From c24f72cc7ca829bbad0532ddf315ace3ae1c359e Mon Sep 17 00:00:00 2001 From: Joel Becker Date: Sat, 3 Feb 2007 03:14:30 -0800 Subject: ocfs2: Proper cleanup in case of error in ocfs2_register_hb_callbacks() If ocfs2_register_hb_callbacks() succeeds on its first callback but fails its second, it doesn't release the first on the way out. Fix that. While we're at it, o2hb_unregister_callback() never returns anything but 0, so let's make it void. Signed-off-by: Joel Becker Signed-off-by: Mark Fasheh --- fs/ocfs2/cluster/heartbeat.c | 6 ++---- fs/ocfs2/cluster/heartbeat.h | 2 +- fs/ocfs2/cluster/tcp.c | 13 ++----------- fs/ocfs2/heartbeat.c | 15 +++++---------- 4 files changed, 10 insertions(+), 26 deletions(-) (limited to 'fs') diff --git a/fs/ocfs2/cluster/heartbeat.c b/fs/ocfs2/cluster/heartbeat.c index 5a9779bb923..0f2cfecd42c 100644 --- a/fs/ocfs2/cluster/heartbeat.c +++ b/fs/ocfs2/cluster/heartbeat.c @@ -1682,7 +1682,7 @@ out: } EXPORT_SYMBOL_GPL(o2hb_register_callback); -int o2hb_unregister_callback(struct o2hb_callback_func *hc) +void o2hb_unregister_callback(struct o2hb_callback_func *hc) { BUG_ON(hc->hc_magic != O2HB_CB_MAGIC); @@ -1690,15 +1690,13 @@ int o2hb_unregister_callback(struct o2hb_callback_func *hc) __builtin_return_address(0), hc); if (list_empty(&hc->hc_item)) - return 0; + return; down_write(&o2hb_callback_sem); list_del_init(&hc->hc_item); up_write(&o2hb_callback_sem); - - return 0; } EXPORT_SYMBOL_GPL(o2hb_unregister_callback); diff --git a/fs/ocfs2/cluster/heartbeat.h b/fs/ocfs2/cluster/heartbeat.h index cac6223206a..cc6d40b3977 100644 --- a/fs/ocfs2/cluster/heartbeat.h +++ b/fs/ocfs2/cluster/heartbeat.h @@ -70,7 +70,7 @@ void o2hb_setup_callback(struct o2hb_callback_func *hc, void *data, int priority); int o2hb_register_callback(struct o2hb_callback_func *hc); -int o2hb_unregister_callback(struct o2hb_callback_func *hc); +void o2hb_unregister_callback(struct o2hb_callback_func *hc); void o2hb_fill_node_map(unsigned long *map, unsigned bytes); void o2hb_init(void); diff --git a/fs/ocfs2/cluster/tcp.c b/fs/ocfs2/cluster/tcp.c index 1718215fc01..69caf3e12fe 100644 --- a/fs/ocfs2/cluster/tcp.c +++ b/fs/ocfs2/cluster/tcp.c @@ -1638,17 +1638,8 @@ static void o2net_hb_node_up_cb(struct o2nm_node *node, int node_num, void o2net_unregister_hb_callbacks(void) { - int ret; - - ret = o2hb_unregister_callback(&o2net_hb_up); - if (ret < 0) - mlog(ML_ERROR, "Status return %d unregistering heartbeat up " - "callback!\n", ret); - - ret = o2hb_unregister_callback(&o2net_hb_down); - if (ret < 0) - mlog(ML_ERROR, "Status return %d unregistering heartbeat down " - "callback!\n", ret); + o2hb_unregister_callback(&o2net_hb_up); + o2hb_unregister_callback(&o2net_hb_down); } int o2net_register_hb_callbacks(void) diff --git a/fs/ocfs2/heartbeat.c b/fs/ocfs2/heartbeat.c index 8fc52d6d0ce..b25ef63781b 100644 --- a/fs/ocfs2/heartbeat.c +++ b/fs/ocfs2/heartbeat.c @@ -164,8 +164,10 @@ int ocfs2_register_hb_callbacks(struct ocfs2_super *osb) } status = o2hb_register_callback(&osb->osb_hb_up); - if (status < 0) + if (status < 0) { mlog_errno(status); + o2hb_unregister_callback(&osb->osb_hb_down); + } bail: return status; @@ -173,18 +175,11 @@ bail: void ocfs2_clear_hb_callbacks(struct ocfs2_super *osb) { - int status; - if (ocfs2_mount_local(osb)) return; - status = o2hb_unregister_callback(&osb->osb_hb_down); - if (status < 0) - mlog_errno(status); - - status = o2hb_unregister_callback(&osb->osb_hb_up); - if (status < 0) - mlog_errno(status); + o2hb_unregister_callback(&osb->osb_hb_down); + o2hb_unregister_callback(&osb->osb_hb_up); } void ocfs2_stop_heartbeat(struct ocfs2_super *osb) -- cgit v1.2.3 From e6c352dbc0f4dc7e3f82feafb9e6207c5814a189 Mon Sep 17 00:00:00 2001 From: Joel Becker Date: Sat, 3 Feb 2007 03:04:20 -0800 Subject: ocfs2: Concurrent access of o2hb_region->hr_task was not locked This means that a build-up and a teardown could race which would result in a double-kthread_stop(). Protect the setting and clearing of hr_task with o2hb_live_lock, as it's not a common thing and not performance critical. Signed-off-by: Joel Becker Signed-off-by: Mark Fasheh --- fs/ocfs2/cluster/heartbeat.c | 44 ++++++++++++++++++++++++++++++++------------ 1 file changed, 32 insertions(+), 12 deletions(-) (limited to 'fs') diff --git a/fs/ocfs2/cluster/heartbeat.c b/fs/ocfs2/cluster/heartbeat.c index 0f2cfecd42c..eba282da500 100644 --- a/fs/ocfs2/cluster/heartbeat.c +++ b/fs/ocfs2/cluster/heartbeat.c @@ -1234,6 +1234,7 @@ static ssize_t o2hb_region_dev_write(struct o2hb_region *reg, const char *page, size_t count) { + struct task_struct *hb_task; long fd; int sectsize; char *p = (char *)page; @@ -1319,20 +1320,28 @@ static ssize_t o2hb_region_dev_write(struct o2hb_region *reg, */ atomic_set(®->hr_steady_iterations, O2HB_LIVE_THRESHOLD + 1); - reg->hr_task = kthread_run(o2hb_thread, reg, "o2hb-%s", - reg->hr_item.ci_name); - if (IS_ERR(reg->hr_task)) { - ret = PTR_ERR(reg->hr_task); + hb_task = kthread_run(o2hb_thread, reg, "o2hb-%s", + reg->hr_item.ci_name); + if (IS_ERR(hb_task)) { + ret = PTR_ERR(hb_task); mlog_errno(ret); - reg->hr_task = NULL; goto out; } + spin_lock(&o2hb_live_lock); + reg->hr_task = hb_task; + spin_unlock(&o2hb_live_lock); + ret = wait_event_interruptible(o2hb_steady_queue, atomic_read(®->hr_steady_iterations) == 0); if (ret) { - kthread_stop(reg->hr_task); + spin_lock(&o2hb_live_lock); + hb_task = reg->hr_task; reg->hr_task = NULL; + spin_unlock(&o2hb_live_lock); + + if (hb_task) + kthread_stop(hb_task); goto out; } @@ -1354,10 +1363,17 @@ out: static ssize_t o2hb_region_pid_read(struct o2hb_region *reg, char *page) { - if (!reg->hr_task) + pid_t pid = 0; + + spin_lock(&o2hb_live_lock); + if (reg->hr_task) + pid = reg->hr_task->pid; + spin_unlock(&o2hb_live_lock); + + if (!pid) return 0; - return sprintf(page, "%u\n", reg->hr_task->pid); + return sprintf(page, "%u\n", pid); } struct o2hb_region_attribute { @@ -1495,13 +1511,17 @@ out: static void o2hb_heartbeat_group_drop_item(struct config_group *group, struct config_item *item) { + struct task_struct *hb_task; struct o2hb_region *reg = to_o2hb_region(item); /* stop the thread when the user removes the region dir */ - if (reg->hr_task) { - kthread_stop(reg->hr_task); - reg->hr_task = NULL; - } + spin_lock(&o2hb_live_lock); + hb_task = reg->hr_task; + reg->hr_task = NULL; + spin_unlock(&o2hb_live_lock); + + if (hb_task) + kthread_stop(hb_task); config_item_put(item); } -- cgit v1.2.3 From 03f981cf2ec95dd8bc43d2ecccaec4e83c8375e2 Mon Sep 17 00:00:00 2001 From: Joel Becker Date: Thu, 4 Jan 2007 14:54:41 -0800 Subject: ocfs2: add some missing address space callbacks Under load, OCFS2 would crash in invalidate_inode_pages2_range() because invalidate_complete_page2() was unable to invalidate a page. It would appear that JBD is holding on to the page. ext3 has a specific ->releasepage() handler to cover this case. Steal ext3's ->releasepage(), ->invalidatepage(), and ->migratepage(), as they appear completely appropriate for OCFS2. Signed-off-by: Joel Becker Signed-off-by: Mark Fasheh --- fs/ocfs2/aops.c | 26 +++++++++++++++++++++++++- 1 file changed, 25 insertions(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/ocfs2/aops.c b/fs/ocfs2/aops.c index 93628b02ef5..875c1144381 100644 --- a/fs/ocfs2/aops.c +++ b/fs/ocfs2/aops.c @@ -614,6 +614,27 @@ static void ocfs2_dio_end_io(struct kiocb *iocb, ocfs2_rw_unlock(inode, 0); } +/* + * ocfs2_invalidatepage() and ocfs2_releasepage() are shamelessly stolen + * from ext3. PageChecked() bits have been removed as OCFS2 does not + * do journalled data. + */ +static void ocfs2_invalidatepage(struct page *page, unsigned long offset) +{ + journal_t *journal = OCFS2_SB(page->mapping->host->i_sb)->journal->j_journal; + + journal_invalidatepage(journal, page, offset); +} + +static int ocfs2_releasepage(struct page *page, gfp_t wait) +{ + journal_t *journal = OCFS2_SB(page->mapping->host->i_sb)->journal->j_journal; + + if (!page_has_buffers(page)) + return 0; + return journal_try_to_free_buffers(journal, page, wait); +} + static ssize_t ocfs2_direct_IO(int rw, struct kiocb *iocb, const struct iovec *iov, @@ -661,5 +682,8 @@ const struct address_space_operations ocfs2_aops = { .commit_write = ocfs2_commit_write, .bmap = ocfs2_bmap, .sync_page = block_sync_page, - .direct_IO = ocfs2_direct_IO + .direct_IO = ocfs2_direct_IO, + .invalidatepage = ocfs2_invalidatepage, + .releasepage = ocfs2_releasepage, + .migratepage = buffer_migrate_page, }; -- cgit v1.2.3 From afdf04ea098139e86147f63aad9c383cad3b6f37 Mon Sep 17 00:00:00 2001 From: Joel Becker Date: Mon, 5 Mar 2007 15:49:49 -0800 Subject: configfs: add missing mutex_unlock() d_alloc() failure in configfs_register_subsystem() would fail to unlock the mutex taken above. Reorganize the exit path to ensure the unlock happens. Reported-by: Akinobu Mita Signed-off-by: Joel Becker Signed-off-by: Mark Fasheh --- fs/configfs/dir.c | 27 ++++++++++++--------------- 1 file changed, 12 insertions(+), 15 deletions(-) (limited to 'fs') diff --git a/fs/configfs/dir.c b/fs/configfs/dir.c index 34750d5e4ff..5e6e37e58f3 100644 --- a/fs/configfs/dir.c +++ b/fs/configfs/dir.c @@ -1141,25 +1141,22 @@ int configfs_register_subsystem(struct configfs_subsystem *subsys) err = -ENOMEM; dentry = d_alloc(configfs_sb->s_root, &name); - if (!dentry) - goto out_release; - - d_add(dentry, NULL); + if (dentry) { + d_add(dentry, NULL); - err = configfs_attach_group(sd->s_element, &group->cg_item, - dentry); - if (!err) - dentry = NULL; - else - d_delete(dentry); + err = configfs_attach_group(sd->s_element, &group->cg_item, + dentry); + if (err) { + d_delete(dentry); + dput(dentry); + } + } mutex_unlock(&configfs_sb->s_root->d_inode->i_mutex); - if (dentry) { - dput(dentry); -out_release: - unlink_group(group); - configfs_release_fs(); + if (err) { + unlink_group(group); + configfs_release_fs(); } return err; -- cgit v1.2.3 From 3fca0894a4b5e52c278421b04435b88e32b423ad Mon Sep 17 00:00:00 2001 From: Sunil Mushran Date: Mon, 12 Mar 2007 13:24:34 -0700 Subject: ocfs2_dlm: Missing get/put lockres in dlm_run_purge_lockres In some circumstances, this was causing us to reference freed memory. Signed-off-by: Sunil Mushran Signed-off-by: Mark Fasheh --- fs/ocfs2/dlm/dlmthread.c | 2 ++ 1 file changed, 2 insertions(+) (limited to 'fs') diff --git a/fs/ocfs2/dlm/dlmthread.c b/fs/ocfs2/dlm/dlmthread.c index 8ffa0916eb8..6421a8fae1d 100644 --- a/fs/ocfs2/dlm/dlmthread.c +++ b/fs/ocfs2/dlm/dlmthread.c @@ -265,8 +265,10 @@ static void dlm_run_purge_list(struct dlm_ctxt *dlm, /* This may drop and reacquire the dlm spinlock if it * has to do migration. */ mlog(0, "calling dlm_purge_lockres!\n"); + dlm_lockres_get(lockres); if (dlm_purge_lockres(dlm, lockres)) BUG(); + dlm_lockres_put(lockres); mlog(0, "DONE calling dlm_purge_lockres!\n"); /* Avoid adding any scheduling latencies */ -- cgit v1.2.3 From b36c3f84988eebf38acaccc756e05f6b70e333ab Mon Sep 17 00:00:00 2001 From: Sunil Mushran Date: Mon, 12 Mar 2007 13:25:44 -0700 Subject: ocfs2_dlm: Add missing locks in dlm_empty_lockres __dlm_lockres_unused() expects the caller to take the lockres spinlock. Signed-off-by: Sunil Mushran Signed-off-by: Mark Fasheh --- fs/ocfs2/dlm/dlmmaster.c | 3 +++ 1 file changed, 3 insertions(+) (limited to 'fs') diff --git a/fs/ocfs2/dlm/dlmmaster.c b/fs/ocfs2/dlm/dlmmaster.c index 77e4e6169a0..9229e04362f 100644 --- a/fs/ocfs2/dlm/dlmmaster.c +++ b/fs/ocfs2/dlm/dlmmaster.c @@ -2730,14 +2730,17 @@ int dlm_empty_lockres(struct dlm_ctxt *dlm, struct dlm_lock_resource *res) int ret; int lock_dropped = 0; + spin_lock(&res->spinlock); if (res->owner != dlm->node_num) { if (!__dlm_lockres_unused(res)) { mlog(ML_ERROR, "%s:%.*s: this node is not master, " "trying to free this but locks remain\n", dlm->name, res->lockname.len, res->lockname.name); } + spin_unlock(&res->spinlock); goto leave; } + spin_unlock(&res->spinlock); /* Wheee! Migrate lockres here! Will sleep so drop spinlock. */ spin_unlock(&dlm->spinlock); -- cgit v1.2.3