| lists.openwall.net | lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC | |
|
Open Source and information security mailing list archives
| ||
|
Message-Id: <1475461717-21631-37-git-send-email-jsimmons@infradead.org>
Date: Sun, 2 Oct 2016 22:28:32 -0400
From: James Simmons <jsimmons@...radead.org>
To: Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
devel@...verdev.osuosl.org,
Andreas Dilger <andreas.dilger@...el.com>,
Oleg Drokin <oleg.drokin@...el.com>
Cc: Linux Kernel Mailing List <linux-kernel@...r.kernel.org>,
Lustre Development List <lustre-devel@...ts.lustre.org>,
Vitaly Fertman <vitaly_fertman@...atex.com>,
James Simmons <jsimmons@...radead.org>
Subject: [PATCH 36/41] staging: lustre: ldlm: interval tree search in ldlm_lock_match()
From: Vitaly Fertman <vitaly_fertman@...atex.com>
replace the linear search by interval_tree one for granted list
in ldlm_lock_match()
Signed-off-by: Vitaly Fertman <vitaly_fertman@...atex.com>
Intel-bug-id: https://jira.hpdd.intel.com/browse/LU-5739
Xyratex-bug-id: MRP-2089
Reviewed-on: http://review.whamcloud.com/12294
Reviewed-by: Andreas Dilger <andreas.dilger@...el.com>
Reviewed-by: Jinshan Xiong <jinshan.xiong@...el.com>
Reviewed-by: Oleg Drokin <oleg.drokin@...el.com>
Signed-off-by: James Simmons <jsimmons@...radead.org>
---
drivers/staging/lustre/lustre/ldlm/ldlm_lock.c | 249 ++++++++++++++++--------
1 files changed, 171 insertions(+), 78 deletions(-)
diff --git a/drivers/staging/lustre/lustre/ldlm/ldlm_lock.c b/drivers/staging/lustre/lustre/ldlm/ldlm_lock.c
index 22b4a52..f2044ec 100644
--- a/drivers/staging/lustre/lustre/ldlm/ldlm_lock.c
+++ b/drivers/staging/lustre/lustre/ldlm/ldlm_lock.c
@@ -1043,88 +1043,173 @@ void ldlm_grant_lock(struct ldlm_lock *lock, struct list_head *work_list)
}
/**
- * Search for a lock with given properties in a queue.
+ * Describe the overlap between two locks. itree_overlap_cb data.
+ */
+struct lock_match_data {
+ struct ldlm_lock *lmd_old;
+ struct ldlm_lock *lmd_lock;
+ enum ldlm_mode *lmd_mode;
+ ldlm_policy_data_t *lmd_policy;
+ __u64 lmd_flags;
+ int lmd_unref;
+};
+
+/**
+ * Check if the given @lock meets the criteria for a match.
+ * A reference on the lock is taken if matched.
*
- * \retval a referenced lock or NULL. See the flag descriptions below, in the
- * comment above ldlm_lock_match
+ * \param lock test-against this lock
+ * \param data parameters
*/
-static struct ldlm_lock *search_queue(struct list_head *queue,
- enum ldlm_mode *mode,
- ldlm_policy_data_t *policy,
- struct ldlm_lock *old_lock,
- __u64 flags, int unref)
+static int lock_matches(struct ldlm_lock *lock, struct lock_match_data *data)
{
- struct ldlm_lock *lock;
- struct list_head *tmp;
+ ldlm_policy_data_t *lpol = &lock->l_policy_data;
+ enum ldlm_mode match;
- list_for_each(tmp, queue) {
- enum ldlm_mode match;
+ if (lock == data->lmd_old)
+ return INTERVAL_ITER_STOP;
- lock = list_entry(tmp, struct ldlm_lock, l_res_link);
+ /*
+ * Check if this lock can be matched.
+ * Used by LU-2919(exclusive open) for open lease lock
+ */
+ if (ldlm_is_excl(lock))
+ return INTERVAL_ITER_CONT;
- if (lock == old_lock)
- break;
+ /*
+ * llite sometimes wants to match locks that will be
+ * canceled when their users drop, but we allow it to match
+ * if it passes in CBPENDING and the lock still has users.
+ * this is generally only going to be used by children
+ * whose parents already hold a lock so forward progress
+ * can still happen.
+ */
+ if (ldlm_is_cbpending(lock) &&
+ !(data->lmd_flags & LDLM_FL_CBPENDING))
+ return INTERVAL_ITER_CONT;
- /* Check if this lock can be matched.
- * Used by LU-2919(exclusive open) for open lease lock
- */
- if (ldlm_is_excl(lock))
- continue;
+ if (!data->lmd_unref && ldlm_is_cbpending(lock) &&
+ !lock->l_readers && !lock->l_writers)
+ return INTERVAL_ITER_CONT;
- /* llite sometimes wants to match locks that will be
- * canceled when their users drop, but we allow it to match
- * if it passes in CBPENDING and the lock still has users.
- * this is generally only going to be used by children
- * whose parents already hold a lock so forward progress
- * can still happen.
- */
- if (ldlm_is_cbpending(lock) && !(flags & LDLM_FL_CBPENDING))
- continue;
- if (!unref && ldlm_is_cbpending(lock) &&
- lock->l_readers == 0 && lock->l_writers == 0)
- continue;
+ if (!(lock->l_req_mode & *data->lmd_mode))
+ return INTERVAL_ITER_CONT;
+ match = lock->l_req_mode;
- if (!(lock->l_req_mode & *mode))
- continue;
- match = lock->l_req_mode;
-
- if (lock->l_resource->lr_type == LDLM_EXTENT &&
- (lock->l_policy_data.l_extent.start >
- policy->l_extent.start ||
- lock->l_policy_data.l_extent.end < policy->l_extent.end))
- continue;
+ switch (lock->l_resource->lr_type) {
+ case LDLM_EXTENT:
+ if (lpol->l_extent.start > data->lmd_policy->l_extent.start ||
+ lpol->l_extent.end < data->lmd_policy->l_extent.end)
+ return INTERVAL_ITER_CONT;
if (unlikely(match == LCK_GROUP) &&
- lock->l_resource->lr_type == LDLM_EXTENT &&
- policy->l_extent.gid != LDLM_GID_ANY &&
- lock->l_policy_data.l_extent.gid != policy->l_extent.gid)
- continue;
-
- /* We match if we have existing lock with same or wider set
+ data->lmd_policy->l_extent.gid != LDLM_GID_ANY &&
+ lpol->l_extent.gid != data->lmd_policy->l_extent.gid)
+ return INTERVAL_ITER_CONT;
+ break;
+ case LDLM_IBITS:
+ /*
+ * We match if we have existing lock with same or wider set
* of bits.
*/
- if (lock->l_resource->lr_type == LDLM_IBITS &&
- ((lock->l_policy_data.l_inodebits.bits &
- policy->l_inodebits.bits) !=
- policy->l_inodebits.bits))
- continue;
+ if ((lpol->l_inodebits.bits &
+ data->lmd_policy->l_inodebits.bits) !=
+ data->lmd_policy->l_inodebits.bits)
+ return INTERVAL_ITER_CONT;
+ break;
+ default:
+ break;
+ }
+ /*
+ * We match if we have existing lock with same or wider set
+ * of bits.
+ */
+ if (!data->lmd_unref && LDLM_HAVE_MASK(lock, GONE))
+ return INTERVAL_ITER_CONT;
+
+ if ((data->lmd_flags & LDLM_FL_LOCAL_ONLY) &&
+ !ldlm_is_local(lock))
+ return INTERVAL_ITER_CONT;
- if (!unref && LDLM_HAVE_MASK(lock, GONE))
+ if (data->lmd_flags & LDLM_FL_TEST_LOCK) {
+ LDLM_LOCK_GET(lock);
+ ldlm_lock_touch_in_lru(lock);
+ } else {
+ ldlm_lock_addref_internal_nolock(lock, match);
+ }
+
+ *data->lmd_mode = match;
+ data->lmd_lock = lock;
+
+ return INTERVAL_ITER_STOP;
+}
+
+static unsigned int itree_overlap_cb(struct interval_node *in, void *args)
+{
+ struct ldlm_interval *node = to_ldlm_interval(in);
+ struct lock_match_data *data = args;
+ struct ldlm_lock *lock;
+ int rc;
+
+ list_for_each_entry(lock, &node->li_group, l_sl_policy) {
+ rc = lock_matches(lock, data);
+ if (rc == INTERVAL_ITER_STOP)
+ return INTERVAL_ITER_STOP;
+ }
+ return INTERVAL_ITER_CONT;
+}
+
+/**
+ * Search for a lock with given parameters in interval trees.
+ *
+ * \param res search for a lock in this resource
+ * \param data parameters
+ *
+ * \retval a referenced lock or NULL.
+ */
+static struct ldlm_lock *search_itree(struct ldlm_resource *res,
+ struct lock_match_data *data)
+{
+ struct interval_node_extent ext = {
+ .start = data->lmd_policy->l_extent.start,
+ .end = data->lmd_policy->l_extent.end
+ };
+ int idx;
+
+ for (idx = 0; idx < LCK_MODE_NUM; idx++) {
+ struct ldlm_interval_tree *tree = &res->lr_itree[idx];
+
+ if (!tree->lit_root)
continue;
- if ((flags & LDLM_FL_LOCAL_ONLY) && !ldlm_is_local(lock))
+ if (!(tree->lit_mode & *data->lmd_mode))
continue;
- if (flags & LDLM_FL_TEST_LOCK) {
- LDLM_LOCK_GET(lock);
- ldlm_lock_touch_in_lru(lock);
- } else {
- ldlm_lock_addref_internal_nolock(lock, match);
- }
- *mode = match;
- return lock;
+ interval_search(tree->lit_root, &ext,
+ itree_overlap_cb, data);
}
+ return data->lmd_lock;
+}
+/**
+ * Search for a lock with given properties in a queue.
+ *
+ * \param queue search for a lock in this queue
+ * \param data parameters
+ *
+ * \retval a referenced lock or NULL.
+ */
+static struct ldlm_lock *search_queue(struct list_head *queue,
+ struct lock_match_data *data)
+{
+ struct ldlm_lock *lock;
+ int rc;
+
+ list_for_each_entry(lock, queue, l_res_link) {
+ rc = lock_matches(lock, data);
+ if (rc == INTERVAL_ITER_STOP)
+ return data->lmd_lock;
+ }
return NULL;
}
@@ -1199,31 +1284,41 @@ enum ldlm_mode ldlm_lock_match(struct ldlm_namespace *ns, __u64 flags,
enum ldlm_mode mode,
struct lustre_handle *lockh, int unref)
{
+ struct lock_match_data data = {
+ .lmd_old = NULL,
+ .lmd_lock = NULL,
+ .lmd_mode = &mode,
+ .lmd_policy = policy,
+ .lmd_flags = flags,
+ .lmd_unref = unref,
+ };
struct ldlm_resource *res;
- struct ldlm_lock *lock, *old_lock = NULL;
+ struct ldlm_lock *lock;
int rc = 0;
if (!ns) {
- old_lock = ldlm_handle2lock(lockh);
- LASSERT(old_lock);
+ data.lmd_old = ldlm_handle2lock(lockh);
+ LASSERT(data.lmd_old);
- ns = ldlm_lock_to_ns(old_lock);
- res_id = &old_lock->l_resource->lr_name;
- type = old_lock->l_resource->lr_type;
- mode = old_lock->l_req_mode;
+ ns = ldlm_lock_to_ns(data.lmd_old);
+ res_id = &data.lmd_old->l_resource->lr_name;
+ type = data.lmd_old->l_resource->lr_type;
+ *data.lmd_mode = data.lmd_old->l_req_mode;
}
res = ldlm_resource_get(ns, NULL, res_id, type, 0);
if (IS_ERR(res)) {
- LASSERT(!old_lock);
+ LASSERT(!data.lmd_old);
return 0;
}
LDLM_RESOURCE_ADDREF(res);
lock_res(res);
- lock = search_queue(&res->lr_granted, &mode, policy, old_lock,
- flags, unref);
+ if (res->lr_type == LDLM_EXTENT)
+ lock = search_itree(res, &data);
+ else
+ lock = search_queue(&res->lr_granted, &data);
if (lock) {
rc = 1;
goto out;
@@ -1232,14 +1327,12 @@ enum ldlm_mode ldlm_lock_match(struct ldlm_namespace *ns, __u64 flags,
rc = 0;
goto out;
}
- lock = search_queue(&res->lr_waiting, &mode, policy, old_lock,
- flags, unref);
+ lock = search_queue(&res->lr_waiting, &data);
if (lock) {
rc = 1;
goto out;
}
-
- out:
+out:
unlock_res(res);
LDLM_RESOURCE_DELREF(res);
ldlm_resource_putref(res);
@@ -1311,8 +1404,8 @@ enum ldlm_mode ldlm_lock_match(struct ldlm_namespace *ns, __u64 flags,
(type == LDLM_PLAIN || type == LDLM_IBITS) ?
res_id->name[3] : policy->l_extent.end);
}
- if (old_lock)
- LDLM_LOCK_PUT(old_lock);
+ if (data.lmd_old)
+ LDLM_LOCK_PUT(data.lmd_old);
return rc ? mode : 0;
}
--
1.7.1
Powered by blists - more mailing lists