| Message ID | 1370056054-25449-9-git-send-email-jlayton@redhat.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
On Fri, May 31, 2013 at 11:07:31PM -0400, Jeff Layton wrote: > Testing has shown that iterating over the blocked_list for deadlock > detection turns out to be a bottleneck. In order to alleviate that, > begin the process of turning it into a hashtable. We start by turning > the fl_link into a hlist_node and the global lists into hlists. A later > patch will do the conversion of the blocked_list to a hashtable. Even simpler would be if we could add a pointer to the (well, a) lock that a lockowner is blocking on, and then we'd just have to follow a pointer. I haven't thought that through, though, perhaps that's hard ot make work.... --b. > > Signed-off-by: Jeff Layton <jlayton@redhat.com> > --- > fs/locks.c | 32 ++++++++++++++++---------------- > include/linux/fs.h | 2 +- > 2 files changed, 17 insertions(+), 17 deletions(-) > > diff --git a/fs/locks.c b/fs/locks.c > index fc35b9e..5ed056b 100644 > --- a/fs/locks.c > +++ b/fs/locks.c > @@ -163,8 +163,8 @@ int lease_break_time = 45; > #define for_each_lock(inode, lockp) \ > for (lockp = &inode->i_flock; *lockp != NULL; lockp = &(*lockp)->fl_next) > > -static LIST_HEAD(file_lock_list); > -static LIST_HEAD(blocked_list); > +static HLIST_HEAD(file_lock_list); > +static HLIST_HEAD(blocked_list); > > /* Protects the two list heads above */ > static DEFINE_SPINLOCK(file_lock_lock); > @@ -173,7 +173,7 @@ static struct kmem_cache *filelock_cache __read_mostly; > > static void locks_init_lock_heads(struct file_lock *fl) > { > - INIT_LIST_HEAD(&fl->fl_link); > + INIT_HLIST_NODE(&fl->fl_link); > INIT_LIST_HEAD(&fl->fl_block); > init_waitqueue_head(&fl->fl_wait); > } > @@ -207,7 +207,7 @@ void locks_free_lock(struct file_lock *fl) > { > BUG_ON(waitqueue_active(&fl->fl_wait)); > BUG_ON(!list_empty(&fl->fl_block)); > - BUG_ON(!list_empty(&fl->fl_link)); > + BUG_ON(!hlist_unhashed(&fl->fl_link)); > > locks_release_private(fl); > kmem_cache_free(filelock_cache, fl); > @@ -486,7 +486,7 @@ static inline void > locks_insert_global_blocked(struct file_lock *waiter) > { > spin_lock(&file_lock_lock); > - list_add(&waiter->fl_link, &blocked_list); > + hlist_add_head(&waiter->fl_link, &blocked_list); > spin_unlock(&file_lock_lock); > } > > @@ -494,7 +494,7 @@ static inline void > locks_delete_global_blocked(struct file_lock *waiter) > { > spin_lock(&file_lock_lock); > - list_del_init(&waiter->fl_link); > + hlist_del_init(&waiter->fl_link); > spin_unlock(&file_lock_lock); > } > > @@ -502,7 +502,7 @@ static inline void > locks_insert_global_locks(struct file_lock *waiter) > { > spin_lock(&file_lock_lock); > - list_add_tail(&waiter->fl_link, &file_lock_list); > + hlist_add_head(&waiter->fl_link, &file_lock_list); > spin_unlock(&file_lock_lock); > } > > @@ -510,7 +510,7 @@ static inline void > locks_delete_global_locks(struct file_lock *waiter) > { > spin_lock(&file_lock_lock); > - list_del_init(&waiter->fl_link); > + hlist_del_init(&waiter->fl_link); > spin_unlock(&file_lock_lock); > } > > @@ -705,7 +705,7 @@ static struct file_lock *what_owner_is_waiting_for(struct file_lock *block_fl) > { > struct file_lock *fl, *ret = NULL; > > - list_for_each_entry(fl, &blocked_list, fl_link) { > + hlist_for_each_entry(fl, &blocked_list, fl_link) { > if (posix_same_owner(fl, block_fl)) { > ret = fl->fl_next; > if (likely(ret)) > @@ -867,7 +867,7 @@ static int __posix_lock_file(struct inode *inode, struct file_lock *request, str > goto out; > error = FILE_LOCK_DEFERRED; > locks_insert_block(fl, request); > - if (list_empty(&request->fl_link)) > + if (hlist_unhashed(&request->fl_link)) > locks_insert_global_blocked(request); > goto out; > } > @@ -882,10 +882,10 @@ static int __posix_lock_file(struct inode *inode, struct file_lock *request, str > * Now that we know the request is no longer blocked, we can take it > * off the global list. Some callers send down partially initialized > * requests, so we only do this if FL_SLEEP is set. Also, avoid taking > - * the lock if the list is empty, as that indicates a request that > + * the lock if the hlist is unhashed, as that indicates a request that > * never blocked. > */ > - if ((request->fl_flags & FL_SLEEP) && !list_empty(&request->fl_link)) > + if ((request->fl_flags & FL_SLEEP) && !hlist_unhashed(&request->fl_link)) > locks_delete_global_blocked(request); > > /* > @@ -2277,11 +2277,11 @@ static int locks_show(struct seq_file *f, void *v) > { > struct file_lock *fl, *bfl; > > - fl = list_entry(v, struct file_lock, fl_link); > + fl = hlist_entry(v, struct file_lock, fl_link); > > lock_get_status(f, fl, *((loff_t *)f->private), ""); > > - list_for_each_entry(bfl, &blocked_list, fl_link) { > + hlist_for_each_entry(bfl, &blocked_list, fl_link) { > if (bfl->fl_next == fl) > lock_get_status(f, bfl, *((loff_t *)f->private), " ->"); > } > @@ -2295,14 +2295,14 @@ static void *locks_start(struct seq_file *f, loff_t *pos) > > spin_lock(&file_lock_lock); > *p = (*pos + 1); > - return seq_list_start(&file_lock_list, *pos); > + return seq_hlist_start(&file_lock_list, *pos); > } > > static void *locks_next(struct seq_file *f, void *v, loff_t *pos) > { > loff_t *p = f->private; > ++*p; > - return seq_list_next(v, &file_lock_list, pos); > + return seq_hlist_next(v, &file_lock_list, pos); > } > > static void locks_stop(struct seq_file *f, void *v) > diff --git a/include/linux/fs.h b/include/linux/fs.h > index ccb44ea..07a009e 100644 > --- a/include/linux/fs.h > +++ b/include/linux/fs.h > @@ -934,7 +934,7 @@ int locks_in_grace(struct net *); > */ > struct file_lock { > struct file_lock *fl_next; /* singly linked list for this inode */ > - struct list_head fl_link; /* doubly linked list of all locks */ > + struct hlist_node fl_link; /* node in global lists */ > struct list_head fl_block; /* circular list of blocked processes */ > fl_owner_t fl_owner; > unsigned int fl_flags; > -- > 1.7.1 > -- To unsubscribe from this list: send the line "unsubscribe ceph-devel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
On Tue, 4 Jun 2013 17:59:50 -0400 "J. Bruce Fields" <bfields@fieldses.org> wrote: > On Fri, May 31, 2013 at 11:07:31PM -0400, Jeff Layton wrote: > > Testing has shown that iterating over the blocked_list for deadlock > > detection turns out to be a bottleneck. In order to alleviate that, > > begin the process of turning it into a hashtable. We start by turning > > the fl_link into a hlist_node and the global lists into hlists. A later > > patch will do the conversion of the blocked_list to a hashtable. > > Even simpler would be if we could add a pointer to the (well, a) lock > that a lockowner is blocking on, and then we'd just have to follow a > pointer. I haven't thought that through, though, perhaps that's hard ot > make work.... > > --b. > I considered that as well and it makes sense for the simple local filesystem case where you just track ownership based on fl_owner_t. But...what about lockd? It considers ownership to be a tuple of the nlm_host and the pid sent in a lock request. I can't seem to wrap my brain around how to make such an approach work there. I'll confess though that I haven't tried *too* hard yet though since I had bigger problems to sort through at the time. Maybe we can consider that for a later set? > > > > Signed-off-by: Jeff Layton <jlayton@redhat.com> > > --- > > fs/locks.c | 32 ++++++++++++++++---------------- > > include/linux/fs.h | 2 +- > > 2 files changed, 17 insertions(+), 17 deletions(-) > > > > diff --git a/fs/locks.c b/fs/locks.c > > index fc35b9e..5ed056b 100644 > > --- a/fs/locks.c > > +++ b/fs/locks.c > > @@ -163,8 +163,8 @@ int lease_break_time = 45; > > #define for_each_lock(inode, lockp) \ > > for (lockp = &inode->i_flock; *lockp != NULL; lockp = &(*lockp)->fl_next) > > > > -static LIST_HEAD(file_lock_list); > > -static LIST_HEAD(blocked_list); > > +static HLIST_HEAD(file_lock_list); > > +static HLIST_HEAD(blocked_list); > > > > /* Protects the two list heads above */ > > static DEFINE_SPINLOCK(file_lock_lock); > > @@ -173,7 +173,7 @@ static struct kmem_cache *filelock_cache __read_mostly; > > > > static void locks_init_lock_heads(struct file_lock *fl) > > { > > - INIT_LIST_HEAD(&fl->fl_link); > > + INIT_HLIST_NODE(&fl->fl_link); > > INIT_LIST_HEAD(&fl->fl_block); > > init_waitqueue_head(&fl->fl_wait); > > } > > @@ -207,7 +207,7 @@ void locks_free_lock(struct file_lock *fl) > > { > > BUG_ON(waitqueue_active(&fl->fl_wait)); > > BUG_ON(!list_empty(&fl->fl_block)); > > - BUG_ON(!list_empty(&fl->fl_link)); > > + BUG_ON(!hlist_unhashed(&fl->fl_link)); > > > > locks_release_private(fl); > > kmem_cache_free(filelock_cache, fl); > > @@ -486,7 +486,7 @@ static inline void > > locks_insert_global_blocked(struct file_lock *waiter) > > { > > spin_lock(&file_lock_lock); > > - list_add(&waiter->fl_link, &blocked_list); > > + hlist_add_head(&waiter->fl_link, &blocked_list); > > spin_unlock(&file_lock_lock); > > } > > > > @@ -494,7 +494,7 @@ static inline void > > locks_delete_global_blocked(struct file_lock *waiter) > > { > > spin_lock(&file_lock_lock); > > - list_del_init(&waiter->fl_link); > > + hlist_del_init(&waiter->fl_link); > > spin_unlock(&file_lock_lock); > > } > > > > @@ -502,7 +502,7 @@ static inline void > > locks_insert_global_locks(struct file_lock *waiter) > > { > > spin_lock(&file_lock_lock); > > - list_add_tail(&waiter->fl_link, &file_lock_list); > > + hlist_add_head(&waiter->fl_link, &file_lock_list); > > spin_unlock(&file_lock_lock); > > } > > > > @@ -510,7 +510,7 @@ static inline void > > locks_delete_global_locks(struct file_lock *waiter) > > { > > spin_lock(&file_lock_lock); > > - list_del_init(&waiter->fl_link); > > + hlist_del_init(&waiter->fl_link); > > spin_unlock(&file_lock_lock); > > } > > > > @@ -705,7 +705,7 @@ static struct file_lock *what_owner_is_waiting_for(struct file_lock *block_fl) > > { > > struct file_lock *fl, *ret = NULL; > > > > - list_for_each_entry(fl, &blocked_list, fl_link) { > > + hlist_for_each_entry(fl, &blocked_list, fl_link) { > > if (posix_same_owner(fl, block_fl)) { > > ret = fl->fl_next; > > if (likely(ret)) > > @@ -867,7 +867,7 @@ static int __posix_lock_file(struct inode *inode, struct file_lock *request, str > > goto out; > > error = FILE_LOCK_DEFERRED; > > locks_insert_block(fl, request); > > - if (list_empty(&request->fl_link)) > > + if (hlist_unhashed(&request->fl_link)) > > locks_insert_global_blocked(request); > > goto out; > > } > > @@ -882,10 +882,10 @@ static int __posix_lock_file(struct inode *inode, struct file_lock *request, str > > * Now that we know the request is no longer blocked, we can take it > > * off the global list. Some callers send down partially initialized > > * requests, so we only do this if FL_SLEEP is set. Also, avoid taking > > - * the lock if the list is empty, as that indicates a request that > > + * the lock if the hlist is unhashed, as that indicates a request that > > * never blocked. > > */ > > - if ((request->fl_flags & FL_SLEEP) && !list_empty(&request->fl_link)) > > + if ((request->fl_flags & FL_SLEEP) && !hlist_unhashed(&request->fl_link)) > > locks_delete_global_blocked(request); > > > > /* > > @@ -2277,11 +2277,11 @@ static int locks_show(struct seq_file *f, void *v) > > { > > struct file_lock *fl, *bfl; > > > > - fl = list_entry(v, struct file_lock, fl_link); > > + fl = hlist_entry(v, struct file_lock, fl_link); > > > > lock_get_status(f, fl, *((loff_t *)f->private), ""); > > > > - list_for_each_entry(bfl, &blocked_list, fl_link) { > > + hlist_for_each_entry(bfl, &blocked_list, fl_link) { > > if (bfl->fl_next == fl) > > lock_get_status(f, bfl, *((loff_t *)f->private), " ->"); > > } > > @@ -2295,14 +2295,14 @@ static void *locks_start(struct seq_file *f, loff_t *pos) > > > > spin_lock(&file_lock_lock); > > *p = (*pos + 1); > > - return seq_list_start(&file_lock_list, *pos); > > + return seq_hlist_start(&file_lock_list, *pos); > > } > > > > static void *locks_next(struct seq_file *f, void *v, loff_t *pos) > > { > > loff_t *p = f->private; > > ++*p; > > - return seq_list_next(v, &file_lock_list, pos); > > + return seq_hlist_next(v, &file_lock_list, pos); > > } > > > > static void locks_stop(struct seq_file *f, void *v) > > diff --git a/include/linux/fs.h b/include/linux/fs.h > > index ccb44ea..07a009e 100644 > > --- a/include/linux/fs.h > > +++ b/include/linux/fs.h > > @@ -934,7 +934,7 @@ int locks_in_grace(struct net *); > > */ > > struct file_lock { > > struct file_lock *fl_next; /* singly linked list for this inode */ > > - struct list_head fl_link; /* doubly linked list of all locks */ > > + struct hlist_node fl_link; /* node in global lists */ > > struct list_head fl_block; /* circular list of blocked processes */ > > fl_owner_t fl_owner; > > unsigned int fl_flags; > > -- > > 1.7.1 > >
On Wed, Jun 05, 2013 at 07:43:09AM -0400, Jeff Layton wrote: > On Tue, 4 Jun 2013 17:59:50 -0400 > "J. Bruce Fields" <bfields@fieldses.org> wrote: > > > On Fri, May 31, 2013 at 11:07:31PM -0400, Jeff Layton wrote: > > > Testing has shown that iterating over the blocked_list for deadlock > > > detection turns out to be a bottleneck. In order to alleviate that, > > > begin the process of turning it into a hashtable. We start by turning > > > the fl_link into a hlist_node and the global lists into hlists. A later > > > patch will do the conversion of the blocked_list to a hashtable. > > > > Even simpler would be if we could add a pointer to the (well, a) lock > > that a lockowner is blocking on, and then we'd just have to follow a > > pointer. I haven't thought that through, though, perhaps that's hard ot > > make work.... > > > > --b. > > > > I considered that as well and it makes sense for the simple local > filesystem case where you just track ownership based on fl_owner_t. > > But...what about lockd? It considers ownership to be a tuple of the > nlm_host and the pid sent in a lock request. I can't seem to wrap my > brain around how to make such an approach work there. I wonder if we could do something vaguely like struct lock_owner_common { struct file_lock *blocker; }; struct nlmsvc_lock_owner { struct lock_owner_common owner; unsigned int client_pid; }; and make fl_owner a (struct lock_owner_common *) and have lockd create nlmsvc_lock_owners as necessary on the fly. The lm_compare_owner callback could then be replaced by a pointer comparison. I'm not sure what kind of locking or refcounting might be needed. But... > I'll confess though that I haven't tried *too* hard yet ... me neither, so... > though since I had bigger problems to sort through at the time. Maybe > we can consider that for a later set? sounds fine. --b. -- To unsubscribe from this list: send the line "unsubscribe ceph-devel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
diff --git a/fs/locks.c b/fs/locks.c index fc35b9e..5ed056b 100644 --- a/fs/locks.c +++ b/fs/locks.c @@ -163,8 +163,8 @@ int lease_break_time = 45; #define for_each_lock(inode, lockp) \ for (lockp = &inode->i_flock; *lockp != NULL; lockp = &(*lockp)->fl_next) -static LIST_HEAD(file_lock_list); -static LIST_HEAD(blocked_list); +static HLIST_HEAD(file_lock_list); +static HLIST_HEAD(blocked_list); /* Protects the two list heads above */ static DEFINE_SPINLOCK(file_lock_lock); @@ -173,7 +173,7 @@ static struct kmem_cache *filelock_cache __read_mostly; static void locks_init_lock_heads(struct file_lock *fl) { - INIT_LIST_HEAD(&fl->fl_link); + INIT_HLIST_NODE(&fl->fl_link); INIT_LIST_HEAD(&fl->fl_block); init_waitqueue_head(&fl->fl_wait); } @@ -207,7 +207,7 @@ void locks_free_lock(struct file_lock *fl) { BUG_ON(waitqueue_active(&fl->fl_wait)); BUG_ON(!list_empty(&fl->fl_block)); - BUG_ON(!list_empty(&fl->fl_link)); + BUG_ON(!hlist_unhashed(&fl->fl_link)); locks_release_private(fl); kmem_cache_free(filelock_cache, fl); @@ -486,7 +486,7 @@ static inline void locks_insert_global_blocked(struct file_lock *waiter) { spin_lock(&file_lock_lock); - list_add(&waiter->fl_link, &blocked_list); + hlist_add_head(&waiter->fl_link, &blocked_list); spin_unlock(&file_lock_lock); } @@ -494,7 +494,7 @@ static inline void locks_delete_global_blocked(struct file_lock *waiter) { spin_lock(&file_lock_lock); - list_del_init(&waiter->fl_link); + hlist_del_init(&waiter->fl_link); spin_unlock(&file_lock_lock); } @@ -502,7 +502,7 @@ static inline void locks_insert_global_locks(struct file_lock *waiter) { spin_lock(&file_lock_lock); - list_add_tail(&waiter->fl_link, &file_lock_list); + hlist_add_head(&waiter->fl_link, &file_lock_list); spin_unlock(&file_lock_lock); } @@ -510,7 +510,7 @@ static inline void locks_delete_global_locks(struct file_lock *waiter) { spin_lock(&file_lock_lock); - list_del_init(&waiter->fl_link); + hlist_del_init(&waiter->fl_link); spin_unlock(&file_lock_lock); } @@ -705,7 +705,7 @@ static struct file_lock *what_owner_is_waiting_for(struct file_lock *block_fl) { struct file_lock *fl, *ret = NULL; - list_for_each_entry(fl, &blocked_list, fl_link) { + hlist_for_each_entry(fl, &blocked_list, fl_link) { if (posix_same_owner(fl, block_fl)) { ret = fl->fl_next; if (likely(ret)) @@ -867,7 +867,7 @@ static int __posix_lock_file(struct inode *inode, struct file_lock *request, str goto out; error = FILE_LOCK_DEFERRED; locks_insert_block(fl, request); - if (list_empty(&request->fl_link)) + if (hlist_unhashed(&request->fl_link)) locks_insert_global_blocked(request); goto out; } @@ -882,10 +882,10 @@ static int __posix_lock_file(struct inode *inode, struct file_lock *request, str * Now that we know the request is no longer blocked, we can take it * off the global list. Some callers send down partially initialized * requests, so we only do this if FL_SLEEP is set. Also, avoid taking - * the lock if the list is empty, as that indicates a request that + * the lock if the hlist is unhashed, as that indicates a request that * never blocked. */ - if ((request->fl_flags & FL_SLEEP) && !list_empty(&request->fl_link)) + if ((request->fl_flags & FL_SLEEP) && !hlist_unhashed(&request->fl_link)) locks_delete_global_blocked(request); /* @@ -2277,11 +2277,11 @@ static int locks_show(struct seq_file *f, void *v) { struct file_lock *fl, *bfl; - fl = list_entry(v, struct file_lock, fl_link); + fl = hlist_entry(v, struct file_lock, fl_link); lock_get_status(f, fl, *((loff_t *)f->private), ""); - list_for_each_entry(bfl, &blocked_list, fl_link) { + hlist_for_each_entry(bfl, &blocked_list, fl_link) { if (bfl->fl_next == fl) lock_get_status(f, bfl, *((loff_t *)f->private), " ->"); } @@ -2295,14 +2295,14 @@ static void *locks_start(struct seq_file *f, loff_t *pos) spin_lock(&file_lock_lock); *p = (*pos + 1); - return seq_list_start(&file_lock_list, *pos); + return seq_hlist_start(&file_lock_list, *pos); } static void *locks_next(struct seq_file *f, void *v, loff_t *pos) { loff_t *p = f->private; ++*p; - return seq_list_next(v, &file_lock_list, pos); + return seq_hlist_next(v, &file_lock_list, pos); } static void locks_stop(struct seq_file *f, void *v) diff --git a/include/linux/fs.h b/include/linux/fs.h index ccb44ea..07a009e 100644 --- a/include/linux/fs.h +++ b/include/linux/fs.h @@ -934,7 +934,7 @@ int locks_in_grace(struct net *); */ struct file_lock { struct file_lock *fl_next; /* singly linked list for this inode */ - struct list_head fl_link; /* doubly linked list of all locks */ + struct hlist_node fl_link; /* node in global lists */ struct list_head fl_block; /* circular list of blocked processes */ fl_owner_t fl_owner; unsigned int fl_flags;
Testing has shown that iterating over the blocked_list for deadlock detection turns out to be a bottleneck. In order to alleviate that, begin the process of turning it into a hashtable. We start by turning the fl_link into a hlist_node and the global lists into hlists. A later patch will do the conversion of the blocked_list to a hashtable. Signed-off-by: Jeff Layton <jlayton@redhat.com> --- fs/locks.c | 32 ++++++++++++++++---------------- include/linux/fs.h | 2 +- 2 files changed, 17 insertions(+), 17 deletions(-)