->l_lock);
lock_res(lock->l_resource);
- lock->l_res_locked = 1;
+ lock->l_flags |= LDLM_FL_RES_LOCKED;
return lock->l_resource;
}
EXPORT_SYMBOL(lock_res_and_lock);
@@ -67,10 +67,10 @@ EXPORT_SYMBOL(lock_res_and_lock);
void unlock_res_and_lock(struct ldlm_lock *lock)
{
/* on server-side resource of lock doesn't change */
- lock->l_res_locked = 0;
+ lock->l_flags &= ~LDLM_FL_RES_LOCKED;
unlock_res(lock->l_resource);
- if (!lock->l_ns_srv)
+ if ((lock->l_flags & LDLM_FL_NS_SRV) == 0)
spin_unlock(&lock->l_lock);
}
EXPORT_SYMBOL(unlock_res_and_lock);
diff --git a/drivers/staging/lustre/lustre/ldlm/ldlm_flock.c b/drivers/staging/lustre/lustre/ldlm/ldlm_flock.c
index f100a84..aca1073 100644
--- a/drivers/staging/lustre/lustre/ldlm/ldlm_flock.c
+++ b/drivers/staging/lustre/lustre/ldlm/ldlm_flock.c
@@ -639,7 +639,7 @@ ldlm_flock_completion_ast(struct ldlm_lock *lock, __u64 flags, void *data)
granted:
OBD_FAIL_TIMEOUT(OBD_FAIL_LDLM_CP_CB_WAIT, 10);
- if (lock->l_destroyed) {
+ if (lock->l_flags & LDLM_FL_DESTROYED) {
LDLM_DEBUG(lock, "client-side enqueue waking up: destroyed");
RETURN(0);
}
diff --git a/drivers/staging/lustre/lustre/ldlm/ldlm_lock.c b/drivers/staging/lustre/lustre/ldlm/ldlm_lock.c
index 0b3ea88..c10ba9c 100644
--- a/drivers/staging/lustre/lustre/ldlm/ldlm_lock.c
+++ b/drivers/staging/lustre/lustre/ldlm/ldlm_lock.c
@@ -199,7 +199,7 @@ void ldlm_lock_put(struct ldlm_lock *lock)
"final lock_put on destroyed lock, freeing it.");
res = lock->l_resource;
- LASSERT(lock->l_destroyed);
+ LASSERT(lock->l_flags & LDLM_FL_DESTROYED);
LASSERT(list_empty(&lock->l_res_link));
LASSERT(list_empty(&lock->l_pending_chain));
@@ -254,7 +254,7 @@ int ldlm_lock_remove_from_lru(struct ldlm_lock *lock)
int rc;
ENTRY;
- if (lock->l_ns_srv) {
+ if (lock->l_flags & LDLM_FL_NS_SRV) {
LASSERT(list_empty(&lock->l_lru));
RETURN(0);
}
@@ -305,7 +305,7 @@ void ldlm_lock_touch_in_lru(struct ldlm_lock *lock)
struct ldlm_namespace *ns = ldlm_lock_to_ns(lock);
ENTRY;
- if (lock->l_ns_srv) {
+ if (lock->l_flags & LDLM_FL_NS_SRV) {
LASSERT(list_empty(&lock->l_lru));
EXIT;
return;
@@ -353,12 +353,12 @@ int ldlm_lock_destroy_internal(struct ldlm_lock *lock)
LBUG();
}
- if (lock->l_destroyed) {
+ if (lock->l_flags & LDLM_FL_DESTROYED) {
LASSERT(list_empty(&lock->l_lru));
EXIT;
return 0;
}
- lock->l_destroyed = 1;
+ lock->l_flags |= LDLM_FL_DESTROYED;
if (lock->l_export && lock->l_export->exp_lock_hash) {
/* NB: it's safe to call cfs_hash_del() even lock isn't
@@ -596,7 +596,7 @@ struct ldlm_lock *__ldlm_handle2lock(const struct lustre_handle *handle,
/* It's unlikely but possible that someone marked the lock as
* destroyed after we did handle2object on it */
- if (flags == 0 && !lock->l_destroyed) {
+ if (flags == 0 && ((lock->l_flags & LDLM_FL_DESTROYED)== 0)) {
lu_ref_add(&lock->l_reference, "handle", current);
RETURN(lock);
}
@@ -606,7 +606,7 @@ struct ldlm_lock *__ldlm_handle2lock(const struct lustre_handle *handle,
LASSERT(lock->l_resource != NULL);
lu_ref_add_atomic(&lock->l_reference, "handle", current);
- if (unlikely(lock->l_destroyed)) {
+ if (unlikely(lock->l_flags & LDLM_FL_DESTROYED)) {
unlock_res_and_lock(lock);
CDEBUG(D_INFO, "lock already destroyed: lock %p\n", lock);
LDLM_LOCK_PUT(lock);
@@ -695,7 +695,7 @@ void ldlm_add_bl_work_item(struct ldlm_lock *lock, struct ldlm_lock *new,
lock->l_flags |= LDLM_FL_AST_SENT;
/* If the enqueuing client said so, tell the AST recipient to
* discard dirty data, rather than writing back. */
- if (new->l_flags & LDLM_AST_DISCARD_DATA)
+ if (new->l_flags & LDLM_FL_AST_DISCARD_DATA)
lock->l_flags |= LDLM_FL_DISCARD_DATA;
LASSERT(list_empty(&lock->l_bl_ast));
list_add(&lock->l_bl_ast, work_list);
@@ -873,7 +873,7 @@ void ldlm_lock_decref_internal(struct ldlm_lock *lock, __u32 mode)
(lock->l_flags & LDLM_FL_CBPENDING)) {
/* If we received a blocked AST and this was the last reference,
* run the callback. */
- if (lock->l_ns_srv && lock->l_export)
+ if ((lock->l_flags & LDLM_FL_NS_SRV) && lock->l_export)
CERROR("FL_CBPENDING set on non-local lock--just a "
"warning\n");
@@ -1069,7 +1069,7 @@ static void ldlm_granted_list_add_lock(struct ldlm_lock *lock,
ldlm_resource_dump(D_INFO, res);
LDLM_DEBUG(lock, "About to add lock:");
- if (lock->l_destroyed) {
+ if (lock->l_flags & LDLM_FL_DESTROYED) {
CDEBUG(D_OTHER, "Lock destroyed, not adding to resource\n");
return;
}
@@ -1203,9 +1203,7 @@ static struct ldlm_lock *search_queue(struct list_head *queue,
policy->l_inodebits.bits))
continue;
- if (!unref &&
- (lock->l_destroyed || lock->l_flags & LDLM_FL_FAILED ||
- lock->l_failed))
+ if (!unref && (lock->l_flags & LDLM_FL_GONE_MASK))
continue;
if ((flags & LDLM_FL_LOCAL_ONLY) &&
@@ -1227,8 +1225,8 @@ static struct ldlm_lock *search_queue(struct list_head *queue,
void ldlm_lock_fail_match_locked(struct ldlm_lock *lock)
{
- if (!lock->l_failed) {
- lock->l_failed = 1;
+ if ((lock->l_flags & LDLM_FL_FAIL_NOTIFIED) == 0) {
+ lock->l_flags |= LDLM_FL_FAIL_NOTIFIED;
wake_up_all(&lock->l_waitq);
}
}
@@ -1352,6 +1350,8 @@ ldlm_mode_t ldlm_lock_match(struct ldlm_namespace *ns, __u64 flags,
ldlm_lock2handle(lock, lockh);
if ((flags & LDLM_FL_LVB_READY) &&
(!(lock->l_flags & LDLM_FL_LVB_READY))) {
+ __u64 wait_flags = LDLM_FL_LVB_READY |
+ LDLM_FL_DESTROYED | LDLM_FL_FAIL_NOTIFIED;
struct l_wait_info lwi;
if (lock->l_completion_ast) {
int err = lock->l_completion_ast(lock,
@@ -1373,8 +1373,7 @@ ldlm_mode_t ldlm_lock_match(struct ldlm_namespace *ns, __u64 flags,
/* XXX FIXME see comment on CAN_MATCH in lustre_dlm.h */
l_wait_event(lock->l_waitq,
- lock->l_flags & LDLM_FL_LVB_READY ||
- lock->l_destroyed || lock->l_failed,
+ lock->l_flags & wait_flags,
&lwi);
if (!(lock->l_flags & LDLM_FL_LVB_READY)) {
if (flags & LDLM_FL_TEST_LOCK)
@@ -1431,8 +1430,7 @@ ldlm_mode_t ldlm_revalidate_lock_handle(struct lustre_handle *lockh,
lock = ldlm_handle2lock(lockh);
if (lock != NULL) {
lock_res_and_lock(lock);
- if (lock->l_destroyed || lock->l_flags & LDLM_FL_FAILED ||
- lock->l_failed)
+ if (lock->l_flags & LDLM_FL_GONE_MASK)
GOTO(out, mode);
if (lock->l_flags & LDLM_FL_CBPENDING &&
@@ -1583,12 +1581,12 @@ struct ldlm_lock *ldlm_lock_create(struct ldlm_namespace *ns,
lock->l_req_mode = mode;
lock->l_ast_data = data;
lock->l_pid = current_pid();
- lock->l_ns_srv = !!ns_is_server(ns);
+ if (ns_is_server(ns))
+ lock->l_flags |= LDLM_FL_NS_SRV;
if (cbs) {
lock->l_blocking_ast = cbs->lcs_blocking;
lock->l_completion_ast = cbs->lcs_completion;
lock->l_glimpse_ast = cbs->lcs_glimpse;
- lock->l_weigh_ast = cbs->lcs_weigh;
}
lock->l_tree_node = NULL;
@@ -1693,7 +1691,7 @@ ldlm_error_t ldlm_lock_enqueue(struct ldlm_namespace *ns,
/* Some flags from the enqueue want to make it into the AST, via the
* lock's l_flags. */
- lock->l_flags |= *flags & LDLM_AST_DISCARD_DATA;
+ lock->l_flags |= *flags & LDLM_FL_AST_DISCARD_DATA;
/* This distinction between local lock trees is very important; a client
* namespace only has information about locks taken by that client, and
@@ -2046,15 +2044,15 @@ void ldlm_lock_cancel(struct ldlm_lock *lock)
LBUG();
}
- if (lock->l_waited)
+ if (lock->l_flags & LDLM_FL_WAITED)
ldlm_del_waiting_lock(lock);
/* Releases cancel callback. */
ldlm_cancel_callback(lock);
/* Yes, second time, just in case it was added again while we were
- running with no res lock in ldlm_cancel_callback */
- if (lock->l_waited)
+ * running with no res lock in ldlm_cancel_callback */
+ if (lock->l_flags & LDLM_FL_WAITED)
ldlm_del_waiting_lock(lock);
ldlm_resource_unlink_lock(lock);
diff --git a/drivers/staging/lustre/lustre/ldlm/ldlm_lockd.c b/drivers/staging/lustre/lustre/ldlm/ldlm_lockd.c
index f79b244..d18936a 100644
--- a/drivers/staging/lustre/lustre/ldlm/ldlm_lockd.c
+++ b/drivers/staging/lustre/lustre/ldlm/ldlm_lockd.c
@@ -198,7 +198,7 @@ static void ldlm_handle_cp_callback(struct ptlrpc_request *req,
schedule_timeout_and_set_state(
TASK_INTERRUPTIBLE, to);
if (lock->l_granted_mode == lock->l_req_mode ||
- lock->l_destroyed)
+ lock->l_flags & LDLM_FL_DESTROYED)
break;
}
}
@@ -238,7 +238,7 @@ static void ldlm_handle_cp_callback(struct ptlrpc_request *req,
}
lock_res_and_lock(lock);
- if (lock->l_destroyed ||
+ if ((lock->l_flags & LDLM_FL_DESTROYED) ||
lock->l_granted_mode == lock->l_req_mode) {
/* bug 11300: the lock has already been granted */
unlock_res_and_lock(lock);
diff --git a/drivers/staging/lustre/lustre/ldlm/ldlm_request.c b/drivers/staging/lustre/lustre/ldlm/ldlm_request.c
index ab41fea..c900706 100644
--- a/drivers/staging/lustre/lustre/ldlm/ldlm_request.c
+++ b/drivers/staging/lustre/lustre/ldlm/ldlm_request.c
@@ -160,7 +160,7 @@ static int ldlm_completion_tail(struct ldlm_lock *lock)
long delay;
int result;
- if (lock->l_destroyed || lock->l_flags & LDLM_FL_FAILED) {
+ if (lock->l_flags & (LDLM_FL_DESTROYED | LDLM_FL_FAILED)) {
LDLM_DEBUG(lock, "client-side enqueue: destroyed");
result = -EIO;
} else {
@@ -888,9 +888,8 @@ int ldlm_cli_enqueue(struct obd_export *exp, struct ptlrpc_request **reqp,
} else {
const struct ldlm_callback_suite cbs = {
.lcs_completion = einfo->ei_cb_cp,
- .lcs_blocking = einfo->ei_cb_bl,
- .lcs_glimpse = einfo->ei_cb_gl,
- .lcs_weigh = einfo->ei_cb_wg
+ .lcs_blocking = einfo->ei_cb_bl,
+ .lcs_glimpse = einfo->ei_cb_gl
};
lock = ldlm_lock_create(ns, res_id, einfo->ei_type,
einfo->ei_mode, &cbs, einfo->ei_cbdata,
diff --git a/drivers/staging/lustre/lustre/ldlm/ldlm_resource.c b/drivers/staging/lustre/lustre/ldlm/ldlm_resource.c
index 17eab79..616cb17 100644
--- a/drivers/staging/lustre/lustre/ldlm/ldlm_resource.c
+++ b/drivers/staging/lustre/lustre/ldlm/ldlm_resource.c
@@ -1283,7 +1283,7 @@ void ldlm_resource_add_lock(struct ldlm_resource *res, struct list_head *head,
LDLM_DEBUG(lock, "About to add this lock:\n");
- if (lock->l_destroyed) {
+ if (lock->l_flags & LDLM_FL_DESTROYED) {
CDEBUG(D_OTHER, "Lock destroyed, not adding to resource\n");
return;
}
@@ -1308,7 +1308,7 @@ void ldlm_resource_insert_lock_after(struct ldlm_lock *original,
ldlm_resource_dump(D_INFO, res);
LDLM_DEBUG(new, "About to insert this lock after %p:\n", original);
- if (new->l_destroyed) {
+ if (new->l_flags & LDLM_FL_DESTROYED) {
CDEBUG(D_OTHER, "Lock destroyed, not adding to resource\n");
goto out;
}
diff --git a/drivers/staging/lustre/lustre/llite/dir.c b/drivers/staging/lustre/lustre/llite/dir.c
index bfc914e..a94e463 100644
--- a/drivers/staging/lustre/lustre/llite/dir.c
+++ b/drivers/staging/lustre/lustre/llite/dir.c
@@ -356,15 +356,12 @@ struct page *ll_get_dir_page(struct inode *dir, __u64 hash,
rc = md_lock_match(ll_i2sbi(dir)->ll_md_exp, LDLM_FL_BLOCK_GRANTED,
ll_inode2fid(dir), LDLM_IBITS, &policy, mode, &lockh);
if (!rc) {
- struct ldlm_enqueue_info einfo = {.ei_type = LDLM_IBITS,
- .ei_mode = mode,
- .ei_cb_bl =
- ll_md_blocking_ast,
- .ei_cb_cp =
- ldlm_completion_ast,
- .ei_cb_gl = NULL,
- .ei_cb_wg = NULL,
- .ei_cbdata = NULL};
+ struct ldlm_enqueue_info einfo = {
+ .ei_type = LDLM_IBITS,
+ .ei_mode = mode,
+ .ei_cb_bl = ll_md_blocking_ast,
+ .ei_cb_cp = ldlm_completion_ast,
+ };
struct lookup_intent it = { .it_op = IT_READDIR };
struct ptlrpc_request *request;
struct md_op_data *op_data;
diff --git a/drivers/staging/lustre/lustre/llite/file.c b/drivers/staging/lustre/lustre/llite/file.c
index 50ef6a5..d51f1ee 100644
--- a/drivers/staging/lustre/lustre/llite/file.c
+++ b/drivers/staging/lustre/lustre/llite/file.c
@@ -2290,9 +2290,11 @@ int ll_file_flock(struct file *file, int cmd, struct file_lock *file_lock)
{
struct inode *inode = file->f_dentry->d_inode;
struct ll_sb_info *sbi = ll_i2sbi(inode);
- struct ldlm_enqueue_info einfo = { .ei_type = LDLM_FLOCK,
- .ei_cb_cp =ldlm_flock_completion_ast,
- .ei_cbdata = file_lock };
+ struct ldlm_enqueue_info einfo = {
+ .ei_type = LDLM_FLOCK,
+ .ei_cb_cp = ldlm_flock_completion_ast,
+ .ei_cbdata = file_lock,
+ };
struct md_op_data *op_data;
struct lustre_handle lockh = {0};
ldlm_policy_data_t flock = {{0}};
@@ -3116,11 +3118,12 @@ int ll_layout_refresh(struct inode *inode, __u32 *gen)
struct lookup_intent it;
struct lustre_handle lockh;
ldlm_mode_t mode;
- struct ldlm_enqueue_info einfo = { .ei_type = LDLM_IBITS,
- .ei_mode = LCK_CR,
- .ei_cb_bl = ll_md_blocking_ast,
- .ei_cb_cp = ldlm_completion_ast,
- .ei_cbdata = NULL };
+ struct ldlm_enqueue_info einfo = {
+ .ei_type = LDLM_IBITS,
+ .ei_mode = LCK_CR,
+ .ei_cb_bl = ll_md_blocking_ast,
+ .ei_cb_cp = ldlm_completion_ast,
+ };
int rc;
ENTRY;
diff --git a/drivers/staging/lustre/lustre/mdc/mdc_locks.c b/drivers/staging/lustre/lustre/mdc/mdc_locks.c
index 9532134..ab88a46 100644
--- a/drivers/staging/lustre/lustre/mdc/mdc_locks.c
+++ b/drivers/staging/lustre/lustre/mdc/mdc_locks.c
@@ -1102,9 +1102,12 @@ int mdc_intent_lock(struct obd_export *exp, struct md_op_data *op_data,
* 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) {
diff --git a/drivers/staging/lustre/lustre/mgc/mgc_request.c b/drivers/staging/lustre/lustre/mgc/mgc_request.c
index c6c84d9..5167767 100644
--- a/drivers/staging/lustre/lustre/mgc/mgc_request.c
+++ b/drivers/staging/lustre/lustre/mgc/mgc_request.c
@@ -900,8 +900,12 @@ static int mgc_enqueue(struct obd_export *exp, struct lov_stripe_md *lsm,
struct lustre_handle *lockh)
{
struct config_llog_data *cld = (struct config_llog_data *)data;
- struct ldlm_enqueue_info einfo = { type, mode, mgc_blocking_ast,
- ldlm_completion_ast, NULL, NULL, NULL };
+ struct ldlm_enqueue_info einfo = {
+ .ei_type = type,
+ .ei_mode = mode,
+ .ei_cb_bl = mgc_blocking_ast,
+ .ei_cb_cp = ldlm_completion_ast,
+ };
struct ptlrpc_request *req;
int short_limit = cld_is_sptlrpc(cld);
int rc;
diff --git a/drivers/staging/lustre/lustre/osc/osc_lock.c b/drivers/staging/lustre/lustre/osc/osc_lock.c
index 640bc3d..98478d2 100644
--- a/drivers/staging/lustre/lustre/osc/osc_lock.c
+++ b/drivers/staging/lustre/lustre/osc/osc_lock.c
@@ -89,35 +89,49 @@ static struct ldlm_lock *osc_handle_ptr(struct lustre_handle *handle)
*/
static int osc_lock_invariant(struct osc_lock *ols)
{
- struct ldlm_lock *lock = osc_handle_ptr(&ols->ols_handle);
- struct ldlm_lock *olock = ols->ols_lock;
- int handle_used = lustre_handle_is_used(&ols->ols_handle);
-
- return
- ergo(osc_lock_is_lockless(ols),
- ols->ols_locklessable && ols->ols_lock == NULL) ||
- (ergo(olock != NULL, handle_used) &&
- ergo(olock != NULL,
- olock->l_handle.h_cookie == ols->ols_handle.cookie) &&
- /*
- * Check that ->ols_handle and ->ols_lock are consistent, but
- * take into account that they are set at the different time.
- */
- ergo(handle_used,
- ergo(lock != NULL && olock != NULL, lock == olock) &&
- ergo(lock == NULL, olock == NULL)) &&
- ergo(ols->ols_state == OLS_CANCELLED,
- olock == NULL && !handle_used) &&
- /*
- * DLM lock is destroyed only after we have seen cancellation
- * ast.
- */
- ergo(olock != NULL && ols->ols_state < OLS_CANCELLED,
- !olock->l_destroyed) &&
- ergo(ols->ols_state == OLS_GRANTED,
- olock != NULL &&
- olock->l_req_mode == olock->l_granted_mode &&
- ols->ols_hold));
+ struct ldlm_lock *lock = osc_handle_ptr(&ols->ols_handle);
+ struct ldlm_lock *olock = ols->ols_lock;
+ int handle_used = lustre_handle_is_used(&ols->ols_handle);
+
+ if (ergo(osc_lock_is_lockless(ols),
+ ols->ols_locklessable && ols->ols_lock == NULL))
+ return 1;
+
+ /*
+ * If all the following "ergo"s are true, return 1, otherwise 0
+ */
+ if (! ergo(olock != NULL, handle_used))
+ return 0;
+
+ if (! ergo(olock != NULL,
+ olock->l_handle.h_cookie == ols->ols_handle.cookie))
+ return 0;
+
+ if (! ergo(handle_used,
+ ergo(lock != NULL && olock != NULL, lock == olock) &&
+ ergo(lock == NULL, olock == NULL)))
+ return 0;
+ /*
+ * Check that ->ols_handle and ->ols_lock are consistent, but
+ * take into account that they are set at the different time.
+ */
+ if (! ergo(ols->ols_state == OLS_CANCELLED,
+ olock == NULL && !handle_used))
+ return 0;
+ /*
+ * DLM lock is destroyed only after we have seen cancellation
+ * ast.
+ */
+ if (! ergo(olock != NULL && ols->ols_state < OLS_CANCELLED,
+ ((olock->l_flags & LDLM_FL_DESTROYED) == 0)))
+ return 0;
+
+ if (! ergo(ols->ols_state == OLS_GRANTED,
+ olock != NULL &&
+ olock->l_req_mode == olock->l_granted_mode &&
+ ols->ols_hold))
+ return 0;
+ return 1;
}
/*****************************************************************************
@@ -261,7 +275,7 @@ static __u64 osc_enq2ldlm_flags(__u32 enqflags)
if (enqflags & CEF_ASYNC)
result |= LDLM_FL_HAS_INTENT;
if (enqflags & CEF_DISCARD_DATA)
- result |= LDLM_AST_DISCARD_DATA;
+ result |= LDLM_FL_AST_DISCARD_DATA;
return result;
}
@@ -896,55 +910,6 @@ static unsigned long osc_lock_weigh(const struct lu_env *env,
return cl_object_header(slice->cls_obj)->coh_pages;
}
-/**
- * Get the weight of dlm lock for early cancellation.
- *
- * XXX: it should return the pages covered by this \a dlmlock.
- */
-static unsigned long osc_ldlm_weigh_ast(struct ldlm_lock *dlmlock)
-{
- struct cl_env_nest nest;
- struct lu_env *env;
- struct osc_lock *lock;
- struct cl_lock *cll;
- unsigned long weight;
- ENTRY;
-
- might_sleep();
- /*
- * osc_ldlm_weigh_ast has a complex context since it might be called
- * because of lock canceling, or from user's input. We have to make
- * a new environment for it. Probably it is implementation safe to use
- * the upper context because cl_lock_put don't modify environment
- * variables. But in case of ..
- */
- env = cl_env_nested_get(&nest);
- if (IS_ERR(env))
- /* Mostly because lack of memory, tend to eliminate this lock*/
- RETURN(0);
-
- LASSERT(dlmlock->l_resource->lr_type == LDLM_EXTENT);
- lock = osc_ast_data_get(dlmlock);
- if (lock == NULL) {
- /* cl_lock was destroyed because of memory pressure.
- * It is much reasonable to assign this type of lock
- * a lower cost.
- */
- GOTO(out, weight = 0);
- }
-
- cll = lock->ols_cl.cls_lock;
- cl_lock_mutex_get(env, cll);
- weight = cl_lock_weigh(env, cll);
- cl_lock_mutex_put(env, cll);
- osc_ast_data_put(env, lock);
- EXIT;
-
-out:
- cl_env_nested_put(&nest, env);
- return weight;
-}
-
static void osc_lock_build_einfo(const struct lu_env *env,
const struct cl_lock *clock,
struct osc_lock *lock,
@@ -966,7 +931,6 @@ static void osc_lock_build_einfo(const struct lu_env *env,
einfo->ei_cb_bl = osc_ldlm_blocking_ast;
einfo->ei_cb_cp = osc_ldlm_completion_ast;
einfo->ei_cb_gl = osc_ldlm_glimpse_ast;
- einfo->ei_cb_wg = osc_ldlm_weigh_ast;
einfo->ei_cbdata = lock; /* value to be put into ->l_ast_data */
}
--
1.7.9.5
^ permalink raw reply related [flat|nested] 62+ messages in thread* [PATCH 17/48] staging/lustre/md: fix lu_ucred.c boilerplate
2013-07-22 16:06 [PATCH 00/48] staging/lustre: minor cleanup and Intel Lustre sync Peng Tao
` (15 preceding siblings ...)
2013-07-22 16:06 ` [PATCH 16/48] staging/lustre/dlmlock: compress out unused space Peng Tao
@ 2013-07-22 16:06 ` Peng Tao
2013-07-22 16:06 ` [PATCH 18/48] staging/lustre/layout: introduce new layout for released files Peng Tao
` (31 subsequent siblings)
48 siblings, 0 replies; 62+ messages in thread
From: Peng Tao @ 2013-07-22 16:06 UTC (permalink / raw)
To: Greg Kroah-Hartman
Cc: linux-kernel, John L. Hammond, Peng Tao, Andreas Dilger
From: "John L. Hammond" <john.hammond@intel.com>
In preparing Ie3a3cd99 (LU-1330 obdclass: splits server-side object
stack from client) the lu_ucred infrastructure was put in its own
file. Fixup the boilerplate of this file to give the proper path,
short description, and authors.
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-1330
Lustre-change: http://review.whamcloud.com/5910
Signed-off-by: John L. Hammond <john.hammond@intel.com>
Reviewed-by: Fan Yong <fan.yong@intel.com>
Reviewed-by: Vitaly Fertman <vitaly_fertman@xyratex.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>
Signed-off-by: Peng Tao <tao.peng@emc.com>
Signed-off-by: Andreas Dilger <andreas.dilger@intel.com>
---
drivers/staging/lustre/lustre/obdclass/lu_ucred.c | 8 ++++----
1 file changed, 4 insertions(+), 4 deletions(-)
diff --git a/drivers/staging/lustre/lustre/obdclass/lu_ucred.c b/drivers/staging/lustre/lustre/obdclass/lu_ucred.c
index 229db6c..e23e545 100644
--- a/drivers/staging/lustre/lustre/obdclass/lu_ucred.c
+++ b/drivers/staging/lustre/lustre/obdclass/lu_ucred.c
@@ -33,13 +33,13 @@
* This file is part of Lustre, http://www.lustre.org/
* Lustre is a trademark of Sun Microsystems, Inc.
*
- * lustre/obdclass/lu_object.c
+ * lustre/obdclass/lu_ucred.c
*
- * Lustre Object.
- * These are the only exported functions, they provide some generic
- * infrastructure for managing object devices
+ * Lustre user credentials context infrastructure.
*
* Author: Nikita Danilov <nikita.danilov@sun.com>
+ * Author: Fan Yong <fan.yong@intel.com>
+ * Author: Vitaly Fertman <vitaly_fertman@xyratex.com>
*/
#define DEBUG_SUBSYSTEM S_CLASS
--
1.7.9.5
^ permalink raw reply related [flat|nested] 62+ messages in thread* [PATCH 18/48] staging/lustre/layout: introduce new layout for released files
2013-07-22 16:06 [PATCH 00/48] staging/lustre: minor cleanup and Intel Lustre sync Peng Tao
` (16 preceding siblings ...)
2013-07-22 16:06 ` [PATCH 17/48] staging/lustre/md: fix lu_ucred.c boilerplate Peng Tao
@ 2013-07-22 16:06 ` Peng Tao
2013-07-22 16:06 ` [PATCH 19/48] staging/lustre/mdt: add macros for fid string len Peng Tao
` (30 subsequent siblings)
48 siblings, 0 replies; 62+ messages in thread
From: Peng Tao @ 2013-07-22 16:06 UTC (permalink / raw)
To: Greg Kroah-Hartman
Cc: linux-kernel, Jinshan Xiong, Johann Lombardi, Peng Tao, Andreas Dilger
From: Jinshan Xiong <jinshan.xiong@intel.com>
Released files now have a standard layout (with generation, pool, ...)
and a stripe count 0 and lmm_pattern flag LOV_PATTERN_F_RELEASED.
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-2482
Lustre-change: http://review.whamcloud.com/4816
Signed-off-by: Jinshan Xiong <jinshan.xiong@intel.com>
Signed-off-by: Johann Lombardi <johann.lombardi@intel.com>
Reviewed-by: Keith Mannthey <keith.mannthey@intel.com>
Reviewed-by: Fan Yong <fan.yong@intel.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Signed-off-by: Peng Tao <tao.peng@emc.com>
Signed-off-by: Andreas Dilger <andreas.dilger@intel.com>
---
.../lustre/lustre/include/lustre/lustre_idl.h | 14 ++--
drivers/staging/lustre/lustre/include/obd.h | 14 ++++
drivers/staging/lustre/lustre/lclient/lcommon_cl.c | 2 +-
drivers/staging/lustre/lustre/llite/file.c | 27 +++++---
drivers/staging/lustre/lustre/llite/llite_lib.c | 22 +++---
drivers/staging/lustre/lustre/llite/vvp_io.c | 10 +--
.../staging/lustre/lustre/lov/lov_cl_internal.h | 11 +--
drivers/staging/lustre/lustre/lov/lov_ea.c | 23 ++++--
drivers/staging/lustre/lustre/lov/lov_io.c | 34 +++++++++
drivers/staging/lustre/lustre/lov/lov_obd.c | 2 +-
drivers/staging/lustre/lustre/lov/lov_object.c | 73 +++++++++++++++++---
drivers/staging/lustre/lustre/lov/lov_pack.c | 19 +++--
12 files changed, 197 insertions(+), 54 deletions(-)
diff --git a/drivers/staging/lustre/lustre/include/lustre/lustre_idl.h b/drivers/staging/lustre/lustre/include/lustre/lustre_idl.h
index 4829419..30c0331 100644
--- a/drivers/staging/lustre/lustre/include/lustre/lustre_idl.h
+++ b/drivers/staging/lustre/lustre/include/lustre/lustre_idl.h
@@ -1546,10 +1546,16 @@ enum obdo_flags {
#define LOV_MAGIC_V1_DEF 0x0CD10BD0
#define LOV_MAGIC_V3_DEF 0x0CD30BD0
-#define LOV_PATTERN_RAID0 0x001 /* stripes are used round-robin */
-#define LOV_PATTERN_RAID1 0x002 /* stripes are mirrors of each other */
-#define LOV_PATTERN_FIRST 0x100 /* first stripe is not in round-robin */
-#define LOV_PATTERN_CMOBD 0x200
+#define LOV_PATTERN_RAID0 0x001 /* stripes are used round-robin */
+#define LOV_PATTERN_RAID1 0x002 /* stripes are mirrors of each other */
+#define LOV_PATTERN_FIRST 0x100 /* first stripe is not in round-robin */
+#define LOV_PATTERN_CMOBD 0x200
+
+#define LOV_PATTERN_F_MASK 0xffff0000
+#define LOV_PATTERN_F_RELEASED 0x80000000 /* HSM released file */
+
+#define lov_pattern(pattern) (pattern & ~LOV_PATTERN_F_MASK)
+#define lov_pattern_flags(pattern) (pattern & LOV_PATTERN_F_MASK)
#define lov_ost_data lov_ost_data_v1
struct lov_ost_data_v1 { /* per-stripe data structure (little-endian)*/
diff --git a/drivers/staging/lustre/lustre/include/obd.h b/drivers/staging/lustre/lustre/include/obd.h
index 0a251fd..11d10fe 100644
--- a/drivers/staging/lustre/lustre/include/obd.h
+++ b/drivers/staging/lustre/lustre/include/obd.h
@@ -119,6 +119,20 @@ struct lov_stripe_md {
#define lsm_stripe_count lsm_wire.lw_stripe_count
#define lsm_pool_name lsm_wire.lw_pool_name
+static inline bool lsm_is_released(struct lov_stripe_md *lsm)
+{
+ return !!(lsm->lsm_pattern & LOV_PATTERN_F_RELEASED);
+}
+
+static inline bool lsm_has_objects(struct lov_stripe_md *lsm)
+{
+ if (lsm == NULL)
+ return false;
+ if (lsm_is_released(lsm))
+ return false;
+ return true;
+}
+
struct obd_info;
typedef int (*obd_enqueue_update_f)(void *cookie, int rc);
diff --git a/drivers/staging/lustre/lustre/lclient/lcommon_cl.c b/drivers/staging/lustre/lustre/lclient/lcommon_cl.c
index 4a01666..c8690cb 100644
--- a/drivers/staging/lustre/lustre/lclient/lcommon_cl.c
+++ b/drivers/staging/lustre/lustre/lclient/lcommon_cl.c
@@ -1166,7 +1166,7 @@ int cl_file_inode_init(struct inode *inode, struct lustre_md *md)
* locked by I_NEW bit.
*/
lli->lli_clob = clob;
- lli->lli_has_smd = md->lsm != NULL;
+ lli->lli_has_smd = lsm_has_objects(md->lsm);
lu_object_ref_add(&clob->co_lu, "inode", inode);
} else
result = PTR_ERR(clob);
diff --git a/drivers/staging/lustre/lustre/llite/file.c b/drivers/staging/lustre/lustre/llite/file.c
index d51f1ee..5493845 100644
--- a/drivers/staging/lustre/lustre/llite/file.c
+++ b/drivers/staging/lustre/lustre/llite/file.c
@@ -1143,7 +1143,7 @@ static int ll_lov_recreate(struct inode *inode, struct ost_id *oi,
RETURN(-ENOMEM);
lsm = ccc_inode_lsm_get(inode);
- if (lsm == NULL)
+ if (!lsm_has_objects(lsm))
GOTO(out, rc = -ENOENT);
lsm_size = sizeof(*lsm) + (sizeof(struct lov_oinfo) *
@@ -1299,6 +1299,12 @@ int ll_lov_getstripe_ea_info(struct inode *inode, const char *filename,
* passing it to userspace.
*/
if (LOV_MAGIC != cpu_to_le32(LOV_MAGIC)) {
+ int stripe_count;
+
+ stripe_count = le16_to_cpu(lmm->lmm_stripe_count);
+ if (le32_to_cpu(lmm->lmm_pattern) & LOV_PATTERN_F_RELEASED)
+ stripe_count = 0;
+
/* if function called for directory - we should
* avoid swab not existent lsm objects */
if (lmm->lmm_magic == cpu_to_le32(LOV_MAGIC_V1)) {
@@ -1306,13 +1312,13 @@ int ll_lov_getstripe_ea_info(struct inode *inode, const char *filename,
if (S_ISREG(body->mode))
lustre_swab_lov_user_md_objects(
((struct lov_user_md_v1 *)lmm)->lmm_objects,
- ((struct lov_user_md_v1 *)lmm)->lmm_stripe_count);
+ stripe_count);
} else if (lmm->lmm_magic == cpu_to_le32(LOV_MAGIC_V3)) {
lustre_swab_lov_user_md_v3((struct lov_user_md_v3 *)lmm);
if (S_ISREG(body->mode))
lustre_swab_lov_user_md_objects(
((struct lov_user_md_v3 *)lmm)->lmm_objects,
- ((struct lov_user_md_v3 *)lmm)->lmm_stripe_count);
+ stripe_count);
}
}
@@ -1698,20 +1704,18 @@ int ll_data_version(struct inode *inode, __u64 *data_version,
/* If no stripe, we consider version is 0. */
lsm = ccc_inode_lsm_get(inode);
- if (lsm == NULL) {
+ if (!lsm_has_objects(lsm)) {
*data_version = 0;
CDEBUG(D_INODE, "No object for inode\n");
- RETURN(0);
+ GOTO(out, rc = 0);
}
OBD_ALLOC_PTR(obdo);
- if (obdo == NULL) {
- ccc_inode_lsm_put(inode, lsm);
- RETURN(-ENOMEM);
- }
+ if (obdo == NULL)
+ GOTO(out, rc = -ENOMEM);
rc = ll_lsm_getattr(lsm, sbi->ll_dt_exp, NULL, obdo, 0, extent_lock);
- if (!rc) {
+ if (rc == 0) {
if (!(obdo->o_valid & OBD_MD_FLDATAVERSION))
rc = -EOPNOTSUPP;
else
@@ -1719,8 +1723,9 @@ int ll_data_version(struct inode *inode, __u64 *data_version,
}
OBD_FREE_PTR(obdo);
+ EXIT;
+out:
ccc_inode_lsm_put(inode, lsm);
-
RETURN(rc);
}
diff --git a/drivers/staging/lustre/lustre/llite/llite_lib.c b/drivers/staging/lustre/lustre/llite/llite_lib.c
index 1eb5779..b9cbe95 100644
--- a/drivers/staging/lustre/lustre/llite/llite_lib.c
+++ b/drivers/staging/lustre/lustre/llite/llite_lib.c
@@ -1282,13 +1282,6 @@ int ll_md_setattr(struct dentry *dentry, struct md_op_data *op_data,
RETURN(rc);
}
- ia_valid = op_data->op_attr.ia_valid;
- /* inode size will be in ll_setattr_ost, can't do it now since dirty
- * cache is not cleared yet. */
- op_data->op_attr.ia_valid &= ~(TIMES_SET_FLAGS | ATTR_SIZE);
- rc = simple_setattr(dentry, &op_data->op_attr);
- op_data->op_attr.ia_valid = ia_valid;
-
/* Extract epoch data if obtained. */
op_data->op_handle = md.body->handle;
op_data->op_ioepoch = md.body->ioepoch;
@@ -1371,6 +1364,7 @@ int ll_setattr_raw(struct dentry *dentry, struct iattr *attr)
struct ll_inode_info *lli = ll_i2info(inode);
struct md_op_data *op_data = NULL;
struct md_open_data *mod = NULL;
+ unsigned int valid = attr->ia_valid;
int rc = 0, rc1 = 0;
ENTRY;
@@ -1482,6 +1476,13 @@ int ll_setattr_raw(struct dentry *dentry, struct iattr *attr)
rc = ll_setattr_ost(inode, attr);
EXIT;
out:
+ if (rc == 0) {
+ /* Update inode attribute after dirty cache is cleaned
+ * by truncating OST objects. */
+ attr->ia_valid |= ATTR_FORCE;
+ rc = simple_setattr(dentry, attr);
+ LASSERT(rc == 0);
+ }
if (op_data) {
if (op_data->op_ioepoch) {
rc1 = ll_setattr_done_writing(inode, op_data, mod);
@@ -1500,7 +1501,8 @@ out:
ll_stats_ops_tally(ll_i2sbi(inode), (attr->ia_valid & ATTR_SIZE) ?
LPROC_LL_TRUNC : LPROC_LL_SETATTR, 1);
- return rc;
+ attr->ia_valid = valid;
+ RETURN(rc);
}
int ll_setattr(struct dentry *de, struct iattr *attr)
@@ -1923,8 +1925,10 @@ int ll_iocontrol(struct inode *inode, struct file *file,
inode->i_flags = ll_ext_to_inode_flags(flags);
lsm = ccc_inode_lsm_get(inode);
- if (lsm == NULL)
+ if (!lsm_has_objects(lsm)) {
+ ccc_inode_lsm_put(inode, lsm);
RETURN(0);
+ }
OBDO_ALLOC(oinfo.oi_oa);
if (!oinfo.oi_oa) {
diff --git a/drivers/staging/lustre/lustre/llite/vvp_io.c b/drivers/staging/lustre/lustre/llite/vvp_io.c
index c24dc268..158b27e 100644
--- a/drivers/staging/lustre/lustre/llite/vvp_io.c
+++ b/drivers/staging/lustre/lustre/llite/vvp_io.c
@@ -407,13 +407,15 @@ static int vvp_io_setattr_start(const struct lu_env *env,
{
struct cl_io *io = ios->cis_io;
struct inode *inode = ccc_object_inode(io->ci_obj);
+ int result = 0;
mutex_lock(&inode->i_mutex);
if (cl_io_is_trunc(io))
- return vvp_io_setattr_trunc(env, ios, inode,
- io->u.ci_setattr.sa_attr.lvb_size);
- else
- return vvp_io_setattr_time(env, ios);
+ result = vvp_io_setattr_trunc(env, ios, inode,
+ io->u.ci_setattr.sa_attr.lvb_size);
+ if (result == 0)
+ result = vvp_io_setattr_time(env, ios);
+ return result;
}
static void vvp_io_setattr_end(const struct lu_env *env,
diff --git a/drivers/staging/lustre/lustre/lov/lov_cl_internal.h b/drivers/staging/lustre/lustre/lov/lov_cl_internal.h
index 28801b8..a82dfea 100644
--- a/drivers/staging/lustre/lustre/lov/lov_cl_internal.h
+++ b/drivers/staging/lustre/lustre/lov/lov_cl_internal.h
@@ -162,10 +162,9 @@ struct lov_device {
* Layout type.
*/
enum lov_layout_type {
- /** empty file without body */
- LLT_EMPTY,
- /** striped file */
- LLT_RAID0,
+ LLT_EMPTY, /** empty file without body (mknod + truncate) */
+ LLT_RAID0, /** striped file */
+ LLT_RELEASED, /** file with no objects (data in HSM) */
LLT_NR
};
@@ -255,6 +254,8 @@ struct lov_object {
} raid0;
struct lov_layout_state_empty {
} empty;
+ struct lov_layout_state_released {
+ } released;
} u;
/**
* Thread that acquired lov_object::lo_type_guard in an exclusive
@@ -582,6 +583,8 @@ int lov_io_init_raid0 (const struct lu_env *env, struct cl_object *obj,
struct cl_io *io);
int lov_io_init_empty (const struct lu_env *env, struct cl_object *obj,
struct cl_io *io);
+int lov_io_init_released(const struct lu_env *env, struct cl_object *obj,
+ struct cl_io *io);
void lov_lock_unlink (const struct lu_env *env, struct lov_lock_link *link,
struct lovsub_lock *sub);
diff --git a/drivers/staging/lustre/lustre/lov/lov_ea.c b/drivers/staging/lustre/lustre/lov/lov_ea.c
index 340dbcf..e6c6015 100644
--- a/drivers/staging/lustre/lustre/lov/lov_ea.c
+++ b/drivers/staging/lustre/lustre/lov/lov_ea.c
@@ -57,7 +57,7 @@ struct lovea_unpack_args {
static int lsm_lmm_verify_common(struct lov_mds_md *lmm, int lmm_bytes,
__u16 stripe_count)
{
- if (stripe_count == 0 || stripe_count > LOV_V1_INSANE_STRIPE_COUNT) {
+ if (stripe_count > LOV_V1_INSANE_STRIPE_COUNT) {
CERROR("bad stripe count %d\n", stripe_count);
lov_dump_lmm_common(D_WARNING, lmm);
return -EINVAL;
@@ -69,7 +69,7 @@ static int lsm_lmm_verify_common(struct lov_mds_md *lmm, int lmm_bytes,
return -EINVAL;
}
- if (lmm->lmm_pattern != cpu_to_le32(LOV_PATTERN_RAID0)) {
+ if (lov_pattern(le32_to_cpu(lmm->lmm_pattern)) != LOV_PATTERN_RAID0) {
CERROR("bad striping pattern\n");
lov_dump_lmm_common(D_WARNING, lmm);
return -EINVAL;
@@ -197,6 +197,8 @@ static int lsm_lmm_verify_v1(struct lov_mds_md_v1 *lmm, int lmm_bytes,
}
*stripe_count = le16_to_cpu(lmm->lmm_stripe_count);
+ if (le32_to_cpu(lmm->lmm_pattern) & LOV_PATTERN_F_RELEASED)
+ *stripe_count = 0;
if (lmm_bytes < lov_mds_md_size(*stripe_count, LOV_MAGIC_V1)) {
CERROR("LOV EA V1 too small: %d, need %d\n",
@@ -213,11 +215,14 @@ int lsm_unpackmd_v1(struct lov_obd *lov, struct lov_stripe_md *lsm,
{
struct lov_oinfo *loi;
int i;
+ int stripe_count;
__u64 stripe_maxbytes = OBD_OBJECT_EOF;
lsm_unpackmd_common(lsm, lmm);
- for (i = 0; i < lsm->lsm_stripe_count; i++) {
+ stripe_count = lsm_is_released(lsm) ? 0 : lsm->lsm_stripe_count;
+
+ for (i = 0; i < stripe_count; i++) {
/* XXX LOV STACKING call down to osc_unpackmd() */
loi = lsm->lsm_oinfo[i];
ostid_le_to_cpu(&lmm->lmm_objects[i].l_ost_oi, &loi->loi_oi);
@@ -240,6 +245,8 @@ int lsm_unpackmd_v1(struct lov_obd *lov, struct lov_stripe_md *lsm,
}
lsm->lsm_maxbytes = stripe_maxbytes * lsm->lsm_stripe_count;
+ if (lsm->lsm_stripe_count == 0)
+ lsm->lsm_maxbytes = stripe_maxbytes * lov->desc.ld_tgt_count;
return 0;
}
@@ -267,6 +274,8 @@ static int lsm_lmm_verify_v3(struct lov_mds_md *lmmv1, int lmm_bytes,
}
*stripe_count = le16_to_cpu(lmm->lmm_stripe_count);
+ if (le32_to_cpu(lmm->lmm_pattern) & LOV_PATTERN_F_RELEASED)
+ *stripe_count = 0;
if (lmm_bytes < lov_mds_md_size(*stripe_count, LOV_MAGIC_V3)) {
CERROR("LOV EA V3 too small: %d, need %d\n",
@@ -285,18 +294,22 @@ int lsm_unpackmd_v3(struct lov_obd *lov, struct lov_stripe_md *lsm,
struct lov_mds_md_v3 *lmm;
struct lov_oinfo *loi;
int i;
+ int stripe_count;
__u64 stripe_maxbytes = OBD_OBJECT_EOF;
int cplen = 0;
lmm = (struct lov_mds_md_v3 *)lmmv1;
lsm_unpackmd_common(lsm, (struct lov_mds_md_v1 *)lmm);
+
+ stripe_count = lsm_is_released(lsm) ? 0 : lsm->lsm_stripe_count;
+
cplen = strlcpy(lsm->lsm_pool_name, lmm->lmm_pool_name,
sizeof(lsm->lsm_pool_name));
if (cplen >= sizeof(lsm->lsm_pool_name))
return -E2BIG;
- for (i = 0; i < lsm->lsm_stripe_count; i++) {
+ for (i = 0; i < stripe_count; i++) {
/* XXX LOV STACKING call down to osc_unpackmd() */
loi = lsm->lsm_oinfo[i];
ostid_le_to_cpu(&lmm->lmm_objects[i].l_ost_oi, &loi->loi_oi);
@@ -319,6 +332,8 @@ int lsm_unpackmd_v3(struct lov_obd *lov, struct lov_stripe_md *lsm,
}
lsm->lsm_maxbytes = stripe_maxbytes * lsm->lsm_stripe_count;
+ if (lsm->lsm_stripe_count == 0)
+ lsm->lsm_maxbytes = stripe_maxbytes * lov->desc.ld_tgt_count;
return 0;
}
diff --git a/drivers/staging/lustre/lustre/lov/lov_io.c b/drivers/staging/lustre/lustre/lov/lov_io.c
index 022e0f4..e78cdb7 100644
--- a/drivers/staging/lustre/lustre/lov/lov_io.c
+++ b/drivers/staging/lustre/lustre/lov/lov_io.c
@@ -964,4 +964,38 @@ int lov_io_init_empty(const struct lu_env *env, struct cl_object *obj,
RETURN(result != 0);
}
+int lov_io_init_released(const struct lu_env *env, struct cl_object *obj,
+ struct cl_io *io)
+{
+ struct lov_object *lov = cl2lov(obj);
+ struct lov_io *lio = lov_env_io(env);
+ int result;
+ ENTRY;
+
+ LASSERT(lov->lo_lsm != NULL);
+ lio->lis_object = lov;
+
+ switch (io->ci_type) {
+ default:
+ LASSERTF(0, "invalid type %d\n", io->ci_type);
+ case CIT_MISC:
+ case CIT_FSYNC:
+ case CIT_SETATTR:
+ result = +1;
+ break;
+ case CIT_READ:
+ case CIT_WRITE:
+ case CIT_FAULT:
+ /* TODO: need to restore the file. */
+ result = -EBADF;
+ break;
+ }
+ if (result == 0) {
+ cl_io_slice_add(io, &lio->lis_cl, obj, &lov_empty_io_ops);
+ atomic_inc(&lov->lo_active_ios);
+ }
+
+ io->ci_result = result < 0 ? result : 0;
+ RETURN(result != 0);
+}
/** @} lov */
diff --git a/drivers/staging/lustre/lustre/lov/lov_obd.c b/drivers/staging/lustre/lustre/lov/lov_obd.c
index ef7ff09..0a9e40d 100644
--- a/drivers/staging/lustre/lustre/lov/lov_obd.c
+++ b/drivers/staging/lustre/lustre/lov/lov_obd.c
@@ -2259,7 +2259,7 @@ static int lov_fiemap(struct lov_obd *lov, __u32 keylen, void *key,
int cur_stripe = 0, cur_stripe_wrap = 0, stripe_count;
unsigned int buffer_size = FIEMAP_BUFFER_SIZE;
- if (lsm == NULL)
+ if (!lsm_has_objects(lsm))
GOTO(out, rc = 0);
if (fiemap_count_to_size(fm_key->fiemap.fm_extent_count) < buffer_size)
diff --git a/drivers/staging/lustre/lustre/lov/lov_object.c b/drivers/staging/lustre/lustre/lov/lov_object.c
index aa8ae80..e7cc650 100644
--- a/drivers/staging/lustre/lustre/lov/lov_object.c
+++ b/drivers/staging/lustre/lustre/lov/lov_object.c
@@ -258,10 +258,25 @@ out:
RETURN(result);
}
+static int lov_init_released(const struct lu_env *env,
+ struct lov_device *dev, struct lov_object *lov,
+ const struct cl_object_conf *conf,
+ union lov_layout_state *state)
+{
+ struct lov_stripe_md *lsm = conf->u.coc_md->lsm;
+
+ LASSERT(lsm != NULL);
+ LASSERT(lsm_is_released(lsm));
+ LASSERT(lov->lo_lsm == NULL);
+
+ lov->lo_lsm = lsm_addref(lsm);
+ return 0;
+}
+
static int lov_delete_empty(const struct lu_env *env, struct lov_object *lov,
union lov_layout_state *state)
{
- LASSERT(lov->lo_type == LLT_EMPTY);
+ LASSERT(lov->lo_type == LLT_EMPTY || lov->lo_type == LLT_RELEASED);
lov_layout_wait(env, lov);
@@ -349,7 +364,7 @@ static int lov_delete_raid0(const struct lu_env *env, struct lov_object *lov,
static void lov_fini_empty(const struct lu_env *env, struct lov_object *lov,
union lov_layout_state *state)
{
- LASSERT(lov->lo_type == LLT_EMPTY);
+ LASSERT(lov->lo_type == LLT_EMPTY || lov->lo_type == LLT_RELEASED);
}
static void lov_fini_raid0(const struct lu_env *env, struct lov_object *lov,
@@ -369,6 +384,15 @@ static void lov_fini_raid0(const struct lu_env *env, struct lov_object *lov,
EXIT;
}
+static void lov_fini_released(const struct lu_env *env, struct lov_object *lov,
+ union lov_layout_state *state)
+{
+ ENTRY;
+ dump_lsm(D_INODE, lov->lo_lsm);
+ lov_free_memmd(&lov->lo_lsm);
+ EXIT;
+}
+
static int lov_print_empty(const struct lu_env *env, void *cookie,
lu_printer_t p, const struct lu_object *o)
{
@@ -400,6 +424,13 @@ static int lov_print_raid0(const struct lu_env *env, void *cookie,
return 0;
}
+static int lov_print_released(const struct lu_env *env, void *cookie,
+ lu_printer_t p, const struct lu_object *o)
+{
+ (*p)(env, cookie, "released\n");
+ return 0;
+}
+
/**
* Implements cl_object_operations::coo_attr_get() method for an object
* without stripes (LLT_EMPTY layout type).
@@ -503,10 +534,20 @@ const static struct lov_layout_operations lov_dispatch[] = {
.llo_lock_init = lov_lock_init_raid0,
.llo_io_init = lov_io_init_raid0,
.llo_getattr = lov_attr_get_raid0
+ },
+ [LLT_RELEASED] = {
+ .llo_init = lov_init_released,
+ .llo_delete = lov_delete_empty,
+ .llo_fini = lov_fini_released,
+ .llo_install = lov_install_empty,
+ .llo_print = lov_print_released,
+ .llo_page_init = lov_page_init_empty,
+ .llo_lock_init = lov_lock_init_empty,
+ .llo_io_init = lov_io_init_released,
+ .llo_getattr = lov_attr_get_empty
}
};
-
/**
* Performs a double-dispatch based on the layout type of an object.
*/
@@ -520,6 +561,18 @@ const static struct lov_layout_operations lov_dispatch[] = {
lov_dispatch[__llt].op(__VA_ARGS__); \
})
+/**
+ * Return lov_layout_type associated with a given lsm
+ */
+enum lov_layout_type lov_type(struct lov_stripe_md *lsm)
+{
+ if (lsm == NULL)
+ return LLT_EMPTY;
+ if (lsm_is_released(lsm))
+ return LLT_RELEASED;
+ return LLT_RAID0;
+}
+
static inline void lov_conf_freeze(struct lov_object *lov)
{
if (lov->lo_owner != current)
@@ -612,8 +665,8 @@ static int lov_layout_change(const struct lu_env *unused,
LASSERT(0 <= lov->lo_type && lov->lo_type < ARRAY_SIZE(lov_dispatch));
- if (conf->u.coc_md != NULL && conf->u.coc_md->lsm != NULL)
- llt = LLT_RAID0; /* only raid0 is supported. */
+ if (conf->u.coc_md != NULL)
+ llt = lov_type(conf->u.coc_md->lsm);
LASSERT(0 <= llt && llt < ARRAY_SIZE(lov_dispatch));
cookie = cl_env_reenter();
@@ -658,7 +711,6 @@ static int lov_layout_change(const struct lu_env *unused,
* Lov object operations.
*
*/
-
int lov_object_init(const struct lu_env *env, struct lu_object *obj,
const struct lu_object_conf *conf)
{
@@ -677,7 +729,7 @@ int lov_object_init(const struct lu_env *env, struct lu_object *obj,
cl_object_page_init(lu2cl(obj), sizeof(struct lov_page));
/* no locking is necessary, as object is being created */
- lov->lo_type = cconf->u.coc_md->lsm != NULL ? LLT_RAID0 : LLT_EMPTY;
+ lov->lo_type = lov_type(cconf->u.coc_md->lsm);
ops = &lov_dispatch[lov->lo_type];
result = ops->llo_init(env, dev, lov, cconf, set);
if (result == 0)
@@ -688,9 +740,9 @@ int lov_object_init(const struct lu_env *env, struct lu_object *obj,
static int lov_conf_set(const struct lu_env *env, struct cl_object *obj,
const struct cl_object_conf *conf)
{
- struct lov_stripe_md *lsm = NULL;
- struct lov_object *lov = cl2lov(obj);
- int result = 0;
+ struct lov_stripe_md *lsm = NULL;
+ struct lov_object *lov = cl2lov(obj);
+ int result = 0;
ENTRY;
lov_conf_lock(lov);
@@ -928,6 +980,7 @@ int lov_read_and_clear_async_rc(struct cl_object *clob)
loi->loi_ar.ar_rc = 0;
}
}
+ case LLT_RELEASED:
case LLT_EMPTY:
break;
default:
diff --git a/drivers/staging/lustre/lustre/lov/lov_pack.c b/drivers/staging/lustre/lustre/lov/lov_pack.c
index 492948a..9c99e4a 100644
--- a/drivers/staging/lustre/lustre/lov/lov_pack.c
+++ b/drivers/staging/lustre/lustre/lov/lov_pack.c
@@ -168,10 +168,12 @@ int lov_packmd(struct obd_export *exp, struct lov_mds_md **lmmp,
* to the actual number of OSTs in this filesystem. */
if (!lmmp) {
stripe_count = lov_get_stripecnt(lov, lmm_magic,
- lsm->lsm_stripe_count);
+ lsm->lsm_stripe_count);
lsm->lsm_stripe_count = stripe_count;
- } else {
+ } else if (!lsm_is_released(lsm)) {
stripe_count = lsm->lsm_stripe_count;
+ } else {
+ stripe_count = 0;
}
} else {
/* No need to allocate more than maximum supported stripes.
@@ -325,7 +327,8 @@ int lov_alloc_memmd(struct lov_stripe_md **lsmp, __u16 stripe_count,
(*lsmp)->lsm_pattern = pattern;
(*lsmp)->lsm_pool_name[0] = '\0';
(*lsmp)->lsm_layout_gen = 0;
- (*lsmp)->lsm_oinfo[0]->loi_ost_idx = ~0;
+ if (stripe_count > 0)
+ (*lsmp)->lsm_oinfo[0]->loi_ost_idx = ~0;
for (i = 0; i < stripe_count; i++)
loi_init((*lsmp)->lsm_oinfo[i]);
@@ -359,6 +362,7 @@ int lov_unpackmd(struct obd_export *exp, struct lov_stripe_md **lsmp,
int rc = 0, lsm_size;
__u16 stripe_count;
__u32 magic;
+ __u32 pattern;
ENTRY;
/* If passed an MDS struct use values from there, otherwise defaults */
@@ -384,8 +388,8 @@ int lov_unpackmd(struct obd_export *exp, struct lov_stripe_md **lsmp,
RETURN(0);
}
- lsm_size = lov_alloc_memmd(lsmp, stripe_count, LOV_PATTERN_RAID0,
- magic);
+ pattern = le32_to_cpu(lmm->lmm_pattern);
+ lsm_size = lov_alloc_memmd(lsmp, stripe_count, pattern, magic);
if (lsm_size < 0)
RETURN(lsm_size);
@@ -430,7 +434,7 @@ static int __lov_setstripe(struct obd_export *exp, int max_lmm_size,
lov->desc.ld_pattern : LOV_PATTERN_RAID0;
}
- if (lumv1->lmm_pattern != LOV_PATTERN_RAID0) {
+ if (lov_pattern(lumv1->lmm_pattern) != LOV_PATTERN_RAID0) {
CDEBUG(D_IOCTL, "bad userland stripe pattern: %#x\n",
lumv1->lmm_pattern);
RETURN(-EINVAL);
@@ -490,6 +494,9 @@ static int __lov_setstripe(struct obd_export *exp, int max_lmm_size,
}
}
+ if (lumv1->lmm_pattern & LOV_PATTERN_F_RELEASED)
+ stripe_count = 0;
+
rc = lov_alloc_memmd(lsmp, stripe_count, lumv1->lmm_pattern, lmm_magic);
if (rc >= 0) {
--
1.7.9.5
^ permalink raw reply related [flat|nested] 62+ messages in thread* [PATCH 19/48] staging/lustre/mdt: add macros for fid string len
2013-07-22 16:06 [PATCH 00/48] staging/lustre: minor cleanup and Intel Lustre sync Peng Tao
` (17 preceding siblings ...)
2013-07-22 16:06 ` [PATCH 18/48] staging/lustre/layout: introduce new layout for released files Peng Tao
@ 2013-07-22 16:06 ` Peng Tao
2013-07-22 16:06 ` [PATCH 20/48] staging/lustre/llapi: add user space method for lov_user_md Peng Tao
` (29 subsequent siblings)
48 siblings, 0 replies; 62+ messages in thread
From: Peng Tao @ 2013-07-22 16:06 UTC (permalink / raw)
To: Greg Kroah-Hartman; +Cc: linux-kernel, jcl, Peng Tao, Andreas Dilger
From: jcl <jacques-charles.lafoucriere@cea.fr>
add 2 macros for the length of a fid string 0xSEQ:0xOID:0xVER
and it's brace version (FID_NOBRACE_LEN, and FID_LEN)
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-2782
Lustre-change: http://review.whamcloud.com/5299
Signed-off-by: JC Lafoucriere <jacques-charles.lafoucriere@cea.fr>
Reviewed-by: Jinshan Xiong <jinshan.xiong@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>
Signed-off-by: Peng Tao <tao.peng@emc.com>
Signed-off-by: Andreas Dilger <andreas.dilger@intel.com>
---
.../lustre/lustre/include/lustre/lustre_user.h | 2 ++
1 file changed, 2 insertions(+)
diff --git a/drivers/staging/lustre/lustre/include/lustre/lustre_user.h b/drivers/staging/lustre/lustre/include/lustre/lustre_user.h
index 7e9f575..f72b2e4 100644
--- a/drivers/staging/lustre/lustre/include/lustre/lustre_user.h
+++ b/drivers/staging/lustre/lustre/include/lustre/lustre_user.h
@@ -462,6 +462,8 @@ static inline void obd_uuid2fsname(char *buf, char *uuid, int buflen)
/* printf display format
e.g. printf("file FID is "DFID"\n", PFID(fid)); */
+#define FID_NOBRACE_LEN 40
+#define FID_LEN (FID_NOBRACE_LEN + 2)
#define DFID_NOBRACE LPX64":0x%x:0x%x"
#define DFID "["DFID_NOBRACE"]"
#define PFID(fid) \
--
1.7.9.5
^ permalink raw reply related [flat|nested] 62+ messages in thread* [PATCH 20/48] staging/lustre/llapi: add user space method for lov_user_md
2013-07-22 16:06 [PATCH 00/48] staging/lustre: minor cleanup and Intel Lustre sync Peng Tao
` (18 preceding siblings ...)
2013-07-22 16:06 ` [PATCH 19/48] staging/lustre/mdt: add macros for fid string len Peng Tao
@ 2013-07-22 16:06 ` Peng Tao
2013-07-22 16:06 ` [PATCH 21/48] staging/lustre/obdclass: use a dummy structure for lu_ref_link Peng Tao
` (28 subsequent siblings)
48 siblings, 0 replies; 62+ messages in thread
From: Peng Tao @ 2013-07-22 16:06 UTC (permalink / raw)
To: Greg Kroah-Hartman; +Cc: linux-kernel, jcl, Peng Tao, Andreas Dilger
From: jcl <jacques-charles.lafoucriere@cea.fr>
move lov_mds_md_size from obd_lov.h to lustre_idl.h
to have it close to lov_mds_md definition.
add lov_user_md_size() to compute lum size so
llapi and user space utils do not use kernel internal
definitions/methods
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-3345
Lustre-change: http://review.whamcloud.com/6345
Signed-off-by: JC Lafoucriere <jacques-charles.lafoucriere@cea.fr>
Reviewed-by: Jinshan Xiong <jinshan.xiong@intel.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Signed-off-by: Peng Tao <tao.peng@emc.com>
Signed-off-by: Andreas Dilger <andreas.dilger@intel.com>
---
.../lustre/lustre/include/lustre/lustre_idl.h | 11 +++++++++++
.../lustre/lustre/include/lustre/lustre_user.h | 10 ++++++++++
drivers/staging/lustre/lustre/include/obd_lov.h | 10 ----------
3 files changed, 21 insertions(+), 10 deletions(-)
diff --git a/drivers/staging/lustre/lustre/include/lustre/lustre_idl.h b/drivers/staging/lustre/lustre/include/lustre/lustre_idl.h
index 30c0331..ace4e18 100644
--- a/drivers/staging/lustre/lustre/include/lustre/lustre_idl.h
+++ b/drivers/staging/lustre/lustre/include/lustre/lustre_idl.h
@@ -1670,6 +1670,17 @@ struct lov_mds_md_v3 { /* LOV EA mds/wire data (little-endian) */
struct lov_ost_data_v1 lmm_objects[0]; /* per-stripe data */
};
+static inline __u32 lov_mds_md_size(__u16 stripes, __u32 lmm_magic)
+{
+ if (lmm_magic == LOV_MAGIC_V3)
+ return sizeof(struct lov_mds_md_v3) +
+ stripes * sizeof(struct lov_ost_data_v1);
+ else
+ return sizeof(struct lov_mds_md_v1) +
+ stripes * sizeof(struct lov_ost_data_v1);
+}
+
+
#define OBD_MD_FLID (0x00000001ULL) /* object ID */
#define OBD_MD_FLATIME (0x00000002ULL) /* access time */
#define OBD_MD_FLMTIME (0x00000004ULL) /* data modification time */
diff --git a/drivers/staging/lustre/lustre/include/lustre/lustre_user.h b/drivers/staging/lustre/lustre/include/lustre/lustre_user.h
index f72b2e4..c7bd447 100644
--- a/drivers/staging/lustre/lustre/include/lustre/lustre_user.h
+++ b/drivers/staging/lustre/lustre/include/lustre/lustre_user.h
@@ -347,6 +347,16 @@ struct lov_user_md_v3 { /* LOV EA user data (host-endian) */
struct lov_user_ost_data_v1 lmm_objects[0]; /* per-stripe data */
} __attribute__((packed));
+static inline __u32 lov_user_md_size(__u16 stripes, __u32 lmm_magic)
+{
+ if (lmm_magic == LOV_USER_MAGIC_V3)
+ return sizeof(struct lov_user_md_v3) +
+ stripes * sizeof(struct lov_user_ost_data_v1);
+ else
+ return sizeof(struct lov_user_md_v1) +
+ stripes * sizeof(struct lov_user_ost_data_v1);
+}
+
/* Compile with -D_LARGEFILE64_SOURCE or -D_GNU_SOURCE (or #define) to
* use this. It is unsafe to #define those values in this header as it
* is possible the application has already #included <sys/stat.h>. */
diff --git a/drivers/staging/lustre/lustre/include/obd_lov.h b/drivers/staging/lustre/lustre/include/obd_lov.h
index d82f334..235718b 100644
--- a/drivers/staging/lustre/lustre/include/obd_lov.h
+++ b/drivers/staging/lustre/lustre/include/obd_lov.h
@@ -44,16 +44,6 @@ static inline int lov_stripe_md_size(__u16 stripes)
return sizeof(struct lov_stripe_md) + stripes*sizeof(struct lov_oinfo*);
}
-static inline __u32 lov_mds_md_size(__u16 stripes, __u32 lmm_magic)
-{
- if (lmm_magic == LOV_MAGIC_V3)
- return sizeof(struct lov_mds_md_v3) +
- stripes * sizeof(struct lov_ost_data_v1);
- else
- return sizeof(struct lov_mds_md_v1) +
- stripes * sizeof(struct lov_ost_data_v1);
-}
-
struct lov_version_size {
__u32 lvs_magic;
size_t lvs_lmm_size;
--
1.7.9.5
^ permalink raw reply related [flat|nested] 62+ messages in thread* [PATCH 21/48] staging/lustre/obdclass: use a dummy structure for lu_ref_link
2013-07-22 16:06 [PATCH 00/48] staging/lustre: minor cleanup and Intel Lustre sync Peng Tao
` (19 preceding siblings ...)
2013-07-22 16:06 ` [PATCH 20/48] staging/lustre/llapi: add user space method for lov_user_md Peng Tao
@ 2013-07-22 16:06 ` Peng Tao
2013-07-23 20:35 ` Greg Kroah-Hartman
2013-07-22 16:06 ` [PATCH 22/48] staging/lustre: fix 'code maintainability' errors Peng Tao
` (27 subsequent siblings)
48 siblings, 1 reply; 62+ messages in thread
From: Peng Tao @ 2013-07-22 16:06 UTC (permalink / raw)
To: Greg Kroah-Hartman
Cc: linux-kernel, John L. Hammond, Peng Tao, Andreas Dilger
From: "John L. Hammond" <john.hammond@intel.com>
Move the definition of struct lu_ref_link to lu_ref.h. If USE_LU_REF
is not defined then define it to be the empty struct. Change the
return type of lu_ref_add() and lu_ref_add_atomic() to void. Add
lu_ref_add_at() taking same arguments as lu_ref_add() togerther with a
pointer to a struct lu_ref_link and returning void. Adjust all
structures containing a lu_ref_link pointer to contain a struct
lu_ref_link instead. Use lu_ref_add_at() and lu_ref_del_at() to
handle embedded lu_ref_links.
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-3059
Lustre-change: http://review.whamcloud.com/5920
Signed-off-by: John L. Hammond <john.hammond@intel.com>
Reviewed-by: Alex Zhuravlev <alexey.zhuravlev@intel.com>
Reviewed-by: Jinshan Xiong <jinshan.xiong@intel.com>
Reviewed-by: Mike Pershin <mike.pershin@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>
Signed-off-by: Peng Tao <tao.peng@emc.com>
Signed-off-by: Andreas Dilger <andreas.dilger@intel.com>
---
drivers/staging/lustre/lustre/include/cl_object.h | 10 +--
drivers/staging/lustre/lustre/include/lu_object.h | 18 +++--
drivers/staging/lustre/lustre/include/lu_ref.h | 77 ++++++++++++--------
drivers/staging/lustre/lustre/obdclass/cl_io.c | 17 +++--
drivers/staging/lustre/lustre/obdclass/cl_lock.c | 6 +-
drivers/staging/lustre/lustre/obdclass/cl_page.c | 5 +-
drivers/staging/lustre/lustre/obdclass/lu_object.c | 15 ++--
drivers/staging/lustre/lustre/obdclass/lu_ref.c | 74 +++++++++----------
drivers/staging/lustre/lustre/osc/osc_cache.c | 8 +-
9 files changed, 130 insertions(+), 100 deletions(-)
diff --git a/drivers/staging/lustre/lustre/include/cl_object.h b/drivers/staging/lustre/lustre/include/cl_object.h
index 4bb6880..4a47850 100644
--- a/drivers/staging/lustre/lustre/include/cl_object.h
+++ b/drivers/staging/lustre/lustre/include/cl_object.h
@@ -768,11 +768,11 @@ struct cl_page {
/** List of references to this page, for debugging. */
struct lu_ref cp_reference;
/** Link to an object, for debugging. */
- struct lu_ref_link *cp_obj_ref;
+ struct lu_ref_link cp_obj_ref;
/** Link to a queue, for debugging. */
- struct lu_ref_link *cp_queue_ref;
+ struct lu_ref_link cp_queue_ref;
/** Per-page flags from enum cl_page_flags. Protected by a VM lock. */
- unsigned cp_flags;
+ unsigned cp_flags;
/** Assigned if doing a sync_io */
struct cl_sync_io *cp_sync_io;
};
@@ -1625,7 +1625,7 @@ struct cl_lock {
/**
* A reference for cl_lock::cll_descr::cld_obj. For debugging.
*/
- struct lu_ref_link *cll_obj_ref;
+ struct lu_ref_link cll_obj_ref;
#ifdef CONFIG_LOCKDEP
/* "dep_map" name is assumed by lockdep.h macros. */
struct lockdep_map dep_map;
@@ -2517,7 +2517,7 @@ struct cl_req_obj {
/** object itself */
struct cl_object *ro_obj;
/** reference to cl_req_obj::ro_obj. For debugging. */
- struct lu_ref_link *ro_obj_ref;
+ struct lu_ref_link ro_obj_ref;
/* something else? Number of pages for a given object? */
};
diff --git a/drivers/staging/lustre/lustre/include/lu_object.h b/drivers/staging/lustre/lustre/include/lu_object.h
index d40ad81..f0d6fb1 100644
--- a/drivers/staging/lustre/lustre/include/lu_object.h
+++ b/drivers/staging/lustre/lustre/include/lu_object.h
@@ -496,7 +496,7 @@ struct lu_object {
/**
* Link to the device, for debugging.
*/
- struct lu_ref_link *lo_dev_ref;
+ struct lu_ref_link lo_dev_ref;
};
enum lu_object_header_flags {
@@ -868,11 +868,19 @@ static inline __u32 lu_object_attr(const struct lu_object *o)
return o->lo_header->loh_attr;
}
-static inline struct lu_ref_link *lu_object_ref_add(struct lu_object *o,
- const char *scope,
- const void *source)
+static inline void lu_object_ref_add(struct lu_object *o,
+ const char *scope,
+ const void *source)
{
- return lu_ref_add(&o->lo_header->loh_reference, scope, source);
+ lu_ref_add(&o->lo_header->loh_reference, scope, source);
+}
+
+static inline void lu_object_ref_add_at(struct lu_object *o,
+ struct lu_ref_link *link,
+ const char *scope,
+ const void *source)
+{
+ lu_ref_add_at(&o->lo_header->loh_reference, link, scope, source);
}
static inline void lu_object_ref_del(struct lu_object *o,
diff --git a/drivers/staging/lustre/lustre/include/lu_ref.h b/drivers/staging/lustre/lustre/include/lu_ref.h
index adfbf51..3b0b483 100644
--- a/drivers/staging/lustre/lustre/include/lu_ref.h
+++ b/drivers/staging/lustre/lustre/include/lu_ref.h
@@ -109,9 +109,6 @@
#ifdef CONFIG_LUSTRE_DEBUG_LU_REF_CHECK
-/* An incomplete type (defined locally in lu_ref.c) */
-struct lu_ref_link;
-
/**
* Data-structure to keep track of references to a given object. This is used
* for debugging.
@@ -153,31 +150,48 @@ struct lu_ref {
struct list_head lf_linkage;
};
-int lu_ref_global_init(void);
-void lu_ref_global_fini(void);
+struct lu_ref_link {
+ struct lu_ref *ll_ref;
+ struct list_head ll_linkage;
+ const char *ll_scope;
+ const void *ll_source;
+};
+
void lu_ref_init_loc(struct lu_ref *ref, const char *func, const int line);
void lu_ref_fini (struct lu_ref *ref);
#define lu_ref_init(ref) lu_ref_init_loc(ref, __FUNCTION__, __LINE__)
-struct lu_ref_link *lu_ref_add (struct lu_ref *ref, const char *scope,
- const void *source);
-struct lu_ref_link *lu_ref_add_atomic(struct lu_ref *ref, const char *scope,
- const void *source);
-void lu_ref_del (struct lu_ref *ref, const char *scope,
- const void *source);
-void lu_ref_set_at (struct lu_ref *ref,
- struct lu_ref_link *link,
- const char *scope, const void *source0,
- const void *source1);
-void lu_ref_del_at (struct lu_ref *ref,
- struct lu_ref_link *link,
- const char *scope, const void *source);
-void lu_ref_print (const struct lu_ref *ref);
-void lu_ref_print_all (void);
+void lu_ref_add(struct lu_ref *ref, const char *scope, const void *source);
+
+void lu_ref_add_atomic(struct lu_ref *ref, const char *scope,
+ const void *source);
+
+void lu_ref_add_at(struct lu_ref *ref, struct lu_ref_link *link,
+ const char *scope, const void *source);
+
+void lu_ref_del(struct lu_ref *ref, const char *scope, const void *source);
+
+void lu_ref_set_at(struct lu_ref *ref, struct lu_ref_link *link,
+ const char *scope, const void *source0, const void *source1);
+
+void lu_ref_del_at(struct lu_ref *ref, struct lu_ref_link *link,
+ const char *scope, const void *source);
+
+void lu_ref_print(const struct lu_ref *ref);
+
+void lu_ref_print_all(void);
+
+int lu_ref_global_init(void);
+
+void lu_ref_global_fini(void);
#else /* !CONFIG_LUSTRE_DEBUG_LU_REF_CHECK */
-struct lu_ref {};
+struct lu_ref {
+};
+
+struct lu_ref_link {
+};
static inline void lu_ref_init(struct lu_ref *ref)
{
@@ -187,18 +201,23 @@ static inline void lu_ref_fini(struct lu_ref *ref)
{
}
-static inline struct lu_ref_link *lu_ref_add(struct lu_ref *ref,
- const char *scope,
- const void *source)
+static inline void lu_ref_add(struct lu_ref *ref,
+ const char *scope,
+ const void *source)
+{
+}
+
+static inline void lu_ref_add_atomic(struct lu_ref *ref,
+ const char *scope,
+ const void *source)
{
- return NULL;
}
-static inline struct lu_ref_link *lu_ref_add_atomic(struct lu_ref *ref,
- const char *scope,
- const void *source)
+static inline void lu_ref_add_at(struct lu_ref *ref,
+ struct lu_ref_link *link,
+ const char *scope,
+ const void *source)
{
- return NULL;
}
static inline void lu_ref_del(struct lu_ref *ref, const char *scope,
diff --git a/drivers/staging/lustre/lustre/obdclass/cl_io.c b/drivers/staging/lustre/lustre/obdclass/cl_io.c
index 75c9be8..f3f0315 100644
--- a/drivers/staging/lustre/lustre/obdclass/cl_io.c
+++ b/drivers/staging/lustre/lustre/obdclass/cl_io.c
@@ -1105,7 +1105,7 @@ void cl_page_list_add(struct cl_page_list *plist, struct cl_page *page)
LASSERT(list_empty(&page->cp_batch));
list_add_tail(&page->cp_batch, &plist->pl_pages);
++plist->pl_nr;
- page->cp_queue_ref = lu_ref_add(&page->cp_reference, "queue", plist);
+ lu_ref_add_at(&page->cp_reference, &page->cp_queue_ref, "queue", plist);
cl_page_get(page);
EXIT;
}
@@ -1126,7 +1126,7 @@ void cl_page_list_del(const struct lu_env *env,
mutex_unlock(&page->cp_mutex);
lockdep_on();
--plist->pl_nr;
- lu_ref_del_at(&page->cp_reference, page->cp_queue_ref, "queue", plist);
+ lu_ref_del_at(&page->cp_reference, &page->cp_queue_ref, "queue", plist);
cl_page_put(env, page);
EXIT;
}
@@ -1146,8 +1146,8 @@ void cl_page_list_move(struct cl_page_list *dst, struct cl_page_list *src,
list_move_tail(&page->cp_batch, &dst->pl_pages);
--src->pl_nr;
++dst->pl_nr;
- lu_ref_set_at(&page->cp_reference,
- page->cp_queue_ref, "queue", src, dst);
+ lu_ref_set_at(&page->cp_reference, &page->cp_queue_ref, "queue",
+ src, dst);
EXIT;
}
EXPORT_SYMBOL(cl_page_list_move);
@@ -1202,7 +1202,8 @@ void cl_page_list_disown(const struct lu_env *env,
* XXX cl_page_disown0() will fail if page is not locked.
*/
cl_page_disown0(env, io, page);
- lu_ref_del(&page->cp_reference, "queue", plist);
+ lu_ref_del_at(&page->cp_reference, &page->cp_queue_ref, "queue",
+ plist);
cl_page_put(env, page);
}
EXIT;
@@ -1449,7 +1450,7 @@ static void cl_req_free(const struct lu_env *env, struct cl_req *req)
struct cl_object *obj = req->crq_o[i].ro_obj;
if (obj != NULL) {
lu_object_ref_del_at(&obj->co_lu,
- req->crq_o[i].ro_obj_ref,
+ &req->crq_o[i].ro_obj_ref,
"cl_req", req);
cl_object_put(env, obj);
}
@@ -1570,8 +1571,8 @@ void cl_req_page_add(const struct lu_env *env,
if (rqo->ro_obj == NULL) {
rqo->ro_obj = obj;
cl_object_get(obj);
- rqo->ro_obj_ref = lu_object_ref_add(&obj->co_lu,
- "cl_req", req);
+ lu_object_ref_add_at(&obj->co_lu, &rqo->ro_obj_ref,
+ "cl_req", req);
break;
}
}
diff --git a/drivers/staging/lustre/lustre/obdclass/cl_lock.c b/drivers/staging/lustre/lustre/obdclass/cl_lock.c
index d34e044..958c1e5 100644
--- a/drivers/staging/lustre/lustre/obdclass/cl_lock.c
+++ b/drivers/staging/lustre/lustre/obdclass/cl_lock.c
@@ -267,7 +267,7 @@ static void cl_lock_free(const struct lu_env *env, struct cl_lock *lock)
}
CS_LOCK_DEC(obj, total);
CS_LOCKSTATE_DEC(obj, lock->cll_state);
- lu_object_ref_del_at(&obj->co_lu, lock->cll_obj_ref, "cl_lock", lock);
+ lu_object_ref_del_at(&obj->co_lu, &lock->cll_obj_ref, "cl_lock", lock);
cl_object_put(env, obj);
lu_ref_fini(&lock->cll_reference);
lu_ref_fini(&lock->cll_holders);
@@ -373,8 +373,8 @@ static struct cl_lock *cl_lock_alloc(const struct lu_env *env,
lock->cll_descr = *descr;
lock->cll_state = CLS_NEW;
cl_object_get(obj);
- lock->cll_obj_ref = lu_object_ref_add(&obj->co_lu,
- "cl_lock", lock);
+ lu_object_ref_add_at(&obj->co_lu, &lock->cll_obj_ref, "cl_lock",
+ lock);
INIT_LIST_HEAD(&lock->cll_layers);
INIT_LIST_HEAD(&lock->cll_linkage);
INIT_LIST_HEAD(&lock->cll_inclosure);
diff --git a/drivers/staging/lustre/lustre/obdclass/cl_page.c b/drivers/staging/lustre/lustre/obdclass/cl_page.c
index bb93359..d9ca5d5 100644
--- a/drivers/staging/lustre/lustre/obdclass/cl_page.c
+++ b/drivers/staging/lustre/lustre/obdclass/cl_page.c
@@ -270,7 +270,7 @@ static void cl_page_free(const struct lu_env *env, struct cl_page *page)
}
CS_PAGE_DEC(obj, total);
CS_PAGESTATE_DEC(obj, page->cp_state);
- lu_object_ref_del_at(&obj->co_lu, page->cp_obj_ref, "cl_page", page);
+ lu_object_ref_del_at(&obj->co_lu, &page->cp_obj_ref, "cl_page", page);
cl_object_put(env, obj);
lu_ref_fini(&page->cp_reference);
OBD_FREE(page, pagesize);
@@ -305,7 +305,8 @@ static struct cl_page *cl_page_alloc(const struct lu_env *env,
atomic_inc(&page->cp_ref);
page->cp_obj = o;
cl_object_get(o);
- page->cp_obj_ref = lu_object_ref_add(&o->co_lu, "cl_page",page);
+ lu_object_ref_add_at(&o->co_lu, &page->cp_obj_ref, "cl_page",
+ page);
page->cp_index = ind;
cl_page_state_set_trust(page, CPS_CACHED);
page->cp_type = type;
diff --git a/drivers/staging/lustre/lustre/obdclass/lu_object.c b/drivers/staging/lustre/lustre/obdclass/lu_object.c
index 5559732..2fb65f6 100644
--- a/drivers/staging/lustre/lustre/obdclass/lu_object.c
+++ b/drivers/staging/lustre/lustre/obdclass/lu_object.c
@@ -1147,15 +1147,16 @@ EXPORT_SYMBOL(lu_device_fini);
* Initialize object \a o that is part of compound object \a h and was created
* by device \a d.
*/
-int lu_object_init(struct lu_object *o,
- struct lu_object_header *h, struct lu_device *d)
+int lu_object_init(struct lu_object *o, struct lu_object_header *h,
+ struct lu_device *d)
{
- memset(o, 0, sizeof *o);
+ memset(o, 0, sizeof(*o));
o->lo_header = h;
- o->lo_dev = d;
+ o->lo_dev = d;
lu_device_get(d);
- o->lo_dev_ref = lu_ref_add(&d->ld_reference, "lu_object", o);
+ lu_ref_add_at(&d->ld_reference, &o->lo_dev_ref, "lu_object", o);
INIT_LIST_HEAD(&o->lo_linkage);
+
return 0;
}
EXPORT_SYMBOL(lu_object_init);
@@ -1170,8 +1171,8 @@ void lu_object_fini(struct lu_object *o)
LASSERT(list_empty(&o->lo_linkage));
if (dev != NULL) {
- lu_ref_del_at(&dev->ld_reference,
- o->lo_dev_ref , "lu_object", o);
+ lu_ref_del_at(&dev->ld_reference, &o->lo_dev_ref,
+ "lu_object", o);
lu_device_put(dev);
o->lo_dev = NULL;
}
diff --git a/drivers/staging/lustre/lustre/obdclass/lu_ref.c b/drivers/staging/lustre/lustre/obdclass/lu_ref.c
index 79d25c5..b5f1bda 100644
--- a/drivers/staging/lustre/lustre/obdclass/lu_ref.c
+++ b/drivers/staging/lustre/lustre/obdclass/lu_ref.c
@@ -43,7 +43,7 @@
#define DEBUG_SUBSYSTEM S_CLASS
#include <linux/gfp.h>
-#include <linux/libcfs.h>
+#include <linux/libcfs/libcfs.h>
#include <obd.h>
#include <obd_class.h>
#include <obd_support.h>
@@ -65,13 +65,6 @@
} \
} while (0)
-struct lu_ref_link {
- struct lu_ref *ll_ref;
- struct list_head ll_linkage;
- const char *ll_scope;
- const void *ll_source;
-};
-
static struct kmem_cache *lu_ref_link_kmem;
static struct lu_kmem_descr lu_ref_caches[] = {
@@ -93,9 +86,9 @@ static struct lu_kmem_descr lu_ref_caches[] = {
static LIST_HEAD(lu_ref_refs);
static spinlock_t lu_ref_refs_guard;
static struct lu_ref lu_ref_marker = {
- .lf_guard = __SPIN_LOCK_UNLOCKED(lu_ref_marker.lf_guard),
- .lf_list = LIST_HEAD_INIT(lu_ref_marker.lf_list),
- .lf_linkage = LIST_HEAD_INIT(lu_ref_marker.lf_linkage)
+ .lf_guard = __SPIN_LOCK_UNLOCKED(lu_ref_marker.lf_guard),
+ .lf_list = LIST_HEAD_INIT(lu_ref_marker.lf_list),
+ .lf_linkage = LIST_HEAD_INIT(lu_ref_marker.lf_linkage)
};
void lu_ref_print(const struct lu_ref *ref)
@@ -186,21 +179,33 @@ static struct lu_ref_link *lu_ref_add_context(struct lu_ref *ref,
return link;
}
-struct lu_ref_link *lu_ref_add(struct lu_ref *ref, const char *scope,
- const void *source)
+void lu_ref_add(struct lu_ref *ref, const char *scope, const void *source)
{
might_sleep();
- return lu_ref_add_context(ref, GFP_IOFS, scope, source);
+ lu_ref_add_context(ref, GFP_IOFS, scope, source);
}
EXPORT_SYMBOL(lu_ref_add);
+void lu_ref_add_at(struct lu_ref *ref, struct lu_ref_link *link,
+ const char *scope, const void *source)
+{
+ link->ll_ref = ref;
+ link->ll_scope = scope;
+ link->ll_source = source;
+ spin_lock(&ref->lf_guard);
+ list_add_tail(&link->ll_linkage, &ref->lf_list);
+ ref->lf_refs++;
+ spin_unlock(&ref->lf_guard);
+}
+EXPORT_SYMBOL(lu_ref_add_at);
+
/**
* Version of lu_ref_add() to be used in non-blockable contexts.
*/
-struct lu_ref_link *lu_ref_add_atomic(struct lu_ref *ref, const char *scope,
- const void *source)
+void lu_ref_add_atomic(struct lu_ref *ref, const char *scope,
+ const void *source)
{
- return lu_ref_add_context(ref, GFP_ATOMIC, scope, source);
+ lu_ref_add_context(ref, GFP_ATOMIC, scope, source);
}
EXPORT_SYMBOL(lu_ref_add_atomic);
@@ -262,14 +267,12 @@ void lu_ref_set_at(struct lu_ref *ref, struct lu_ref_link *link,
const char *scope,
const void *source0, const void *source1)
{
+ REFASSERT(ref, link != NULL && !IS_ERR(link));
+
spin_lock(&ref->lf_guard);
- if (link != ERR_PTR(-ENOMEM)) {
- REFASSERT(ref, link->ll_ref == ref);
- REFASSERT(ref, lu_ref_link_eq(link, scope, source0));
- link->ll_source = source1;
- } else {
- REFASSERT(ref, ref->lf_failed > 0);
- }
+ REFASSERT(ref, link->ll_ref == ref);
+ REFASSERT(ref, lu_ref_link_eq(link, scope, source0));
+ link->ll_source = source1;
spin_unlock(&ref->lf_guard);
}
EXPORT_SYMBOL(lu_ref_set_at);
@@ -277,20 +280,13 @@ EXPORT_SYMBOL(lu_ref_set_at);
void lu_ref_del_at(struct lu_ref *ref, struct lu_ref_link *link,
const char *scope, const void *source)
{
- if (link != ERR_PTR(-ENOMEM)) {
- spin_lock(&ref->lf_guard);
- REFASSERT(ref, link->ll_ref == ref);
- REFASSERT(ref, lu_ref_link_eq(link, scope, source));
- list_del(&link->ll_linkage);
- ref->lf_refs--;
- spin_unlock(&ref->lf_guard);
- OBD_SLAB_FREE(link, lu_ref_link_kmem, sizeof(*link));
- } else {
- spin_lock(&ref->lf_guard);
- REFASSERT(ref, ref->lf_failed > 0);
- ref->lf_failed--;
- spin_unlock(&ref->lf_guard);
- }
+ REFASSERT(ref, link != NULL && !IS_ERR(link));
+ spin_lock(&ref->lf_guard);
+ REFASSERT(ref, link->ll_ref == ref);
+ REFASSERT(ref, lu_ref_link_eq(link, scope, source));
+ list_del(&link->ll_linkage);
+ ref->lf_refs--;
+ spin_unlock(&ref->lf_guard);
}
EXPORT_SYMBOL(lu_ref_del_at);
@@ -418,6 +414,8 @@ static struct file_operations lu_ref_dump_fops = {
.release = lu_ref_seq_release
};
+#endif
+
int lu_ref_global_init(void)
{
int result;
diff --git a/drivers/staging/lustre/lustre/osc/osc_cache.c b/drivers/staging/lustre/lustre/osc/osc_cache.c
index 0a0ec6f..6b0a842 100644
--- a/drivers/staging/lustre/lustre/osc/osc_cache.c
+++ b/drivers/staging/lustre/lustre/osc/osc_cache.c
@@ -2064,7 +2064,7 @@ static void osc_check_rpcs(const struct lu_env *env, struct client_obd *cli,
while ((osc = osc_next_obj(cli)) != NULL) {
struct cl_object *obj = osc2cl(osc);
- struct lu_ref_link *link;
+ struct lu_ref_link link;
OSC_IO_DEBUG(osc, "%lu in flight\n", rpcs_in_flight(cli));
@@ -2075,7 +2075,8 @@ static void osc_check_rpcs(const struct lu_env *env, struct client_obd *cli,
cl_object_get(obj);
client_obd_list_unlock(&cli->cl_loi_list_lock);
- link = lu_object_ref_add(&obj->co_lu, "check", current);
+ lu_object_ref_add_at(&obj->co_lu, &link, "check",
+ current);
/* attempt some read/write balancing by alternating between
* reads and writes in an object. The makes_rpc checks here
@@ -2116,7 +2117,8 @@ static void osc_check_rpcs(const struct lu_env *env, struct client_obd *cli,
osc_object_unlock(osc);
osc_list_maint(cli, osc);
- lu_object_ref_del_at(&obj->co_lu, link, "check", current);
+ lu_object_ref_del_at(&obj->co_lu, &link, "check",
+ current);
cl_object_put(env, obj);
client_obd_list_lock(&cli->cl_loi_list_lock);
--
1.7.9.5
^ permalink raw reply related [flat|nested] 62+ messages in thread* Re: [PATCH 21/48] staging/lustre/obdclass: use a dummy structure for lu_ref_link
2013-07-22 16:06 ` [PATCH 21/48] staging/lustre/obdclass: use a dummy structure for lu_ref_link Peng Tao
@ 2013-07-23 20:35 ` Greg Kroah-Hartman
2013-07-24 6:10 ` Peng Tao
0 siblings, 1 reply; 62+ messages in thread
From: Greg Kroah-Hartman @ 2013-07-23 20:35 UTC (permalink / raw)
To: Peng Tao; +Cc: linux-kernel, John L. Hammond, Peng Tao, Andreas Dilger
On Tue, Jul 23, 2013 at 12:06:42AM +0800, Peng Tao wrote:
> From: "John L. Hammond" <john.hammond@intel.com>
>
> Move the definition of struct lu_ref_link to lu_ref.h. If USE_LU_REF
> is not defined then define it to be the empty struct. Change the
> return type of lu_ref_add() and lu_ref_add_atomic() to void. Add
> lu_ref_add_at() taking same arguments as lu_ref_add() togerther with a
> pointer to a struct lu_ref_link and returning void. Adjust all
> structures containing a lu_ref_link pointer to contain a struct
> lu_ref_link instead. Use lu_ref_add_at() and lu_ref_del_at() to
> handle embedded lu_ref_links.
>
> Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-3059
> Lustre-change: http://review.whamcloud.com/5920
> Signed-off-by: John L. Hammond <john.hammond@intel.com>
> Reviewed-by: Alex Zhuravlev <alexey.zhuravlev@intel.com>
> Reviewed-by: Jinshan Xiong <jinshan.xiong@intel.com>
> Reviewed-by: Mike Pershin <mike.pershin@intel.com>
> Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>
> Signed-off-by: Peng Tao <tao.peng@emc.com>
> Signed-off-by: Andreas Dilger <andreas.dilger@intel.com>
This patch fails to apply :(
^ permalink raw reply [flat|nested] 62+ messages in thread
* Re: [PATCH 21/48] staging/lustre/obdclass: use a dummy structure for lu_ref_link
2013-07-23 20:35 ` Greg Kroah-Hartman
@ 2013-07-24 6:10 ` Peng Tao
0 siblings, 0 replies; 62+ messages in thread
From: Peng Tao @ 2013-07-24 6:10 UTC (permalink / raw)
To: Greg Kroah-Hartman
Cc: Linux Kernel Mailing List, John L. Hammond, Andreas Dilger
On Wed, Jul 24, 2013 at 4:35 AM, Greg Kroah-Hartman
<gregkh@linuxfoundation.org> wrote:
> On Tue, Jul 23, 2013 at 12:06:42AM +0800, Peng Tao wrote:
>> From: "John L. Hammond" <john.hammond@intel.com>
>>
>> Move the definition of struct lu_ref_link to lu_ref.h. If USE_LU_REF
>> is not defined then define it to be the empty struct. Change the
>> return type of lu_ref_add() and lu_ref_add_atomic() to void. Add
>> lu_ref_add_at() taking same arguments as lu_ref_add() togerther with a
>> pointer to a struct lu_ref_link and returning void. Adjust all
>> structures containing a lu_ref_link pointer to contain a struct
>> lu_ref_link instead. Use lu_ref_add_at() and lu_ref_del_at() to
>> handle embedded lu_ref_links.
>>
>> Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-3059
>> Lustre-change: http://review.whamcloud.com/5920
>> Signed-off-by: John L. Hammond <john.hammond@intel.com>
>> Reviewed-by: Alex Zhuravlev <alexey.zhuravlev@intel.com>
>> Reviewed-by: Jinshan Xiong <jinshan.xiong@intel.com>
>> Reviewed-by: Mike Pershin <mike.pershin@intel.com>
>> Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>
>> Signed-off-by: Peng Tao <tao.peng@emc.com>
>> Signed-off-by: Andreas Dilger <andreas.dilger@intel.com>
>
> This patch fails to apply :(
>
It relies on one previous patch ([PATCH 02/48]
staging/lustre/obdclass: restore lu_ref), which you have suggested to
drop. I will remove unnecessary part of the patch and resend (if
there's still anything left).
Thanks,
Tao
^ permalink raw reply [flat|nested] 62+ messages in thread
* [PATCH 22/48] staging/lustre: fix 'code maintainability' errors
2013-07-22 16:06 [PATCH 00/48] staging/lustre: minor cleanup and Intel Lustre sync Peng Tao
` (20 preceding siblings ...)
2013-07-22 16:06 ` [PATCH 21/48] staging/lustre/obdclass: use a dummy structure for lu_ref_link Peng Tao
@ 2013-07-22 16:06 ` Peng Tao
2013-07-22 16:06 ` [PATCH 23/48] staging/lustre/fid: prepare FID module for client server split Peng Tao
` (26 subsequent siblings)
48 siblings, 0 replies; 62+ messages in thread
From: Peng Tao @ 2013-07-22 16:06 UTC (permalink / raw)
To: Greg Kroah-Hartman
Cc: linux-kernel, Sebastien Buisson, Peng Tao, Andreas Dilger
From: Sebastien Buisson <sebastien.buisson@bull.net>
Fix 'code maintainability' issues found by Coverity version 6.5.1:
Unused pointer value (UNUSED_VALUE)
Pointer returned by function is never used.
Missing varargs init or cleanup (VARARGS)
va_end was not called for variable.
Signed-off-by: Sebastien Buisson <sebastien.buisson@bull.net>
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-3107
Lustre-change: http://review.whamcloud.com/5944
Reviewed-by: Dmitry Eremin <dmitry.eremin@intel.com>
Reviewed-by: John Hammond <johnlockwoodhammond@gmail.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Signed-off-by: Peng Tao <tao.peng@emc.com>
Signed-off-by: Andreas Dilger <andreas.dilger@intel.com>
---
.../lustre/lustre/obdclass/lprocfs_jobstats.c | 2 --
.../staging/lustre/lustre/ptlrpc/pack_generic.c | 1 +
2 files changed, 1 insertion(+), 2 deletions(-)
diff --git a/drivers/staging/lustre/lustre/obdclass/lprocfs_jobstats.c b/drivers/staging/lustre/lustre/obdclass/lprocfs_jobstats.c
index e2d57fe..7f6072c 100644
--- a/drivers/staging/lustre/lustre/obdclass/lprocfs_jobstats.c
+++ b/drivers/staging/lustre/lustre/obdclass/lprocfs_jobstats.c
@@ -366,7 +366,6 @@ static int lprocfs_jobstats_seq_show(struct seq_file *p, void *v)
struct job_stat *job = v;
struct lprocfs_stats *s;
struct lprocfs_counter ret;
- struct lprocfs_counter *cntr;
struct lprocfs_counter_header *cntr_header;
int i;
@@ -380,7 +379,6 @@ static int lprocfs_jobstats_seq_show(struct seq_file *p, void *v)
s = job->js_stats;
for (i = 0; i < s->ls_num; i++) {
- cntr = lprocfs_stats_counter_get(s, 0, i);
cntr_header = &s->ls_cnt_header[i];
lprocfs_stats_collect(s, i, &ret);
diff --git a/drivers/staging/lustre/lustre/ptlrpc/pack_generic.c b/drivers/staging/lustre/lustre/ptlrpc/pack_generic.c
index be246d4..ffff648 100644
--- a/drivers/staging/lustre/lustre/ptlrpc/pack_generic.c
+++ b/drivers/staging/lustre/lustre/ptlrpc/pack_generic.c
@@ -2462,6 +2462,7 @@ void _debug_req(struct ptlrpc_request *req,
rep_ok ? lustre_msg_get_flags(req->rq_repmsg) : -1,
req->rq_status,
rep_ok ? lustre_msg_get_status(req->rq_repmsg) : -1);
+ va_end(args);
}
EXPORT_SYMBOL(_debug_req);
--
1.7.9.5
^ permalink raw reply related [flat|nested] 62+ messages in thread* [PATCH 23/48] staging/lustre/fid: prepare FID module for client server split
2013-07-22 16:06 [PATCH 00/48] staging/lustre: minor cleanup and Intel Lustre sync Peng Tao
` (21 preceding siblings ...)
2013-07-22 16:06 ` [PATCH 22/48] staging/lustre: fix 'code maintainability' errors Peng Tao
@ 2013-07-22 16:06 ` Peng Tao
2013-07-22 16:06 ` [PATCH 24/48] staging/lustre/llite: force lvb_data update after layout change Peng Tao
` (25 subsequent siblings)
48 siblings, 0 replies; 62+ messages in thread
From: Peng Tao @ 2013-07-22 16:06 UTC (permalink / raw)
To: Greg Kroah-Hartman
Cc: linux-kernel, Liu Xuezhao, John L. Hammond, Peng Tao, Andreas Dilger
From: Liu Xuezhao <xuezhao.liu@emc.com>
Split FID server from client, fid_{handler,store,lib}.c are not
compliled unless server support is enabled. Generally cleanup
includes in lustre/fid/ and reduce the need for client code to
directly or indirectly include {dt,md}_object.h.
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-1330
Lustre-change: http://review.whamcloud.com/2673
Signed-off-by: Liu Xuezhao <xuezhao.liu@emc.com>
Signed-off-by: John L. Hammond <john.hammond@intel.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Reviewed-by: Nathaniel Clark <nathaniel.l.clark@intel.com>
Signed-off-by: Peng Tao <tao.peng@emc.com>
Signed-off-by: Andreas Dilger <andreas.dilger@intel.com>
---
drivers/staging/lustre/lustre/fid/Makefile | 2 +-
drivers/staging/lustre/lustre/fid/fid_handler.c | 661 --------------------
drivers/staging/lustre/lustre/fid/fid_internal.h | 36 +-
drivers/staging/lustre/lustre/fid/fid_lib.c | 14 +-
drivers/staging/lustre/lustre/fid/fid_request.c | 133 ++--
drivers/staging/lustre/lustre/fid/fid_store.c | 259 --------
drivers/staging/lustre/lustre/fid/lproc_fid.c | 3 -
drivers/staging/lustre/lustre/include/lu_object.h | 5 +
drivers/staging/lustre/lustre/include/lustre_fid.h | 24 +-
drivers/staging/lustre/lustre/include/md_object.h | 5 -
drivers/staging/lustre/lustre/include/obd.h | 7 +-
11 files changed, 131 insertions(+), 1018 deletions(-)
delete mode 100644 drivers/staging/lustre/lustre/fid/fid_handler.c
delete mode 100644 drivers/staging/lustre/lustre/fid/fid_store.c
diff --git a/drivers/staging/lustre/lustre/fid/Makefile b/drivers/staging/lustre/lustre/fid/Makefile
index b8d6d21..ed21bea 100644
--- a/drivers/staging/lustre/lustre/fid/Makefile
+++ b/drivers/staging/lustre/lustre/fid/Makefile
@@ -1,5 +1,5 @@
obj-$(CONFIG_LUSTRE_FS) += fid.o
-fid-y := fid_handler.o fid_store.o fid_request.o lproc_fid.o fid_lib.o
+fid-y := fid_request.o lproc_fid.o fid_lib.o
ccflags-y := -I$(src)/../include
diff --git a/drivers/staging/lustre/lustre/fid/fid_handler.c b/drivers/staging/lustre/lustre/fid/fid_handler.c
deleted file mode 100644
index bbbb3cf..0000000
--- a/drivers/staging/lustre/lustre/fid/fid_handler.c
+++ /dev/null
@@ -1,661 +0,0 @@
-/*
- * GPL HEADER START
- *
- * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
- *
- * 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 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).
- *
- * 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.
- *
- * GPL HEADER END
- */
-/*
- * Copyright (c) 2007, 2010, Oracle and/or its affiliates. All rights reserved.
- * Use is subject to license terms.
- *
- * Copyright (c) 2011, 2012, Intel Corporation.
- */
-/*
- * This file is part of Lustre, http://www.lustre.org/
- * Lustre is a trademark of Sun Microsystems, Inc.
- *
- * lustre/fid/fid_handler.c
- *
- * Lustre Sequence Manager
- *
- * Author: Yury Umanets <umka@clusterfs.com>
- */
-
-#define DEBUG_SUBSYSTEM S_FID
-
-# include <linux/libcfs/libcfs.h>
-# include <linux/module.h>
-
-#include <obd.h>
-#include <obd_class.h>
-#include <dt_object.h>
-#include <md_object.h>
-#include <obd_support.h>
-#include <lustre_req_layout.h>
-#include <lustre_fid.h>
-#include "fid_internal.h"
-
-int client_fid_init(struct obd_device *obd,
- struct obd_export *exp, enum lu_cli_type type)
-{
- struct client_obd *cli = &obd->u.cli;
- char *prefix;
- int rc;
- ENTRY;
-
- OBD_ALLOC_PTR(cli->cl_seq);
- if (cli->cl_seq == NULL)
- RETURN(-ENOMEM);
-
- OBD_ALLOC(prefix, MAX_OBD_NAME + 5);
- if (prefix == NULL)
- GOTO(out_free_seq, rc = -ENOMEM);
-
- snprintf(prefix, MAX_OBD_NAME + 5, "cli-%s", obd->obd_name);
-
- /* Init client side sequence-manager */
- rc = seq_client_init(cli->cl_seq, exp, type, prefix, NULL);
- OBD_FREE(prefix, MAX_OBD_NAME + 5);
- if (rc)
- GOTO(out_free_seq, rc);
-
- RETURN(rc);
-out_free_seq:
- OBD_FREE_PTR(cli->cl_seq);
- cli->cl_seq = NULL;
- return rc;
-}
-EXPORT_SYMBOL(client_fid_init);
-
-int client_fid_fini(struct obd_device *obd)
-{
- struct client_obd *cli = &obd->u.cli;
- ENTRY;
-
- if (cli->cl_seq != NULL) {
- seq_client_fini(cli->cl_seq);
- OBD_FREE_PTR(cli->cl_seq);
- cli->cl_seq = NULL;
- }
-
- RETURN(0);
-}
-EXPORT_SYMBOL(client_fid_fini);
-
-static void seq_server_proc_fini(struct lu_server_seq *seq);
-
-/* Assigns client to sequence controller node. */
-int seq_server_set_cli(struct lu_server_seq *seq,
- struct lu_client_seq *cli,
- const struct lu_env *env)
-{
- int rc = 0;
- ENTRY;
-
- /*
- * Ask client for new range, assign that range to ->seq_space and write
- * seq state to backing store should be atomic.
- */
- mutex_lock(&seq->lss_mutex);
-
- if (cli == NULL) {
- CDEBUG(D_INFO, "%s: Detached sequence client %s\n",
- seq->lss_name, cli->lcs_name);
- seq->lss_cli = cli;
- GOTO(out_up, rc = 0);
- }
-
- if (seq->lss_cli != NULL) {
- CDEBUG(D_HA, "%s: Sequence controller is already "
- "assigned\n", seq->lss_name);
- GOTO(out_up, rc = -EEXIST);
- }
-
- CDEBUG(D_INFO, "%s: Attached sequence controller %s\n",
- seq->lss_name, cli->lcs_name);
-
- seq->lss_cli = cli;
- cli->lcs_space.lsr_index = seq->lss_site->ss_node_id;
- EXIT;
-out_up:
- mutex_unlock(&seq->lss_mutex);
- return rc;
-}
-EXPORT_SYMBOL(seq_server_set_cli);
-/*
- * allocate \a w units of sequence from range \a from.
- */
-static inline void range_alloc(struct lu_seq_range *to,
- struct lu_seq_range *from,
- __u64 width)
-{
- width = min(range_space(from), width);
- to->lsr_start = from->lsr_start;
- to->lsr_end = from->lsr_start + width;
- from->lsr_start += width;
-}
-
-/**
- * On controller node, allocate new super sequence for regular sequence server.
- * As this super sequence controller, this node suppose to maintain fld
- * and update index.
- * \a out range always has currect mds node number of requester.
- */
-
-static int __seq_server_alloc_super(struct lu_server_seq *seq,
- struct lu_seq_range *out,
- const struct lu_env *env)
-{
- struct lu_seq_range *space = &seq->lss_space;
- int rc;
- ENTRY;
-
- LASSERT(range_is_sane(space));
-
- if (range_is_exhausted(space)) {
- CERROR("%s: Sequences space is exhausted\n",
- seq->lss_name);
- RETURN(-ENOSPC);
- } else {
- range_alloc(out, space, seq->lss_width);
- }
-
- rc = seq_store_update(env, seq, out, 1 /* sync */);
-
- LCONSOLE_INFO("%s: super-sequence allocation rc = %d " DRANGE"\n",
- seq->lss_name, rc, PRANGE(out));
-
- RETURN(rc);
-}
-
-int seq_server_alloc_super(struct lu_server_seq *seq,
- struct lu_seq_range *out,
- const struct lu_env *env)
-{
- int rc;
- ENTRY;
-
- mutex_lock(&seq->lss_mutex);
- rc = __seq_server_alloc_super(seq, out, env);
- mutex_unlock(&seq->lss_mutex);
-
- RETURN(rc);
-}
-
-static int __seq_set_init(const struct lu_env *env,
- struct lu_server_seq *seq)
-{
- struct lu_seq_range *space = &seq->lss_space;
- int rc;
-
- range_alloc(&seq->lss_lowater_set, space, seq->lss_set_width);
- range_alloc(&seq->lss_hiwater_set, space, seq->lss_set_width);
-
- rc = seq_store_update(env, seq, NULL, 1);
-
- return rc;
-}
-
-/*
- * This function implements new seq allocation algorithm using async
- * updates to seq file on disk. ref bug 18857 for details.
- * there are four variable to keep track of this process
- *
- * lss_space; - available lss_space
- * lss_lowater_set; - lu_seq_range for all seqs before barrier, i.e. safe to use
- * lss_hiwater_set; - lu_seq_range after barrier, i.e. allocated but may be
- * not yet committed
- *
- * when lss_lowater_set reaches the end it is replaced with hiwater one and
- * a write operation is initiated to allocate new hiwater range.
- * if last seq write opearion is still not commited, current operation is
- * flaged as sync write op.
- */
-static int range_alloc_set(const struct lu_env *env,
- struct lu_seq_range *out,
- struct lu_server_seq *seq)
-{
- struct lu_seq_range *space = &seq->lss_space;
- struct lu_seq_range *loset = &seq->lss_lowater_set;
- struct lu_seq_range *hiset = &seq->lss_hiwater_set;
- int rc = 0;
-
- if (range_is_zero(loset))
- __seq_set_init(env, seq);
-
- if (OBD_FAIL_CHECK(OBD_FAIL_SEQ_ALLOC)) /* exhaust set */
- loset->lsr_start = loset->lsr_end;
-
- if (range_is_exhausted(loset)) {
- /* reached high water mark. */
- struct lu_device *dev = seq->lss_site->ss_lu->ls_top_dev;
- int obd_num_clients = dev->ld_obd->obd_num_exports;
- __u64 set_sz;
-
- /* calculate new seq width based on number of clients */
- set_sz = max(seq->lss_set_width,
- obd_num_clients * seq->lss_width);
- set_sz = min(range_space(space), set_sz);
-
- /* Switch to hiwater range now */
- *loset = *hiset;
- /* allocate new hiwater range */
- range_alloc(hiset, space, set_sz);
-
- /* update ondisk seq with new *space */
- rc = seq_store_update(env, seq, NULL, seq->lss_need_sync);
- }
-
- LASSERTF(!range_is_exhausted(loset) || range_is_sane(loset),
- DRANGE"\n", PRANGE(loset));
-
- if (rc == 0)
- range_alloc(out, loset, seq->lss_width);
-
- RETURN(rc);
-}
-
-static int __seq_server_alloc_meta(struct lu_server_seq *seq,
- struct lu_seq_range *out,
- const struct lu_env *env)
-{
- struct lu_seq_range *space = &seq->lss_space;
- int rc = 0;
-
- ENTRY;
-
- LASSERT(range_is_sane(space));
-
- /* Check if available space ends and allocate new super seq */
- if (range_is_exhausted(space)) {
- if (!seq->lss_cli) {
- CERROR("%s: No sequence controller is attached.\n",
- seq->lss_name);
- RETURN(-ENODEV);
- }
-
- rc = seq_client_alloc_super(seq->lss_cli, env);
- if (rc) {
- CERROR("%s: Can't allocate super-sequence, rc %d\n",
- seq->lss_name, rc);
- RETURN(rc);
- }
-
- /* Saving new range to allocation space. */
- *space = seq->lss_cli->lcs_space;
- LASSERT(range_is_sane(space));
- }
-
- rc = range_alloc_set(env, out, seq);
- if (rc != 0) {
- CERROR("%s: Allocated meta-sequence failed: rc = %d\n",
- seq->lss_name, rc);
- RETURN(rc);
- }
-
- CDEBUG(D_INFO, "%s: Allocated meta-sequence " DRANGE"\n",
- seq->lss_name, PRANGE(out));
-
- RETURN(rc);
-}
-
-int seq_server_alloc_meta(struct lu_server_seq *seq,
- struct lu_seq_range *out,
- const struct lu_env *env)
-{
- int rc;
- ENTRY;
-
- mutex_lock(&seq->lss_mutex);
- rc = __seq_server_alloc_meta(seq, out, env);
- mutex_unlock(&seq->lss_mutex);
-
- RETURN(rc);
-}
-EXPORT_SYMBOL(seq_server_alloc_meta);
-
-static int seq_server_handle(struct lu_site *site,
- const struct lu_env *env,
- __u32 opc, struct lu_seq_range *out)
-{
- int rc;
- struct seq_server_site *ss_site;
- ENTRY;
-
- ss_site = lu_site2seq(site);
-
- switch (opc) {
- case SEQ_ALLOC_META:
- if (!ss_site->ss_server_seq) {
- CERROR("Sequence server is not "
- "initialized\n");
- RETURN(-EINVAL);
- }
- rc = seq_server_alloc_meta(ss_site->ss_server_seq, out, env);
- break;
- case SEQ_ALLOC_SUPER:
- if (!ss_site->ss_control_seq) {
- CERROR("Sequence controller is not "
- "initialized\n");
- RETURN(-EINVAL);
- }
- rc = seq_server_alloc_super(ss_site->ss_control_seq, out, env);
- break;
- default:
- rc = -EINVAL;
- break;
- }
-
- RETURN(rc);
-}
-
-static int seq_req_handle(struct ptlrpc_request *req,
- const struct lu_env *env,
- struct seq_thread_info *info)
-{
- struct lu_seq_range *out, *tmp;
- struct lu_site *site;
- int rc = -EPROTO;
- __u32 *opc;
- ENTRY;
-
- LASSERT(!(lustre_msg_get_flags(req->rq_reqmsg) & MSG_REPLAY));
- site = req->rq_export->exp_obd->obd_lu_dev->ld_site;
- LASSERT(site != NULL);
-
- rc = req_capsule_server_pack(info->sti_pill);
- if (rc)
- RETURN(err_serious(rc));
-
- opc = req_capsule_client_get(info->sti_pill, &RMF_SEQ_OPC);
- if (opc != NULL) {
- out = req_capsule_server_get(info->sti_pill, &RMF_SEQ_RANGE);
- if (out == NULL)
- RETURN(err_serious(-EPROTO));
-
- tmp = req_capsule_client_get(info->sti_pill, &RMF_SEQ_RANGE);
-
- /* seq client passed mdt id, we need to pass that using out
- * range parameter */
-
- out->lsr_index = tmp->lsr_index;
- out->lsr_flags = tmp->lsr_flags;
- rc = seq_server_handle(site, env, *opc, out);
- } else
- rc = err_serious(-EPROTO);
-
- RETURN(rc);
-}
-
-/* context key constructor/destructor: seq_key_init, seq_key_fini */
-LU_KEY_INIT_FINI(seq, struct seq_thread_info);
-
-/* context key: seq_thread_key */
-LU_CONTEXT_KEY_DEFINE(seq, LCT_MD_THREAD | LCT_DT_THREAD);
-
-static void seq_thread_info_init(struct ptlrpc_request *req,
- struct seq_thread_info *info)
-{
- info->sti_pill = &req->rq_pill;
- /* Init request capsule */
- req_capsule_init(info->sti_pill, req, RCL_SERVER);
- req_capsule_set(info->sti_pill, &RQF_SEQ_QUERY);
-}
-
-static void seq_thread_info_fini(struct seq_thread_info *info)
-{
- req_capsule_fini(info->sti_pill);
-}
-
-int seq_handle(struct ptlrpc_request *req)
-{
- const struct lu_env *env;
- struct seq_thread_info *info;
- int rc;
-
- env = req->rq_svc_thread->t_env;
- LASSERT(env != NULL);
-
- info = lu_context_key_get(&env->le_ctx, &seq_thread_key);
- LASSERT(info != NULL);
-
- seq_thread_info_init(req, info);
- rc = seq_req_handle(req, env, info);
- /* XXX: we don't need replay but MDT assign transno in any case,
- * remove it manually before reply*/
- lustre_msg_set_transno(req->rq_repmsg, 0);
- seq_thread_info_fini(info);
-
- return rc;
-}
-EXPORT_SYMBOL(seq_handle);
-
-/*
- * Entry point for handling FLD RPCs called from MDT.
- */
-int seq_query(struct com_thread_info *info)
-{
- return seq_handle(info->cti_pill->rc_req);
-}
-EXPORT_SYMBOL(seq_query);
-
-
-#ifdef LPROCFS
-static int seq_server_proc_init(struct lu_server_seq *seq)
-{
- int rc;
- ENTRY;
-
- seq->lss_proc_dir = lprocfs_register(seq->lss_name,
- seq_type_proc_dir,
- NULL, NULL);
- if (IS_ERR(seq->lss_proc_dir)) {
- rc = PTR_ERR(seq->lss_proc_dir);
- RETURN(rc);
- }
-
- rc = lprocfs_add_vars(seq->lss_proc_dir,
- seq_server_proc_list, seq);
- if (rc) {
- CERROR("%s: Can't init sequence manager "
- "proc, rc %d\n", seq->lss_name, rc);
- GOTO(out_cleanup, rc);
- }
-
- RETURN(0);
-
-out_cleanup:
- seq_server_proc_fini(seq);
- return rc;
-}
-
-static void seq_server_proc_fini(struct lu_server_seq *seq)
-{
- ENTRY;
- if (seq->lss_proc_dir != NULL) {
- if (!IS_ERR(seq->lss_proc_dir))
- lprocfs_remove(&seq->lss_proc_dir);
- seq->lss_proc_dir = NULL;
- }
- EXIT;
-}
-#else
-static int seq_server_proc_init(struct lu_server_seq *seq)
-{
- return 0;
-}
-
-static void seq_server_proc_fini(struct lu_server_seq *seq)
-{
- return;
-}
-#endif
-
-
-int seq_server_init(struct lu_server_seq *seq,
- struct dt_device *dev,
- const char *prefix,
- enum lu_mgr_type type,
- struct seq_server_site *ss,
- const struct lu_env *env)
-{
- int rc, is_srv = (type == LUSTRE_SEQ_SERVER);
- ENTRY;
-
- LASSERT(dev != NULL);
- LASSERT(prefix != NULL);
- LASSERT(ss != NULL);
- LASSERT(ss->ss_lu != NULL);
-
- seq->lss_cli = NULL;
- seq->lss_type = type;
- seq->lss_site = ss;
- range_init(&seq->lss_space);
-
- range_init(&seq->lss_lowater_set);
- range_init(&seq->lss_hiwater_set);
- seq->lss_set_width = LUSTRE_SEQ_BATCH_WIDTH;
-
- mutex_init(&seq->lss_mutex);
-
- seq->lss_width = is_srv ?
- LUSTRE_SEQ_META_WIDTH : LUSTRE_SEQ_SUPER_WIDTH;
-
- snprintf(seq->lss_name, sizeof(seq->lss_name),
- "%s-%s", (is_srv ? "srv" : "ctl"), prefix);
-
- rc = seq_store_init(seq, env, dev);
- if (rc)
- GOTO(out, rc);
- /* Request backing store for saved sequence info. */
- rc = seq_store_read(seq, env);
- if (rc == -ENODATA) {
-
- /* Nothing is read, init by default value. */
- seq->lss_space = is_srv ?
- LUSTRE_SEQ_ZERO_RANGE:
- LUSTRE_SEQ_SPACE_RANGE;
-
- LASSERT(ss != NULL);
- seq->lss_space.lsr_index = ss->ss_node_id;
- LCONSOLE_INFO("%s: No data found "
- "on store. Initialize space\n",
- seq->lss_name);
-
- rc = seq_store_update(env, seq, NULL, 0);
- if (rc) {
- CERROR("%s: Can't write space data, "
- "rc %d\n", seq->lss_name, rc);
- }
- } else if (rc) {
- CERROR("%s: Can't read space data, rc %d\n",
- seq->lss_name, rc);
- GOTO(out, rc);
- }
-
- if (is_srv) {
- LASSERT(range_is_sane(&seq->lss_space));
- } else {
- LASSERT(!range_is_zero(&seq->lss_space) &&
- range_is_sane(&seq->lss_space));
- }
-
- rc = seq_server_proc_init(seq);
- if (rc)
- GOTO(out, rc);
-
- EXIT;
-out:
- if (rc)
- seq_server_fini(seq, env);
- return rc;
-}
-EXPORT_SYMBOL(seq_server_init);
-
-void seq_server_fini(struct lu_server_seq *seq,
- const struct lu_env *env)
-{
- ENTRY;
-
- seq_server_proc_fini(seq);
- seq_store_fini(seq, env);
-
- EXIT;
-}
-EXPORT_SYMBOL(seq_server_fini);
-
-int seq_site_fini(const struct lu_env *env, struct seq_server_site *ss)
-{
- if (ss == NULL)
- RETURN(0);
-
- if (ss->ss_server_seq) {
- seq_server_fini(ss->ss_server_seq, env);
- OBD_FREE_PTR(ss->ss_server_seq);
- ss->ss_server_seq = NULL;
- }
-
- if (ss->ss_control_seq) {
- seq_server_fini(ss->ss_control_seq, env);
- OBD_FREE_PTR(ss->ss_control_seq);
- ss->ss_control_seq = NULL;
- }
-
- if (ss->ss_client_seq) {
- seq_client_fini(ss->ss_client_seq);
- OBD_FREE_PTR(ss->ss_client_seq);
- ss->ss_client_seq = NULL;
- }
-
- RETURN(0);
-}
-EXPORT_SYMBOL(seq_site_fini);
-
-proc_dir_entry_t *seq_type_proc_dir = NULL;
-
-static int __init fid_mod_init(void)
-{
- seq_type_proc_dir = lprocfs_register(LUSTRE_SEQ_NAME,
- proc_lustre_root,
- NULL, NULL);
- if (IS_ERR(seq_type_proc_dir))
- return PTR_ERR(seq_type_proc_dir);
-
- LU_CONTEXT_KEY_INIT(&seq_thread_key);
- lu_context_key_register(&seq_thread_key);
- return 0;
-}
-
-static void __exit fid_mod_exit(void)
-{
- lu_context_key_degister(&seq_thread_key);
- if (seq_type_proc_dir != NULL && !IS_ERR(seq_type_proc_dir)) {
- lprocfs_remove(&seq_type_proc_dir);
- seq_type_proc_dir = NULL;
- }
-}
-
-MODULE_AUTHOR("Sun Microsystems, Inc. <http://www.lustre.org/>");
-MODULE_DESCRIPTION("Lustre FID Module");
-MODULE_LICENSE("GPL");
-
-cfs_module(fid, "0.1.0", fid_mod_init, fid_mod_exit);
diff --git a/drivers/staging/lustre/lustre/fid/fid_internal.h b/drivers/staging/lustre/lustre/fid/fid_internal.h
index 407a743..1dbe46b 100644
--- a/drivers/staging/lustre/lustre/fid/fid_internal.h
+++ b/drivers/staging/lustre/lustre/fid/fid_internal.h
@@ -41,44 +41,16 @@
#define __FID_INTERNAL_H
#include <lustre/lustre_idl.h>
-#include <dt_object.h>
-
#include <linux/libcfs/libcfs.h>
-struct seq_thread_info {
- struct req_capsule *sti_pill;
- struct lu_seq_range sti_space;
- struct lu_buf sti_buf;
-};
-
-enum {
- SEQ_TXN_STORE_CREDITS = 20
-};
-
-extern struct lu_context_key seq_thread_key;
-
+/* Functions used internally in module. */
int seq_client_alloc_super(struct lu_client_seq *seq,
const struct lu_env *env);
-/* Store API functions. */
-int seq_store_init(struct lu_server_seq *seq,
- const struct lu_env *env,
- struct dt_device *dt);
-void seq_store_fini(struct lu_server_seq *seq,
- const struct lu_env *env);
-
-int seq_store_read(struct lu_server_seq *seq,
- const struct lu_env *env);
-
-int seq_store_update(const struct lu_env *env, struct lu_server_seq *seq,
- struct lu_seq_range *out, int sync);
-
-#ifdef LPROCFS
-extern struct lprocfs_vars seq_server_proc_list[];
+# ifdef LPROCFS
extern struct lprocfs_vars seq_client_proc_list[];
-#endif
-
+# endif
-extern proc_dir_entry_t *seq_type_proc_dir;
+extern struct proc_dir_entry *seq_type_proc_dir;
#endif /* __FID_INTERNAL_H */
diff --git a/drivers/staging/lustre/lustre/fid/fid_lib.c b/drivers/staging/lustre/lustre/fid/fid_lib.c
index eaff51a..f03afde 100644
--- a/drivers/staging/lustre/lustre/fid/fid_lib.c
+++ b/drivers/staging/lustre/lustre/fid/fid_lib.c
@@ -43,11 +43,9 @@
#define DEBUG_SUBSYSTEM S_FID
-# include <linux/libcfs/libcfs.h>
-# include <linux/module.h>
-
-#include <obd.h>
-#include <lu_object.h>
+#include <linux/libcfs/libcfs.h>
+#include <linux/module.h>
+#include <lustre/lustre_idl.h>
#include <lustre_fid.h>
/**
@@ -56,9 +54,9 @@
*
* Fid namespace:
* <pre>
- * Normal FID: seq:64 [2^33,2^64-1] oid:32 ver:32
- * IGIF : 0:32, ino:32 gen:32 0:32
- * IDIF : 0:31, 1:1, ost-index:16, objd:48 0:32
+ * Normal FID: seq:64 [2^33,2^64-1] oid:32 ver:32
+ * IGIF : 0:32, ino:32 gen:32 0:32
+ * IDIF : 0:31, 1:1, ost-index:16, objd:48 0:32
* </pre>
*
* The first 0x400 sequences of normal FID are reserved for special purpose.
diff --git a/drivers/staging/lustre/lustre/fid/fid_request.c b/drivers/staging/lustre/lustre/fid/fid_request.c
index fcaaca7..a279b5a 100644
--- a/drivers/staging/lustre/lustre/fid/fid_request.c
+++ b/drivers/staging/lustre/lustre/fid/fid_request.c
@@ -27,7 +27,7 @@
* Copyright (c) 2007, 2010, Oracle and/or its affiliates. All rights reserved.
* Use is subject to license terms.
*
- * Copyright (c) 2011, 2012, Intel Corporation.
+ * Copyright (c) 2011, 2013, Intel Corporation.
*/
/*
* This file is part of Lustre, http://www.lustre.org/
@@ -42,15 +42,12 @@
#define DEBUG_SUBSYSTEM S_FID
-# include <linux/libcfs/libcfs.h>
-# include <linux/module.h>
+#include <linux/libcfs/libcfs.h>
+#include <linux/module.h>
#include <obd.h>
#include <obd_class.h>
-#include <dt_object.h>
-#include <md_object.h>
#include <obd_support.h>
-#include <lustre_req_layout.h>
#include <lustre_fid.h>
/* mdc RPC locks */
#include <lustre_mdc.h>
@@ -63,9 +60,9 @@ static int seq_client_rpc(struct lu_client_seq *seq,
struct obd_export *exp = seq->lcs_exp;
struct ptlrpc_request *req;
struct lu_seq_range *out, *in;
- __u32 *op;
- unsigned int debug_mask;
- int rc;
+ __u32 *op;
+ unsigned int debug_mask;
+ int rc;
ENTRY;
req = ptlrpc_request_alloc_pack(class_exp2cliimp(exp), &RQF_SEQ_QUERY,
@@ -153,9 +150,7 @@ int seq_client_alloc_super(struct lu_client_seq *seq,
mutex_lock(&seq->lcs_mutex);
if (seq->lcs_srv) {
- LASSERT(env != NULL);
- rc = seq_server_alloc_super(seq->lcs_srv, &seq->lcs_space,
- env);
+ rc = 0;
} else {
/* Check whether the connection to seq controller has been
* setup (lcs_exp != NULL) */
@@ -179,8 +174,7 @@ static int seq_client_alloc_meta(const struct lu_env *env,
ENTRY;
if (seq->lcs_srv) {
- LASSERT(env != NULL);
- rc = seq_server_alloc_meta(seq->lcs_srv, &seq->lcs_space, env);
+ rc = 0;
} else {
do {
/* If meta server return -EINPROGRESS or EAGAIN,
@@ -191,6 +185,7 @@ static int seq_client_alloc_meta(const struct lu_env *env,
SEQ_ALLOC_META, "meta");
} while (rc == -EINPROGRESS || rc == -EAGAIN);
}
+
RETURN(rc);
}
@@ -409,11 +404,22 @@ void seq_client_flush(struct lu_client_seq *seq)
}
EXPORT_SYMBOL(seq_client_flush);
-static void seq_client_proc_fini(struct lu_client_seq *seq);
-
+static void seq_client_proc_fini(struct lu_client_seq *seq)
+{
#ifdef LPROCFS
+ ENTRY;
+ if (seq->lcs_proc_dir) {
+ if (!IS_ERR(seq->lcs_proc_dir))
+ lprocfs_remove(&seq->lcs_proc_dir);
+ seq->lcs_proc_dir = NULL;
+ }
+ EXIT;
+#endif /* LPROCFS */
+}
+
static int seq_client_proc_init(struct lu_client_seq *seq)
{
+#ifdef LPROCFS
int rc;
ENTRY;
@@ -441,29 +447,11 @@ static int seq_client_proc_init(struct lu_client_seq *seq)
out_cleanup:
seq_client_proc_fini(seq);
return rc;
-}
-static void seq_client_proc_fini(struct lu_client_seq *seq)
-{
- ENTRY;
- if (seq->lcs_proc_dir) {
- if (!IS_ERR(seq->lcs_proc_dir))
- lprocfs_remove(&seq->lcs_proc_dir);
- seq->lcs_proc_dir = NULL;
- }
- EXIT;
-}
-#else
-static int seq_client_proc_init(struct lu_client_seq *seq)
-{
+#else /* LPROCFS */
return 0;
-}
-
-static void seq_client_proc_fini(struct lu_client_seq *seq)
-{
- return;
-}
#endif
+}
int seq_client_init(struct lu_client_seq *seq,
struct obd_export *exp,
@@ -520,3 +508,76 @@ void seq_client_fini(struct lu_client_seq *seq)
EXIT;
}
EXPORT_SYMBOL(seq_client_fini);
+
+int client_fid_init(struct obd_device *obd,
+ struct obd_export *exp, enum lu_cli_type type)
+{
+ struct client_obd *cli = &obd->u.cli;
+ char *prefix;
+ int rc;
+ ENTRY;
+
+ OBD_ALLOC_PTR(cli->cl_seq);
+ if (cli->cl_seq == NULL)
+ RETURN(-ENOMEM);
+
+ OBD_ALLOC(prefix, MAX_OBD_NAME + 5);
+ if (prefix == NULL)
+ GOTO(out_free_seq, rc = -ENOMEM);
+
+ snprintf(prefix, MAX_OBD_NAME + 5, "cli-%s", obd->obd_name);
+
+ /* Init client side sequence-manager */
+ rc = seq_client_init(cli->cl_seq, exp, type, prefix, NULL);
+ OBD_FREE(prefix, MAX_OBD_NAME + 5);
+ if (rc)
+ GOTO(out_free_seq, rc);
+
+ RETURN(rc);
+out_free_seq:
+ OBD_FREE_PTR(cli->cl_seq);
+ cli->cl_seq = NULL;
+ return rc;
+}
+EXPORT_SYMBOL(client_fid_init);
+
+int client_fid_fini(struct obd_device *obd)
+{
+ struct client_obd *cli = &obd->u.cli;
+ ENTRY;
+
+ if (cli->cl_seq != NULL) {
+ seq_client_fini(cli->cl_seq);
+ OBD_FREE_PTR(cli->cl_seq);
+ cli->cl_seq = NULL;
+ }
+
+ RETURN(0);
+}
+EXPORT_SYMBOL(client_fid_fini);
+
+struct proc_dir_entry *seq_type_proc_dir;
+
+static int __init fid_mod_init(void)
+{
+ seq_type_proc_dir = lprocfs_register(LUSTRE_SEQ_NAME,
+ proc_lustre_root,
+ NULL, NULL);
+ if (IS_ERR(seq_type_proc_dir))
+ return PTR_ERR(seq_type_proc_dir);
+ return 0;
+}
+
+static void __exit fid_mod_exit(void)
+{
+ if (seq_type_proc_dir != NULL && !IS_ERR(seq_type_proc_dir)) {
+ lprocfs_remove(&seq_type_proc_dir);
+ seq_type_proc_dir = NULL;
+ }
+}
+
+MODULE_AUTHOR("Sun Microsystems, Inc. <http://www.lustre.org/>");
+MODULE_DESCRIPTION("Lustre FID Module");
+MODULE_LICENSE("GPL");
+
+cfs_module(fid, "0.1.0", fid_mod_init, fid_mod_exit);
diff --git a/drivers/staging/lustre/lustre/fid/fid_store.c b/drivers/staging/lustre/lustre/fid/fid_store.c
deleted file mode 100644
index a90e6e3..0000000
--- a/drivers/staging/lustre/lustre/fid/fid_store.c
+++ /dev/null
@@ -1,259 +0,0 @@
-/*
- * GPL HEADER START
- *
- * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
- *
- * 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 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).
- *
- * 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.
- *
- * 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.
- */
-/*
- * This file is part of Lustre, http://www.lustre.org/
- * Lustre is a trademark of Sun Microsystems, Inc.
- *
- * lustre/fid/fid_store.c
- *
- * Lustre Sequence Manager
- *
- * Author: Yury Umanets <umka@clusterfs.com>
- */
-
-#define DEBUG_SUBSYSTEM S_FID
-
-# include <linux/libcfs/libcfs.h>
-# include <linux/module.h>
-
-#include <obd.h>
-#include <obd_class.h>
-#include <dt_object.h>
-#include <md_object.h>
-#include <obd_support.h>
-#include <lustre_req_layout.h>
-#include <lustre_fid.h>
-#include "fid_internal.h"
-
-
-static struct lu_buf *seq_store_buf(struct seq_thread_info *info)
-{
- struct lu_buf *buf;
-
- buf = &info->sti_buf;
- buf->lb_buf = &info->sti_space;
- buf->lb_len = sizeof(info->sti_space);
- return buf;
-}
-
-struct seq_update_callback {
- struct dt_txn_commit_cb suc_cb;
- struct lu_server_seq *suc_seq;
-};
-
-void seq_update_cb(struct lu_env *env, struct thandle *th,
- struct dt_txn_commit_cb *cb, int err)
-{
- struct seq_update_callback *ccb;
-
- ccb = container_of0(cb, struct seq_update_callback, suc_cb);
-
- LASSERT(ccb->suc_seq != NULL);
-
- ccb->suc_seq->lss_need_sync = 0;
- OBD_FREE_PTR(ccb);
-}
-
-int seq_update_cb_add(struct thandle *th, struct lu_server_seq *seq)
-{
- struct seq_update_callback *ccb;
- struct dt_txn_commit_cb *dcb;
- int rc;
-
- OBD_ALLOC_PTR(ccb);
- if (ccb == NULL)
- return -ENOMEM;
-
- ccb->suc_seq = seq;
- seq->lss_need_sync = 1;
-
- dcb = &ccb->suc_cb;
- dcb->dcb_func = seq_update_cb;
- INIT_LIST_HEAD(&dcb->dcb_linkage);
- strncpy(dcb->dcb_name, "seq_update_cb", MAX_COMMIT_CB_STR_LEN);
- dcb->dcb_name[MAX_COMMIT_CB_STR_LEN - 1] = '\0';
-
- rc = dt_trans_cb_add(th, dcb);
- if (rc)
- OBD_FREE_PTR(ccb);
- return rc;
-}
-
-/* This function implies that caller takes care about locking. */
-int seq_store_update(const struct lu_env *env, struct lu_server_seq *seq,
- struct lu_seq_range *out, int sync)
-{
- struct dt_device *dt_dev = lu2dt_dev(seq->lss_obj->do_lu.lo_dev);
- struct seq_thread_info *info;
- struct thandle *th;
- loff_t pos = 0;
- int rc;
-
- info = lu_context_key_get(&env->le_ctx, &seq_thread_key);
- LASSERT(info != NULL);
-
- th = dt_trans_create(env, dt_dev);
- if (IS_ERR(th))
- RETURN(PTR_ERR(th));
-
- rc = dt_declare_record_write(env, seq->lss_obj,
- sizeof(struct lu_seq_range), 0, th);
- if (rc)
- GOTO(exit, rc);
-
- if (out != NULL) {
- rc = fld_declare_server_create(env,
- seq->lss_site->ss_server_fld,
- out, th);
- if (rc)
- GOTO(exit, rc);
- }
-
- rc = dt_trans_start_local(env, dt_dev, th);
- if (rc)
- GOTO(exit, rc);
-
- /* Store ranges in le format. */
- range_cpu_to_le(&info->sti_space, &seq->lss_space);
-
- rc = dt_record_write(env, seq->lss_obj, seq_store_buf(info), &pos, th);
- if (rc) {
- CERROR("%s: Can't write space data, rc %d\n",
- seq->lss_name, rc);
- GOTO(exit, rc);
- } else if (out != NULL) {
- rc = fld_server_create(env, seq->lss_site->ss_server_fld, out,
- th);
- if (rc) {
- CERROR("%s: Can't Update fld database, rc %d\n",
- seq->lss_name, rc);
- GOTO(exit, rc);
- }
- }
- /* next sequence update will need sync until this update is committed
- * in case of sync operation this is not needed obviously */
- if (!sync)
- /* if callback can't be added then sync always */
- sync = !!seq_update_cb_add(th, seq);
-
- th->th_sync |= sync;
-exit:
- dt_trans_stop(env, dt_dev, th);
- return rc;
-}
-
-/*
- * This function implies that caller takes care about locking or locking is not
- * needed (init time).
- */
-int seq_store_read(struct lu_server_seq *seq,
- const struct lu_env *env)
-{
- struct seq_thread_info *info;
- loff_t pos = 0;
- int rc;
- ENTRY;
-
- info = lu_context_key_get(&env->le_ctx, &seq_thread_key);
- LASSERT(info != NULL);
-
- rc = seq->lss_obj->do_body_ops->dbo_read(env, seq->lss_obj,
- seq_store_buf(info),
- &pos, BYPASS_CAPA);
-
- if (rc == sizeof(info->sti_space)) {
- range_le_to_cpu(&seq->lss_space, &info->sti_space);
- CDEBUG(D_INFO, "%s: Space - "DRANGE"\n",
- seq->lss_name, PRANGE(&seq->lss_space));
- rc = 0;
- } else if (rc == 0) {
- rc = -ENODATA;
- } else if (rc > 0) {
- CERROR("%s: Read only %d bytes of %d\n", seq->lss_name,
- rc, (int)sizeof(info->sti_space));
- rc = -EIO;
- }
-
- RETURN(rc);
-}
-
-int seq_store_init(struct lu_server_seq *seq,
- const struct lu_env *env,
- struct dt_device *dt)
-{
- struct dt_object *dt_obj;
- struct lu_fid fid;
- struct lu_attr attr;
- struct dt_object_format dof;
- const char *name;
- int rc;
- ENTRY;
-
- name = seq->lss_type == LUSTRE_SEQ_SERVER ?
- LUSTRE_SEQ_SRV_NAME : LUSTRE_SEQ_CTL_NAME;
-
- if (seq->lss_type == LUSTRE_SEQ_SERVER)
- lu_local_obj_fid(&fid, FID_SEQ_SRV_OID);
- else
- lu_local_obj_fid(&fid, FID_SEQ_CTL_OID);
-
- memset(&attr, 0, sizeof(attr));
- attr.la_valid = LA_MODE;
- attr.la_mode = S_IFREG | 0666;
- dof.dof_type = DFT_REGULAR;
-
- dt_obj = dt_find_or_create(env, dt, &fid, &dof, &attr);
- if (!IS_ERR(dt_obj)) {
- seq->lss_obj = dt_obj;
- rc = 0;
- } else {
- CERROR("%s: Can't find \"%s\" obj %d\n",
- seq->lss_name, name, (int)PTR_ERR(dt_obj));
- rc = PTR_ERR(dt_obj);
- }
-
- RETURN(rc);
-}
-
-void seq_store_fini(struct lu_server_seq *seq,
- const struct lu_env *env)
-{
- ENTRY;
-
- if (seq->lss_obj != NULL) {
- if (!IS_ERR(seq->lss_obj))
- lu_object_put(env, &seq->lss_obj->do_lu);
- seq->lss_obj = NULL;
- }
-
- EXIT;
-}
diff --git a/drivers/staging/lustre/lustre/fid/lproc_fid.c b/drivers/staging/lustre/lustre/fid/lproc_fid.c
index af817a8..20d0448 100644
--- a/drivers/staging/lustre/lustre/fid/lproc_fid.c
+++ b/drivers/staging/lustre/lustre/fid/lproc_fid.c
@@ -204,9 +204,6 @@ lprocfs_fid_server_seq_show(struct seq_file *m, void *unused)
RETURN(rc);
}
-struct lprocfs_vars seq_server_proc_list[] = {
-};
-
LPROC_SEQ_FOPS(lprocfs_fid_space);
LPROC_SEQ_FOPS(lprocfs_fid_width);
LPROC_SEQ_FOPS_RO(lprocfs_fid_server);
diff --git a/drivers/staging/lustre/lustre/include/lu_object.h b/drivers/staging/lustre/lustre/include/lu_object.h
index f0d6fb1..8077898 100644
--- a/drivers/staging/lustre/lustre/include/lu_object.h
+++ b/drivers/staging/lustre/lustre/include/lu_object.h
@@ -665,6 +665,11 @@ lu_site_bkt_from_fid(struct lu_site *site, struct lu_fid *fid)
return cfs_hash_bd_extra_get(site->ls_obj_hash, &bd);
}
+static inline struct seq_server_site *lu_site2seq(const struct lu_site *s)
+{
+ return s->ld_seq_site;
+}
+
/** \name ctors
* Constructors/destructors.
* @{
diff --git a/drivers/staging/lustre/lustre/include/lustre_fid.h b/drivers/staging/lustre/lustre/include/lustre_fid.h
index 7523b40..f8f35c3 100644
--- a/drivers/staging/lustre/lustre/include/lustre_fid.h
+++ b/drivers/staging/lustre/lustre/include/lustre_fid.h
@@ -38,8 +38,8 @@
* Author: Yury Umanets <umka@clusterfs.com>
*/
-#ifndef __LINUX_FID_H
-#define __LINUX_FID_H
+#ifndef __LUSTRE_FID_H
+#define __LUSTRE_FID_H
/** \defgroup fid fid
*
@@ -154,13 +154,12 @@
#include <linux/libcfs/libcfs.h>
#include <lustre/lustre_idl.h>
-#include <lustre_req_layout.h>
-#include <lustre_mdt.h>
-#include <obd.h>
-
+struct lu_env;
struct lu_site;
struct lu_context;
+struct obd_device;
+struct obd_export;
/* Whole sequences space range and zero range definitions */
extern const struct lu_seq_range LUSTRE_SEQ_SPACE_RANGE;
@@ -320,6 +319,12 @@ static inline void lu_last_id_fid(struct lu_fid *fid, __u64 seq)
fid->f_ver = 0;
}
+/* seq client type */
+enum lu_cli_type {
+ LUSTRE_SEQ_METADATA = 1,
+ LUSTRE_SEQ_DATA
+};
+
enum lu_mgr_type {
LUSTRE_SEQ_SERVER,
LUSTRE_SEQ_CONTROLLER
@@ -426,10 +431,14 @@ struct lu_server_seq {
struct seq_server_site *lss_site;
};
+struct com_thread_info;
int seq_query(struct com_thread_info *info);
+
+struct ptlrpc_request;
int seq_handle(struct ptlrpc_request *req);
/* Server methods */
+
int seq_server_init(struct lu_server_seq *seq,
struct dt_device *dev,
const char *prefix,
@@ -472,6 +481,7 @@ int seq_site_fini(const struct lu_env *env, struct seq_server_site *ss);
int fid_is_local(const struct lu_env *env,
struct lu_site *site, const struct lu_fid *fid);
+enum lu_cli_type;
int client_fid_init(struct obd_device *obd, struct obd_export *exp,
enum lu_cli_type type);
int client_fid_fini(struct obd_device *obd);
@@ -760,4 +770,4 @@ static inline void range_be_to_cpu(struct lu_seq_range *dst, const struct lu_seq
/** @} fid */
-#endif /* __LINUX_FID_H */
+#endif /* __LUSTRE_FID_H */
diff --git a/drivers/staging/lustre/lustre/include/md_object.h b/drivers/staging/lustre/lustre/include/md_object.h
index 92d6420..65b270a 100644
--- a/drivers/staging/lustre/lustre/include/md_object.h
+++ b/drivers/staging/lustre/lustre/include/md_object.h
@@ -503,11 +503,6 @@ static inline struct md_device *md_obj2dev(const struct md_object *o)
return container_of0(o->mo_lu.lo_dev, struct md_device, md_lu_dev);
}
-static inline struct seq_server_site *lu_site2seq(const struct lu_site *s)
-{
- return s->ld_seq_site;
-}
-
static inline int md_device_init(struct md_device *md, struct lu_device_type *t)
{
return lu_device_init(&md->md_lu_dev, t);
diff --git a/drivers/staging/lustre/lustre/include/obd.h b/drivers/staging/lustre/lustre/include/obd.h
index 11d10fe..fbf997f 100644
--- a/drivers/staging/lustre/lustre/include/obd.h
+++ b/drivers/staging/lustre/lustre/include/obd.h
@@ -52,6 +52,7 @@
#include <lu_ref.h>
#include <lustre_lib.h>
#include <lustre_export.h>
+#include <lustre_fid.h>
#include <lustre_fld.h>
#include <lustre_capa.h>
@@ -1232,12 +1233,6 @@ typedef int (* md_enqueue_cb_t)(struct ptlrpc_request *req,
struct md_enqueue_info *minfo,
int rc);
-/* seq client type */
-enum lu_cli_type {
- LUSTRE_SEQ_METADATA = 1,
- LUSTRE_SEQ_DATA
-};
-
struct md_enqueue_info {
struct md_op_data mi_data;
struct lookup_intent mi_it;
--
1.7.9.5
^ permalink raw reply related [flat|nested] 62+ messages in thread* [PATCH 24/48] staging/lustre/llite: force lvb_data update after layout change
2013-07-22 16:06 [PATCH 00/48] staging/lustre: minor cleanup and Intel Lustre sync Peng Tao
` (22 preceding siblings ...)
2013-07-22 16:06 ` [PATCH 23/48] staging/lustre/fid: prepare FID module for client server split Peng Tao
@ 2013-07-22 16:06 ` Peng Tao
2013-07-22 16:06 ` [PATCH 25/48] staging/lustre/lfsck: LFSCK 1.5 technical debts (3) Peng Tao
` (24 subsequent siblings)
48 siblings, 0 replies; 62+ messages in thread
From: Peng Tao @ 2013-07-22 16:06 UTC (permalink / raw)
To: Greg Kroah-Hartman; +Cc: linux-kernel, jcl, Peng Tao, Andreas Dilger
From: jcl <jacques-charles.lafoucriere@cea.fr>
When a file is restored the layout lock is first
associated with the released layout and after restore
it has to be assocaited with the new layout. This patch
forces lvb_data update in ll_layout_fetch() even if one
is present (case for released->normal state change)
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-3299
Lustre-change: http://review.whamcloud.com/6291
Signed-off-by: JC Lafoucriere <jacques-charles.lafoucriere@cea.fr>
Reviewed-by: Jinshan Xiong <jinshan.xiong@intel.com>
Reviewed-by: Johann Lombardi <johann.lombardi@intel.com>
Signed-off-by: Peng Tao <tao.peng@emc.com>
Signed-off-by: Andreas Dilger <andreas.dilger@intel.com>
---
drivers/staging/lustre/lustre/llite/file.c | 20 +++++++++++---------
1 file changed, 11 insertions(+), 9 deletions(-)
diff --git a/drivers/staging/lustre/lustre/llite/file.c b/drivers/staging/lustre/lustre/llite/file.c
index 5493845..b1df7f5 100644
--- a/drivers/staging/lustre/lustre/llite/file.c
+++ b/drivers/staging/lustre/lustre/llite/file.c
@@ -2941,7 +2941,11 @@ static int ll_layout_fetch(struct inode *inode, struct ldlm_lock *lock)
int rc;
ENTRY;
- if (lock->l_lvb_data != NULL)
+ CDEBUG(D_INODE, DFID" LVB_READY=%d l_lvb_data=%p l_lvb_len=%d\n",
+ PFID(ll_inode2fid(inode)), !!(lock->l_flags & LDLM_FL_LVB_READY),
+ lock->l_lvb_data, lock->l_lvb_len);
+
+ if ((lock->l_lvb_data != NULL) && (lock->l_flags & LDLM_FL_LVB_READY))
RETURN(0);
/* if layout lock was granted right away, the layout is returned
@@ -2977,15 +2981,13 @@ static int ll_layout_fetch(struct inode *inode, struct ldlm_lock *lock)
memcpy(lvbdata, lmm, lmmsize);
lock_res_and_lock(lock);
- if (lock->l_lvb_data == NULL) {
- lock->l_lvb_data = lvbdata;
- lock->l_lvb_len = lmmsize;
- lvbdata = NULL;
- }
+ if (lock->l_lvb_data != NULL)
+ OBD_FREE_LARGE(lock->l_lvb_data, lock->l_lvb_len);
+
+ lock->l_lvb_data = lvbdata;
+ lock->l_lvb_len = lmmsize;
unlock_res_and_lock(lock);
- if (lvbdata != NULL)
- OBD_FREE_LARGE(lvbdata, lmmsize);
EXIT;
out:
@@ -3017,7 +3019,7 @@ static int ll_layout_lock_set(struct lustre_handle *lockh, ldlm_mode_t mode,
LASSERT(ldlm_has_layout(lock));
LDLM_DEBUG(lock, "File %p/"DFID" being reconfigured: %d.\n",
- inode, PFID(&lli->lli_fid), reconf);
+ inode, PFID(&lli->lli_fid), reconf);
/* in case this is a caching lock and reinstate with new inode */
md_set_lock_data(sbi->ll_md_exp, &lockh->cookie, inode, NULL);
--
1.7.9.5
^ permalink raw reply related [flat|nested] 62+ messages in thread* [PATCH 25/48] staging/lustre/lfsck: LFSCK 1.5 technical debts (3)
2013-07-22 16:06 [PATCH 00/48] staging/lustre: minor cleanup and Intel Lustre sync Peng Tao
` (23 preceding siblings ...)
2013-07-22 16:06 ` [PATCH 24/48] staging/lustre/llite: force lvb_data update after layout change Peng Tao
@ 2013-07-22 16:06 ` Peng Tao
2013-07-22 16:06 ` [PATCH 26/48] staging/lustre/osc: Check return code for lu_kmem_init Peng Tao
` (23 subsequent siblings)
48 siblings, 0 replies; 62+ messages in thread
From: Peng Tao @ 2013-07-22 16:06 UTC (permalink / raw)
To: Greg Kroah-Hartman
Cc: linux-kernel, Fan Yong, Fan Yong, Peng Tao, Andreas Dilger
From: Fan Yong <yong.fan@whamcloud.com>
Original patch resolves some LFSCK 1.5 technical debts, including:
1) Check and remove repeated linkea entries.
2) Merge some "goto" branches to make the code more readable.
3) Some comments about object's nlink inconsistency processing.
This patch picks the obd flags change.
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-2915
Lustre-change: http://review.whamcloud.com/6344
Signed-off-by: Fan Yong <fan.yong@intel.com>
Reviewed-by: Alex Zhuravlev <alexey.zhuravlev@intel.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>
Signed-off-by: Peng Tao <tao.peng@emc.com>
Signed-off-by: Andreas Dilger <andreas.dilger@intel.com>
---
.../staging/lustre/lustre/include/obd_support.h | 1 +
1 file changed, 1 insertion(+)
diff --git a/drivers/staging/lustre/lustre/include/obd_support.h b/drivers/staging/lustre/lustre/include/obd_support.h
index b5d40af..03e6133 100644
--- a/drivers/staging/lustre/lustre/include/obd_support.h
+++ b/drivers/staging/lustre/lustre/include/obd_support.h
@@ -470,6 +470,7 @@ int obd_alloc_fail(const void *ptr, const char *name, const char *type,
#define OBD_FAIL_LFSCK_DELAY3 0x1602
#define OBD_FAIL_LFSCK_LINKEA_CRASH 0x1603
#define OBD_FAIL_LFSCK_LINKEA_MORE 0x1604
+#define OBD_FAIL_LFSCK_LINKEA_MORE2 0x1605
#define OBD_FAIL_LFSCK_FATAL1 0x1608
#define OBD_FAIL_LFSCK_FATAL2 0x1609
#define OBD_FAIL_LFSCK_CRASH 0x160a
--
1.7.9.5
^ permalink raw reply related [flat|nested] 62+ messages in thread* [PATCH 26/48] staging/lustre/osc: Check return code for lu_kmem_init
2013-07-22 16:06 [PATCH 00/48] staging/lustre: minor cleanup and Intel Lustre sync Peng Tao
` (24 preceding siblings ...)
2013-07-22 16:06 ` [PATCH 25/48] staging/lustre/lfsck: LFSCK 1.5 technical debts (3) Peng Tao
@ 2013-07-22 16:06 ` Peng Tao
2013-07-22 16:06 ` [PATCH 27/48] staging/lustre/ptlrpc: Race between start and stop service threads Peng Tao
` (22 subsequent siblings)
48 siblings, 0 replies; 62+ messages in thread
From: Peng Tao @ 2013-07-22 16:06 UTC (permalink / raw)
To: Greg Kroah-Hartman; +Cc: linux-kernel, Keith Mannthey, Peng Tao, Andreas Dilger
From: Keith Mannthey <keith.mannthey@intel.com>
lu_kmem_init can fail and returns has a return code.
Check for this return code in lu_kmem_init.
This issue was found during 2gb VM Racer testing
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-3063
Lustre-change: http://review.whamcloud.com/6514
Signed-off-by: Keith Mannthey <keith.mannthey@intel.com>
Reviewed-by: Mike Pershin <mike.pershin@intel.com>
Reviewed-by: Nathaniel Clark <nathaniel.l.clark@intel.com>
Reviewed-by: Bob Glossman <bob.glossman@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>
Signed-off-by: Peng Tao <tao.peng@emc.com>
Signed-off-by: Andreas Dilger <andreas.dilger@intel.com>
---
drivers/staging/lustre/lustre/osc/osc_request.c | 2 ++
1 file changed, 2 insertions(+)
diff --git a/drivers/staging/lustre/lustre/osc/osc_request.c b/drivers/staging/lustre/lustre/osc/osc_request.c
index 0c0e07b..fea04ca 100644
--- a/drivers/staging/lustre/lustre/osc/osc_request.c
+++ b/drivers/staging/lustre/lustre/osc/osc_request.c
@@ -3681,6 +3681,8 @@ int __init osc_init(void)
CDEBUG(D_INFO, "Lustre OSC module (%p).\n", &osc_caches);
rc = lu_kmem_init(osc_caches);
+ if (rc)
+ RETURN(rc);
lprocfs_osc_init_vars(&lvars);
--
1.7.9.5
^ permalink raw reply related [flat|nested] 62+ messages in thread* [PATCH 27/48] staging/lustre/ptlrpc: Race between start and stop service threads
2013-07-22 16:06 [PATCH 00/48] staging/lustre: minor cleanup and Intel Lustre sync Peng Tao
` (25 preceding siblings ...)
2013-07-22 16:06 ` [PATCH 26/48] staging/lustre/osc: Check return code for lu_kmem_init Peng Tao
@ 2013-07-22 16:06 ` Peng Tao
2013-07-22 16:06 ` [PATCH 28/48] staging/lustre/crypto: add crc32c module loading to libcfs Peng Tao
` (21 subsequent siblings)
48 siblings, 0 replies; 62+ messages in thread
From: Peng Tao @ 2013-07-22 16:06 UTC (permalink / raw)
To: Greg Kroah-Hartman; +Cc: linux-kernel, Hiroya Nozaki, Peng Tao, Andreas Dilger
From: Hiroya Nozaki <nozaki.hiroya@jp.fujitsu.com>
When ptlrpc_start_thread fails to create a new thread, it will
finalize and free a struct ptlrpc_thread created and used here.
Considering this, it can be a problem when ptlrpc_svcpt_stop_thread
is driven and handles the struct ptlrpc_thread right after or right
before failure of cfs_create_thread. Because this situation let
the both of ptlrpc_start_thread and ptlrpc_svcpt_stop_threads
access the freed ptlrpc_thread and cause OS panic. Or, it may
happen that ptlrpc_svcpt_stop_threads waits forever holding an
already-freed waitq.
This patch adds an error handling into ptlrpc_start_thread to fix
this problem.
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-2889
Lustre-change: http://review.whamcloud.com/5552
Signed-off-by: Hiroya Nozaki <nozaki.hiroya@jp.fujitsu.com>
Reviewed-by: Liang Zhen <liang.zhen@intel.com>
Reviewed-by: Nikitas Angelinas <nikitas_angelinas@xyratex.com>
Reviewed-by: Keith Mannthey <keith.mannthey@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>
Signed-off-by: Peng Tao <tao.peng@emc.com>
Signed-off-by: Andreas Dilger <andreas.dilger@intel.com>
---
drivers/staging/lustre/lustre/ptlrpc/service.c | 16 ++++++++++++----
1 file changed, 12 insertions(+), 4 deletions(-)
diff --git a/drivers/staging/lustre/lustre/ptlrpc/service.c b/drivers/staging/lustre/lustre/ptlrpc/service.c
index 6871711..87bd638 100644
--- a/drivers/staging/lustre/lustre/ptlrpc/service.c
+++ b/drivers/staging/lustre/lustre/ptlrpc/service.c
@@ -2765,11 +2765,19 @@ int ptlrpc_start_thread(struct ptlrpc_service_part *svcpt, int wait)
CERROR("cannot start thread '%s': rc %d\n",
thread->t_name, rc);
spin_lock(&svcpt->scp_lock);
- list_del(&thread->t_link);
--svcpt->scp_nthrs_starting;
- spin_unlock(&svcpt->scp_lock);
-
- OBD_FREE(thread, sizeof(*thread));
+ if (thread_is_stopping(thread)) {
+ /* this ptlrpc_thread is being hanled
+ * by ptlrpc_svcpt_stop_threads now
+ */
+ thread_add_flags(thread, SVC_STOPPED);
+ wake_up(&thread->t_ctl_waitq);
+ spin_unlock(&svcpt->scp_lock);
+ } else {
+ list_del(&thread->t_link);
+ spin_unlock(&svcpt->scp_lock);
+ OBD_FREE_PTR(thread);
+ }
RETURN(rc);
}
--
1.7.9.5
^ permalink raw reply related [flat|nested] 62+ messages in thread* [PATCH 28/48] staging/lustre/crypto: add crc32c module loading to libcfs
2013-07-22 16:06 [PATCH 00/48] staging/lustre: minor cleanup and Intel Lustre sync Peng Tao
` (26 preceding siblings ...)
2013-07-22 16:06 ` [PATCH 27/48] staging/lustre/ptlrpc: Race between start and stop service threads Peng Tao
@ 2013-07-22 16:06 ` Peng Tao
2013-07-22 16:06 ` [PATCH 29/48] staging/lustre/mdt: duplicate link names in directory Peng Tao
` (20 subsequent siblings)
48 siblings, 0 replies; 62+ messages in thread
From: Peng Tao @ 2013-07-22 16:06 UTC (permalink / raw)
To: Greg Kroah-Hartman
Cc: linux-kernel, Alexander.Boyko, Peng Tao, Andreas Dilger
From: "Alexander.Boyko" <alexander_boyko@xyratex.com>
This patch add automatically module loading for crc32c
when libcfs is starting.
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-2212
Lustre-change: http://review.whamcloud.com/4372
Signed-off-by: Alexander Boyko <alexander_boyko@xyratex.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Reviewed-by: Shuichi Ihara <sihara@ddn.com>
Reviewed-by: James Simmons <uja.ornl@gmail.com>
Signed-off-by: Peng Tao <tao.peng@emc.com>
Signed-off-by: Andreas Dilger <andreas.dilger@intel.com>
---
.../lustre/lustre/libcfs/linux/linux-crypto.c | 2 ++
1 file changed, 2 insertions(+)
diff --git a/drivers/staging/lustre/lustre/libcfs/linux/linux-crypto.c b/drivers/staging/lustre/lustre/libcfs/linux/linux-crypto.c
index 8e35777..b6c79bc 100644
--- a/drivers/staging/lustre/lustre/libcfs/linux/linux-crypto.c
+++ b/drivers/staging/lustre/lustre/libcfs/linux/linux-crypto.c
@@ -274,6 +274,8 @@ static int adler32;
int cfs_crypto_register(void)
{
+ request_module("crc32c");
+
adler32 = cfs_crypto_adler32_register();
/* check all algorithms and do performance test */
--
1.7.9.5
^ permalink raw reply related [flat|nested] 62+ messages in thread* [PATCH 29/48] staging/lustre/mdt: duplicate link names in directory
2013-07-22 16:06 [PATCH 00/48] staging/lustre: minor cleanup and Intel Lustre sync Peng Tao
` (27 preceding siblings ...)
2013-07-22 16:06 ` [PATCH 28/48] staging/lustre/crypto: add crc32c module loading to libcfs Peng Tao
@ 2013-07-22 16:06 ` Peng Tao
2013-07-22 16:06 ` [PATCH 30/48] staging/lustre/llite: call simple_setattr() from ll_md_setattr() Peng Tao
` (19 subsequent siblings)
48 siblings, 0 replies; 62+ messages in thread
From: Peng Tao @ 2013-07-22 16:06 UTC (permalink / raw)
To: Greg Kroah-Hartman; +Cc: linux-kernel, Andreas Dilger, Lai Siyao, Peng Tao
From: Andreas Dilger <andreas.dilger@intel.com>
When creating a hard link to a file, the MDT/MDD/OSD code does
not verify whether the target link name already exists in the
directory. The ZFS ZAP code checks for duplicate entries. The
add_dirent_to_buf() function in ldiskfs only checks entries for
duplicates while it is traversing the leaf block looking for free
space. Even if it scanned the whole leaf block, this would not
work for non-htree directories since there is no guarantee that
the name is being inserted into the same leaf block.
To fix this, link should check target object doesn't exist as
other creat operations.
Add sanity.sh test_31o with multiple threads racing to link a new
name into the directory, while ensuring that there is a free entry
in the leaf block that is large enough to hold the duplicate name.
This needs to be racy, because otherwise the client VFS will see
the existing name and not send the RPC to the MDS, hiding the bug.
Add DLDLMRES/PLDLMRES macros for printing the whole lock resource
name (including the name hash) in LDLM_DEBUG() messages in a format
similar to DFID/PFID so they can be found in debug logs more easily.
The patch pickes client side change of the original patch, which only
contains the DLM printk part.
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-2901
Lustre-change: http://review.whamcloud.com/6591
Signed-off-by: Andreas Dilger <andreas.dilger@intel.com>
Signed-off-by: Lai Siyao <lai.siyao@intel.com>
Reviewed-by: Alex Zhuravlev <alexey.zhuravlev@intel.com>
Reviewed-by: Mike Pershin <mike.pershin@intel.com>
Signed-off-by: Peng Tao <tao.peng@emc.com>
Signed-off-by: Andreas Dilger <andreas.dilger@intel.com>
---
.../lustre/lustre/include/lustre/lustre_idl.h | 4 +
drivers/staging/lustre/lustre/ldlm/ldlm_lock.c | 145 ++++++++++----------
2 files changed, 76 insertions(+), 73 deletions(-)
diff --git a/drivers/staging/lustre/lustre/include/lustre/lustre_idl.h b/drivers/staging/lustre/lustre/include/lustre/lustre_idl.h
index ace4e18..3aaa869 100644
--- a/drivers/staging/lustre/lustre/include/lustre/lustre_idl.h
+++ b/drivers/staging/lustre/lustre/include/lustre/lustre_idl.h
@@ -2690,6 +2690,10 @@ struct ldlm_res_id {
__u64 name[RES_NAME_SIZE];
};
+#define DLDLMRES "["LPX64":"LPX64":"LPX64"]."LPX64i
+#define PLDLMRES(res) (res)->lr_name.name[0], (res)->lr_name.name[1], \
+ (res)->lr_name.name[2], (res)->lr_name.name[3]
+
extern void lustre_swab_ldlm_res_id (struct ldlm_res_id *id);
static inline int ldlm_res_eq(const struct ldlm_res_id *res0,
diff --git a/drivers/staging/lustre/lustre/ldlm/ldlm_lock.c b/drivers/staging/lustre/lustre/ldlm/ldlm_lock.c
index c10ba9c..ce1add1 100644
--- a/drivers/staging/lustre/lustre/ldlm/ldlm_lock.c
+++ b/drivers/staging/lustre/lustre/ldlm/ldlm_lock.c
@@ -2336,91 +2336,90 @@ void _ldlm_lock_debug(struct ldlm_lock *lock,
switch (resource->lr_type) {
case LDLM_EXTENT:
libcfs_debug_vmsg2(msgdata, fmt, args,
- " ns: %s lock: %p/"LPX64" lrc: %d/%d,%d mode: %s/%s "
- "res: "LPU64"/"LPU64" rrc: %d type: %s ["LPU64"->"LPU64
- "] (req "LPU64"->"LPU64") flags: "LPX64" nid: %s remote:"
- " "LPX64" expref: %d pid: %u timeout: %lu lvb_type: %d\n",
- ldlm_lock_to_ns_name(lock), lock,
- lock->l_handle.h_cookie, atomic_read(&lock->l_refc),
- lock->l_readers, lock->l_writers,
- ldlm_lockname[lock->l_granted_mode],
- ldlm_lockname[lock->l_req_mode],
- resource->lr_name.name[0],
- resource->lr_name.name[1],
- atomic_read(&resource->lr_refcount),
- ldlm_typename[resource->lr_type],
- lock->l_policy_data.l_extent.start,
- lock->l_policy_data.l_extent.end,
- lock->l_req_extent.start, lock->l_req_extent.end,
- lock->l_flags, nid, lock->l_remote_handle.cookie,
- exp ? atomic_read(&exp->exp_refcount) : -99,
- lock->l_pid, lock->l_callback_timeout, lock->l_lvb_type);
+ " ns: %s lock: %p/"LPX64" lrc: %d/%d,%d mode: %s/%s "
+ "res: "DLDLMRES" rrc: %d type: %s ["LPU64"->"LPU64"] "
+ "(req "LPU64"->"LPU64") flags: "LPX64" nid: %s remote: "
+ LPX64" expref: %d pid: %u timeout: %lu lvb_type: %d\n",
+ ldlm_lock_to_ns_name(lock), lock,
+ lock->l_handle.h_cookie, atomic_read(&lock->l_refc),
+ lock->l_readers, lock->l_writers,
+ ldlm_lockname[lock->l_granted_mode],
+ ldlm_lockname[lock->l_req_mode],
+ PLDLMRES(resource),
+ atomic_read(&resource->lr_refcount),
+ ldlm_typename[resource->lr_type],
+ lock->l_policy_data.l_extent.start,
+ lock->l_policy_data.l_extent.end,
+ lock->l_req_extent.start, lock->l_req_extent.end,
+ lock->l_flags, nid, lock->l_remote_handle.cookie,
+ exp ? atomic_read(&exp->exp_refcount) : -99,
+ lock->l_pid, lock->l_callback_timeout,
+ lock->l_lvb_type);
break;
case LDLM_FLOCK:
libcfs_debug_vmsg2(msgdata, fmt, args,
- " ns: %s lock: %p/"LPX64" lrc: %d/%d,%d mode: %s/%s "
- "res: "LPU64"/"LPU64" rrc: %d type: %s pid: %d "
- "["LPU64"->"LPU64"] flags: "LPX64" nid: %s remote: "LPX64
- " expref: %d pid: %u timeout: %lu\n",
- ldlm_lock_to_ns_name(lock), lock,
- lock->l_handle.h_cookie, atomic_read(&lock->l_refc),
- lock->l_readers, lock->l_writers,
- ldlm_lockname[lock->l_granted_mode],
- ldlm_lockname[lock->l_req_mode],
- resource->lr_name.name[0],
- resource->lr_name.name[1],
- atomic_read(&resource->lr_refcount),
- ldlm_typename[resource->lr_type],
- lock->l_policy_data.l_flock.pid,
- lock->l_policy_data.l_flock.start,
- lock->l_policy_data.l_flock.end,
- lock->l_flags, nid, lock->l_remote_handle.cookie,
- exp ? atomic_read(&exp->exp_refcount) : -99,
- lock->l_pid, lock->l_callback_timeout);
+ " ns: %s lock: %p/"LPX64" lrc: %d/%d,%d mode: %s/%s "
+ "res: "DLDLMRES" rrc: %d type: %s pid: %d "
+ "["LPU64"->"LPU64"] flags: "LPX64" nid: %s "
+ "remote: "LPX64" expref: %d pid: %u timeout: %lu\n",
+ ldlm_lock_to_ns_name(lock), lock,
+ lock->l_handle.h_cookie, atomic_read(&lock->l_refc),
+ lock->l_readers, lock->l_writers,
+ ldlm_lockname[lock->l_granted_mode],
+ ldlm_lockname[lock->l_req_mode],
+ PLDLMRES(resource),
+ atomic_read(&resource->lr_refcount),
+ ldlm_typename[resource->lr_type],
+ lock->l_policy_data.l_flock.pid,
+ lock->l_policy_data.l_flock.start,
+ lock->l_policy_data.l_flock.end,
+ lock->l_flags, nid, lock->l_remote_handle.cookie,
+ exp ? atomic_read(&exp->exp_refcount) : -99,
+ lock->l_pid, lock->l_callback_timeout);
break;
case LDLM_IBITS:
libcfs_debug_vmsg2(msgdata, fmt, args,
- " ns: %s lock: %p/"LPX64" lrc: %d/%d,%d mode: %s/%s "
- "res: "LPU64"/"LPU64" bits "LPX64" rrc: %d type: %s "
- "flags: "LPX64" nid: %s remote: "LPX64" expref: %d "
- "pid: %u timeout: %lu lvb_type: %d\n",
- ldlm_lock_to_ns_name(lock),
- lock, lock->l_handle.h_cookie,
- atomic_read (&lock->l_refc),
- lock->l_readers, lock->l_writers,
- ldlm_lockname[lock->l_granted_mode],
- ldlm_lockname[lock->l_req_mode],
- resource->lr_name.name[0],
- resource->lr_name.name[1],
- lock->l_policy_data.l_inodebits.bits,
- atomic_read(&resource->lr_refcount),
- ldlm_typename[resource->lr_type],
- lock->l_flags, nid, lock->l_remote_handle.cookie,
- exp ? atomic_read(&exp->exp_refcount) : -99,
- lock->l_pid, lock->l_callback_timeout, lock->l_lvb_type);
+ " ns: %s lock: %p/"LPX64" lrc: %d/%d,%d mode: %s/%s "
+ "res: "DLDLMRES" bits "LPX64" rrc: %d type: %s "
+ "flags: "LPX64" nid: %s remote: "LPX64" expref: %d "
+ "pid: %u timeout: %lu lvb_type: %d\n",
+ ldlm_lock_to_ns_name(lock),
+ lock, lock->l_handle.h_cookie,
+ atomic_read(&lock->l_refc),
+ lock->l_readers, lock->l_writers,
+ ldlm_lockname[lock->l_granted_mode],
+ ldlm_lockname[lock->l_req_mode],
+ PLDLMRES(resource),
+ lock->l_policy_data.l_inodebits.bits,
+ atomic_read(&resource->lr_refcount),
+ ldlm_typename[resource->lr_type],
+ lock->l_flags, nid, lock->l_remote_handle.cookie,
+ exp ? atomic_read(&exp->exp_refcount) : -99,
+ lock->l_pid, lock->l_callback_timeout,
+ lock->l_lvb_type);
break;
default:
libcfs_debug_vmsg2(msgdata, fmt, args,
- " ns: %s lock: %p/"LPX64" lrc: %d/%d,%d mode: %s/%s "
- "res: "LPU64"/"LPU64" rrc: %d type: %s flags: "LPX64" "
- "nid: %s remote: "LPX64" expref: %d pid: %u timeout: %lu"
- "lvb_type: %d\n",
- ldlm_lock_to_ns_name(lock),
- lock, lock->l_handle.h_cookie,
- atomic_read (&lock->l_refc),
- lock->l_readers, lock->l_writers,
- ldlm_lockname[lock->l_granted_mode],
- ldlm_lockname[lock->l_req_mode],
- resource->lr_name.name[0],
- resource->lr_name.name[1],
- atomic_read(&resource->lr_refcount),
- ldlm_typename[resource->lr_type],
- lock->l_flags, nid, lock->l_remote_handle.cookie,
- exp ? atomic_read(&exp->exp_refcount) : -99,
- lock->l_pid, lock->l_callback_timeout, lock->l_lvb_type);
+ " ns: %s lock: %p/"LPX64" lrc: %d/%d,%d mode: %s/%s "
+ "res: "DLDLMRES" rrc: %d type: %s flags: "LPX64" "
+ "nid: %s remote: "LPX64" expref: %d pid: %u "
+ "timeout: %lu lvb_type: %d\n",
+ ldlm_lock_to_ns_name(lock),
+ lock, lock->l_handle.h_cookie,
+ atomic_read(&lock->l_refc),
+ lock->l_readers, lock->l_writers,
+ ldlm_lockname[lock->l_granted_mode],
+ ldlm_lockname[lock->l_req_mode],
+ PLDLMRES(resource),
+ atomic_read(&resource->lr_refcount),
+ ldlm_typename[resource->lr_type],
+ lock->l_flags, nid, lock->l_remote_handle.cookie,
+ exp ? atomic_read(&exp->exp_refcount) : -99,
+ lock->l_pid, lock->l_callback_timeout,
+ lock->l_lvb_type);
break;
}
va_end(args);
--
1.7.9.5
^ permalink raw reply related [flat|nested] 62+ messages in thread* [PATCH 30/48] staging/lustre/llite: call simple_setattr() from ll_md_setattr()
2013-07-22 16:06 [PATCH 00/48] staging/lustre: minor cleanup and Intel Lustre sync Peng Tao
` (28 preceding siblings ...)
2013-07-22 16:06 ` [PATCH 29/48] staging/lustre/mdt: duplicate link names in directory Peng Tao
@ 2013-07-22 16:06 ` Peng Tao
2013-07-22 16:06 ` [PATCH 31/48] staging/lustre/ldlm: Fix flock deadlock detection race Peng Tao
` (18 subsequent siblings)
48 siblings, 0 replies; 62+ messages in thread
From: Peng Tao @ 2013-07-22 16:06 UTC (permalink / raw)
To: Greg Kroah-Hartman
Cc: linux-kernel, John L. Hammond, Peng Tao, Andreas Dilger
From: "John L. Hammond" <john.hammond@intel.com>
This partially reverts the change from "LU-2482 layout: introduce new
layout for released files" by calling simple_setattr() from
ll_md_setattr() without ATTR_SIZE set. Doing so avoids failed
assertions in osc_page_delete(). Disable truncates on released files
and modify sanity 229 accordingly.
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-3448
Lustre-change: http://review.whamcloud.com/6643
Signed-off-by: John L. Hammond <john.hammond@intel.com>
Reviewed-by: jacques-Charles Lafoucriere <jacques-charles.lafoucriere@cea.fr>
Reviewed-by: Jinshan Xiong <jinshan.xiong@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>
Signed-off-by: Peng Tao <tao.peng@emc.com>
Signed-off-by: Andreas Dilger <andreas.dilger@intel.com>
---
drivers/staging/lustre/lustre/llite/llite_lib.c | 18 ++++++++----------
drivers/staging/lustre/lustre/lov/lov_io.c | 2 +-
2 files changed, 9 insertions(+), 11 deletions(-)
diff --git a/drivers/staging/lustre/lustre/llite/llite_lib.c b/drivers/staging/lustre/lustre/llite/llite_lib.c
index b9cbe95..39d519b 100644
--- a/drivers/staging/lustre/lustre/llite/llite_lib.c
+++ b/drivers/staging/lustre/lustre/llite/llite_lib.c
@@ -1282,6 +1282,13 @@ int ll_md_setattr(struct dentry *dentry, struct md_op_data *op_data,
RETURN(rc);
}
+ ia_valid = op_data->op_attr.ia_valid;
+ /* inode size will be in ll_setattr_ost, can't do it now since dirty
+ * cache is not cleared yet. */
+ op_data->op_attr.ia_valid &= ~(TIMES_SET_FLAGS | ATTR_SIZE);
+ rc = simple_setattr(dentry, &op_data->op_attr);
+ op_data->op_attr.ia_valid = ia_valid;
+
/* Extract epoch data if obtained. */
op_data->op_handle = md.body->handle;
op_data->op_ioepoch = md.body->ioepoch;
@@ -1364,7 +1371,6 @@ int ll_setattr_raw(struct dentry *dentry, struct iattr *attr)
struct ll_inode_info *lli = ll_i2info(inode);
struct md_op_data *op_data = NULL;
struct md_open_data *mod = NULL;
- unsigned int valid = attr->ia_valid;
int rc = 0, rc1 = 0;
ENTRY;
@@ -1476,13 +1482,6 @@ int ll_setattr_raw(struct dentry *dentry, struct iattr *attr)
rc = ll_setattr_ost(inode, attr);
EXIT;
out:
- if (rc == 0) {
- /* Update inode attribute after dirty cache is cleaned
- * by truncating OST objects. */
- attr->ia_valid |= ATTR_FORCE;
- rc = simple_setattr(dentry, attr);
- LASSERT(rc == 0);
- }
if (op_data) {
if (op_data->op_ioepoch) {
rc1 = ll_setattr_done_writing(inode, op_data, mod);
@@ -1501,8 +1500,7 @@ out:
ll_stats_ops_tally(ll_i2sbi(inode), (attr->ia_valid & ATTR_SIZE) ?
LPROC_LL_TRUNC : LPROC_LL_SETATTR, 1);
- attr->ia_valid = valid;
- RETURN(rc);
+ return rc;
}
int ll_setattr(struct dentry *de, struct iattr *attr)
diff --git a/drivers/staging/lustre/lustre/lov/lov_io.c b/drivers/staging/lustre/lustre/lov/lov_io.c
index e78cdb7..b316ce8 100644
--- a/drivers/staging/lustre/lustre/lov/lov_io.c
+++ b/drivers/staging/lustre/lustre/lov/lov_io.c
@@ -980,9 +980,9 @@ int lov_io_init_released(const struct lu_env *env, struct cl_object *obj,
LASSERTF(0, "invalid type %d\n", io->ci_type);
case CIT_MISC:
case CIT_FSYNC:
- case CIT_SETATTR:
result = +1;
break;
+ case CIT_SETATTR:
case CIT_READ:
case CIT_WRITE:
case CIT_FAULT:
--
1.7.9.5
^ permalink raw reply related [flat|nested] 62+ messages in thread* [PATCH 31/48] staging/lustre/ldlm: Fix flock deadlock detection race
2013-07-22 16:06 [PATCH 00/48] staging/lustre: minor cleanup and Intel Lustre sync Peng Tao
` (29 preceding siblings ...)
2013-07-22 16:06 ` [PATCH 30/48] staging/lustre/llite: call simple_setattr() from ll_md_setattr() Peng Tao
@ 2013-07-22 16:06 ` Peng Tao
2013-07-22 16:06 ` [PATCH 32/48] staging/lustre/lnet: remove empty file lnet/lnet/api-errno.c Peng Tao
` (17 subsequent siblings)
48 siblings, 0 replies; 62+ messages in thread
From: Peng Tao @ 2013-07-22 16:06 UTC (permalink / raw)
To: Greg Kroah-Hartman; +Cc: linux-kernel, Andriy Skulysh, Peng Tao, Andreas Dilger
From: Andriy Skulysh <Andriy_Skulysh@xyratex.com>
Deadlock isn't detected if 2 threads are trying to
grant 2 locks which deadlock on each other.
They call ldlm_flock_deadlock() simultaneously
and deadlock ins't detected.
The soulition is to add lock to blocking list before
calling ldlm_flock_deadlock()
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-1602
Lustre-change: http://review.whamcloud.com/3277
Signed-off-by: Andriy Skulysh <Andriy_Skulysh@xyratex.com>
Reviewed-by: Vitaly Fertman <vitaly_fertman@xyratex.com>
Reviewed-by: Bruce Korb <bruce_korb@xyratex.com>
Reviewed-by: Keith Mannthey <keith.mannthey@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>
Signed-off-by: Peng Tao <tao.peng@emc.com>
Signed-off-by: Andreas Dilger <andreas.dilger@intel.com>
---
drivers/staging/lustre/lustre/ldlm/ldlm_flock.c | 16 ++++++++++------
1 file changed, 10 insertions(+), 6 deletions(-)
diff --git a/drivers/staging/lustre/lustre/ldlm/ldlm_flock.c b/drivers/staging/lustre/lustre/ldlm/ldlm_flock.c
index aca1073..39973cb 100644
--- a/drivers/staging/lustre/lustre/ldlm/ldlm_flock.c
+++ b/drivers/staging/lustre/lustre/ldlm/ldlm_flock.c
@@ -198,6 +198,7 @@ ldlm_flock_deadlock(struct ldlm_lock *req, struct ldlm_lock *bl_lock)
if (lock == NULL)
break;
+ LASSERT(req != lock);
flock = &lock->l_policy_data.l_flock;
LASSERT(flock->owner == bl_owner);
bl_owner = flock->blocking_owner;
@@ -329,18 +330,21 @@ reprocess:
RETURN(LDLM_ITER_STOP);
}
- if (ldlm_flock_deadlock(req, lock)) {
- ldlm_flock_destroy(req, mode, *flags);
- *err = -EDEADLK;
- RETURN(LDLM_ITER_STOP);
- }
-
+ /* add lock to blocking list before deadlock
+ * check to prevent race */
rc = ldlm_flock_blocking_link(req, lock);
if (rc) {
ldlm_flock_destroy(req, mode, *flags);
*err = rc;
RETURN(LDLM_ITER_STOP);
}
+ if (ldlm_flock_deadlock(req, lock)) {
+ ldlm_flock_blocking_unlink(req);
+ ldlm_flock_destroy(req, mode, *flags);
+ *err = -EDEADLK;
+ RETURN(LDLM_ITER_STOP);
+ }
+
ldlm_resource_add_lock(res, &res->lr_waiting, req);
*flags |= LDLM_FL_BLOCK_GRANTED;
RETURN(LDLM_ITER_STOP);
--
1.7.9.5
^ permalink raw reply related [flat|nested] 62+ messages in thread* [PATCH 32/48] staging/lustre/lnet: remove empty file lnet/lnet/api-errno.c
2013-07-22 16:06 [PATCH 00/48] staging/lustre: minor cleanup and Intel Lustre sync Peng Tao
` (30 preceding siblings ...)
2013-07-22 16:06 ` [PATCH 31/48] staging/lustre/ldlm: Fix flock deadlock detection race Peng Tao
@ 2013-07-22 16:06 ` Peng Tao
2013-07-22 16:06 ` [PATCH 33/48] staging/lustre/fld: prepare FLD module for client server split Peng Tao
` (16 subsequent siblings)
48 siblings, 0 replies; 62+ messages in thread
From: Peng Tao @ 2013-07-22 16:06 UTC (permalink / raw)
To: Greg Kroah-Hartman; +Cc: linux-kernel, Peng Tao, Andreas Dilger
From: Peng Tao <tao.peng@emc.com>
The file is empty. We can just remove it.
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-2335
Lustre-change: http://review.whamcloud.com/5880
Reviewed-by: Liang Zhen <liang.zhen@intel.com>
Reviewed-by: Li Wei <wei.g.li@intel.com>
Reviewed-by: Isaac Huang <he.huang@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>
Signed-off-by: Peng Tao <tao.peng@emc.com>
Signed-off-by: Andreas Dilger <andreas.dilger@intel.com>
---
drivers/staging/lustre/lnet/lnet/Makefile | 2 +-
drivers/staging/lustre/lnet/lnet/api-errno.c | 39 --------------------------
2 files changed, 1 insertion(+), 40 deletions(-)
delete mode 100644 drivers/staging/lustre/lnet/lnet/api-errno.c
diff --git a/drivers/staging/lustre/lnet/lnet/Makefile b/drivers/staging/lustre/lnet/lnet/Makefile
index 1bd9ef7..b815fe1 100644
--- a/drivers/staging/lustre/lnet/lnet/Makefile
+++ b/drivers/staging/lustre/lnet/lnet/Makefile
@@ -1,6 +1,6 @@
obj-$(CONFIG_LNET) += lnet.o
-lnet-y := api-errno.o api-ni.o config.o lib-me.o lib-msg.o lib-eq.o \
+lnet-y := api-ni.o config.o lib-me.o lib-msg.o lib-eq.o \
lib-md.o lib-ptl.o lib-move.o module.o lo.o router.o \
router_proc.o acceptor.o peer.o
diff --git a/drivers/staging/lustre/lnet/lnet/api-errno.c b/drivers/staging/lustre/lnet/lnet/api-errno.c
deleted file mode 100644
index 695b272..0000000
--- a/drivers/staging/lustre/lnet/lnet/api-errno.c
+++ /dev/null
@@ -1,39 +0,0 @@
-/*
- * GPL HEADER START
- *
- * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
- *
- * 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 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).
- *
- * 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.
- *
- * GPL HEADER END
- */
-/*
- * Copyright (c) 2003, 2010, Oracle and/or its affiliates. All rights reserved.
- * Use is subject to license terms.
- */
-/*
- * This file is part of Lustre, http://www.lustre.org/
- * Lustre is a trademark of Sun Microsystems, Inc.
- *
- * lnet/lnet/api-errno.c
- *
- * Instantiate the string table of errors
- */
-
-/* If you change these, you must update the number table in portals/errno.h */
--
1.7.9.5
^ permalink raw reply related [flat|nested] 62+ messages in thread* [PATCH 33/48] staging/lustre/fld: prepare FLD module for client server split
2013-07-22 16:06 [PATCH 00/48] staging/lustre: minor cleanup and Intel Lustre sync Peng Tao
` (31 preceding siblings ...)
2013-07-22 16:06 ` [PATCH 32/48] staging/lustre/lnet: remove empty file lnet/lnet/api-errno.c Peng Tao
@ 2013-07-22 16:06 ` Peng Tao
2013-07-22 16:06 ` [PATCH 34/48] staging/lustre/lmv: support DNE with HSM Peng Tao
` (15 subsequent siblings)
48 siblings, 0 replies; 62+ messages in thread
From: Peng Tao @ 2013-07-22 16:06 UTC (permalink / raw)
To: Greg Kroah-Hartman
Cc: linux-kernel, Liu Xuezhao, John L. Hammond, Peng Tao, Andreas Dilger
From: Liu Xuezhao <xuezhao.liu@emc.com>
Split FLD server from client, fld_{handler,index}.c are not compliled
unless server support is enabled. Do not include dt_object.h or
lustre_mdt.h in lustre_fld.h and fix the minor breakages caused by
this elsewhere. Generally cleanup includes in lustre/fld.
Signed-off-by: Liu Xuezhao <xuezhao.liu@emc.com>
Signed-off-by: John L. Hammond <john.hammond@intel.com>
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-1330
Lustre-change: http://review.whamcloud.com/2675
Reviewed-by: Nathaniel Clark <nathaniel.l.clark@intel.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Signed-off-by: Peng Tao <tao.peng@emc.com>
Signed-off-by: Andreas Dilger <andreas.dilger@intel.com>
---
drivers/staging/lustre/lustre/fld/Makefile | 2 +-
drivers/staging/lustre/lustre/fld/fld_handler.c | 447 --------------------
drivers/staging/lustre/lustre/fld/fld_index.c | 426 -------------------
drivers/staging/lustre/lustre/fld/fld_internal.h | 29 --
drivers/staging/lustre/lustre/fld/fld_request.c | 40 +-
drivers/staging/lustre/lustre/fld/lproc_fld.c | 205 +--------
drivers/staging/lustre/lustre/include/lustre_fld.h | 41 --
drivers/staging/lustre/lustre/include/lustre_mdt.h | 84 ----
.../staging/lustre/lustre/obdecho/echo_client.c | 1 +
9 files changed, 37 insertions(+), 1238 deletions(-)
delete mode 100644 drivers/staging/lustre/lustre/fld/fld_handler.c
delete mode 100644 drivers/staging/lustre/lustre/fld/fld_index.c
delete mode 100644 drivers/staging/lustre/lustre/include/lustre_mdt.h
diff --git a/drivers/staging/lustre/lustre/fld/Makefile b/drivers/staging/lustre/lustre/fld/Makefile
index e7f2881..90d46d8 100644
--- a/drivers/staging/lustre/lustre/fld/Makefile
+++ b/drivers/staging/lustre/lustre/fld/Makefile
@@ -1,5 +1,5 @@
obj-$(CONFIG_LUSTRE_FS) += fld.o
-fld-y := fld_handler.o fld_request.o fld_cache.o fld_index.o lproc_fld.o
+fld-y := fld_request.o fld_cache.o lproc_fld.o
ccflags-y := -I$(src)/../include
diff --git a/drivers/staging/lustre/lustre/fld/fld_handler.c b/drivers/staging/lustre/lustre/fld/fld_handler.c
deleted file mode 100644
index d2707ae..0000000
--- a/drivers/staging/lustre/lustre/fld/fld_handler.c
+++ /dev/null
@@ -1,447 +0,0 @@
-/*
- * GPL HEADER START
- *
- * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
- *
- * 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 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).
- *
- * 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.
- *
- * 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.
- */
-/*
- * This file is part of Lustre, http://www.lustre.org/
- * Lustre is a trademark of Sun Microsystems, Inc.
- *
- * lustre/fld/fld_handler.c
- *
- * FLD (Fids Location Database)
- *
- * Author: Yury Umanets <umka@clusterfs.com>
- * Author: WangDi <wangdi@clusterfs.com>
- * Author: Pravin Shelar <pravin.shelar@sun.com>
- */
-
-#define DEBUG_SUBSYSTEM S_FLD
-
-# include <linux/libcfs/libcfs.h>
-# include <linux/module.h>
-# include <linux/jbd.h>
-# include <asm/div64.h>
-
-#include <obd.h>
-#include <obd_class.h>
-#include <lustre_ver.h>
-#include <obd_support.h>
-#include <lprocfs_status.h>
-
-#include <md_object.h>
-#include <lustre_fid.h>
-#include <lustre_req_layout.h>
-#include "fld_internal.h"
-#include <lustre_fid.h>
-
-
-/* context key constructor/destructor: fld_key_init, fld_key_fini */
-LU_KEY_INIT_FINI(fld, struct fld_thread_info);
-
-/* context key: fld_thread_key */
-LU_CONTEXT_KEY_DEFINE(fld, LCT_MD_THREAD | LCT_DT_THREAD | LCT_MG_THREAD);
-
-proc_dir_entry_t *fld_type_proc_dir = NULL;
-
-static int __init fld_mod_init(void)
-{
- 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);
-
- LU_CONTEXT_KEY_INIT(&fld_thread_key);
- lu_context_key_register(&fld_thread_key);
- return 0;
-}
-
-static void __exit fld_mod_exit(void)
-{
- lu_context_key_degister(&fld_thread_key);
- if (fld_type_proc_dir != NULL && !IS_ERR(fld_type_proc_dir)) {
- lprocfs_remove(&fld_type_proc_dir);
- fld_type_proc_dir = NULL;
- }
-}
-
-int fld_declare_server_create(const struct lu_env *env,
- struct lu_server_fld *fld,
- struct lu_seq_range *range,
- struct thandle *th)
-{
- int rc;
-
- rc = fld_declare_index_create(env, fld, range, th);
- RETURN(rc);
-}
-EXPORT_SYMBOL(fld_declare_server_create);
-
-/**
- * Insert FLD index entry and update FLD cache.
- *
- * This function is called from the sequence allocator when a super-sequence
- * is granted to a server.
- */
-int fld_server_create(const struct lu_env *env, struct lu_server_fld *fld,
- struct lu_seq_range *range, struct thandle *th)
-{
- int rc;
-
- mutex_lock(&fld->lsf_lock);
- rc = fld_index_create(env, fld, range, th);
- mutex_unlock(&fld->lsf_lock);
-
- RETURN(rc);
-}
-EXPORT_SYMBOL(fld_server_create);
-
-/**
- * Lookup mds by seq, returns a range for given seq.
- *
- * If that entry is not cached in fld cache, request is sent to super
- * sequence controller node (MDT0). All other MDT[1...N] and client
- * cache fld entries, but this cache is not persistent.
- */
-int fld_server_lookup(const struct lu_env *env, struct lu_server_fld *fld,
- seqno_t seq, struct lu_seq_range *range)
-{
- struct lu_seq_range *erange;
- struct fld_thread_info *info;
- int rc;
- ENTRY;
-
- info = lu_context_key_get(&env->le_ctx, &fld_thread_key);
- LASSERT(info != NULL);
- erange = &info->fti_lrange;
-
- /* Lookup it in the cache. */
- rc = fld_cache_lookup(fld->lsf_cache, seq, erange);
- if (rc == 0) {
- if (unlikely(fld_range_type(erange) != fld_range_type(range) &&
- !fld_range_is_any(range))) {
- CERROR("%s: FLD cache range "DRANGE" does not match"
- "requested flag %x: rc = %d\n", fld->lsf_name,
- PRANGE(erange), range->lsr_flags, -EIO);
- RETURN(-EIO);
- }
- *range = *erange;
- RETURN(0);
- }
-
- if (fld->lsf_obj) {
- /* On server side, all entries should be in cache.
- * If we can not find it in cache, just return error */
- CERROR("%s: Cannot find sequence "LPX64": rc = %d\n",
- fld->lsf_name, seq, -EIO);
- RETURN(-EIO);
- } else {
- LASSERT(fld->lsf_control_exp);
- /* send request to mdt0 i.e. super seq. controller.
- * This is temporary solution, long term solution is fld
- * replication on all mdt servers.
- */
- range->lsr_start = seq;
- rc = fld_client_rpc(fld->lsf_control_exp,
- range, FLD_LOOKUP);
- if (rc == 0)
- fld_cache_insert(fld->lsf_cache, range);
- }
- RETURN(rc);
-}
-EXPORT_SYMBOL(fld_server_lookup);
-
-/**
- * All MDT server handle fld lookup operation. But only MDT0 has fld index.
- * if entry is not found in cache we need to forward lookup request to MDT0
- */
-
-static int fld_server_handle(struct lu_server_fld *fld,
- const struct lu_env *env,
- __u32 opc, struct lu_seq_range *range,
- struct fld_thread_info *info)
-{
- int rc;
- ENTRY;
-
- switch (opc) {
- case FLD_LOOKUP:
- rc = fld_server_lookup(env, fld, range->lsr_start, range);
- break;
- default:
- rc = -EINVAL;
- break;
- }
-
- CDEBUG(D_INFO, "%s: FLD req handle: error %d (opc: %d, range: "
- DRANGE"\n", fld->lsf_name, rc, opc, PRANGE(range));
-
- RETURN(rc);
-
-}
-
-static int fld_req_handle(struct ptlrpc_request *req,
- struct fld_thread_info *info)
-{
- struct obd_export *exp = req->rq_export;
- struct lu_site *site = exp->exp_obd->obd_lu_dev->ld_site;
- struct lu_seq_range *in;
- struct lu_seq_range *out;
- int rc;
- __u32 *opc;
- ENTRY;
-
- rc = req_capsule_server_pack(info->fti_pill);
- if (rc)
- RETURN(err_serious(rc));
-
- opc = req_capsule_client_get(info->fti_pill, &RMF_FLD_OPC);
- if (opc != NULL) {
- in = req_capsule_client_get(info->fti_pill, &RMF_FLD_MDFLD);
- if (in == NULL)
- RETURN(err_serious(-EPROTO));
- out = req_capsule_server_get(info->fti_pill, &RMF_FLD_MDFLD);
- if (out == NULL)
- RETURN(err_serious(-EPROTO));
- *out = *in;
-
- /* For old 2.0 client, the 'lsr_flags' is uninitialized.
- * Set it as 'LU_SEQ_RANGE_MDT' by default. */
- if (!(exp_connect_flags(exp) & OBD_CONNECT_64BITHASH) &&
- !(exp_connect_flags(exp) & OBD_CONNECT_MDS_MDS) &&
- !(exp_connect_flags(exp) & OBD_CONNECT_LIGHTWEIGHT) &&
- !exp->exp_libclient)
- fld_range_set_mdt(out);
-
- rc = fld_server_handle(lu_site2seq(site)->ss_server_fld,
- req->rq_svc_thread->t_env,
- *opc, out, info);
- } else {
- rc = err_serious(-EPROTO);
- }
-
- RETURN(rc);
-}
-
-static void fld_thread_info_init(struct ptlrpc_request *req,
- struct fld_thread_info *info)
-{
- info->fti_pill = &req->rq_pill;
- /* Init request capsule. */
- req_capsule_init(info->fti_pill, req, RCL_SERVER);
- req_capsule_set(info->fti_pill, &RQF_FLD_QUERY);
-}
-
-static void fld_thread_info_fini(struct fld_thread_info *info)
-{
- req_capsule_fini(info->fti_pill);
-}
-
-static int fld_handle(struct ptlrpc_request *req)
-{
- struct fld_thread_info *info;
- const struct lu_env *env;
- int rc;
-
- env = req->rq_svc_thread->t_env;
- LASSERT(env != NULL);
-
- info = lu_context_key_get(&env->le_ctx, &fld_thread_key);
- LASSERT(info != NULL);
-
- fld_thread_info_init(req, info);
- rc = fld_req_handle(req, info);
- fld_thread_info_fini(info);
-
- return rc;
-}
-
-/*
- * Entry point for handling FLD RPCs called from MDT.
- */
-int fld_query(struct com_thread_info *info)
-{
- return fld_handle(info->cti_pill->rc_req);
-}
-EXPORT_SYMBOL(fld_query);
-
-/*
- * Returns true, if fid is local to this server node.
- *
- * WARNING: this function is *not* guaranteed to return false if fid is
- * remote: it makes an educated conservative guess only.
- *
- * fid_is_local() is supposed to be used in assertion checks only.
- */
-int fid_is_local(const struct lu_env *env,
- struct lu_site *site, const struct lu_fid *fid)
-{
- int result;
- struct seq_server_site *ss_site;
- struct lu_seq_range *range;
- struct fld_thread_info *info;
- ENTRY;
-
- info = lu_context_key_get(&env->le_ctx, &fld_thread_key);
- range = &info->fti_lrange;
-
- result = 1; /* conservatively assume fid is local */
- ss_site = lu_site2seq(site);
- if (ss_site->ss_client_fld != NULL) {
- int rc;
-
- rc = fld_cache_lookup(ss_site->ss_client_fld->lcf_cache,
- fid_seq(fid), range);
- if (rc == 0)
- result = (range->lsr_index == ss_site->ss_node_id);
- }
- return result;
-}
-EXPORT_SYMBOL(fid_is_local);
-
-static void fld_server_proc_fini(struct lu_server_fld *fld);
-
-#ifdef LPROCFS
-static int fld_server_proc_init(struct lu_server_fld *fld)
-{
- int rc = 0;
- ENTRY;
-
- fld->lsf_proc_dir = lprocfs_register(fld->lsf_name,
- fld_type_proc_dir,
- fld_server_proc_list, fld);
- if (IS_ERR(fld->lsf_proc_dir)) {
- rc = PTR_ERR(fld->lsf_proc_dir);
- RETURN(rc);
- }
-
- rc = lprocfs_seq_create(fld->lsf_proc_dir, "fldb", 0444,
- &fld_proc_seq_fops, fld);
- if (rc) {
- lprocfs_remove(&fld->lsf_proc_dir);
- fld->lsf_proc_dir = NULL;
- }
-
- RETURN(rc);
-}
-
-static void fld_server_proc_fini(struct lu_server_fld *fld)
-{
- ENTRY;
- if (fld->lsf_proc_dir != NULL) {
- if (!IS_ERR(fld->lsf_proc_dir))
- lprocfs_remove(&fld->lsf_proc_dir);
- fld->lsf_proc_dir = NULL;
- }
- EXIT;
-}
-#else
-static int fld_server_proc_init(struct lu_server_fld *fld)
-{
- return 0;
-}
-
-static void fld_server_proc_fini(struct lu_server_fld *fld)
-{
- return;
-}
-#endif
-
-int fld_server_init(const struct lu_env *env, struct lu_server_fld *fld,
- struct dt_device *dt, const char *prefix, int mds_node_id,
- int type)
-{
- int cache_size, cache_threshold;
- int rc;
- ENTRY;
-
- snprintf(fld->lsf_name, sizeof(fld->lsf_name),
- "srv-%s", prefix);
-
- cache_size = FLD_SERVER_CACHE_SIZE /
- sizeof(struct fld_cache_entry);
-
- cache_threshold = cache_size *
- FLD_SERVER_CACHE_THRESHOLD / 100;
-
- mutex_init(&fld->lsf_lock);
- fld->lsf_cache = fld_cache_init(fld->lsf_name,
- cache_size, cache_threshold);
- if (IS_ERR(fld->lsf_cache)) {
- rc = PTR_ERR(fld->lsf_cache);
- fld->lsf_cache = NULL;
- GOTO(out, rc);
- }
-
- if (!mds_node_id && type == LU_SEQ_RANGE_MDT) {
- rc = fld_index_init(env, fld, dt);
- if (rc)
- GOTO(out, rc);
- } else {
- fld->lsf_obj = NULL;
- }
-
- rc = fld_server_proc_init(fld);
- if (rc)
- GOTO(out, rc);
-
- fld->lsf_control_exp = NULL;
-
- GOTO(out, rc);
-
-out:
- if (rc)
- fld_server_fini(env, fld);
- return rc;
-}
-EXPORT_SYMBOL(fld_server_init);
-
-void fld_server_fini(const struct lu_env *env, struct lu_server_fld *fld)
-{
- ENTRY;
-
- fld_server_proc_fini(fld);
- fld_index_fini(env, fld);
-
- if (fld->lsf_cache != NULL) {
- if (!IS_ERR(fld->lsf_cache))
- fld_cache_fini(fld->lsf_cache);
- fld->lsf_cache = NULL;
- }
-
- EXIT;
-}
-EXPORT_SYMBOL(fld_server_fini);
-
-MODULE_AUTHOR("Sun Microsystems, Inc. <http://www.lustre.org/>");
-MODULE_DESCRIPTION("Lustre FLD");
-MODULE_LICENSE("GPL");
-
-cfs_module(mdd, "0.1.0", fld_mod_init, fld_mod_exit);
diff --git a/drivers/staging/lustre/lustre/fld/fld_index.c b/drivers/staging/lustre/lustre/fld/fld_index.c
deleted file mode 100644
index ec68a54..0000000
--- a/drivers/staging/lustre/lustre/fld/fld_index.c
+++ /dev/null
@@ -1,426 +0,0 @@
-/*
- * GPL HEADER START
- *
- * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
- *
- * 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 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).
- *
- * 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.
- *
- * 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.
- */
-/*
- * This file is part of Lustre, http://www.lustre.org/
- * Lustre is a trademark of Sun Microsystems, Inc.
- *
- * lustre/fld/fld_index.c
- *
- * Author: WangDi <wangdi@clusterfs.com>
- * Author: Yury Umanets <umka@clusterfs.com>
- */
-
-#define DEBUG_SUBSYSTEM S_FLD
-
-# include <linux/libcfs/libcfs.h>
-# include <linux/module.h>
-# include <linux/jbd.h>
-
-#include <obd.h>
-#include <obd_class.h>
-#include <lustre_ver.h>
-#include <obd_support.h>
-#include <lprocfs_status.h>
-
-#include <dt_object.h>
-#include <md_object.h>
-#include <lustre_mdc.h>
-#include <lustre_fid.h>
-#include <lustre_fld.h>
-#include "fld_internal.h"
-
-const char fld_index_name[] = "fld";
-
-static const struct lu_seq_range IGIF_FLD_RANGE = {
- .lsr_start = FID_SEQ_IGIF,
- .lsr_end = FID_SEQ_IGIF_MAX + 1,
- .lsr_index = 0,
- .lsr_flags = LU_SEQ_RANGE_MDT
-};
-
-static const struct lu_seq_range DOT_LUSTRE_FLD_RANGE = {
- .lsr_start = FID_SEQ_DOT_LUSTRE,
- .lsr_end = FID_SEQ_DOT_LUSTRE + 1,
- .lsr_index = 0,
- .lsr_flags = LU_SEQ_RANGE_MDT
-};
-
-static const struct lu_seq_range ROOT_FLD_RANGE = {
- .lsr_start = FID_SEQ_ROOT,
- .lsr_end = FID_SEQ_ROOT + 1,
- .lsr_index = 0,
- .lsr_flags = LU_SEQ_RANGE_MDT
-};
-
-const struct dt_index_features fld_index_features = {
- .dif_flags = DT_IND_UPDATE,
- .dif_keysize_min = sizeof(seqno_t),
- .dif_keysize_max = sizeof(seqno_t),
- .dif_recsize_min = sizeof(struct lu_seq_range),
- .dif_recsize_max = sizeof(struct lu_seq_range),
- .dif_ptrsize = 4
-};
-
-extern struct lu_context_key fld_thread_key;
-
-int fld_declare_index_create(const struct lu_env *env,
- struct lu_server_fld *fld,
- const struct lu_seq_range *new_range,
- struct thandle *th)
-{
- struct lu_seq_range *tmp;
- struct lu_seq_range *range;
- struct fld_thread_info *info;
- int rc = 0;
-
- ENTRY;
-
- info = lu_context_key_get(&env->le_ctx, &fld_thread_key);
- range = &info->fti_lrange;
- tmp = &info->fti_irange;
- memset(range, 0, sizeof(*range));
-
- rc = fld_index_lookup(env, fld, new_range->lsr_start, range);
- if (rc == 0) {
- /* In case of duplicate entry, the location must be same */
- LASSERT((range_compare_loc(new_range, range) == 0));
- GOTO(out, rc = -EEXIST);
- }
-
- if (rc != -ENOENT) {
- CERROR("%s: lookup range "DRANGE" error: rc = %d\n",
- fld->lsf_name, PRANGE(range), rc);
- GOTO(out, rc);
- }
-
- /* Check for merge case, since the fld entry can only be increamental,
- * so we will only check whether it can be merged from the left. */
- if (new_range->lsr_start == range->lsr_end && range->lsr_end != 0 &&
- range_compare_loc(new_range, range) == 0) {
- range_cpu_to_be(tmp, range);
- rc = dt_declare_delete(env, fld->lsf_obj,
- (struct dt_key *)&tmp->lsr_start, th);
- if (rc) {
- CERROR("%s: declare record "DRANGE" failed: rc = %d\n",
- fld->lsf_name, PRANGE(range), rc);
- GOTO(out, rc);
- }
- memcpy(tmp, new_range, sizeof(*new_range));
- tmp->lsr_start = range->lsr_start;
- } else {
- memcpy(tmp, new_range, sizeof(*new_range));
- }
-
- range_cpu_to_be(tmp, tmp);
- rc = dt_declare_insert(env, fld->lsf_obj, (struct dt_rec *)tmp,
- (struct dt_key *)&tmp->lsr_start, th);
-out:
- RETURN(rc);
-}
-
-/**
- * insert range in fld store.
- *
- * \param range range to be inserted
- * \param th transaction for this operation as it could compound
- * transaction.
- *
- * \retval 0 success
- * \retval -ve error
- *
- * The whole fld index insertion is protected by seq->lss_mutex (see
- * seq_server_alloc_super), i.e. only one thread will access fldb each
- * time, so we do not need worry the fld file and cache will being
- * changed between declare and create.
- * Because the fld entry can only be increamental, so we will only check
- * whether it can be merged from the left.
- **/
-int fld_index_create(const struct lu_env *env, struct lu_server_fld *fld,
- const struct lu_seq_range *new_range, struct thandle *th)
-{
- struct lu_seq_range *range;
- struct lu_seq_range *tmp;
- struct fld_thread_info *info;
- int rc = 0;
- int deleted = 0;
- struct fld_cache_entry *flde;
- ENTRY;
-
- info = lu_context_key_get(&env->le_ctx, &fld_thread_key);
-
- LASSERT(mutex_is_locked(&fld->lsf_lock));
-
- range = &info->fti_lrange;
- memset(range, 0, sizeof(*range));
- tmp = &info->fti_irange;
- rc = fld_index_lookup(env, fld, new_range->lsr_start, range);
- if (rc != -ENOENT) {
- rc = rc == 0 ? -EEXIST : rc;
- GOTO(out, rc);
- }
-
- if (new_range->lsr_start == range->lsr_end && range->lsr_end != 0 &&
- range_compare_loc(new_range, range) == 0) {
- range_cpu_to_be(tmp, range);
- rc = dt_delete(env, fld->lsf_obj,
- (struct dt_key *)&tmp->lsr_start, th,
- BYPASS_CAPA);
- if (rc != 0)
- GOTO(out, rc);
- memcpy(tmp, new_range, sizeof(*new_range));
- tmp->lsr_start = range->lsr_start;
- deleted = 1;
- } else {
- memcpy(tmp, new_range, sizeof(*new_range));
- }
-
- range_cpu_to_be(tmp, tmp);
- rc = dt_insert(env, fld->lsf_obj, (struct dt_rec *)tmp,
- (struct dt_key *)&tmp->lsr_start, th, BYPASS_CAPA, 1);
- if (rc != 0) {
- CERROR("%s: insert range "DRANGE" failed: rc = %d\n",
- fld->lsf_name, PRANGE(new_range), rc);
- GOTO(out, rc);
- }
-
- flde = fld_cache_entry_create(new_range);
- if (IS_ERR(flde))
- GOTO(out, rc = PTR_ERR(flde));
-
- write_lock(&fld->lsf_cache->fci_lock);
- if (deleted)
- fld_cache_delete_nolock(fld->lsf_cache, new_range);
- rc = fld_cache_insert_nolock(fld->lsf_cache, flde);
- write_unlock(&fld->lsf_cache->fci_lock);
- if (rc)
- OBD_FREE_PTR(flde);
-out:
- RETURN(rc);
-}
-
-/**
- * lookup range for a seq passed. note here we only care about the start/end,
- * caller should handle the attached location data (flags, index).
- *
- * \param seq seq for lookup.
- * \param range result of lookup.
- *
- * \retval 0 found, \a range is the matched range;
- * \retval -ENOENT not found, \a range is the left-side range;
- * \retval -ve other error;
- */
-int fld_index_lookup(const struct lu_env *env, struct lu_server_fld *fld,
- seqno_t seq, struct lu_seq_range *range)
-{
- struct lu_seq_range *fld_rec;
- struct fld_thread_info *info;
- int rc;
-
- ENTRY;
-
- info = lu_context_key_get(&env->le_ctx, &fld_thread_key);
- fld_rec = &info->fti_rec;
-
- rc = fld_cache_lookup(fld->lsf_cache, seq, fld_rec);
- if (rc == 0) {
- *range = *fld_rec;
- if (range_within(range, seq))
- rc = 0;
- else
- rc = -ENOENT;
- }
-
- CDEBUG(D_INFO, "%s: lookup seq = "LPX64" range : "DRANGE" rc = %d\n",
- fld->lsf_name, seq, PRANGE(range), rc);
-
- RETURN(rc);
-}
-
-int fld_insert_entry(const struct lu_env *env,
- struct lu_server_fld *fld,
- const struct lu_seq_range *range)
-{
- struct thandle *th;
- int rc;
- ENTRY;
-
- th = dt_trans_create(env, lu2dt_dev(fld->lsf_obj->do_lu.lo_dev));
- if (IS_ERR(th))
- RETURN(PTR_ERR(th));
-
- rc = fld_declare_index_create(env, fld, range, th);
- if (rc != 0) {
- if (rc == -EEXIST)
- rc = 0;
- GOTO(out, rc);
- }
-
- rc = dt_trans_start_local(env, lu2dt_dev(fld->lsf_obj->do_lu.lo_dev),
- th);
- if (rc)
- GOTO(out, rc);
-
- rc = fld_index_create(env, fld, range, th);
- if (rc == -EEXIST)
- rc = 0;
-out:
- dt_trans_stop(env, lu2dt_dev(fld->lsf_obj->do_lu.lo_dev), th);
- RETURN(rc);
-}
-EXPORT_SYMBOL(fld_insert_entry);
-
-static int fld_insert_special_entries(const struct lu_env *env,
- struct lu_server_fld *fld)
-{
- int rc;
-
- rc = fld_insert_entry(env, fld, &IGIF_FLD_RANGE);
- if (rc != 0)
- RETURN(rc);
-
- rc = fld_insert_entry(env, fld, &DOT_LUSTRE_FLD_RANGE);
- if (rc != 0)
- RETURN(rc);
-
- rc = fld_insert_entry(env, fld, &ROOT_FLD_RANGE);
-
- RETURN(rc);
-}
-
-int fld_index_init(const struct lu_env *env, struct lu_server_fld *fld,
- struct dt_device *dt)
-{
- struct dt_object *dt_obj = NULL;
- struct lu_fid fid;
- struct lu_attr *attr = NULL;
- struct lu_seq_range *range = NULL;
- struct fld_thread_info *info;
- struct dt_object_format dof;
- struct dt_it *it;
- const struct dt_it_ops *iops;
- int rc;
- ENTRY;
-
- info = lu_context_key_get(&env->le_ctx, &fld_thread_key);
- LASSERT(info != NULL);
-
- lu_local_obj_fid(&fid, FLD_INDEX_OID);
- OBD_ALLOC_PTR(attr);
- if (attr == NULL)
- RETURN(-ENOMEM);
-
- memset(attr, 0, sizeof(*attr));
- attr->la_valid = LA_MODE;
- attr->la_mode = S_IFREG | 0666;
- dof.dof_type = DFT_INDEX;
- dof.u.dof_idx.di_feat = &fld_index_features;
-
- dt_obj = dt_find_or_create(env, dt, &fid, &dof, attr);
- if (IS_ERR(dt_obj)) {
- rc = PTR_ERR(dt_obj);
- CERROR("%s: Can't find \"%s\" obj %d\n", fld->lsf_name,
- fld_index_name, rc);
- dt_obj = NULL;
- GOTO(out, rc);
- }
-
- fld->lsf_obj = dt_obj;
- rc = dt_obj->do_ops->do_index_try(env, dt_obj, &fld_index_features);
- if (rc != 0) {
- CERROR("%s: File \"%s\" is not an index: rc = %d!\n",
- fld->lsf_name, fld_index_name, rc);
- GOTO(out, rc);
- }
-
- range = &info->fti_rec;
- /* Load fld entry to cache */
- iops = &dt_obj->do_index_ops->dio_it;
- it = iops->init(env, dt_obj, 0, NULL);
- if (IS_ERR(it))
- GOTO(out, rc = PTR_ERR(it));
-
- rc = iops->load(env, it, 0);
- if (rc < 0)
- GOTO(out_it_fini, rc);
-
- if (rc > 0) {
- /* Load FLD entry into server cache */
- do {
- rc = iops->rec(env, it, (struct dt_rec *)range, 0);
- if (rc != 0)
- GOTO(out_it_put, rc);
- LASSERT(range != NULL);
- range_be_to_cpu(range, range);
- rc = fld_cache_insert(fld->lsf_cache, range);
- if (rc != 0)
- GOTO(out_it_put, rc);
- rc = iops->next(env, it);
- } while (rc == 0);
- }
-
- /* Note: fld_insert_entry will detect whether these
- * special entries already exist inside FLDB */
- mutex_lock(&fld->lsf_lock);
- rc = fld_insert_special_entries(env, fld);
- mutex_unlock(&fld->lsf_lock);
- if (rc != 0) {
- CERROR("%s: insert special entries failed!: rc = %d\n",
- fld->lsf_name, rc);
- GOTO(out_it_put, rc);
- }
-
-out_it_put:
- iops->put(env, it);
-out_it_fini:
- iops->fini(env, it);
-out:
- if (attr != NULL)
- OBD_FREE_PTR(attr);
-
- if (rc != 0) {
- if (dt_obj != NULL)
- lu_object_put(env, &dt_obj->do_lu);
- fld->lsf_obj = NULL;
- }
- RETURN(rc);
-}
-
-void fld_index_fini(const struct lu_env *env, struct lu_server_fld *fld)
-{
- ENTRY;
- if (fld->lsf_obj != NULL) {
- if (!IS_ERR(fld->lsf_obj))
- lu_object_put(env, &fld->lsf_obj->do_lu);
- fld->lsf_obj = NULL;
- }
- EXIT;
-}
diff --git a/drivers/staging/lustre/lustre/fld/fld_internal.h b/drivers/staging/lustre/lustre/fld/fld_internal.h
index 9fa9e01..6e7ca9e 100644
--- a/drivers/staging/lustre/lustre/fld/fld_internal.h
+++ b/drivers/staging/lustre/lustre/fld/fld_internal.h
@@ -139,38 +139,10 @@ enum {
extern struct lu_fld_hash fld_hash[];
-
-struct fld_thread_info {
- struct req_capsule *fti_pill;
- __u64 fti_key;
- struct lu_seq_range fti_rec;
- struct lu_seq_range fti_lrange;
- struct lu_seq_range fti_irange;
-};
-
-extern struct lu_context_key fld_thread_key;
-
-int fld_index_init(const struct lu_env *env, struct lu_server_fld *fld,
- struct dt_device *dt);
-
-void fld_index_fini(const struct lu_env *env, struct lu_server_fld *fld);
-
-int fld_declare_index_create(const struct lu_env *env,
- struct lu_server_fld *fld,
- const struct lu_seq_range *new,
- struct thandle *th);
-
-int fld_index_create(const struct lu_env *env, struct lu_server_fld *fld,
- const struct lu_seq_range *new, struct thandle *th);
-
-int fld_index_lookup(const struct lu_env *env, struct lu_server_fld *fld,
- seqno_t seq, struct lu_seq_range *range);
-
int fld_client_rpc(struct obd_export *exp,
struct lu_seq_range *range, __u32 fld_op);
#ifdef LPROCFS
-extern struct lprocfs_vars fld_server_proc_list[];
extern struct lprocfs_vars fld_client_proc_list[];
#endif
@@ -219,5 +191,4 @@ fld_target_name(struct lu_fld_target *tar)
}
extern proc_dir_entry_t *fld_type_proc_dir;
-extern struct file_operations fld_proc_seq_fops;
#endif /* __FLD_INTERNAL_H */
diff --git a/drivers/staging/lustre/lustre/fld/fld_request.c b/drivers/staging/lustre/lustre/fld/fld_request.c
index e9f0739..c99b945 100644
--- a/drivers/staging/lustre/lustre/fld/fld_request.c
+++ b/drivers/staging/lustre/lustre/fld/fld_request.c
@@ -60,6 +60,8 @@
#include <lustre_mdc.h>
#include "fld_internal.h"
+struct lu_context_key fld_thread_key;
+
/* 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)
@@ -280,6 +282,8 @@ int fld_client_del_target(struct lu_client_fld *fld, __u64 idx)
EXPORT_SYMBOL(fld_client_del_target);
#ifdef LPROCFS
+proc_dir_entry_t *fld_type_proc_dir = NULL;
+
static int fld_client_proc_init(struct lu_client_fld *fld)
{
int rc;
@@ -496,12 +500,7 @@ int fld_client_lookup(struct lu_client_fld *fld, seqno_t seq, mdsno_t *mds,
res.lsr_start = seq;
fld_range_set_type(&res, flags);
- if (target->ft_srv != NULL) {
- LASSERT(env != NULL);
- rc = fld_server_lookup(env, target->ft_srv, seq, &res);
- } else {
- rc = fld_client_rpc(target->ft_exp, &res, FLD_LOOKUP);
- }
+ rc = fld_client_rpc(target->ft_exp, &res, FLD_LOOKUP);
if (rc == 0) {
*mds = res.lsr_index;
@@ -517,3 +516,32 @@ void fld_client_flush(struct lu_client_fld *fld)
fld_cache_flush(fld->lcf_cache);
}
EXPORT_SYMBOL(fld_client_flush);
+
+static int __init fld_mod_init(void)
+{
+ 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);
+
+ LU_CONTEXT_KEY_INIT(&fld_thread_key);
+ lu_context_key_register(&fld_thread_key);
+ return 0;
+}
+
+static void __exit fld_mod_exit(void)
+{
+ lu_context_key_degister(&fld_thread_key);
+ if (fld_type_proc_dir != NULL && !IS_ERR(fld_type_proc_dir)) {
+ lprocfs_remove(&fld_type_proc_dir);
+ fld_type_proc_dir = NULL;
+ }
+}
+
+MODULE_AUTHOR("Sun Microsystems, Inc. <http://www.lustre.org/>");
+MODULE_DESCRIPTION("Lustre FLD");
+MODULE_LICENSE("GPL");
+
+module_init(fld_mod_init)
+module_exit(fld_mod_exit)
diff --git a/drivers/staging/lustre/lustre/fld/lproc_fld.c b/drivers/staging/lustre/lustre/fld/lproc_fld.c
index c1bd803..77322ba 100644
--- a/drivers/staging/lustre/lustre/fld/lproc_fld.c
+++ b/drivers/staging/lustre/lustre/fld/lproc_fld.c
@@ -158,202 +158,6 @@ struct file_operations fld_proc_cache_flush_fops = {
.release = fld_proc_cache_flush_release,
};
-struct fld_seq_param {
- struct lu_env fsp_env;
- struct dt_it *fsp_it;
- struct lu_server_fld *fsp_fld;
- unsigned int fsp_stop:1;
-};
-
-static void *fldb_seq_start(struct seq_file *p, loff_t *pos)
-{
- struct fld_seq_param *param = p->private;
- struct lu_server_fld *fld;
- struct dt_object *obj;
- const struct dt_it_ops *iops;
-
- if (param == NULL || param->fsp_stop)
- return NULL;
-
- fld = param->fsp_fld;
- obj = fld->lsf_obj;
- LASSERT(obj != NULL);
- iops = &obj->do_index_ops->dio_it;
-
- iops->load(¶m->fsp_env, param->fsp_it, *pos);
-
- *pos = be64_to_cpu(*(__u64 *)iops->key(¶m->fsp_env, param->fsp_it));
- return param;
-}
-
-static void fldb_seq_stop(struct seq_file *p, void *v)
-{
- struct fld_seq_param *param = p->private;
- const struct dt_it_ops *iops;
- struct lu_server_fld *fld;
- struct dt_object *obj;
-
- if (param == NULL)
- return;
-
- fld = param->fsp_fld;
- obj = fld->lsf_obj;
- LASSERT(obj != NULL);
- iops = &obj->do_index_ops->dio_it;
-
- iops->put(¶m->fsp_env, param->fsp_it);
-}
-
-static void *fldb_seq_next(struct seq_file *p, void *v, loff_t *pos)
-{
- struct fld_seq_param *param = p->private;
- struct lu_server_fld *fld;
- struct dt_object *obj;
- const struct dt_it_ops *iops;
- int rc;
-
- if (param == NULL || param->fsp_stop)
- return NULL;
-
- fld = param->fsp_fld;
- obj = fld->lsf_obj;
- LASSERT(obj != NULL);
- iops = &obj->do_index_ops->dio_it;
-
- rc = iops->next(¶m->fsp_env, param->fsp_it);
- if (rc > 0) {
- param->fsp_stop = 1;
- return NULL;
- }
-
- *pos = be64_to_cpu(*(__u64 *)iops->key(¶m->fsp_env, param->fsp_it));
- return param;
-}
-
-static int fldb_seq_show(struct seq_file *p, void *v)
-{
- struct fld_seq_param *param = p->private;
- struct lu_server_fld *fld;
- struct dt_object *obj;
- const struct dt_it_ops *iops;
- struct fld_thread_info *info;
- struct lu_seq_range *fld_rec;
- int rc;
-
- if (param == NULL || param->fsp_stop)
- return 0;
-
- fld = param->fsp_fld;
- obj = fld->lsf_obj;
- LASSERT(obj != NULL);
- iops = &obj->do_index_ops->dio_it;
-
- info = lu_context_key_get(¶m->fsp_env.le_ctx,
- &fld_thread_key);
- fld_rec = &info->fti_rec;
- rc = iops->rec(¶m->fsp_env, param->fsp_it,
- (struct dt_rec *)fld_rec, 0);
- if (rc != 0) {
- CERROR("%s:read record error: rc %d\n",
- fld->lsf_name, rc);
- } else if (fld_rec->lsr_start != 0) {
- range_be_to_cpu(fld_rec, fld_rec);
- rc = seq_printf(p, DRANGE"\n", PRANGE(fld_rec));
- }
-
- return rc;
-}
-
-struct seq_operations fldb_sops = {
- .start = fldb_seq_start,
- .stop = fldb_seq_stop,
- .next = fldb_seq_next,
- .show = fldb_seq_show,
-};
-
-static int fldb_seq_open(struct inode *inode, struct file *file)
-{
- struct seq_file *seq;
- struct lu_server_fld *fld = (struct lu_server_fld *)PDE_DATA(inode);
- struct dt_object *obj;
- const struct dt_it_ops *iops;
- struct fld_seq_param *param = NULL;
- int env_init = 0;
- int rc;
-
- rc = seq_open(file, &fldb_sops);
- if (rc)
- GOTO(out, rc);
-
- obj = fld->lsf_obj;
- if (obj == NULL) {
- seq = file->private_data;
- seq->private = NULL;
- return 0;
- }
-
- OBD_ALLOC_PTR(param);
- if (param == NULL)
- GOTO(out, rc = -ENOMEM);
-
- rc = lu_env_init(¶m->fsp_env, LCT_MD_THREAD);
- if (rc != 0)
- GOTO(out, rc);
-
- env_init = 1;
- iops = &obj->do_index_ops->dio_it;
- param->fsp_it = iops->init(¶m->fsp_env, obj, 0, NULL);
- if (IS_ERR(param->fsp_it))
- GOTO(out, rc = PTR_ERR(param->fsp_it));
-
- param->fsp_fld = fld;
- param->fsp_stop = 0;
-
- seq = file->private_data;
- seq->private = param;
-out:
- if (rc != 0) {
- if (env_init == 1)
- lu_env_fini(¶m->fsp_env);
- if (param != NULL)
- OBD_FREE_PTR(param);
- }
- return rc;
-}
-
-static int fldb_seq_release(struct inode *inode, struct file *file)
-{
- struct seq_file *seq = file->private_data;
- struct fld_seq_param *param;
- struct lu_server_fld *fld;
- struct dt_object *obj;
- const struct dt_it_ops *iops;
-
- param = seq->private;
- if (param == NULL) {
- lprocfs_seq_release(inode, file);
- return 0;
- }
-
- fld = param->fsp_fld;
- obj = fld->lsf_obj;
- LASSERT(obj != NULL);
- iops = &obj->do_index_ops->dio_it;
-
- LASSERT(iops != NULL);
- LASSERT(obj != NULL);
- LASSERT(param->fsp_it != NULL);
- iops->fini(¶m->fsp_env, param->fsp_it);
- lu_env_fini(¶m->fsp_env);
- OBD_FREE_PTR(param);
- lprocfs_seq_release(inode, file);
-
- return 0;
-}
-
-struct lprocfs_vars fld_server_proc_list[] = {
- { NULL }};
-
LPROC_SEQ_FOPS_RO(fld_proc_targets);
LPROC_SEQ_FOPS(fld_proc_hash);
@@ -363,11 +167,4 @@ struct lprocfs_vars fld_client_proc_list[] = {
{ "cache_flush", &fld_proc_cache_flush_fops },
{ NULL }};
-struct file_operations fld_proc_seq_fops = {
- .owner = THIS_MODULE,
- .open = fldb_seq_open,
- .read = seq_read,
- .release = fldb_seq_release,
-};
-
-#endif
+#endif /* LPROCFS */
diff --git a/drivers/staging/lustre/lustre/include/lustre_fld.h b/drivers/staging/lustre/lustre/include/lustre_fld.h
index 11e034a..94b697d 100644
--- a/drivers/staging/lustre/lustre/include/lustre_fld.h
+++ b/drivers/staging/lustre/lustre/include/lustre_fld.h
@@ -43,9 +43,6 @@
*/
#include <lustre/lustre_idl.h>
-#include <lustre_mdt.h>
-#include <dt_object.h>
-
#include <linux/libcfs/libcfs.h>
struct lu_client_fld;
@@ -129,47 +126,9 @@ struct lu_client_fld {
* Client fld proc entry name. */
char lcf_name[80];
- const struct lu_context *lcf_ctx;
-
int lcf_flags;
};
-/**
- * number of blocks to reserve for particular operations. Should be function of
- * ... something. Stub for now.
- */
-enum {
- /* one insert operation can involve two delete and one insert */
- FLD_TXN_INDEX_INSERT_CREDITS = 60,
- FLD_TXN_INDEX_DELETE_CREDITS = 20,
-};
-
-int fld_query(struct com_thread_info *info);
-
-/* Server methods */
-int fld_server_init(const struct lu_env *env, struct lu_server_fld *fld,
- struct dt_device *dt, const char *prefix, int mds_node_id,
- int type);
-
-void fld_server_fini(const struct lu_env *env, struct lu_server_fld *fld);
-
-int fld_declare_server_create(const struct lu_env *env,
- struct lu_server_fld *fld,
- struct lu_seq_range *new,
- struct thandle *th);
-
-int fld_server_create(const struct lu_env *env,
- struct lu_server_fld *fld,
- struct lu_seq_range *add_range,
- struct thandle *th);
-
-int fld_insert_entry(const struct lu_env *env,
- struct lu_server_fld *fld,
- const struct lu_seq_range *range);
-
-int fld_server_lookup(const struct lu_env *env, struct lu_server_fld *fld,
- seqno_t seq, struct lu_seq_range *range);
-
/* Client methods */
int fld_client_init(struct lu_client_fld *fld,
const char *prefix, int hash);
diff --git a/drivers/staging/lustre/lustre/include/lustre_mdt.h b/drivers/staging/lustre/lustre/include/lustre_mdt.h
deleted file mode 100644
index dba26a6..0000000
--- a/drivers/staging/lustre/lustre/include/lustre_mdt.h
+++ /dev/null
@@ -1,84 +0,0 @@
-/*
- * GPL HEADER START
- *
- * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
- *
- * 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 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).
- *
- * 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.
- *
- * GPL HEADER END
- */
-/*
- * Copyright (c) 2007, 2010, Oracle and/or its affiliates. All rights reserved.
- * Use is subject to license terms.
- */
-/*
- * This file is part of Lustre, http://www.lustre.org/
- * Lustre is a trademark of Sun Microsystems, Inc.
- */
-
-#ifndef __LINUX_MDT_H
-#define __LINUX_MDT_H
-
-/** \defgroup mdt mdt
- *
- * @{
- */
-
-#include <lustre/lustre_idl.h>
-#include <lustre_req_layout.h>
-#include <md_object.h>
-#include <dt_object.h>
-#include <linux/libcfs/libcfs.h>
-
-/*
- * Common thread info for mdt, seq and fld
- */
-struct com_thread_info {
- /*
- * for req-layout interface.
- */
- struct req_capsule *cti_pill;
-};
-
-enum {
- ESERIOUS = 0x0001000
-};
-
-static inline int err_serious(int rc)
-{
- LASSERT(rc < 0);
- LASSERT(-rc < ESERIOUS);
- return -(-rc | ESERIOUS);
-}
-
-static inline int clear_serious(int rc)
-{
- if (rc < 0)
- rc = -(-rc & ~ESERIOUS);
- return rc;
-}
-
-static inline int is_serious(int rc)
-{
- return (rc < 0 && -rc & ESERIOUS);
-}
-
-/** @} mdt */
-
-#endif
diff --git a/drivers/staging/lustre/lustre/obdecho/echo_client.c b/drivers/staging/lustre/lustre/obdecho/echo_client.c
index ae89ea6f..de8f4ea 100644
--- a/drivers/staging/lustre/lustre/obdecho/echo_client.c
+++ b/drivers/staging/lustre/lustre/obdecho/echo_client.c
@@ -43,6 +43,7 @@
#include <lustre_debug.h>
#include <lprocfs_status.h>
#include <cl_object.h>
+#include <md_object.h>
#include <lustre_fid.h>
#include <lustre_acl.h>
#include <lustre_net.h>
--
1.7.9.5
^ permalink raw reply related [flat|nested] 62+ messages in thread* [PATCH 34/48] staging/lustre/lmv: support DNE with HSM.
2013-07-22 16:06 [PATCH 00/48] staging/lustre: minor cleanup and Intel Lustre sync Peng Tao
` (32 preceding siblings ...)
2013-07-22 16:06 ` [PATCH 33/48] staging/lustre/fld: prepare FLD module for client server split Peng Tao
@ 2013-07-22 16:06 ` Peng Tao
2013-07-22 16:06 ` [PATCH 35/48] " Peng Tao
` (14 subsequent siblings)
48 siblings, 0 replies; 62+ messages in thread
From: Peng Tao @ 2013-07-22 16:06 UTC (permalink / raw)
To: Greg Kroah-Hartman
Cc: linux-kernel, Thomas Leibovici, Peng Tao, Andreas Dilger
From: Thomas Leibovici <thomas.leibovici@cea.fr>
Send HSM requests to the appropriate MDT. Split lists of fids of HSM
actions into one list per MDT.
Move kuc registration/unregistration from MDC to LMV as this is not
MDT related.
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-3365
Lustre-change: http://review.whamcloud.com/6714
Signed-off-by: Thomas Leibovici <thomas.leibovici@cea.fr>
Reviewed-by: John L. Hammond <john.hammond@intel.com>
Reviewed-by: jacques-Charles Lafoucriere <jacques-charles.lafoucriere@cea.fr>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>
Signed-off-by: Peng Tao <tao.peng@emc.com>
Signed-off-by: Andreas Dilger <andreas.dilger@intel.com>
---
drivers/staging/lustre/lustre/lmv/lmv_obd.c | 200 ++++++++++++++++++++++-
drivers/staging/lustre/lustre/mdc/mdc_request.c | 18 +-
2 files changed, 205 insertions(+), 13 deletions(-)
diff --git a/drivers/staging/lustre/lustre/lmv/lmv_obd.c b/drivers/staging/lustre/lustre/lmv/lmv_obd.c
index d10f7fc..9e68eea 100644
--- a/drivers/staging/lustre/lustre/lmv/lmv_obd.c
+++ b/drivers/staging/lustre/lustre/lmv/lmv_obd.c
@@ -44,6 +44,7 @@
#include <asm/div64.h>
#include <linux/seq_file.h>
#include <linux/namei.h>
+#include <asm/uaccess.h>
#include <lustre/lustre_idl.h>
#include <obd_support.h>
@@ -783,6 +784,125 @@ out_fid2path:
RETURN(rc);
}
+static int lmv_hsm_req_count(struct lmv_obd *lmv,
+ const struct hsm_user_request *hur,
+ const struct lmv_tgt_desc *tgt_mds)
+{
+ int i, nr = 0;
+ struct lmv_tgt_desc *curr_tgt;
+
+ /* count how many requests must be sent to the given target */
+ for (i = 0; i < hur->hur_request.hr_itemcount; i++) {
+ curr_tgt = lmv_find_target(lmv, &hur->hur_user_item[i].hui_fid);
+ if (obd_uuid_equals(&curr_tgt->ltd_uuid, &tgt_mds->ltd_uuid))
+ nr++;
+ }
+ return nr;
+}
+
+static void lmv_hsm_req_build(struct lmv_obd *lmv,
+ struct hsm_user_request *hur_in,
+ const struct lmv_tgt_desc *tgt_mds,
+ struct hsm_user_request *hur_out)
+{
+ int i, nr_out;
+ struct lmv_tgt_desc *curr_tgt;
+
+ /* build the hsm_user_request for the given target */
+ hur_out->hur_request = hur_in->hur_request;
+ nr_out = 0;
+ for (i = 0; i < hur_in->hur_request.hr_itemcount; i++) {
+ curr_tgt = lmv_find_target(lmv,
+ &hur_in->hur_user_item[i].hui_fid);
+ if (obd_uuid_equals(&curr_tgt->ltd_uuid, &tgt_mds->ltd_uuid)) {
+ hur_out->hur_user_item[nr_out] =
+ hur_in->hur_user_item[i];
+ nr_out++;
+ }
+ }
+ hur_out->hur_request.hr_itemcount = nr_out;
+ memcpy(hur_data(hur_out), hur_data(hur_in),
+ hur_in->hur_request.hr_data_len);
+}
+
+static int lmv_hsm_ct_unregister(struct lmv_obd *lmv, unsigned int cmd, int len,
+ struct lustre_kernelcomm *lk, void *uarg)
+{
+ int i, rc = 0;
+ ENTRY;
+
+ /* unregister request (call from llapi_hsm_copytool_fini) */
+ for (i = 0; i < lmv->desc.ld_tgt_count; i++) {
+ /* best effort: try to clean as much as possible
+ * (continue on error) */
+ obd_iocontrol(cmd, lmv->tgts[i]->ltd_exp, len, lk, uarg);
+ }
+
+ /* Whatever the result, remove copytool from kuc groups.
+ * Unreached coordinators will get EPIPE on next requests
+ * and will unregister automatically.
+ */
+ rc = libcfs_kkuc_group_rem(lk->lk_uid, lk->lk_group);
+ RETURN(rc);
+}
+
+static int lmv_hsm_ct_register(struct lmv_obd *lmv, unsigned int cmd, int len,
+ struct lustre_kernelcomm *lk, void *uarg)
+{
+ struct file *filp;
+ int i, j, err;
+ int rc = 0;
+ bool any_set = false;
+ ENTRY;
+
+ /* All or nothing: try to register to all MDS.
+ * In case of failure, unregister from previous MDS,
+ * except if it because of inactive target. */
+ for (i = 0; i < lmv->desc.ld_tgt_count; i++) {
+ err = obd_iocontrol(cmd, lmv->tgts[i]->ltd_exp,
+ len, lk, uarg);
+ if (err) {
+ if (lmv->tgts[i]->ltd_active) {
+ /* permanent error */
+ CERROR("error: iocontrol MDC %s on MDT"
+ "idx %d cmd %x: err = %d\n",
+ lmv->tgts[i]->ltd_uuid.uuid,
+ i, cmd, err);
+ rc = err;
+ lk->lk_flags |= LK_FLG_STOP;
+ /* unregister from previous MDS */
+ for (j = 0; j < i; j++)
+ obd_iocontrol(cmd,
+ lmv->tgts[j]->ltd_exp,
+ len, lk, uarg);
+ RETURN(rc);
+ }
+ /* else: transient error.
+ * kuc will register to the missing MDT
+ * when it is back */
+ } else {
+ any_set = true;
+ }
+ }
+
+ if (!any_set)
+ /* no registration done: return error */
+ RETURN(-ENOTCONN);
+
+ /* at least one registration done, with no failure */
+ filp = fget(lk->lk_wfd);
+ if (filp == NULL) {
+ RETURN(-EBADF);
+ }
+ rc = libcfs_kkuc_group_add(filp, lk->lk_uid, lk->lk_group, lk->lk_data);
+ if (rc != 0 && filp != NULL)
+ fput(filp);
+ RETURN(rc);
+}
+
+
+
+
static int lmv_iocontrol(unsigned int cmd, struct obd_export *exp,
int len, void *karg, void *uarg)
{
@@ -908,7 +1028,77 @@ static int lmv_iocontrol(unsigned int cmd, struct obd_export *exp,
}
case LL_IOC_HSM_STATE_GET:
case LL_IOC_HSM_STATE_SET:
- case LL_IOC_HSM_ACTION:
+ case LL_IOC_HSM_ACTION: {
+ struct md_op_data *op_data = karg;
+ struct lmv_tgt_desc *tgt;
+
+ tgt = lmv_find_target(lmv, &op_data->op_fid1);
+ if (IS_ERR(tgt))
+ RETURN(PTR_ERR(tgt));
+
+ if (tgt->ltd_exp == NULL)
+ RETURN(-EINVAL);
+
+ rc = obd_iocontrol(cmd, tgt->ltd_exp, len, karg, uarg);
+ break;
+ }
+ case LL_IOC_HSM_PROGRESS: {
+ const struct hsm_progress_kernel *hpk = karg;
+ struct lmv_tgt_desc *tgt;
+
+ tgt = lmv_find_target(lmv, &hpk->hpk_fid);
+ if (IS_ERR(tgt))
+ RETURN(PTR_ERR(tgt));
+ rc = obd_iocontrol(cmd, tgt->ltd_exp, len, karg, uarg);
+ break;
+ }
+ case LL_IOC_HSM_REQUEST: {
+ struct hsm_user_request *hur = karg;
+ struct lmv_tgt_desc *tgt;
+ unsigned int reqcount = hur->hur_request.hr_itemcount;
+
+ if (reqcount == 0)
+ RETURN(0);
+
+ /* if the request is about a single fid
+ * or if there is a single MDS, no need to split
+ * the request. */
+ if (reqcount == 1 || count == 1) {
+ tgt = lmv_find_target(lmv,
+ &hur->hur_user_item[0].hui_fid);
+ if (IS_ERR(tgt))
+ RETURN(PTR_ERR(tgt));
+ rc = obd_iocontrol(cmd, tgt->ltd_exp, len, karg, uarg);
+ } else {
+ /* split fid list to their respective MDS */
+ for (i = 0; i < count; i++) {
+ unsigned int nr, reqlen;
+ int rc1;
+ struct hsm_user_request *req;
+
+ nr = lmv_hsm_req_count(lmv, hur, lmv->tgts[i]);
+ if (nr == 0) /* nothing for this MDS */
+ continue;
+
+ /* build a request with fids for this MDS */
+ reqlen = offsetof(typeof(*hur),
+ hur_user_item[nr])
+ + hur->hur_request.hr_data_len;
+ OBD_ALLOC_LARGE(req, reqlen);
+ if (req == NULL)
+ RETURN(-ENOMEM);
+
+ lmv_hsm_req_build(lmv, hur, lmv->tgts[i], req);
+
+ rc1 = obd_iocontrol(cmd, lmv->tgts[i]->ltd_exp,
+ reqlen, req, uarg);
+ if (rc1 != 0 && rc == 0)
+ rc = rc1;
+ OBD_FREE_LARGE(req, reqlen);
+ }
+ }
+ break;
+ }
case LL_IOC_LOV_SWAP_LAYOUTS: {
struct md_op_data *op_data = karg;
struct lmv_tgt_desc *tgt1, *tgt2;
@@ -931,6 +1121,14 @@ static int lmv_iocontrol(unsigned int cmd, struct obd_export *exp,
rc = obd_iocontrol(cmd, tgt1->ltd_exp, len, karg, uarg);
break;
}
+ case LL_IOC_HSM_CT_START: {
+ struct lustre_kernelcomm *lk = karg;
+ if (lk->lk_flags & LK_FLG_STOP)
+ rc = lmv_hsm_ct_unregister(lmv, cmd, len, lk, uarg);
+ else
+ rc = lmv_hsm_ct_register(lmv, cmd, len, lk, uarg);
+ break;
+ }
default:
for (i = 0; i < count; i++) {
struct obd_device *mdc_obd;
diff --git a/drivers/staging/lustre/lustre/mdc/mdc_request.c b/drivers/staging/lustre/lustre/mdc/mdc_request.c
index b8bf2dc..191edc6 100644
--- a/drivers/staging/lustre/lustre/mdc/mdc_request.c
+++ b/drivers/staging/lustre/lustre/mdc/mdc_request.c
@@ -1771,6 +1771,9 @@ static int mdc_iocontrol(unsigned int cmd, struct obd_export *exp, int len,
GOTO(out, rc);
case LL_IOC_HSM_CT_START:
rc = mdc_ioc_hsm_ct_start(exp, karg);
+ /* ignore if it was already registered on this MDS. */
+ if (rc == -EEXIST)
+ rc = 0;
GOTO(out, rc);
case LL_IOC_HSM_PROGRESS:
rc = mdc_ioc_hsm_progress(exp, karg);
@@ -1988,19 +1991,10 @@ static int mdc_ioc_hsm_ct_start(struct obd_export *exp,
lk->lk_uid, lk->lk_group, lk->lk_flags);
if (lk->lk_flags & LK_FLG_STOP) {
- rc = libcfs_kkuc_group_rem(lk->lk_uid, lk->lk_group);
/* Unregister with the coordinator */
- if (rc == 0)
- rc = mdc_ioc_hsm_ct_unregister(imp);
+ rc = mdc_ioc_hsm_ct_unregister(imp);
} else {
- struct file *fp = fget(lk->lk_wfd);
-
- rc = libcfs_kkuc_group_add(fp, lk->lk_uid, lk->lk_group,
- lk->lk_data);
- if (rc && fp)
- fput(fp);
- if (rc == 0)
- rc = mdc_ioc_hsm_ct_register(imp, archive);
+ rc = mdc_ioc_hsm_ct_register(imp, archive);
}
return rc;
@@ -2325,7 +2319,7 @@ static int mdc_import_event(struct obd_device *obd, struct obd_import *imp,
}
case IMP_EVENT_ACTIVE:
rc = obd_notify_observer(obd, obd, OBD_NOTIFY_ACTIVE, NULL);
- /* restore re-establish kuc registration after reconnecting */
+ /* redo the kuc registration after reconnecting */
if (rc == 0)
rc = mdc_kuc_reregister(imp);
break;
--
1.7.9.5
^ permalink raw reply related [flat|nested] 62+ messages in thread* [PATCH 35/48] staging/lustre/lmv: support DNE with HSM.
2013-07-22 16:06 [PATCH 00/48] staging/lustre: minor cleanup and Intel Lustre sync Peng Tao
` (33 preceding siblings ...)
2013-07-22 16:06 ` [PATCH 34/48] staging/lustre/lmv: support DNE with HSM Peng Tao
@ 2013-07-22 16:06 ` Peng Tao
2013-07-23 20:39 ` Greg Kroah-Hartman
2013-07-22 16:06 ` [PATCH 36/48] staging/lustre/obdclass: add obd_target.h Peng Tao
` (13 subsequent siblings)
48 siblings, 1 reply; 62+ messages in thread
From: Peng Tao @ 2013-07-22 16:06 UTC (permalink / raw)
To: Greg Kroah-Hartman
Cc: linux-kernel, Thomas Leibovici, Peng Tao, Andreas Dilger
From: Thomas Leibovici <thomas.leibovici@cea.fr>
Send HSM requests to the appropriate MDT. Split lists of fids of HSM
actions into one list per MDT.
Move kuc registration/unregistration from MDC to LMV as this is not
MDT related.
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-3365
Lustre-change: http://review.whamcloud.com/6714
Signed-off-by: Thomas Leibovici <thomas.leibovici@cea.fr>
Reviewed-by: John L. Hammond <john.hammond@intel.com>
Reviewed-by: jacques-Charles Lafoucriere <jacques-charles.lafoucriere@cea.fr>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>
Signed-off-by: Peng Tao <tao.peng@emc.com>
Signed-off-by: Andreas Dilger <andreas.dilger@intel.com>
---
drivers/staging/lustre/lustre/llite/llite_mmap.c | 36 +++++++++++++---------
1 file changed, 21 insertions(+), 15 deletions(-)
diff --git a/drivers/staging/lustre/lustre/llite/llite_mmap.c b/drivers/staging/lustre/lustre/llite/llite_mmap.c
index d9590d8..8aa1dae 100644
--- a/drivers/staging/lustre/lustre/llite/llite_mmap.c
+++ b/drivers/staging/lustre/lustre/llite/llite_mmap.c
@@ -107,11 +107,12 @@ struct cl_io *ll_fault_io_init(struct vm_area_struct *vma,
struct cl_env_nest *nest,
pgoff_t index, unsigned long *ra_flags)
{
- struct file *file = vma->vm_file;
- struct inode *inode = file->f_dentry->d_inode;
- struct cl_io *io;
- struct cl_fault_io *fio;
- struct lu_env *env;
+ struct file *file = vma->vm_file;
+ struct inode *inode = file->f_dentry->d_inode;
+ struct cl_io *io;
+ struct cl_fault_io *fio;
+ struct lu_env *env;
+ int rc;
ENTRY;
*env_ret = NULL;
@@ -152,17 +153,22 @@ struct cl_io *ll_fault_io_init(struct vm_area_struct *vma,
CDEBUG(D_MMAP, "vm_flags: %lx (%lu %d)\n", vma->vm_flags,
fio->ft_index, fio->ft_executable);
- if (cl_io_init(env, io, CIT_FAULT, io->ci_obj) == 0) {
+ rc = cl_io_init(env, io, CIT_FAULT, io->ci_obj);
+ if (rc == 0) {
struct ccc_io *cio = ccc_env_io(env);
struct ll_file_data *fd = LUSTRE_FPRIVATE(file);
LASSERT(cio->cui_cl.cis_io == io);
- /* mmap lock must be MANDATORY
- * it has to cache pages. */
+ /* mmap lock must be MANDATORY it has to cache
+ * pages. */
io->ci_lockreq = CILR_MANDATORY;
-
- cio->cui_fd = fd;
+ cio->cui_fd = fd;
+ } else {
+ LASSERT(rc < 0);
+ cl_io_fini(env, io);
+ cl_env_nested_put(nest, env);
+ io = ERR_PTR(rc);
}
return io;
@@ -190,7 +196,7 @@ static int ll_page_mkwrite0(struct vm_area_struct *vma, struct page *vmpage,
result = io->ci_result;
if (result < 0)
- GOTO(out, result);
+ GOTO(out_io, result);
io->u.ci_fault.ft_mkwrite = 1;
io->u.ci_fault.ft_writable = 1;
@@ -252,14 +258,14 @@ static int ll_page_mkwrite0(struct vm_area_struct *vma, struct page *vmpage,
}
EXIT;
-out:
+out_io:
cl_io_fini(env, io);
cl_env_nested_put(&nest, env);
-
+out:
CDEBUG(D_MMAP, "%s mkwrite with %d\n", current->comm, result);
-
LASSERT(ergo(result == 0, PageLocked(vmpage)));
- return(result);
+
+ return result;
}
--
1.7.9.5
^ permalink raw reply related [flat|nested] 62+ messages in thread* Re: [PATCH 35/48] staging/lustre/lmv: support DNE with HSM.
2013-07-22 16:06 ` [PATCH 35/48] " Peng Tao
@ 2013-07-23 20:39 ` Greg Kroah-Hartman
0 siblings, 0 replies; 62+ messages in thread
From: Greg Kroah-Hartman @ 2013-07-23 20:39 UTC (permalink / raw)
To: Peng Tao; +Cc: linux-kernel, Thomas Leibovici, Peng Tao, Andreas Dilger
On Tue, Jul 23, 2013 at 12:06:56AM +0800, Peng Tao wrote:
> From: Thomas Leibovici <thomas.leibovici@cea.fr>
>
> Send HSM requests to the appropriate MDT. Split lists of fids of HSM
> actions into one list per MDT.
> Move kuc registration/unregistration from MDC to LMV as this is not
> MDT related.
>
> Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-3365
> Lustre-change: http://review.whamcloud.com/6714
> Signed-off-by: Thomas Leibovici <thomas.leibovici@cea.fr>
> Reviewed-by: John L. Hammond <john.hammond@intel.com>
> Reviewed-by: jacques-Charles Lafoucriere <jacques-charles.lafoucriere@cea.fr>
> Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>
> Signed-off-by: Peng Tao <tao.peng@emc.com>
> Signed-off-by: Andreas Dilger <andreas.dilger@intel.com>
> ---
> drivers/staging/lustre/lustre/llite/llite_mmap.c | 36 +++++++++++++---------
> 1 file changed, 21 insertions(+), 15 deletions(-)
This has the exact same Subject: as patch 34, which implies that one of
these is wrong, so I'll not take either, sorry.
Please fix up and resend.
greg k-h
^ permalink raw reply [flat|nested] 62+ messages in thread
* [PATCH 36/48] staging/lustre/obdclass: add obd_target.h
2013-07-22 16:06 [PATCH 00/48] staging/lustre: minor cleanup and Intel Lustre sync Peng Tao
` (34 preceding siblings ...)
2013-07-22 16:06 ` [PATCH 35/48] " Peng Tao
@ 2013-07-22 16:06 ` Peng Tao
2013-07-23 20:41 ` Greg Kroah-Hartman
2013-07-22 16:06 ` [PATCH 37/48] staging/lustre/procfs: return -ENOMEM from lprocfs_register() Peng Tao
` (12 subsequent siblings)
48 siblings, 1 reply; 62+ messages in thread
From: Peng Tao @ 2013-07-22 16:06 UTC (permalink / raw)
To: Greg Kroah-Hartman
Cc: linux-kernel, Liu Xuezhao, John L. Hammond, Peng Tao, Andreas Dilger
From: Liu Xuezhao <xuezhao.liu@emc.com>
Move some server-side data structures form obd.h to obd_target.h. To
remove some code dependencies use HAVE_SERVER_SUPPORT to protect some
llog/fsfilt_* functions.
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-1330
Lustre-change: http://review.whamcloud.com/2678
Signed-off-by: Liu Xuezhao <xuezhao.liu@emc.com>
Signed-off-by: John L. Hammond <john.hammond@intel.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Reviewed-by: Peng Tao <bergwolf@gmail.com>
Reviewed-by: Mike Pershin <mike.pershin@intel.com>
Signed-off-by: Peng Tao <tao.peng@emc.com>
Signed-off-by: Andreas Dilger <andreas.dilger@intel.com>
---
.../lustre/lustre/include/linux/lustre_fsfilt.h | 10 -
.../staging/lustre/lustre/include/lprocfs_status.h | 36 --
drivers/staging/lustre/lustre/include/obd.h | 140 +----
drivers/staging/lustre/lustre/lvfs/lvfs_linux.c | 1 -
drivers/staging/lustre/lustre/obdclass/Makefile | 2 +-
.../lustre/lustre/obdclass/lprocfs_jobstats.c | 560 --------------------
drivers/staging/lustre/lustre/obdclass/obd_mount.c | 1 -
7 files changed, 3 insertions(+), 747 deletions(-)
delete mode 100644 drivers/staging/lustre/lustre/obdclass/lprocfs_jobstats.c
diff --git a/drivers/staging/lustre/lustre/include/linux/lustre_fsfilt.h b/drivers/staging/lustre/lustre/include/linux/lustre_fsfilt.h
index 6c72609..e637006 100644
--- a/drivers/staging/lustre/lustre/include/linux/lustre_fsfilt.h
+++ b/drivers/staging/lustre/lustre/include/linux/lustre_fsfilt.h
@@ -145,16 +145,6 @@ static inline int fsfilt_commit(struct obd_device *obd, struct inode *inode,
return rc;
}
-static inline int fsfilt_map_inode_pages(struct obd_device *obd,
- struct inode *inode,
- struct page **page, int pages,
- unsigned long *blocks,
- int create, struct mutex *mutex)
-{
- return obd->obd_fsops->fs_map_inode_pages(inode, page, pages, blocks,
- create, mutex);
-}
-
static inline int fsfilt_read_record(struct obd_device *obd, struct file *file,
void *buf, loff_t size, loff_t *offs)
{
diff --git a/drivers/staging/lustre/lustre/include/lprocfs_status.h b/drivers/staging/lustre/lustre/include/lprocfs_status.h
index 294fb78..20633d4 100644
--- a/drivers/staging/lustre/lustre/include/lprocfs_status.h
+++ b/drivers/staging/lustre/lustre/include/lprocfs_status.h
@@ -370,18 +370,6 @@ static inline void s2dhms(struct dhms *ts, time_t secs)
#define JOBSTATS_DISABLE "disable"
#define JOBSTATS_PROCNAME_UID "procname_uid"
-typedef void (*cntr_init_callback)(struct lprocfs_stats *stats);
-
-struct obd_job_stats {
- cfs_hash_t *ojs_hash;
- struct list_head ojs_list;
- rwlock_t ojs_lock; /* protect the obj_list */
- cntr_init_callback ojs_cntr_init_fn;
- int ojs_cntr_num;
- int ojs_cleanup_interval;
- time_t ojs_last_cleanup;
-};
-
#ifdef LPROCFS
extern int lprocfs_stats_alloc_one(struct lprocfs_stats *stats,
@@ -748,16 +736,6 @@ struct file_operations name##_fops = { \
.release = lprocfs_single_release, \
};
-/* lprocfs_jobstats.c */
-int lprocfs_job_stats_log(struct obd_device *obd, char *jobid,
- int event, long amount);
-void lprocfs_job_stats_fini(struct obd_device *obd);
-int lprocfs_job_stats_init(struct obd_device *obd, int cntr_num,
- cntr_init_callback fn);
-int lprocfs_rd_job_interval(struct seq_file *m, void *data);
-int lprocfs_wr_job_interval(struct file *file, const char *buffer,
- unsigned long count, void *data);
-
/* lproc_ptlrpc.c */
struct ptlrpc_request;
extern void target_print_req(void *seq_file, struct ptlrpc_request *req);
@@ -1021,20 +999,6 @@ __u64 lprocfs_stats_collector(struct lprocfs_stats *stats, int idx,
#define LPROC_SEQ_FOPS_RW_TYPE(name, type)
#define LPROC_SEQ_FOPS_WR_ONLY(name, type)
-/* lprocfs_jobstats.c */
-static inline
-int lprocfs_job_stats_log(struct obd_device *obd, char *jobid, int event,
- long amount)
-{ return 0; }
-static inline
-void lprocfs_job_stats_fini(struct obd_device *obd)
-{ return; }
-static inline
-int lprocfs_job_stats_init(struct obd_device *obd, int cntr_num,
- cntr_init_callback fn)
-{ return 0; }
-
-
/* lproc_ptlrpc.c */
#define target_print_req NULL
diff --git a/drivers/staging/lustre/lustre/include/obd.h b/drivers/staging/lustre/lustre/include/obd.h
index fbf997f..be3a08f 100644
--- a/drivers/staging/lustre/lustre/include/obd.h
+++ b/drivers/staging/lustre/lustre/include/obd.h
@@ -49,6 +49,8 @@
#define IOC_MDC_MAX_NR 50
#include <lustre/lustre_idl.h>
+#include <lustre_lib.h>
+#include <linux/libcfs/bitmap.h>
#include <lu_ref.h>
#include <lustre_lib.h>
#include <lustre_export.h>
@@ -56,9 +58,6 @@
#include <lustre_fld.h>
#include <lustre_capa.h>
-#include <linux/libcfs/bitmap.h>
-
-
#define MAX_OBD_DEVICES 8192
struct osc_async_rc {
@@ -254,30 +253,6 @@ struct brw_page {
obd_flag flag;
};
-/* Individual type definitions */
-
-struct ost_server_data;
-
-struct osd_properties {
- size_t osd_max_ea_size;
-};
-
-#define OBT_MAGIC 0xBDDECEAE
-/* hold common fields for "target" device */
-struct obd_device_target {
- __u32 obt_magic;
- __u32 obt_instance;
- struct super_block *obt_sb;
- /** last_rcvd file */
- struct file *obt_rcvd_filp;
- __u64 obt_mount_count;
- struct rw_semaphore obt_rwsem;
- struct vfsmount *obt_vfsmnt;
- struct file *obt_health_check_filp;
- struct osd_properties obt_osd_properties;
- struct obd_job_stats obt_jobstats;
-};
-
/* llog contexts */
enum llog_ctxt_id {
LLOG_CONFIG_ORIG_CTXT = 0,
@@ -298,94 +273,6 @@ enum llog_ctxt_id {
LLOG_MAX_CTXTS
};
-#define FILTER_SUBDIR_COUNT 32 /* set to zero for no subdirs */
-
-struct filter_subdirs {
- struct dentry *dentry[FILTER_SUBDIR_COUNT];
-};
-
-
-struct filter_ext {
- __u64 fe_start;
- __u64 fe_end;
-};
-
-struct filter_obd {
- /* NB this field MUST be first */
- struct obd_device_target fo_obt;
- const char *fo_fstype;
-
- int fo_group_count;
- struct dentry *fo_dentry_O;
- struct dentry **fo_dentry_O_groups;
- struct filter_subdirs *fo_dentry_O_sub;
- struct mutex fo_init_lock; /* group initialization lock*/
- int fo_committed_group;
-
- spinlock_t fo_objidlock; /* protect fo_lastobjid */
-
- unsigned long fo_destroys_in_progress;
- struct mutex fo_create_locks[FILTER_SUBDIR_COUNT];
-
- struct list_head fo_export_list;
- int fo_subdir_count;
-
- obd_size fo_tot_dirty; /* protected by obd_osfs_lock */
- obd_size fo_tot_granted; /* all values in bytes */
- obd_size fo_tot_pending;
- int fo_tot_granted_clients;
-
- obd_size fo_readcache_max_filesize;
- spinlock_t fo_flags_lock;
- unsigned int fo_read_cache:1, /**< enable read-only cache */
- fo_writethrough_cache:1,/**< read cache writes */
- fo_mds_ost_sync:1, /**< MDS-OST orphan recovery*/
- fo_raid_degraded:1;/**< RAID device degraded */
-
- struct obd_import *fo_mdc_imp;
- struct obd_uuid fo_mdc_uuid;
- struct lustre_handle fo_mdc_conn;
- struct file **fo_last_objid_files;
- __u64 *fo_last_objids; /* last created objid for groups,
- * protected by fo_objidlock */
-
- struct mutex fo_alloc_lock;
-
- atomic_t fo_r_in_flight;
- atomic_t fo_w_in_flight;
-
- /*
- * per-filter pool of kiobuf's allocated by filter_common_setup() and
- * torn down by filter_cleanup().
- *
- * This pool contains kiobuf used by
- * filter_{prep,commit}rw_{read,write}() and is shared by all OST
- * threads.
- *
- * Locking: protected by internal lock of cfs_hash, pool can be
- * found from this hash table by t_id of ptlrpc_thread.
- */
- struct cfs_hash *fo_iobuf_hash;
-
- struct brw_stats fo_filter_stats;
-
- int fo_fmd_max_num; /* per exp filter_mod_data */
- int fo_fmd_max_age; /* jiffies to fmd expiry */
- unsigned long fo_syncjournal:1, /* sync journal on writes */
- fo_sync_lock_cancel:2;/* sync on lock cancel */
-
-
- /* sptlrpc stuff */
- rwlock_t fo_sptlrpc_lock;
- struct sptlrpc_rule_set fo_sptlrpc_rset;
-
- /* capability related */
- unsigned int fo_fl_oss_capa;
- struct list_head fo_capa_keys;
- struct hlist_head *fo_capa_hash;
- int fo_sec_level;
-};
-
struct timeout_item {
enum timeout_event ti_event;
cfs_time_t ti_timeout;
@@ -551,25 +438,6 @@ struct obd_id_info {
obd_id *data;
};
-/* */
-
-struct echo_obd {
- struct obd_device_target eo_obt;
- struct obdo eo_oa;
- spinlock_t eo_lock;
- __u64 eo_lastino;
- struct lustre_handle eo_nl_lock;
- atomic_t eo_prep;
-};
-
-struct ost_obd {
- struct ptlrpc_service *ost_service;
- struct ptlrpc_service *ost_create_service;
- struct ptlrpc_service *ost_io_service;
- struct ptlrpc_service *ost_seq_service;
- struct mutex ost_health_mutex;
-};
-
struct echo_client_obd {
struct obd_export *ec_exp; /* the local connection to osc/lov */
spinlock_t ec_lock;
@@ -1051,12 +919,8 @@ struct obd_device {
int obd_recovery_stage;
union {
- struct obd_device_target obt;
- struct filter_obd filter;
struct client_obd cli;
- struct ost_obd ost;
struct echo_client_obd echo_client;
- struct echo_obd echo;
struct lov_obd lov;
struct lmv_obd lmv;
} u;
diff --git a/drivers/staging/lustre/lustre/lvfs/lvfs_linux.c b/drivers/staging/lustre/lustre/lvfs/lvfs_linux.c
index e70d8fe..8bfae01 100644
--- a/drivers/staging/lustre/lustre/lvfs/lvfs_linux.c
+++ b/drivers/staging/lustre/lustre/lvfs/lvfs_linux.c
@@ -48,7 +48,6 @@
#include <linux/quotaops.h>
#include <linux/version.h>
#include <linux/libcfs/libcfs.h>
-#include <lustre_fsfilt.h>
#include <obd.h>
#include <linux/module.h>
#include <linux/init.h>
diff --git a/drivers/staging/lustre/lustre/obdclass/Makefile b/drivers/staging/lustre/lustre/obdclass/Makefile
index 9930a9e..023e94f 100644
--- a/drivers/staging/lustre/lustre/obdclass/Makefile
+++ b/drivers/staging/lustre/lustre/obdclass/Makefile
@@ -3,7 +3,7 @@ obj-$(CONFIG_LUSTRE_FS) += obdclass.o llog_test.o
obdclass-y := linux/linux-module.o linux/linux-obdo.o linux/linux-sysctl.o \
llog.o llog_cat.o llog_obd.o llog_swab.o class_obd.o debug.o \
genops.o uuid.o llog_ioctl.o lprocfs_status.o \
- lprocfs_jobstats.o lustre_handles.o lustre_peer.o llog_osd.o \
+ lustre_handles.o lustre_peer.o llog_osd.o \
local_storage.o statfs_pack.o obdo.o obd_config.o obd_mount.o\
mea.o lu_object.o dt_object.o capa.o cl_object.o \
cl_page.o cl_lock.o cl_io.o acl.o idmap.o lu_ucred.o
diff --git a/drivers/staging/lustre/lustre/obdclass/lprocfs_jobstats.c b/drivers/staging/lustre/lustre/obdclass/lprocfs_jobstats.c
deleted file mode 100644
index 7f6072c..0000000
--- a/drivers/staging/lustre/lustre/obdclass/lprocfs_jobstats.c
+++ /dev/null
@@ -1,560 +0,0 @@
-/* GPL HEADER START
- *
- * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
- *
- * 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 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).
- *
- * 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
- *
- * GPL HEADER END
- */
-/*
- * Copyright (c) 2011, 2012, Intel Corporation.
- * Use is subject to license terms.
- *
- * Author: Niu Yawei <niu@whamcloud.com>
- */
-/*
- * lustre/obdclass/lprocfs_jobstats.c
- */
-
-#ifndef EXPORT_SYMTAB
-# define EXPORT_SYMTAB
-#endif
-#define DEBUG_SUBSYSTEM S_CLASS
-
-
-#include <obd_class.h>
-#include <lprocfs_status.h>
-#include <lustre/lustre_idl.h>
-
-#if defined(LPROCFS)
-
-/*
- * JobID formats & JobID environment variable names for supported
- * job schedulers:
- *
- * SLURM:
- * JobID format: 32 bit integer.
- * JobID env var: SLURM_JOB_ID.
- * SGE:
- * JobID format: Decimal integer range to 99999.
- * JobID env var: JOB_ID.
- * LSF:
- * JobID format: 6 digit integer by default (up to 999999), can be
- * increased to 10 digit (up to 2147483646).
- * JobID env var: LSB_JOBID.
- * Loadleveler:
- * JobID format: String of machine_name.cluster_id.process_id, for
- * example: fr2n02.32.0
- * JobID env var: LOADL_STEP_ID.
- * PBS:
- * JobID format: String of sequence_number[.server_name][@server].
- * JobID env var: PBS_JOBID.
- * Maui/MOAB:
- * JobID format: Same as PBS.
- * JobID env var: Same as PBS.
- */
-
-struct job_stat {
- struct hlist_node js_hash;
- struct list_head js_list;
- atomic_t js_refcount;
- char js_jobid[JOBSTATS_JOBID_SIZE];
- time_t js_timestamp; /* seconds */
- struct lprocfs_stats *js_stats;
- struct obd_job_stats *js_jobstats;
-};
-
-static unsigned job_stat_hash(cfs_hash_t *hs, const void *key, unsigned mask)
-{
- return cfs_hash_djb2_hash(key, strlen(key), mask);
-}
-
-static void *job_stat_key(struct hlist_node *hnode)
-{
- struct job_stat *job;
- job = hlist_entry(hnode, struct job_stat, js_hash);
- return job->js_jobid;
-}
-
-static int job_stat_keycmp(const void *key, struct hlist_node *hnode)
-{
- struct job_stat *job;
- job = hlist_entry(hnode, struct job_stat, js_hash);
- return (strlen(job->js_jobid) == strlen(key)) &&
- !strncmp(job->js_jobid, key, strlen(key));
-}
-
-static void *job_stat_object(struct hlist_node *hnode)
-{
- return hlist_entry(hnode, struct job_stat, js_hash);
-}
-
-static void job_stat_get(cfs_hash_t *hs, struct hlist_node *hnode)
-{
- struct job_stat *job;
- job = hlist_entry(hnode, struct job_stat, js_hash);
- atomic_inc(&job->js_refcount);
-}
-
-static void job_free(struct job_stat *job)
-{
- LASSERT(atomic_read(&job->js_refcount) == 0);
- LASSERT(job->js_jobstats);
-
- write_lock(&job->js_jobstats->ojs_lock);
- list_del_init(&job->js_list);
- write_unlock(&job->js_jobstats->ojs_lock);
-
- lprocfs_free_stats(&job->js_stats);
- OBD_FREE_PTR(job);
-}
-
-static void job_putref(struct job_stat *job)
-{
- LASSERT(atomic_read(&job->js_refcount) > 0);
- if (atomic_dec_and_test(&job->js_refcount))
- job_free(job);
-}
-
-static void job_stat_put_locked(cfs_hash_t *hs, struct hlist_node *hnode)
-{
- struct job_stat *job;
- job = hlist_entry(hnode, struct job_stat, js_hash);
- job_putref(job);
-}
-
-static void job_stat_exit(cfs_hash_t *hs, struct hlist_node *hnode)
-{
- CERROR("Should not have any items!");
-}
-
-static cfs_hash_ops_t job_stats_hash_ops = {
- .hs_hash = job_stat_hash,
- .hs_key = job_stat_key,
- .hs_keycmp = job_stat_keycmp,
- .hs_object = job_stat_object,
- .hs_get = job_stat_get,
- .hs_put_locked = job_stat_put_locked,
- .hs_exit = job_stat_exit,
-};
-
-static int job_iter_callback(cfs_hash_t *hs, cfs_hash_bd_t *bd,
- struct hlist_node *hnode, void *data)
-{
- time_t oldest = *((time_t *)data);
- struct job_stat *job;
-
- job = hlist_entry(hnode, struct job_stat, js_hash);
- if (!oldest || job->js_timestamp < oldest)
- cfs_hash_bd_del_locked(hs, bd, hnode);
-
- return 0;
-}
-
-static void lprocfs_job_cleanup(struct obd_job_stats *stats, bool force)
-{
- time_t oldest, now;
-
- if (stats->ojs_cleanup_interval == 0)
- return;
-
- now = cfs_time_current_sec();
- if (!force && now < stats->ojs_last_cleanup +
- stats->ojs_cleanup_interval)
- return;
-
- oldest = now - stats->ojs_cleanup_interval;
- cfs_hash_for_each_safe(stats->ojs_hash, job_iter_callback,
- &oldest);
- stats->ojs_last_cleanup = cfs_time_current_sec();
-}
-
-static struct job_stat *job_alloc(char *jobid, struct obd_job_stats *jobs)
-{
- struct job_stat *job;
-
- LASSERT(jobs->ojs_cntr_num && jobs->ojs_cntr_init_fn);
-
- OBD_ALLOC_PTR(job);
- if (job == NULL)
- return NULL;
-
- job->js_stats = lprocfs_alloc_stats(jobs->ojs_cntr_num, 0);
- if (job->js_stats == NULL) {
- OBD_FREE_PTR(job);
- return NULL;
- }
-
- jobs->ojs_cntr_init_fn(job->js_stats);
-
- memcpy(job->js_jobid, jobid, JOBSTATS_JOBID_SIZE);
- job->js_timestamp = cfs_time_current_sec();
- job->js_jobstats = jobs;
- INIT_HLIST_NODE(&job->js_hash);
- INIT_LIST_HEAD(&job->js_list);
- atomic_set(&job->js_refcount, 1);
-
- return job;
-}
-
-int lprocfs_job_stats_log(struct obd_device *obd, char *jobid,
- int event, long amount)
-{
- struct obd_job_stats *stats = &obd->u.obt.obt_jobstats;
- struct job_stat *job, *job2;
- ENTRY;
-
- LASSERT(stats && stats->ojs_hash);
-
- lprocfs_job_cleanup(stats, false);
-
- if (!jobid || !strlen(jobid))
- RETURN(-EINVAL);
-
- if (strlen(jobid) >= JOBSTATS_JOBID_SIZE) {
- CERROR("Invalid jobid size (%lu), expect(%d)\n",
- (unsigned long)strlen(jobid) + 1, JOBSTATS_JOBID_SIZE);
- RETURN(-EINVAL);
- }
-
- job = cfs_hash_lookup(stats->ojs_hash, jobid);
- if (job)
- goto found;
-
- job = job_alloc(jobid, stats);
- if (job == NULL)
- RETURN(-ENOMEM);
-
- job2 = cfs_hash_findadd_unique(stats->ojs_hash, job->js_jobid,
- &job->js_hash);
- if (job2 != job) {
- job_putref(job);
- job = job2;
- /* We cannot LASSERT(!list_empty(&job->js_list)) here,
- * since we just lost the race for inserting "job" into the
- * ojs_list, and some other thread is doing it _right_now_.
- * Instead, be content the other thread is doing this, since
- * "job2" was initialized in job_alloc() already. LU-2163 */
- } else {
- LASSERT(list_empty(&job->js_list));
- write_lock(&stats->ojs_lock);
- list_add_tail(&job->js_list, &stats->ojs_list);
- write_unlock(&stats->ojs_lock);
- }
-
-found:
- LASSERT(stats == job->js_jobstats);
- LASSERT(stats->ojs_cntr_num > event);
- job->js_timestamp = cfs_time_current_sec();
- lprocfs_counter_add(job->js_stats, event, amount);
-
- job_putref(job);
- RETURN(0);
-}
-EXPORT_SYMBOL(lprocfs_job_stats_log);
-
-void lprocfs_job_stats_fini(struct obd_device *obd)
-{
- struct obd_job_stats *stats = &obd->u.obt.obt_jobstats;
- time_t oldest = 0;
-
- if (stats->ojs_hash == NULL)
- return;
- cfs_hash_for_each_safe(stats->ojs_hash, job_iter_callback, &oldest);
- cfs_hash_putref(stats->ojs_hash);
- stats->ojs_hash = NULL;
- LASSERT(list_empty(&stats->ojs_list));
-}
-EXPORT_SYMBOL(lprocfs_job_stats_fini);
-
-static void *lprocfs_jobstats_seq_start(struct seq_file *p, loff_t *pos)
-{
- struct obd_job_stats *stats = p->private;
- loff_t off = *pos;
- struct job_stat *job;
-
- read_lock(&stats->ojs_lock);
- if (off == 0)
- return SEQ_START_TOKEN;
- off--;
- list_for_each_entry(job, &stats->ojs_list, js_list) {
- if (!off--)
- return job;
- }
- return NULL;
-}
-
-static void lprocfs_jobstats_seq_stop(struct seq_file *p, void *v)
-{
- struct obd_job_stats *stats = p->private;
-
- read_unlock(&stats->ojs_lock);
-}
-
-static void *lprocfs_jobstats_seq_next(struct seq_file *p, void *v, loff_t *pos)
-{
- struct obd_job_stats *stats = p->private;
- struct job_stat *job;
- struct list_head *next;
-
- ++*pos;
- if (v == SEQ_START_TOKEN) {
- next = stats->ojs_list.next;
- } else {
- job = (struct job_stat *)v;
- next = job->js_list.next;
- }
-
- return next == &stats->ojs_list ? NULL :
- list_entry(next, struct job_stat, js_list);
-}
-
-/*
- * Example of output on MDT:
- *
- * job_stats:
- * - job_id: test_id.222.25844
- * snapshot_time: 1322494486
- * open: { samples: 3, unit: reqs }
- * close: { samples: 3, unit: reqs }
- * mknod: { samples: 0, unit: reqs }
- * link: { samples: 0, unit: reqs }
- * unlink: { samples: 0, unit: reqs }
- * mkdir: { samples: 0, unit: reqs }
- * rmdir: { samples: 0, unit: reqs }
- * rename: { samples: 1, unit: reqs }
- * getattr: { samples: 7, unit: reqs }
- * setattr: { samples: 0, unit: reqs }
- * getxattr: { samples: 0, unit: reqs }
- * setxattr: { samples: 0, unit: reqs }
- * statfs: { samples: 0, unit: reqs }
- * sync: { samples: 0, unit: reqs }
- *
- * Example of output on OST:
- *
- * job_stats:
- * - job_id 4854
- * snapshot_time: 1322494602
- * read: { samples: 0, unit: bytes, min: 0, max: 0, sum: 0 }
- * write: { samples: 1, unit: bytes, min: 10, max: 10, sum: 10 }
- * setattr: { samples: 0, unit: reqs }
- * punch: { samples: 0, unit: reqs }
- * sync: { samples: 0, unit: reqs }
- */
-
-static const char spaces[] = " ";
-
-static int inline width(const char *str, int len)
-{
- return len - min((int)strlen(str), 15);
-}
-
-static int lprocfs_jobstats_seq_show(struct seq_file *p, void *v)
-{
- struct job_stat *job = v;
- struct lprocfs_stats *s;
- struct lprocfs_counter ret;
- struct lprocfs_counter_header *cntr_header;
- int i;
-
- if (v == SEQ_START_TOKEN) {
- seq_printf(p, "job_stats:\n");
- return 0;
- }
-
- seq_printf(p, "- %-16s %s\n", "job_id:", job->js_jobid);
- seq_printf(p, " %-16s %ld\n", "snapshot_time:", job->js_timestamp);
-
- s = job->js_stats;
- for (i = 0; i < s->ls_num; i++) {
- cntr_header = &s->ls_cnt_header[i];
- lprocfs_stats_collect(s, i, &ret);
-
- seq_printf(p, " %s:%.*s { samples: %11"LPF64"u",
- cntr_header->lc_name,
- width(cntr_header->lc_name, 15), spaces,
- ret.lc_count);
- if (cntr_header->lc_units[0] != '\0')
- seq_printf(p, ", unit: %5s", cntr_header->lc_units);
-
- if (cntr_header->lc_config & LPROCFS_CNTR_AVGMINMAX) {
- seq_printf(p, ", min:%8"LPF64"u, max:%8"LPF64"u,"
- " sum:%16"LPF64"u",
- ret.lc_count ? ret.lc_min : 0,
- ret.lc_count ? ret.lc_max : 0,
- ret.lc_count ? ret.lc_sum : 0);
- }
- if (cntr_header->lc_config & LPROCFS_CNTR_STDDEV) {
- seq_printf(p, ", sumsq: %18"LPF64"u",
- ret.lc_count ? ret.lc_sumsquare : 0);
- }
-
- seq_printf(p, " }\n");
-
- }
- return 0;
-}
-
-struct seq_operations lprocfs_jobstats_seq_sops = {
- start: lprocfs_jobstats_seq_start,
- stop: lprocfs_jobstats_seq_stop,
- next: lprocfs_jobstats_seq_next,
- show: lprocfs_jobstats_seq_show,
-};
-
-static int lprocfs_jobstats_seq_open(struct inode *inode, struct file *file)
-{
- struct seq_file *seq;
- int rc;
-
- rc = seq_open(file, &lprocfs_jobstats_seq_sops);
- if (rc)
- return rc;
- seq = file->private_data;
- seq->private = PDE_DATA(inode);
- return 0;
-}
-
-static ssize_t lprocfs_jobstats_seq_write(struct file *file, const char *buf,
- size_t len, loff_t *off)
-{
- struct seq_file *seq = file->private_data;
- struct obd_job_stats *stats = seq->private;
- char jobid[JOBSTATS_JOBID_SIZE];
- int all = 0;
- struct job_stat *job;
-
- if (!memcmp(buf, "clear", strlen("clear"))) {
- all = 1;
- } else if (len < JOBSTATS_JOBID_SIZE) {
- memset(jobid, 0, JOBSTATS_JOBID_SIZE);
- /* Trim '\n' if any */
- if (buf[len - 1] == '\n')
- memcpy(jobid, buf, len - 1);
- else
- memcpy(jobid, buf, len);
- } else {
- return -EINVAL;
- }
-
- LASSERT(stats->ojs_hash);
- if (all) {
- time_t oldest = 0;
- cfs_hash_for_each_safe(stats->ojs_hash, job_iter_callback,
- &oldest);
- return len;
- }
-
- if (!strlen(jobid))
- return -EINVAL;
-
- job = cfs_hash_lookup(stats->ojs_hash, jobid);
- if (!job)
- return -EINVAL;
-
- cfs_hash_del_key(stats->ojs_hash, jobid);
-
- job_putref(job);
- return len;
-}
-
-struct file_operations lprocfs_jobstats_seq_fops = {
- .owner = THIS_MODULE,
- .open = lprocfs_jobstats_seq_open,
- .read = seq_read,
- .write = lprocfs_jobstats_seq_write,
- .llseek = seq_lseek,
- .release = lprocfs_seq_release,
-};
-
-int lprocfs_job_stats_init(struct obd_device *obd, int cntr_num,
- cntr_init_callback init_fn)
-{
- struct proc_dir_entry *entry;
- struct obd_job_stats *stats;
- ENTRY;
-
- LASSERT(obd->obd_proc_entry != NULL);
- LASSERT(obd->obd_type->typ_name);
-
- if (strcmp(obd->obd_type->typ_name, LUSTRE_MDT_NAME) &&
- strcmp(obd->obd_type->typ_name, LUSTRE_OST_NAME)) {
- CERROR("Invalid obd device type.\n");
- RETURN(-EINVAL);
- }
- stats = &obd->u.obt.obt_jobstats;
-
- LASSERT(stats->ojs_hash == NULL);
- stats->ojs_hash = cfs_hash_create("JOB_STATS",
- HASH_JOB_STATS_CUR_BITS,
- HASH_JOB_STATS_MAX_BITS,
- HASH_JOB_STATS_BKT_BITS, 0,
- CFS_HASH_MIN_THETA,
- CFS_HASH_MAX_THETA,
- &job_stats_hash_ops,
- CFS_HASH_DEFAULT);
- if (stats->ojs_hash == NULL)
- RETURN(-ENOMEM);
-
- INIT_LIST_HEAD(&stats->ojs_list);
- rwlock_init(&stats->ojs_lock);
- stats->ojs_cntr_num = cntr_num;
- stats->ojs_cntr_init_fn = init_fn;
- stats->ojs_cleanup_interval = 600; /* 10 mins by default */
- stats->ojs_last_cleanup = cfs_time_current_sec();
-
- entry = proc_create_data("job_stats", 0644, obd->obd_proc_entry,
- &lprocfs_jobstats_seq_fops, stats);
- if (entry)
- RETURN(0);
- else
- RETURN(-ENOMEM);
-}
-EXPORT_SYMBOL(lprocfs_job_stats_init);
-
-int lprocfs_rd_job_interval(struct seq_file *m, void *data)
-{
- struct obd_device *obd = (struct obd_device *)data;
- struct obd_job_stats *stats;
-
- LASSERT(obd != NULL);
- stats = &obd->u.obt.obt_jobstats;
- return seq_printf(m, "%d\n", stats->ojs_cleanup_interval);
-}
-EXPORT_SYMBOL(lprocfs_rd_job_interval);
-
-int lprocfs_wr_job_interval(struct file *file, const char *buffer,
- unsigned long count, void *data)
-{
- struct obd_device *obd = (struct obd_device *)data;
- struct obd_job_stats *stats;
- int val, rc;
-
- LASSERT(obd != NULL);
- stats = &obd->u.obt.obt_jobstats;
-
- rc = lprocfs_write_helper(buffer, count, &val);
- if (rc)
- return rc;
-
- stats->ojs_cleanup_interval = val;
- lprocfs_job_cleanup(stats, true);
-
- return count;
-
-}
-EXPORT_SYMBOL(lprocfs_wr_job_interval);
-
-#endif /* LPROCFS*/
diff --git a/drivers/staging/lustre/lustre/obdclass/obd_mount.c b/drivers/staging/lustre/lustre/obdclass/obd_mount.c
index 99adad9..4f20723 100644
--- a/drivers/staging/lustre/lustre/obdclass/obd_mount.c
+++ b/drivers/staging/lustre/lustre/obdclass/obd_mount.c
@@ -47,7 +47,6 @@
#include <obd.h>
#include <lvfs.h>
-#include <lustre_fsfilt.h>
#include <obd_class.h>
#include <lustre/lustre_user.h>
#include <linux/version.h>
--
1.7.9.5
^ permalink raw reply related [flat|nested] 62+ messages in thread* Re: [PATCH 36/48] staging/lustre/obdclass: add obd_target.h
2013-07-22 16:06 ` [PATCH 36/48] staging/lustre/obdclass: add obd_target.h Peng Tao
@ 2013-07-23 20:41 ` Greg Kroah-Hartman
0 siblings, 0 replies; 62+ messages in thread
From: Greg Kroah-Hartman @ 2013-07-23 20:41 UTC (permalink / raw)
To: Peng Tao
Cc: linux-kernel, Liu Xuezhao, John L. Hammond, Peng Tao, Andreas Dilger
On Tue, Jul 23, 2013 at 12:06:57AM +0800, Peng Tao wrote:
> From: Liu Xuezhao <xuezhao.liu@emc.com>
>
> Move some server-side data structures form obd.h to obd_target.h. To
> remove some code dependencies use HAVE_SERVER_SUPPORT to protect some
> llog/fsfilt_* functions.
>
> Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-1330
> Lustre-change: http://review.whamcloud.com/2678
> Signed-off-by: Liu Xuezhao <xuezhao.liu@emc.com>
> Signed-off-by: John L. Hammond <john.hammond@intel.com>
> Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
> Reviewed-by: Peng Tao <bergwolf@gmail.com>
> Reviewed-by: Mike Pershin <mike.pershin@intel.com>
> Signed-off-by: Peng Tao <tao.peng@emc.com>
> Signed-off-by: Andreas Dilger <andreas.dilger@intel.com>
> ---
> .../lustre/lustre/include/linux/lustre_fsfilt.h | 10 -
> .../staging/lustre/lustre/include/lprocfs_status.h | 36 --
> drivers/staging/lustre/lustre/include/obd.h | 140 +----
> drivers/staging/lustre/lustre/lvfs/lvfs_linux.c | 1 -
> drivers/staging/lustre/lustre/obdclass/Makefile | 2 +-
This patch has too much fuzz to apply properly, please refresh and
resend.
thanks,
greg k-h
^ permalink raw reply [flat|nested] 62+ messages in thread
* [PATCH 37/48] staging/lustre/procfs: return -ENOMEM from lprocfs_register()
2013-07-22 16:06 [PATCH 00/48] staging/lustre: minor cleanup and Intel Lustre sync Peng Tao
` (35 preceding siblings ...)
2013-07-22 16:06 ` [PATCH 36/48] staging/lustre/obdclass: add obd_target.h Peng Tao
@ 2013-07-22 16:06 ` Peng Tao
2013-07-22 16:06 ` [PATCH 38/48] staging/lustre/lmv: fix duplicate directory entries Peng Tao
` (11 subsequent siblings)
48 siblings, 0 replies; 62+ messages in thread
From: Peng Tao @ 2013-07-22 16:06 UTC (permalink / raw)
To: Greg Kroah-Hartman
Cc: linux-kernel, John L. Hammond, Peng Tao, Andreas Dilger
From: "John L. Hammond" <john.hammond@intel.com>
In lprocfs_register(), if proc_mkdir() fails then return
ERR_PTR(-ENOMEM) rather than NULL and hold _lprocfs_mutex for the
whole function. In lprocfs_remove_nolock() return early if the entry
is an error pointer. Improve error handling around lprocfs_register()
in a few spots.
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-2650
Lustre-change: http://review.whamcloud.com/5161
Signed-off-by: John L. Hammond <john.hammond@intel.com>
Reviewed-by: Emoly Liu <emoly.liu@intel.com>
Reviewed-by: Keith Mannthey <keith.mannthey@intel.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Signed-off-by: Peng Tao <tao.peng@emc.com>
Signed-off-by: Andreas Dilger <andreas.dilger@intel.com>
---
drivers/staging/lustre/lustre/ldlm/ldlm_pool.c | 1 +
drivers/staging/lustre/lustre/llite/super25.c | 2 +-
drivers/staging/lustre/lustre/lov/lov_obd.c | 6 ++---
.../lustre/lustre/obdclass/linux/linux-module.c | 9 ++++++-
.../lustre/lustre/obdclass/lprocfs_status.c | 28 ++++++++++++--------
.../staging/lustre/lustre/ptlrpc/gss/lproc_gss.c | 6 +++--
6 files changed, 34 insertions(+), 18 deletions(-)
diff --git a/drivers/staging/lustre/lustre/ldlm/ldlm_pool.c b/drivers/staging/lustre/lustre/ldlm/ldlm_pool.c
index 3277a0d..101af4b 100644
--- a/drivers/staging/lustre/lustre/ldlm/ldlm_pool.c
+++ b/drivers/staging/lustre/lustre/ldlm/ldlm_pool.c
@@ -750,6 +750,7 @@ static int ldlm_pool_proc_init(struct ldlm_pool *pl)
if (IS_ERR(pl->pl_proc_dir)) {
CERROR("LProcFS failed in ldlm-pool-init\n");
rc = PTR_ERR(pl->pl_proc_dir);
+ pl->pl_proc_dir = NULL;
GOTO(out_free_name, rc);
}
diff --git a/drivers/staging/lustre/lustre/llite/super25.c b/drivers/staging/lustre/lustre/llite/super25.c
index 82c14a9..ea06f1a 100644
--- a/drivers/staging/lustre/lustre/llite/super25.c
+++ b/drivers/staging/lustre/lustre/llite/super25.c
@@ -214,7 +214,7 @@ static void __exit exit_lustre_lite(void)
ll_remote_perm_cachep = NULL;
kmem_cache_destroy(ll_file_data_slab);
- if (proc_lustre_fs_root)
+ if (proc_lustre_fs_root && !IS_ERR(proc_lustre_fs_root))
lprocfs_remove(&proc_lustre_fs_root);
}
diff --git a/drivers/staging/lustre/lustre/lov/lov_obd.c b/drivers/staging/lustre/lustre/lov/lov_obd.c
index 0a9e40d..c7c1a8c 100644
--- a/drivers/staging/lustre/lustre/lov/lov_obd.c
+++ b/drivers/staging/lustre/lustre/lov/lov_obd.c
@@ -836,11 +836,11 @@ int lov_setup(struct obd_device *obd, struct lustre_cfg *lcfg)
lprocfs_obd_setup(obd, lvars.obd_vars);
#ifdef LPROCFS
{
- int rc;
+ int rc1;
- rc = lprocfs_seq_create(obd->obd_proc_entry, "target_obd",
+ rc1 = lprocfs_seq_create(obd->obd_proc_entry, "target_obd",
0444, &lov_proc_target_fops, obd);
- if (rc)
+ if (rc1)
CWARN("Error adding the target_obd file\n");
}
#endif
diff --git a/drivers/staging/lustre/lustre/obdclass/linux/linux-module.c b/drivers/staging/lustre/lustre/obdclass/linux/linux-module.c
index d2c3072..2abacf2 100644
--- a/drivers/staging/lustre/lustre/obdclass/linux/linux-module.c
+++ b/drivers/staging/lustre/lustre/obdclass/linux/linux-module.c
@@ -385,14 +385,21 @@ struct file_operations obd_device_list_fops = {
int class_procfs_init(void)
{
- int rc;
+ int rc = 0;
ENTRY;
obd_sysctl_init();
proc_lustre_root = lprocfs_register("fs/lustre", NULL,
lprocfs_base, NULL);
+ if (IS_ERR(proc_lustre_root)) {
+ rc = PTR_ERR(proc_lustre_root);
+ proc_lustre_root = NULL;
+ goto out;
+ }
+
rc = lprocfs_seq_create(proc_lustre_root, "devices", 0444,
&obd_device_list_fops, NULL);
+out:
if (rc)
CERROR("error adding /proc/fs/lustre/devices file\n");
RETURN(0);
diff --git a/drivers/staging/lustre/lustre/obdclass/lprocfs_status.c b/drivers/staging/lustre/lustre/obdclass/lprocfs_status.c
index 85163f4..e57a922 100644
--- a/drivers/staging/lustre/lustre/obdclass/lprocfs_status.c
+++ b/drivers/staging/lustre/lustre/obdclass/lprocfs_status.c
@@ -179,17 +179,21 @@ struct proc_dir_entry *lprocfs_register(const char *name,
struct proc_dir_entry *parent,
struct lprocfs_vars *list, void *data)
{
- struct proc_dir_entry *newchild;
-
- newchild = proc_mkdir(name, parent);
- if (newchild != NULL && list != NULL) {
- int rc = lprocfs_add_vars(newchild, list, data);
- if (rc) {
- lprocfs_remove(&newchild);
- return ERR_PTR(rc);
+ struct proc_dir_entry *entry;
+
+ entry = proc_mkdir(name, parent);
+ if (entry == NULL)
+ GOTO(out, entry = ERR_PTR(-ENOMEM));
+
+ if (list != NULL) {
+ int rc = lprocfs_add_vars(entry, list, data);
+ if (rc != 0) {
+ lprocfs_remove(&entry);
+ entry = ERR_PTR(rc);
}
}
- return newchild;
+out:
+ return entry;
}
EXPORT_SYMBOL(lprocfs_register);
@@ -1596,10 +1600,12 @@ int lprocfs_exp_setup(struct obd_export *exp, lnet_nid_t *nid, int *newnid)
NULL, NULL);
OBD_FREE(buffer, LNET_NIDSTR_SIZE);
- if (new_stat->nid_proc == NULL) {
+ if (IS_ERR(new_stat->nid_proc)) {
CERROR("Error making export directory for nid %s\n",
libcfs_nid2str(*nid));
- GOTO(destroy_new_ns, rc = -ENOMEM);
+ rc = PTR_ERR(new_stat->nid_proc);
+ new_stat->nid_proc = NULL;
+ GOTO(destroy_new_ns, rc);
}
entry = lprocfs_add_simple(new_stat->nid_proc, "uuid",
diff --git a/drivers/staging/lustre/lustre/ptlrpc/gss/lproc_gss.c b/drivers/staging/lustre/lustre/ptlrpc/gss/lproc_gss.c
index 3404000..de100a1 100644
--- a/drivers/staging/lustre/lustre/ptlrpc/gss/lproc_gss.c
+++ b/drivers/staging/lustre/lustre/ptlrpc/gss/lproc_gss.c
@@ -199,15 +199,17 @@ int gss_init_lproc(void)
gss_proc_root = lprocfs_register("gss", sptlrpc_proc_root,
gss_lprocfs_vars, NULL);
if (IS_ERR(gss_proc_root)) {
+ rc = PTR_ERR(gss_proc_root);
gss_proc_root = NULL;
- GOTO(err_out, rc = PTR_ERR(gss_proc_root));
+ GOTO(err_out, rc);
}
gss_proc_lk = lprocfs_register("lgss_keyring", gss_proc_root,
gss_lk_lprocfs_vars, NULL);
if (IS_ERR(gss_proc_lk)) {
+ rc = PTR_ERR(gss_proc_lk);
gss_proc_lk = NULL;
- GOTO(err_out, rc = PTR_ERR(gss_proc_root));
+ GOTO(err_out, rc);
}
return 0;
--
1.7.9.5
^ permalink raw reply related [flat|nested] 62+ messages in thread* [PATCH 38/48] staging/lustre/lmv: fix duplicate directory entries
2013-07-22 16:06 [PATCH 00/48] staging/lustre: minor cleanup and Intel Lustre sync Peng Tao
` (36 preceding siblings ...)
2013-07-22 16:06 ` [PATCH 37/48] staging/lustre/procfs: return -ENOMEM from lprocfs_register() Peng Tao
@ 2013-07-22 16:06 ` Peng Tao
2013-07-22 16:07 ` [PATCH 39/48] staging/lustre/obdclass: be more careful processing server name Peng Tao
` (10 subsequent siblings)
48 siblings, 0 replies; 62+ messages in thread
From: Peng Tao @ 2013-07-22 16:06 UTC (permalink / raw)
To: Greg Kroah-Hartman; +Cc: linux-kernel, Ned Bass, Peng Tao, Andreas Dilger
From: Ned Bass <bass6@llnl.gov>
Previously, I accidentally introduced a new way for duplicate
directory entries to be returned from readdir(). That patch fails
to properly decrement the nlupgs counter when breaking out of the
inner-for loop. This accounting error causes an extra iteration
of the inner-for loop when processing the next cfs page and a bad
ldp_hash_end value is then saved in the lu_dirpage. To fix this,
always decrement the nlupgs counter on entry into the inner loop.
Note: this bug only affects architectures with > 4k-sized pages, e.g.
PowerPC.
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-3182
Lustre-change: http://review.whamcloud.com/6405
Signed-off-by: Ned Bass <bass6@llnl.gov>
Reviewed-by: Fan Yong <fan.yong@intel.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Reviewed-by: Bobi Jam <bobijam@gmail.com>
Signed-off-by: Peng Tao <tao.peng@emc.com>
Signed-off-by: Andreas Dilger <andreas.dilger@intel.com>
---
drivers/staging/lustre/lustre/lmv/lmv_obd.c | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/drivers/staging/lustre/lustre/lmv/lmv_obd.c b/drivers/staging/lustre/lustre/lmv/lmv_obd.c
index 9e68eea..11c8d64 100644
--- a/drivers/staging/lustre/lustre/lmv/lmv_obd.c
+++ b/drivers/staging/lustre/lustre/lmv/lmv_obd.c
@@ -2157,7 +2157,7 @@ static void lmv_adjust_dirpages(struct page **pages, int ncfspgs, int nlupgs)
__u64 hash_end = dp->ldp_hash_end;
__u32 flags = dp->ldp_flags;
- for (; nlupgs > 1; nlupgs--) {
+ while (--nlupgs > 0) {
ent = lu_dirent_start(dp);
for (end_dirent = ent; ent != NULL;
end_dirent = ent, ent = lu_dirent_next(ent));
@@ -2191,6 +2191,7 @@ static void lmv_adjust_dirpages(struct page **pages, int ncfspgs, int nlupgs)
kunmap(pages[i]);
}
+ LASSERTF(nlupgs == 0, "left = %d", nlupgs);
}
#else
#define lmv_adjust_dirpages(pages, ncfspgs, nlupgs) do {} while (0)
--
1.7.9.5
^ permalink raw reply related [flat|nested] 62+ messages in thread* [PATCH 39/48] staging/lustre/obdclass: be more careful processing server name
2013-07-22 16:06 [PATCH 00/48] staging/lustre: minor cleanup and Intel Lustre sync Peng Tao
` (37 preceding siblings ...)
2013-07-22 16:06 ` [PATCH 38/48] staging/lustre/lmv: fix duplicate directory entries Peng Tao
@ 2013-07-22 16:07 ` Peng Tao
2013-07-22 16:07 ` [PATCH 40/48] staging/lustre/llite: return valid fsid for statfs Peng Tao
` (9 subsequent siblings)
48 siblings, 0 replies; 62+ messages in thread
From: Peng Tao @ 2013-07-22 16:07 UTC (permalink / raw)
To: Greg Kroah-Hartman
Cc: linux-kernel, Nathaniel Clark, Peng Tao, Andreas Dilger
From: Nathaniel Clark <nathaniel.l.clark@intel.com>
Because whole options line gets passed to exclude processing, don't
search from end of passed in argument to determine fsname at
beginning.
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-2200
Lustre-change: http://review.whamcloud.com/6197
Signed-off-by: Nathaniel Clark <nathaniel.l.clark@intel.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Reviewed-by: Alex Zhuravlev <alexey.zhuravlev@intel.com>
Signed-off-by: Peng Tao <tao.peng@emc.com>
Signed-off-by: Andreas Dilger <andreas.dilger@intel.com>
---
.../staging/lustre/lustre/obdclass/obd_config.c | 11 ++--
drivers/staging/lustre/lustre/obdclass/obd_mount.c | 58 +++++++++++---------
2 files changed, 40 insertions(+), 29 deletions(-)
diff --git a/drivers/staging/lustre/lustre/obdclass/obd_config.c b/drivers/staging/lustre/lustre/obdclass/obd_config.c
index bbf06d0..8b5eb6c 100644
--- a/drivers/staging/lustre/lustre/obdclass/obd_config.c
+++ b/drivers/staging/lustre/lustre/obdclass/obd_config.c
@@ -1426,10 +1426,13 @@ int class_config_llog_handler(const struct lu_env *env,
}
- if ((clli->cfg_flags & CFG_F_EXCLUDE) &&
- (lcfg->lcfg_command == LCFG_LOV_ADD_OBD))
- /* Add inactive instead */
- lcfg->lcfg_command = LCFG_LOV_ADD_INA;
+ if (clli->cfg_flags & CFG_F_EXCLUDE) {
+ CDEBUG(D_CONFIG, "cmd: %x marked EXCLUDED\n",
+ lcfg->lcfg_command);
+ if (lcfg->lcfg_command == LCFG_LOV_ADD_OBD)
+ /* Add inactive instead */
+ lcfg->lcfg_command = LCFG_LOV_ADD_INA;
+ }
lustre_cfg_bufs_init(&bufs, lcfg);
diff --git a/drivers/staging/lustre/lustre/obdclass/obd_mount.c b/drivers/staging/lustre/lustre/obdclass/obd_mount.c
index 4f20723..2cc6a26 100644
--- a/drivers/staging/lustre/lustre/obdclass/obd_mount.c
+++ b/drivers/staging/lustre/lustre/obdclass/obd_mount.c
@@ -649,6 +649,15 @@ int lustre_put_lsi(struct super_block *sb)
RETURN(0);
}
+/*** SERVER NAME ***
+ * <FSNAME><SEPERATOR><TYPE><INDEX>
+ * FSNAME is between 1 and 8 characters (inclusive).
+ * Excluded characters are '/' and ':'
+ * SEPERATOR is either ':' or '-'
+ * TYPE: "OST", "MDT", etc.
+ * INDEX: Hex representation of the index
+ */
+
/** Get the fsname ("lustre") from the server name ("lustre-OST003F").
* @param [in] svname server name including type and index
* @param [out] fsname Buffer to copy filesystem name prefix into.
@@ -658,22 +667,13 @@ int lustre_put_lsi(struct super_block *sb)
*/
int server_name2fsname(const char *svname, char *fsname, const char **endptr)
{
- const char *dash = strrchr(svname, '-');
- if (!dash) {
- dash = strrchr(svname, ':');
- if (!dash)
- return -EINVAL;
- }
+ const char *dash;
- /* interpret <fsname>-MDTXXXXX-mdc as mdt, the better way is to pass
- * in the fsname, then determine the server index */
- if (!strcmp(LUSTRE_MDC_NAME, dash + 1)) {
- dash--;
- for (; dash > svname && *dash != '-' && *dash != ':'; dash--)
- ;
- if (dash == svname)
- return -EINVAL;
- }
+ dash = svname + strnlen(svname, 8); /* max fsname length is 8 */
+ for (; dash > svname && *dash != '-' && *dash != ':'; dash--)
+ ;
+ if (dash == svname)
+ return -EINVAL;
if (fsname != NULL) {
strncpy(fsname, svname, dash - svname);
@@ -696,15 +696,15 @@ int server_name2svname(const char *label, char *svname, const char **endptr,
size_t svsize)
{
int rc;
- const const char *dash;
+ const char *dash;
/* We use server_name2fsname() just for parsing */
rc = server_name2fsname(label, NULL, &dash);
if (rc != 0)
return rc;
- if (*dash != '-')
- return -1;
+ if (endptr != NULL)
+ *endptr = dash;
if (strlcpy(svname, dash + 1, svsize) >= svsize)
return -E2BIG;
@@ -729,9 +729,6 @@ int server_name2index(const char *svname, __u32 *idx, const char **endptr)
if (rc != 0)
return rc;
- if (*dash != '-')
- return -EINVAL;
-
dash++;
if (strncmp(dash, "MDT", 3) == 0)
@@ -743,11 +740,20 @@ int server_name2index(const char *svname, __u32 *idx, const char **endptr)
dash += 3;
- if (strcmp(dash, "all") == 0)
+ if (strncmp(dash, "all", 3) == 0) {
+ if (endptr != NULL)
+ *endptr = dash + 3;
return rc | LDD_F_SV_ALL;
+ }
index = simple_strtoul(dash, (char **)endptr, 16);
- *idx = index;
+ if (idx != NULL)
+ *idx = index;
+
+ /* Account for -mdc after index that is possible when specifying mdt */
+ if (endptr != NULL && strncmp(LUSTRE_MDC_NAME, *endptr + 1,
+ sizeof(LUSTRE_MDC_NAME)-1) == 0)
+ *endptr += sizeof(LUSTRE_MDC_NAME);
return rc;
}
@@ -857,13 +863,15 @@ static int lmd_make_exclusion(struct lustre_mount_data *lmd, const char *ptr)
s1++;
rc = server_name2index(s1, &index, &s2);
if (rc < 0) {
- CERROR("Can't parse server name '%s'\n", s1);
+ CERROR("Can't parse server name '%s': rc = %d\n",
+ s1, rc);
break;
}
if (rc == LDD_F_SV_TYPE_OST)
exclude_list[lmd->lmd_exclude_count++] = index;
else
- CDEBUG(D_MOUNT, "ignoring exclude %.7s\n", s1);
+ CDEBUG(D_MOUNT, "ignoring exclude %.*s: type = %#x\n",
+ (uint)(s2-s1), s1, rc);
s1 = s2;
/* now we are pointing at ':' (next exclude)
or ',' (end of excludes) */
--
1.7.9.5
^ permalink raw reply related [flat|nested] 62+ messages in thread* [PATCH 40/48] staging/lustre/llite: return valid fsid for statfs
2013-07-22 16:06 [PATCH 00/48] staging/lustre: minor cleanup and Intel Lustre sync Peng Tao
` (38 preceding siblings ...)
2013-07-22 16:07 ` [PATCH 39/48] staging/lustre/obdclass: be more careful processing server name Peng Tao
@ 2013-07-22 16:07 ` Peng Tao
2013-07-22 16:07 ` [PATCH 41/48] staging/lustre/llite: error of listxattr when buffer is small Peng Tao
` (8 subsequent siblings)
48 siblings, 0 replies; 62+ messages in thread
From: Peng Tao @ 2013-07-22 16:07 UTC (permalink / raw)
To: Greg Kroah-Hartman
Cc: linux-kernel, Fan Yong, Fan Yong, Peng Tao, Andreas Dilger
From: Fan Yong <yong.fan@whamcloud.com>
Lustre uses 64-bits inode number to identify object on client side.
When re-export Lustre via NFS, NFS will detect whether support fsid
via statfs(). For the non-support case, it will only recognizes and
packs low 32-bits inode number in nfs handle. Such handle cannot be
used to locate the object properly.
To avoid patch linux kernel, Lustre client should generate fsid and
return it via statfs() to up layer.
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-2904
Lustre-change: http://review.whamcloud.com/6493
Signed-off-by: Fan Yong <fan.yong@intel.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Reviewed-by: Lai Siyao <lai.siyao@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>
Signed-off-by: Peng Tao <tao.peng@emc.com>
Signed-off-by: Andreas Dilger <andreas.dilger@intel.com>
---
.../staging/lustre/lustre/llite/llite_internal.h | 2 ++
drivers/staging/lustre/lustre/llite/llite_lib.c | 6 ++++--
drivers/staging/lustre/lustre/llite/llite_nfs.c | 16 ++++++++++++++++
3 files changed, 22 insertions(+), 2 deletions(-)
diff --git a/drivers/staging/lustre/lustre/llite/llite_internal.h b/drivers/staging/lustre/lustre/llite/llite_internal.h
index 0534665..28aed68 100644
--- a/drivers/staging/lustre/lustre/llite/llite_internal.h
+++ b/drivers/staging/lustre/lustre/llite/llite_internal.h
@@ -505,6 +505,7 @@ struct ll_sb_info {
* clustred nfs */
struct rmtacl_ctl_table ll_rct;
struct eacl_table ll_et;
+ __kernel_fsid_t ll_fsid;
};
#define LL_DEFAULT_MAX_RW_CHUNK (32 * 1024 * 1024)
@@ -833,6 +834,7 @@ char *ll_get_fsname(struct super_block *sb, char *buf, int buflen);
/* llite/llite_nfs.c */
extern struct export_operations lustre_export_operations;
__u32 get_uuid2int(const char *name, int len);
+void get_uuid2fsid(const char *name, int len, __kernel_fsid_t *fsid);
struct inode *search_inode_for_lustre(struct super_block *sb,
const struct lu_fid *fid);
diff --git a/drivers/staging/lustre/lustre/llite/llite_lib.c b/drivers/staging/lustre/lustre/llite/llite_lib.c
index 39d519b..d7d87b1 100644
--- a/drivers/staging/lustre/lustre/llite/llite_lib.c
+++ b/drivers/staging/lustre/lustre/llite/llite_lib.c
@@ -583,8 +583,10 @@ static int client_common_fill_super(struct super_block *sb, char *md, char *dt,
/* s_dev is also used in lt_compare() to compare two fs, but that is
* only a node-local comparison. */
uuid = obd_get_uuid(sbi->ll_md_exp);
- if (uuid != NULL)
+ if (uuid != NULL) {
sb->s_dev = get_uuid2int(uuid->uuid, strlen(uuid->uuid));
+ get_uuid2fsid(uuid->uuid, strlen(uuid->uuid), &sbi->ll_fsid);
+ }
if (data != NULL)
OBD_FREE_PTR(data);
@@ -1615,7 +1617,7 @@ int ll_statfs(struct dentry *de, struct kstatfs *sfs)
sfs->f_blocks = osfs.os_blocks;
sfs->f_bfree = osfs.os_bfree;
sfs->f_bavail = osfs.os_bavail;
-
+ sfs->f_fsid = ll_s2sbi(sb)->ll_fsid;
return 0;
}
diff --git a/drivers/staging/lustre/lustre/llite/llite_nfs.c b/drivers/staging/lustre/lustre/llite/llite_nfs.c
index f142a1e..b830004 100644
--- a/drivers/staging/lustre/lustre/llite/llite_nfs.c
+++ b/drivers/staging/lustre/lustre/llite/llite_nfs.c
@@ -58,6 +58,22 @@ __u32 get_uuid2int(const char *name, int len)
return (key0 << 1);
}
+void get_uuid2fsid(const char *name, int len, __kernel_fsid_t *fsid)
+{
+ __u64 key = 0, key0 = 0x12a3fe2d, key1 = 0x37abe8f9;
+
+ while (len--) {
+ key = key1 + (key0 ^ (*name++ * 7152373));
+ if (key & 0x8000000000000000ULL)
+ key -= 0x7fffffffffffffffULL;
+ key1 = key0;
+ key0 = key;
+ }
+
+ fsid->val[0] = key;
+ fsid->val[1] = key >> 32;
+}
+
static int ll_nfs_test_inode(struct inode *inode, void *opaque)
{
return lu_fid_eq(&ll_i2info(inode)->lli_fid,
--
1.7.9.5
^ permalink raw reply related [flat|nested] 62+ messages in thread* [PATCH 41/48] staging/lustre/llite: error of listxattr when buffer is small
2013-07-22 16:06 [PATCH 00/48] staging/lustre: minor cleanup and Intel Lustre sync Peng Tao
` (39 preceding siblings ...)
2013-07-22 16:07 ` [PATCH 40/48] staging/lustre/llite: return valid fsid for statfs Peng Tao
@ 2013-07-22 16:07 ` Peng Tao
2013-07-22 16:07 ` [PATCH 42/48] staging/lustre/llite: Anonymous dentry incorrectly identified as root Peng Tao
` (7 subsequent siblings)
48 siblings, 0 replies; 62+ messages in thread
From: Peng Tao @ 2013-07-22 16:07 UTC (permalink / raw)
To: Greg Kroah-Hartman; +Cc: linux-kernel, Keith Mannthey, Peng Tao, Andreas Dilger
From: Keith Mannthey <keith.mannthey@intel.com>
According to the standard, listxattr(2) should return -1
and errno should be set to ERANGE if the size of the list
buffer is too small to hold the result. However ll_listxattr()
will return a value bigger than the size of buffer in some cases.
Let's assume listxattr(2) returns SIZE when it is called with a
large enough list buffer. If it's called again with a list buffer
whose size is smaller than SIZE but bigger than (SIZE - 12), then
listxattr(2) will return SIZE too. This patch fixes the problem.
Original patch by Li Xi <pkuelelixi@gmail.com>
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-3403
Lustre-change: http://review.whamcloud.com/6463
Signed-off-by: Keith Mannthey <keith.mannthey@intel.com>
Reviewed-by: Li Xi <pkuelelixi@gmail.com>
Reviewed-by: Dmitry Eremin <dmitry.eremin@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>
Signed-off-by: Peng Tao <tao.peng@emc.com>
Signed-off-by: Andreas Dilger <andreas.dilger@intel.com>
---
drivers/staging/lustre/lustre/llite/xattr.c | 7 ++++++-
1 file changed, 6 insertions(+), 1 deletion(-)
diff --git a/drivers/staging/lustre/lustre/llite/xattr.c b/drivers/staging/lustre/lustre/llite/xattr.c
index 0a11902..2e3ff6d 100644
--- a/drivers/staging/lustre/lustre/llite/xattr.c
+++ b/drivers/staging/lustre/lustre/llite/xattr.c
@@ -564,7 +564,12 @@ ssize_t ll_listxattr(struct dentry *dentry, char *buffer, size_t size)
const size_t name_len = sizeof("lov") - 1;
const size_t total_len = prefix_len + name_len + 1;
- if (buffer && (rc + total_len) <= size) {
+ if (((rc + total_len) > size) && (buffer != NULL)) {
+ ptlrpc_req_finished(request);
+ return -ERANGE;
+ }
+
+ if (buffer != NULL) {
buffer += rc;
memcpy(buffer, XATTR_LUSTRE_PREFIX, prefix_len);
memcpy(buffer + prefix_len, "lov", name_len);
--
1.7.9.5
^ permalink raw reply related [flat|nested] 62+ messages in thread* [PATCH 42/48] staging/lustre/llite: Anonymous dentry incorrectly identified as root
2013-07-22 16:06 [PATCH 00/48] staging/lustre: minor cleanup and Intel Lustre sync Peng Tao
` (40 preceding siblings ...)
2013-07-22 16:07 ` [PATCH 41/48] staging/lustre/llite: error of listxattr when buffer is small Peng Tao
@ 2013-07-22 16:07 ` Peng Tao
2013-07-22 16:07 ` [PATCH 43/48] staging/lustre/build: fix 'data race condition' issues Peng Tao
` (6 subsequent siblings)
48 siblings, 0 replies; 62+ messages in thread
From: Peng Tao @ 2013-07-22 16:07 UTC (permalink / raw)
To: Greg Kroah-Hartman
Cc: linux-kernel, Patrick Farrell, Peng Tao, Andreas Dilger
From: Patrick Farrell <paf@cray.com>
When exporting Lustre via NFS on SLES11SP2, this check incorrectly identifies
anonymous dentries as root dentries. This is due to a change in SLES11SP2
which makes the d_name.name for anonymous dentries the same as that for root
dentries. (Details in LU-3484.)
This changes the check to directly compare the value of the dentry pointer
to the root dentry pointer found in the superblock, rather than using the name.
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-3484
Lustre-change: http://review.whamcloud.com/6726
Signed-off-by: Patrick Farrell <paf@cray.com>
Reviewed-by: Bob Glossman <bob.glossman@intel.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Reviewed-by: Peng Tao <bergwolf@gmail.com>
Reviewed-by: James Simmons <uja.ornl@gmail.com>
Reviewed-by: Alexey Shvetsov <alexxy@gentoo.org>
Signed-off-by: Peng Tao <tao.peng@emc.com>
Signed-off-by: Andreas Dilger <andreas.dilger@intel.com>
---
drivers/staging/lustre/lustre/llite/dcache.c | 7 ++++---
1 file changed, 4 insertions(+), 3 deletions(-)
diff --git a/drivers/staging/lustre/lustre/llite/dcache.c b/drivers/staging/lustre/lustre/llite/dcache.c
index a160a84..0d49e1b 100644
--- a/drivers/staging/lustre/lustre/llite/dcache.c
+++ b/drivers/staging/lustre/lustre/llite/dcache.c
@@ -295,9 +295,10 @@ void ll_invalidate_aliases(struct inode *inode)
dentry->d_name.name, dentry, dentry->d_parent,
dentry->d_inode, dentry->d_flags);
- if (dentry->d_name.len == 1 && dentry->d_name.name[0] == '/') {
- CERROR("called on root (?) dentry=%p, inode=%p "
- "ino=%lu\n", dentry, inode, inode->i_ino);
+ if (unlikely(dentry == dentry->d_sb->s_root)) {
+ CERROR("%s: called on root dentry=%p, fid="DFID"\n",
+ ll_get_fsname(dentry->d_sb, NULL, 0),
+ dentry, PFID(ll_inode2fid(inode)));
lustre_dump_dentry(dentry, 1);
dump_stack();
}
--
1.7.9.5
^ permalink raw reply related [flat|nested] 62+ messages in thread* [PATCH 43/48] staging/lustre/build: fix 'data race condition' issues
2013-07-22 16:06 [PATCH 00/48] staging/lustre: minor cleanup and Intel Lustre sync Peng Tao
` (41 preceding siblings ...)
2013-07-22 16:07 ` [PATCH 42/48] staging/lustre/llite: Anonymous dentry incorrectly identified as root Peng Tao
@ 2013-07-22 16:07 ` Peng Tao
2013-07-22 16:07 ` [PATCH 44/48] " Peng Tao
` (5 subsequent siblings)
48 siblings, 0 replies; 62+ messages in thread
From: Peng Tao @ 2013-07-22 16:07 UTC (permalink / raw)
To: Greg Kroah-Hartman
Cc: linux-kernel, Sebastien Buisson, Peng Tao, Andreas Dilger
From: Sebastien Buisson <sebastien.buisson@bull.net>
Fix 'data race condition' defects found by Coverity version
6.5.0:
Data race condition (MISSING_LOCK)
Accessing variable without holding lock. Elsewhere,
this variable is accessed with lock held.
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-2744
Lustre-change: http://review.whamcloud.com/6569
Signed-off-by: Sebastien Buisson <sebastien.buisson@bull.net>
Reviewed-by: Bob Glossman <bob.glossman@intel.com>
Reviewed-by: Keith Mannthey <keith.mannthey@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>
Signed-off-by: Peng Tao <tao.peng@emc.com>
Signed-off-by: Andreas Dilger <andreas.dilger@intel.com>
---
.../staging/lustre/lustre/include/lustre_import.h | 2 ++
1 file changed, 2 insertions(+)
diff --git a/drivers/staging/lustre/lustre/include/lustre_import.h b/drivers/staging/lustre/lustre/include/lustre_import.h
index 3a5dd6a..67259eb 100644
--- a/drivers/staging/lustre/lustre/include/lustre_import.h
+++ b/drivers/staging/lustre/lustre/include/lustre_import.h
@@ -336,9 +336,11 @@ static inline unsigned int at_timeout2est(unsigned int val)
}
static inline void at_reset(struct adaptive_timeout *at, int val) {
+ spin_lock(&at->at_lock);
at->at_current = val;
at->at_worst_ever = val;
at->at_worst_time = cfs_time_current_sec();
+ spin_unlock(&at->at_lock);
}
static inline void at_init(struct adaptive_timeout *at, int val, int flags) {
memset(at, 0, sizeof(*at));
--
1.7.9.5
^ permalink raw reply related [flat|nested] 62+ messages in thread* [PATCH 44/48] staging/lustre/build: fix 'data race condition' issues
2013-07-22 16:06 [PATCH 00/48] staging/lustre: minor cleanup and Intel Lustre sync Peng Tao
` (42 preceding siblings ...)
2013-07-22 16:07 ` [PATCH 43/48] staging/lustre/build: fix 'data race condition' issues Peng Tao
@ 2013-07-22 16:07 ` Peng Tao
2013-07-23 20:44 ` Greg Kroah-Hartman
2013-07-22 16:07 ` [PATCH 45/48] staging/lustre/style: removes obsolete EXPORT_SYMTAB macros v2 Peng Tao
` (4 subsequent siblings)
48 siblings, 1 reply; 62+ messages in thread
From: Peng Tao @ 2013-07-22 16:07 UTC (permalink / raw)
To: Greg Kroah-Hartman
Cc: linux-kernel, Sebastien Buisson, Peng Tao, Andreas Dilger
From: Sebastien Buisson <sebastien.buisson@bull.net>
Fix 'data race condition' defects found by Coverity version
6.5.0:
Data race condition (MISSING_LOCK)
Accessing variable without holding lock. Elsewhere,
this variable is accessed with lock held.
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-2744
Lustre-change: http://review.whamcloud.com/6571
Signed-off-by: Sebastien Buisson <sebastien.buisson@bull.net>
Reviewed-by: Fan Yong <fan.yong@intel.com>
Reviewed-by: Keith Mannthey <keith.mannthey@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>
Signed-off-by: Peng Tao <tao.peng@emc.com>
Signed-off-by: Andreas Dilger <andreas.dilger@intel.com>
---
drivers/staging/lustre/lustre/llite/llite_lib.c | 4 ++++
drivers/staging/lustre/lustre/llite/namei.c | 5 ++++-
2 files changed, 8 insertions(+), 1 deletion(-)
diff --git a/drivers/staging/lustre/lustre/llite/llite_lib.c b/drivers/staging/lustre/lustre/llite/llite_lib.c
index d7d87b1..36a07fa 100644
--- a/drivers/staging/lustre/lustre/llite/llite_lib.c
+++ b/drivers/staging/lustre/lustre/llite/llite_lib.c
@@ -1190,7 +1190,9 @@ void ll_clear_inode(struct inode *inode)
LASSERT(lli->lli_opendir_pid == 0);
}
+ spin_lock(&lli->lli_lock);
ll_i2info(inode)->lli_flags &= ~LLIF_MDS_SIZE_LOCK;
+ spin_unlock(&lli->lli_lock);
md_null_inode(sbi->ll_md_exp, ll_inode2fid(inode));
LASSERT(!lli->lli_open_fd_write_count);
@@ -1757,7 +1759,9 @@ void ll_update_inode(struct inode *inode, struct lustre_md *md)
/* Use old size assignment to avoid
* deadlock bz14138 & bz14326 */
i_size_write(inode, body->size);
+ spin_lock(&lli->lli_lock);
lli->lli_flags |= LLIF_MDS_SIZE_LOCK;
+ spin_unlock(&lli->lli_lock);
}
ldlm_lock_decref(&lockh, mode);
}
diff --git a/drivers/staging/lustre/lustre/llite/namei.c b/drivers/staging/lustre/lustre/llite/namei.c
index 8e948a7..333e55f 100644
--- a/drivers/staging/lustre/lustre/llite/namei.c
+++ b/drivers/staging/lustre/lustre/llite/namei.c
@@ -275,8 +275,11 @@ int ll_md_blocking_ast(struct ldlm_lock *lock, struct ldlm_lock_desc *desc,
CDEBUG(D_INODE, "invaliding layout %d.\n", rc);
}
- if (bits & MDS_INODELOCK_UPDATE)
+ if (bits & MDS_INODELOCK_UPDATE) {
+ spin_lock(&lli->lli_lock);
lli->lli_flags &= ~LLIF_MDS_SIZE_LOCK;
+ spin_unlock(&lli->lli_lock);
+ }
if (S_ISDIR(inode->i_mode) &&
(bits & MDS_INODELOCK_UPDATE)) {
--
1.7.9.5
^ permalink raw reply related [flat|nested] 62+ messages in thread* Re: [PATCH 44/48] staging/lustre/build: fix 'data race condition' issues
2013-07-22 16:07 ` [PATCH 44/48] " Peng Tao
@ 2013-07-23 20:44 ` Greg Kroah-Hartman
0 siblings, 0 replies; 62+ messages in thread
From: Greg Kroah-Hartman @ 2013-07-23 20:44 UTC (permalink / raw)
To: Peng Tao; +Cc: linux-kernel, Sebastien Buisson, Peng Tao, Andreas Dilger
On Tue, Jul 23, 2013 at 12:07:05AM +0800, Peng Tao wrote:
> From: Sebastien Buisson <sebastien.buisson@bull.net>
>
> Fix 'data race condition' defects found by Coverity version
> 6.5.0:
> Data race condition (MISSING_LOCK)
> Accessing variable without holding lock. Elsewhere,
> this variable is accessed with lock held.
Again duplicate Subject: lines with the previous patch, so I've dropped
both.
thanks,
greg k-h
^ permalink raw reply [flat|nested] 62+ messages in thread
* [PATCH 45/48] staging/lustre/style: removes obsolete EXPORT_SYMTAB macros v2
2013-07-22 16:06 [PATCH 00/48] staging/lustre: minor cleanup and Intel Lustre sync Peng Tao
` (43 preceding siblings ...)
2013-07-22 16:07 ` [PATCH 44/48] " Peng Tao
@ 2013-07-22 16:07 ` Peng Tao
2013-07-22 16:07 ` [PATCH 46/48] staging/lustre/mdt: HSM on disk actions record Peng Tao
` (3 subsequent siblings)
48 siblings, 0 replies; 62+ messages in thread
From: Peng Tao @ 2013-07-22 16:07 UTC (permalink / raw)
To: Greg Kroah-Hartman; +Cc: linux-kernel, Keith Mannthey, Peng Tao, Andreas Dilger
From: Keith Mannthey <keith.mannthey@intel.com>
EXPORT_SYMTAB is obsolete after 2.4 kernel, this patch removes
EXPORT_SYMTAB from the source code again.
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-1347
Lustre-change: http://review.whamcloud.com/6739
Signed-off-by: Keith Mannthey <keith.mannthey@intel.com>
Reviewed-by: Peng Tao <bergwolf@gmail.com>
Reviewed-by: Liu Xuezhao <xuezhao.liu@emc.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Signed-off-by: Peng Tao <tao.peng@emc.com>
Signed-off-by: Andreas Dilger <andreas.dilger@intel.com>
---
drivers/staging/lustre/lustre/libcfs/libcfs_cpu.c | 3 ---
drivers/staging/lustre/lustre/libcfs/libcfs_lock.c | 3 ---
drivers/staging/lustre/lustre/libcfs/libcfs_mem.c | 3 ---
drivers/staging/lustre/lustre/obdclass/llog_osd.c | 4 ----
4 files changed, 13 deletions(-)
diff --git a/drivers/staging/lustre/lustre/libcfs/libcfs_cpu.c b/drivers/staging/lustre/lustre/libcfs/libcfs_cpu.c
index 8e88eb5..1fb3700 100644
--- a/drivers/staging/lustre/lustre/libcfs/libcfs_cpu.c
+++ b/drivers/staging/lustre/lustre/libcfs/libcfs_cpu.c
@@ -33,9 +33,6 @@
* Author: liang@whamcloud.com
*/
-#ifndef EXPORT_SYMTAB
-# define EXPORT_SYMTAB
-#endif
#define DEBUG_SUBSYSTEM S_LNET
#include <linux/libcfs/libcfs.h>
diff --git a/drivers/staging/lustre/lustre/libcfs/libcfs_lock.c b/drivers/staging/lustre/lustre/libcfs/libcfs_lock.c
index 8d6c4ad..a2ce4c0 100644
--- a/drivers/staging/lustre/lustre/libcfs/libcfs_lock.c
+++ b/drivers/staging/lustre/lustre/libcfs/libcfs_lock.c
@@ -30,9 +30,6 @@
* Author: liang@whamcloud.com
*/
-#ifndef EXPORT_SYMTAB
-# define EXPORT_SYMTAB
-#endif
#define DEBUG_SUBSYSTEM S_LNET
#include <linux/libcfs/libcfs.h>
diff --git a/drivers/staging/lustre/lustre/libcfs/libcfs_mem.c b/drivers/staging/lustre/lustre/libcfs/libcfs_mem.c
index 8791373..feab537 100644
--- a/drivers/staging/lustre/lustre/libcfs/libcfs_mem.c
+++ b/drivers/staging/lustre/lustre/libcfs/libcfs_mem.c
@@ -31,9 +31,6 @@
* Author: liang@whamcloud.com
*/
-#ifndef EXPORT_SYMTAB
-# define EXPORT_SYMTAB
-#endif
#define DEBUG_SUBSYSTEM S_LNET
#include <linux/libcfs/libcfs.h>
diff --git a/drivers/staging/lustre/lustre/obdclass/llog_osd.c b/drivers/staging/lustre/lustre/obdclass/llog_osd.c
index 6dbd21a..87d5012 100644
--- a/drivers/staging/lustre/lustre/obdclass/llog_osd.c
+++ b/drivers/staging/lustre/lustre/obdclass/llog_osd.c
@@ -41,10 +41,6 @@
#define DEBUG_SUBSYSTEM S_LOG
-#ifndef EXPORT_SYMTAB
-#define EXPORT_SYMTAB
-#endif
-
#include <obd.h>
#include <obd_class.h>
#include <lustre_fid.h>
--
1.7.9.5
^ permalink raw reply related [flat|nested] 62+ messages in thread* [PATCH 46/48] staging/lustre/mdt: HSM on disk actions record
2013-07-22 16:06 [PATCH 00/48] staging/lustre: minor cleanup and Intel Lustre sync Peng Tao
` (44 preceding siblings ...)
2013-07-22 16:07 ` [PATCH 45/48] staging/lustre/style: removes obsolete EXPORT_SYMTAB macros v2 Peng Tao
@ 2013-07-22 16:07 ` Peng Tao
2013-07-22 16:07 ` [PATCH 47/48] staging/lustre/scrub: purge inconsistenct objects after OI scrub Peng Tao
` (2 subsequent siblings)
48 siblings, 0 replies; 62+ messages in thread
From: Peng Tao @ 2013-07-22 16:07 UTC (permalink / raw)
To: Greg Kroah-Hartman; +Cc: linux-kernel, jcl, Peng Tao, Andreas Dilger
From: jcl <jacques-charles.lafoucriere@cea.fr>
HSM coordinator memorizes all actions in a llog
This patch implements the methods needed to create
update display these records.
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-3339
Lustre-change: http://review.whamcloud.com/6529
Signed-off-by: JC Lafoucriere <jacques-charles.lafoucriere@cea.fr>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>
Signed-off-by: Peng Tao <tao.peng@emc.com>
Signed-off-by: Andreas Dilger <andreas.dilger@intel.com>
---
.../staging/lustre/lustre/include/lprocfs_status.h | 3 --
.../lustre/lustre/include/lustre/lustre_idl.h | 47 ++++++++++++++++
.../staging/lustre/lustre/include/lustre_disk.h | 56 ++++++++++----------
drivers/staging/lustre/lustre/include/obd.h | 7 +--
drivers/staging/lustre/lustre/obdclass/llog_swab.c | 17 ++++++
5 files changed, 97 insertions(+), 33 deletions(-)
diff --git a/drivers/staging/lustre/lustre/include/lprocfs_status.h b/drivers/staging/lustre/lustre/include/lprocfs_status.h
index 20633d4..d7f1be9 100644
--- a/drivers/staging/lustre/lustre/include/lprocfs_status.h
+++ b/drivers/staging/lustre/lustre/include/lprocfs_status.h
@@ -804,9 +804,6 @@ extern int lprocfs_quota_rd_qs_factor(char *page, char **start, loff_t off,
extern int lprocfs_quota_wr_qs_factor(struct file *file,
const char *buffer,
unsigned long count, void *data);
-
-
-
#else
/* LPROCFS is not defined */
diff --git a/drivers/staging/lustre/lustre/include/lustre/lustre_idl.h b/drivers/staging/lustre/lustre/include/lustre/lustre_idl.h
index 3aaa869..984235c 100644
--- a/drivers/staging/lustre/lustre/include/lustre/lustre_idl.h
+++ b/drivers/staging/lustre/lustre/include/lustre/lustre_idl.h
@@ -2986,6 +2986,7 @@ typedef enum {
/* LLOG_JOIN_REC = LLOG_OP_MAGIC | 0x50000, obsolete 1.8.0 */
CHANGELOG_REC = LLOG_OP_MAGIC | 0x60000,
CHANGELOG_USER_REC = LLOG_OP_MAGIC | 0x70000,
+ HSM_AGENT_REC = LLOG_OP_MAGIC | 0x80000,
LLOG_HDR_MAGIC = LLOG_OP_MAGIC | 0x45539,
LLOG_LOGID_MAGIC = LLOG_OP_MAGIC | 0x4553b,
} llog_op_type;
@@ -3105,6 +3106,52 @@ struct llog_changelog_user_rec {
struct llog_rec_tail cur_tail;
} __attribute__((packed));
+enum agent_req_status {
+ ARS_WAITING,
+ ARS_STARTED,
+ ARS_FAILED,
+ ARS_CANCELED,
+ ARS_SUCCEED,
+};
+
+static inline char *agent_req_status2name(enum agent_req_status ars)
+{
+ switch (ars) {
+ case ARS_WAITING:
+ return "WAITING";
+ case ARS_STARTED:
+ return "STARTED";
+ case ARS_FAILED:
+ return "FAILED";
+ case ARS_CANCELED:
+ return "CANCELED";
+ case ARS_SUCCEED:
+ return "SUCCEED";
+ default:
+ return "UNKNOWN";
+ }
+}
+
+static inline bool agent_req_in_final_state(enum agent_req_status ars)
+{
+ return ((ars == ARS_SUCCEED) || (ars == ARS_FAILED) ||
+ (ars == ARS_CANCELED));
+}
+
+struct llog_agent_req_rec {
+ struct llog_rec_hdr arr_hdr; /**< record header */
+ __u32 arr_status; /**< status of the request */
+ /* must match enum
+ * agent_req_status */
+ __u32 arr_archive_id; /**< backend archive number */
+ __u64 arr_flags; /**< req flags */
+ __u64 arr_compound_id; /**< compound cookie */
+ __u64 arr_req_create; /**< req. creation time */
+ __u64 arr_req_change; /**< req. status change time */
+ struct hsm_action_item arr_hai; /**< req. to the agent */
+ struct llog_rec_tail arr_tail; /**< record tail for_sizezof_only */
+} __attribute__((packed));
+
/* Old llog gen for compatibility */
struct llog_gen {
__u64 mnt_cnt;
diff --git a/drivers/staging/lustre/lustre/include/lustre_disk.h b/drivers/staging/lustre/lustre/include/lustre_disk.h
index 8db6086..9228b16 100644
--- a/drivers/staging/lustre/lustre/include/lustre_disk.h
+++ b/drivers/staging/lustre/lustre/include/lustre_disk.h
@@ -53,20 +53,21 @@
/****************** on-disk files *********************/
-#define MDT_LOGS_DIR "LOGS" /* COMPAT_146 */
-#define MOUNT_CONFIGS_DIR "CONFIGS"
-#define CONFIGS_FILE "mountdata"
+#define MDT_LOGS_DIR "LOGS" /* COMPAT_146 */
+#define MOUNT_CONFIGS_DIR "CONFIGS"
+#define CONFIGS_FILE "mountdata"
/** Persistent mount data are stored on the disk in this file. */
-#define MOUNT_DATA_FILE MOUNT_CONFIGS_DIR"/"CONFIGS_FILE
-#define LAST_RCVD "last_rcvd"
-#define LOV_OBJID "lov_objid"
+#define MOUNT_DATA_FILE MOUNT_CONFIGS_DIR"/"CONFIGS_FILE
+#define LAST_RCVD "last_rcvd"
+#define LOV_OBJID "lov_objid"
#define LOV_OBJSEQ "lov_objseq"
-#define HEALTH_CHECK "health_check"
-#define CAPA_KEYS "capa_keys"
-#define CHANGELOG_USERS "changelog_users"
-#define MGS_NIDTBL_DIR "NIDTBL_VERSIONS"
-#define QMT_DIR "quota_master"
-#define QSD_DIR "quota_slave"
+#define HEALTH_CHECK "health_check"
+#define CAPA_KEYS "capa_keys"
+#define CHANGELOG_USERS "changelog_users"
+#define MGS_NIDTBL_DIR "NIDTBL_VERSIONS"
+#define QMT_DIR "quota_master"
+#define QSD_DIR "quota_slave"
+#define HSM_ACTIONS "hsm_actions"
/****************** persistent mount data *********************/
@@ -226,21 +227,22 @@ struct lustre_mount_data {
char *lmd_osd_type; /* OSD type */
};
-#define LMD_FLG_SERVER 0x0001 /* Mounting a server */
-#define LMD_FLG_CLIENT 0x0002 /* Mounting a client */
-#define LMD_FLG_ABORT_RECOV 0x0008 /* Abort recovery */
-#define LMD_FLG_NOSVC 0x0010 /* Only start MGS/MGC for servers,
- no other services */
-#define LMD_FLG_NOMGS 0x0020 /* Only start target for servers, reusing
- existing MGS services */
-#define LMD_FLG_WRITECONF 0x0040 /* Rewrite config log */
-#define LMD_FLG_NOIR 0x0080 /* NO imperative recovery */
-#define LMD_FLG_NOSCRUB 0x0100 /* Do not trigger scrub automatically */
-#define LMD_FLG_MGS 0x0200 /* Also start MGS along with server */
-#define LMD_FLG_IAM 0x0400 /* IAM dir */
-#define LMD_FLG_NO_PRIMNODE 0x0800 /* all nodes are service nodes */
-#define LMD_FLG_VIRGIN 0x1000 /* the service registers first time */
-#define LMD_FLG_UPDATE 0x2000 /* update parameters */
+#define LMD_FLG_SERVER 0x0001 /* Mounting a server */
+#define LMD_FLG_CLIENT 0x0002 /* Mounting a client */
+#define LMD_FLG_ABORT_RECOV 0x0008 /* Abort recovery */
+#define LMD_FLG_NOSVC 0x0010 /* Only start MGS/MGC for servers,
+ no other services */
+#define LMD_FLG_NOMGS 0x0020 /* Only start target for servers, reusing
+ existing MGS services */
+#define LMD_FLG_WRITECONF 0x0040 /* Rewrite config log */
+#define LMD_FLG_NOIR 0x0080 /* NO imperative recovery */
+#define LMD_FLG_NOSCRUB 0x0100 /* Do not trigger scrub automatically */
+#define LMD_FLG_MGS 0x0200 /* Also start MGS along with server */
+#define LMD_FLG_IAM 0x0400 /* IAM dir */
+#define LMD_FLG_NO_PRIMNODE 0x0800 /* all nodes are service nodes */
+#define LMD_FLG_VIRGIN 0x1000 /* the service registers first time */
+#define LMD_FLG_UPDATE 0x2000 /* update parameters */
+#define LMD_FLG_HSM 0x4000 /* Start coordinator */
#define lmd_is_client(x) ((x)->lmd_flags & LMD_FLG_CLIENT)
diff --git a/drivers/staging/lustre/lustre/include/obd.h b/drivers/staging/lustre/lustre/include/obd.h
index be3a08f..094eb96 100644
--- a/drivers/staging/lustre/lustre/include/obd.h
+++ b/drivers/staging/lustre/lustre/include/obd.h
@@ -267,9 +267,10 @@ enum llog_ctxt_id {
LLOG_TEST_REPL_CTXT,
LLOG_LOVEA_ORIG_CTXT,
LLOG_LOVEA_REPL_CTXT,
- LLOG_CHANGELOG_ORIG_CTXT, /**< changelog generation on mdd */
- LLOG_CHANGELOG_REPL_CTXT, /**< changelog access on clients */
- LLOG_CHANGELOG_USER_ORIG_CTXT, /**< for multiple changelog consumers */
+ LLOG_CHANGELOG_ORIG_CTXT, /**< changelog generation on mdd */
+ LLOG_CHANGELOG_REPL_CTXT, /**< changelog access on clients */
+ LLOG_CHANGELOG_USER_ORIG_CTXT, /**< for multiple changelog consumers */
+ LLOG_AGENT_ORIG_CTXT, /**< agent requests generation on cdt */
LLOG_MAX_CTXTS
};
diff --git a/drivers/staging/lustre/lustre/obdclass/llog_swab.c b/drivers/staging/lustre/lustre/obdclass/llog_swab.c
index dedfecf..a75b04e 100644
--- a/drivers/staging/lustre/lustre/obdclass/llog_swab.c
+++ b/drivers/staging/lustre/lustre/obdclass/llog_swab.c
@@ -203,6 +203,23 @@ void lustre_swab_llog_rec(struct llog_rec_hdr *rec)
break;
}
+ case HSM_AGENT_REC: {
+ struct llog_agent_req_rec *arr =
+ (struct llog_agent_req_rec *)rec;
+
+ __swab32s(&arr->arr_hai.hai_len);
+ __swab32s(&arr->arr_hai.hai_action);
+ lustre_swab_lu_fid(&arr->arr_hai.hai_fid);
+ lustre_swab_lu_fid(&arr->arr_hai.hai_dfid);
+ __swab64s(&arr->arr_hai.hai_cookie);
+ __swab64s(&arr->arr_hai.hai_extent.offset);
+ __swab64s(&arr->arr_hai.hai_extent.length);
+ __swab64s(&arr->arr_hai.hai_gid);
+ /* no swabing for opaque data */
+ /* hai_data[0]; */
+ break;
+ }
+
case MDS_SETATTR64_REC:
{
struct llog_setattr64_rec *lsr =
--
1.7.9.5
^ permalink raw reply related [flat|nested] 62+ messages in thread* [PATCH 47/48] staging/lustre/scrub: purge inconsistenct objects after OI scrub
2013-07-22 16:06 [PATCH 00/48] staging/lustre: minor cleanup and Intel Lustre sync Peng Tao
` (45 preceding siblings ...)
2013-07-22 16:07 ` [PATCH 46/48] staging/lustre/mdt: HSM on disk actions record Peng Tao
@ 2013-07-22 16:07 ` Peng Tao
2013-07-22 16:07 ` [PATCH 48/48] staging/lustre/mdc: Keep resend FLocks Peng Tao
2013-07-23 20:51 ` [PATCH 00/48] staging/lustre: minor cleanup and Intel Lustre sync Greg Kroah-Hartman
48 siblings, 0 replies; 62+ messages in thread
From: Peng Tao @ 2013-07-22 16:07 UTC (permalink / raw)
To: Greg Kroah-Hartman; +Cc: linux-kernel, Fan Yong, Peng Tao, Andreas Dilger
From: Fan Yong <fan.yong@intel.com>
When OI scrub repaired the found inconsistency, it needs to purge the
old object out of cache; otherwise, others may still use those cached
stale information.
Original patch adds functions in obdclass that is only used by server.
Drop that part. Only merge in error handling change.
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-3335
Lustre-change: http://review.whamcloud.com/6697
Signed-off-by: Fan Yong <fan.yong@intel.com>
Reviewed-by: Alex Zhuravlev <alexey.zhuravlev@intel.com>
Reviewed-by: Andreas Dilger <andreas.dilger@intel.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>
Signed-off-by: Peng Tao <tao.peng@emc.com>
Signed-off-by: Andreas Dilger <andreas.dilger@intel.com>
---
drivers/staging/lustre/lustre/obdclass/lu_object.c | 10 +++++-----
1 file changed, 5 insertions(+), 5 deletions(-)
diff --git a/drivers/staging/lustre/lustre/obdclass/lu_object.c b/drivers/staging/lustre/lustre/obdclass/lu_object.c
index 2fb65f6..896f24a 100644
--- a/drivers/staging/lustre/lustre/obdclass/lu_object.c
+++ b/drivers/staging/lustre/lustre/obdclass/lu_object.c
@@ -538,7 +538,7 @@ static struct lu_object *htable_lookup(struct lu_site *s,
__u64 ver = cfs_hash_bd_version_get(bd);
if (*version == ver)
- return NULL;
+ return ERR_PTR(-ENOENT);
*version = ver;
bkt = cfs_hash_bd_extra_get(s->ls_obj_hash, bd);
@@ -547,7 +547,7 @@ static struct lu_object *htable_lookup(struct lu_site *s,
hnode = cfs_hash_bd_peek_locked(s->ls_obj_hash, bd, (void *)f);
if (hnode == NULL) {
lprocfs_counter_incr(s->ls_stats, LU_SS_CACHE_MISS);
- return NULL;
+ return ERR_PTR(-ENOENT);
}
h = container_of0(hnode, struct lu_object_header, loh_hash);
@@ -651,7 +651,7 @@ static struct lu_object *lu_object_find_try(const struct lu_env *env,
cfs_hash_bd_get_and_lock(hs, (void *)f, &bd, 1);
o = htable_lookup(s, &bd, f, waiter, &version);
cfs_hash_bd_unlock(hs, &bd, 1);
- if (o != NULL)
+ if (!IS_ERR(o) || PTR_ERR(o) != -ENOENT)
return o;
/*
@@ -667,7 +667,7 @@ static struct lu_object *lu_object_find_try(const struct lu_env *env,
cfs_hash_bd_lock(hs, &bd, 1);
shadow = htable_lookup(s, &bd, f, waiter, &version);
- if (likely(shadow == NULL)) {
+ if (likely(IS_ERR(shadow) && PTR_ERR(shadow) == -ENOENT)) {
struct lu_site_bkt_data *bkt;
bkt = cfs_hash_bd_extra_get(hs, &bd);
@@ -2077,7 +2077,7 @@ void lu_object_assign_fid(const struct lu_env *env, struct lu_object *o,
cfs_hash_bd_get_and_lock(hs, (void *)fid, &bd, 1);
shadow = htable_lookup(s, &bd, fid, &waiter, &version);
/* supposed to be unique */
- LASSERT(shadow == NULL);
+ LASSERT(IS_ERR(shadow) && PTR_ERR(shadow) == -ENOENT);
*old = *fid;
bkt = cfs_hash_bd_extra_get(hs, &bd);
cfs_hash_bd_add_locked(hs, &bd, &o->lo_header->loh_hash);
--
1.7.9.5
^ permalink raw reply related [flat|nested] 62+ messages in thread* [PATCH 48/48] staging/lustre/mdc: Keep resend FLocks
2013-07-22 16:06 [PATCH 00/48] staging/lustre: minor cleanup and Intel Lustre sync Peng Tao
` (46 preceding siblings ...)
2013-07-22 16:07 ` [PATCH 47/48] staging/lustre/scrub: purge inconsistenct objects after OI scrub Peng Tao
@ 2013-07-22 16:07 ` Peng Tao
2013-07-23 20:51 ` [PATCH 00/48] staging/lustre: minor cleanup and Intel Lustre sync Greg Kroah-Hartman
48 siblings, 0 replies; 62+ messages in thread
From: Peng Tao @ 2013-07-22 16:07 UTC (permalink / raw)
To: Greg Kroah-Hartman; +Cc: linux-kernel, Bruno Faccini, Peng Tao, Andreas Dilger
From: Bruno Faccini <bruno.faccini@intel.com>
FLocks requests (particulary F_UNLCKs) can't be trashed
upon comm problems with Server/MDS nor upon kill/exit,
thus we need to keep retry/send.
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-2665
Lustre-change: http://review.whamcloud.com/6415
Signed-off-by: Bruno Faccini <bruno.faccini@intel.com>
Reviewed-by: Bobi Jam <bobijam@gmail.com>
Reviewed-by: Oleg Drokin <oleg.drokin@intel.com>
Signed-off-by: Peng Tao <tao.peng@emc.com>
Signed-off-by: Andreas Dilger <andreas.dilger@intel.com>
---
drivers/staging/lustre/lustre/mdc/mdc_locks.c | 8 +++++++-
1 file changed, 7 insertions(+), 1 deletion(-)
diff --git a/drivers/staging/lustre/lustre/mdc/mdc_locks.c b/drivers/staging/lustre/lustre/mdc/mdc_locks.c
index ab88a46..803c390 100644
--- a/drivers/staging/lustre/lustre/mdc/mdc_locks.c
+++ b/drivers/staging/lustre/lustre/mdc/mdc_locks.c
@@ -823,7 +823,13 @@ resend:
/* 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);
}
--
1.7.9.5
^ permalink raw reply related [flat|nested] 62+ messages in thread* Re: [PATCH 00/48] staging/lustre: minor cleanup and Intel Lustre sync
2013-07-22 16:06 [PATCH 00/48] staging/lustre: minor cleanup and Intel Lustre sync Peng Tao
` (47 preceding siblings ...)
2013-07-22 16:07 ` [PATCH 48/48] staging/lustre/mdc: Keep resend FLocks Peng Tao
@ 2013-07-23 20:51 ` Greg Kroah-Hartman
2013-07-24 6:05 ` Peng Tao
48 siblings, 1 reply; 62+ messages in thread
From: Greg Kroah-Hartman @ 2013-07-23 20:51 UTC (permalink / raw)
To: Peng Tao; +Cc: linux-kernel, Andreas Dilger
On Tue, Jul 23, 2013 at 12:06:21AM +0800, Peng Tao wrote:
> Hi Greg,
>
> The first two of the patchset are minor cleanup of lustre code. And the
> rest of it are patches ported from Intel Lustre tree. Please help to
> queue them.
I've applied most of these, see my comments on the ones I didn't.
Feel free to fix those up and resend.
Also please resync on my tree, as I tool some other lustre patches from
different developers that might conflict with your future changes.
thanks,
greg k-h
^ permalink raw reply [flat|nested] 62+ messages in thread* Re: [PATCH 00/48] staging/lustre: minor cleanup and Intel Lustre sync
2013-07-23 20:51 ` [PATCH 00/48] staging/lustre: minor cleanup and Intel Lustre sync Greg Kroah-Hartman
@ 2013-07-24 6:05 ` Peng Tao
0 siblings, 0 replies; 62+ messages in thread
From: Peng Tao @ 2013-07-24 6:05 UTC (permalink / raw)
To: Greg Kroah-Hartman; +Cc: Linux Kernel Mailing List, Andreas Dilger
On Wed, Jul 24, 2013 at 4:51 AM, Greg Kroah-Hartman
<gregkh@linuxfoundation.org> wrote:
> On Tue, Jul 23, 2013 at 12:06:21AM +0800, Peng Tao wrote:
>> Hi Greg,
>>
>> The first two of the patchset are minor cleanup of lustre code. And the
>> rest of it are patches ported from Intel Lustre tree. Please help to
>> queue them.
>
> I've applied most of these, see my comments on the ones I didn't.
>
> Feel free to fix those up and resend.
>
> Also please resync on my tree, as I tool some other lustre patches from
> different developers that might conflict with your future changes.
>
Thanks. I will fix up, rebase and resend.
Thanks,
Tao
^ permalink raw reply [flat|nested] 62+ messages in thread