From 03efed8a2a1b8e00164eb4720a82a7dd5e368a8e Mon Sep 17 00:00:00 2001 From: Tiger Yang Date: Sat, 28 May 2011 00:34:19 +0800 Subject: ocfs2: Bugfix for hard readonly mount ocfs2 cannot currently mount a device that is readonly at the media ("hard readonly"). Fix the broken places. see detail: http://oss.oracle.com/bugzilla/show_bug.cgi?id=1322 [ Description edited -- Joel ] Signed-off-by: Tiger Yang Reviewed-by: Sunil Mushran Signed-off-by: Joel Becker --- fs/ocfs2/dlmglue.c | 21 +++++++++++++++------ fs/ocfs2/super.c | 3 ++- 2 files changed, 17 insertions(+), 7 deletions(-) (limited to 'fs') diff --git a/fs/ocfs2/dlmglue.c b/fs/ocfs2/dlmglue.c index 7642d7ca73e..da103f555a6 100644 --- a/fs/ocfs2/dlmglue.c +++ b/fs/ocfs2/dlmglue.c @@ -1692,7 +1692,7 @@ int ocfs2_open_lock(struct inode *inode) mlog(0, "inode %llu take PRMODE open lock\n", (unsigned long long)OCFS2_I(inode)->ip_blkno); - if (ocfs2_mount_local(osb)) + if (ocfs2_is_hard_readonly(osb) || ocfs2_mount_local(osb)) goto out; lockres = &OCFS2_I(inode)->ip_open_lockres; @@ -1718,6 +1718,12 @@ int ocfs2_try_open_lock(struct inode *inode, int write) (unsigned long long)OCFS2_I(inode)->ip_blkno, write ? "EXMODE" : "PRMODE"); + if (ocfs2_is_hard_readonly(osb)) { + if (write) + status = -EROFS; + goto out; + } + if (ocfs2_mount_local(osb)) goto out; @@ -2298,7 +2304,7 @@ int ocfs2_inode_lock_full_nested(struct inode *inode, if (ocfs2_is_hard_readonly(osb)) { if (ex) status = -EROFS; - goto bail; + goto getbh; } if (ocfs2_mount_local(osb)) @@ -2356,7 +2362,7 @@ local: mlog_errno(status); goto bail; } - +getbh: if (ret_bh) { status = ocfs2_assign_bh(inode, ret_bh, local_bh); if (status < 0) { @@ -2628,8 +2634,11 @@ int ocfs2_dentry_lock(struct dentry *dentry, int ex) BUG_ON(!dl); - if (ocfs2_is_hard_readonly(osb)) - return -EROFS; + if (ocfs2_is_hard_readonly(osb)) { + if (ex) + return -EROFS; + return 0; + } if (ocfs2_mount_local(osb)) return 0; @@ -2647,7 +2656,7 @@ void ocfs2_dentry_unlock(struct dentry *dentry, int ex) struct ocfs2_dentry_lock *dl = dentry->d_fsdata; struct ocfs2_super *osb = OCFS2_SB(dentry->d_sb); - if (!ocfs2_mount_local(osb)) + if (!ocfs2_is_hard_readonly(osb) && !ocfs2_mount_local(osb)) ocfs2_cluster_unlock(osb, &dl->dl_lockres, level); } diff --git a/fs/ocfs2/super.c b/fs/ocfs2/super.c index cdbaf5e9730..029c4cd8a69 100644 --- a/fs/ocfs2/super.c +++ b/fs/ocfs2/super.c @@ -1974,7 +1974,8 @@ static void ocfs2_dismount_volume(struct super_block *sb, int mnt_err) * If we failed before we got a uuid_str yet, we can't stop * heartbeat. Otherwise, do it. */ - if (!mnt_err && !ocfs2_mount_local(osb) && osb->uuid_str) + if (!mnt_err && !ocfs2_mount_local(osb) && osb->uuid_str && + !ocfs2_is_hard_readonly(osb)) hangup_needed = 1; if (osb->cconn) -- cgit v1.2.3-70-g09d2 From 3d75be7c4771c7e4d5b5fa586a599af8473de32c Mon Sep 17 00:00:00 2001 From: Dan Carpenter Date: Sun, 29 May 2011 22:56:31 +0300 Subject: ocfs2: checking the wrong variable in ocfs2_move_extent() "new_phys_cpos" is always a valid pointer here. ocfs2_probe_alloc_group() allocates "*new_phys_cpos". Signed-off-by: Dan Carpenter Signed-off-by: Joel Becker --- fs/ocfs2/move_extents.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/ocfs2/move_extents.c b/fs/ocfs2/move_extents.c index cd9427023d2..d3433d60dbb 100644 --- a/fs/ocfs2/move_extents.c +++ b/fs/ocfs2/move_extents.c @@ -746,7 +746,7 @@ static int ocfs2_move_extent(struct ocfs2_move_extents_context *context, */ ocfs2_probe_alloc_group(inode, gd_bh, &goal_bit, len, move_max_hop, new_phys_cpos); - if (!new_phys_cpos) { + if (!*new_phys_cpos) { ret = -ENOSPC; goto out_commit; } -- cgit v1.2.3-70-g09d2 From 87f0d5c8db7aad85b9120c26723fdc63cd84a460 Mon Sep 17 00:00:00 2001 From: Dan Carpenter Date: Sun, 29 May 2011 22:57:16 +0300 Subject: ocfs2: null deref on allocation error The original code had a null derefence in the error handling. Signed-off-by: Dan Carpenter Signed-off-by: Joel Becker --- fs/ocfs2/ioctl.c | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) (limited to 'fs') diff --git a/fs/ocfs2/ioctl.c b/fs/ocfs2/ioctl.c index bc91072b721..d9a6ce77926 100644 --- a/fs/ocfs2/ioctl.c +++ b/fs/ocfs2/ioctl.c @@ -381,7 +381,7 @@ int ocfs2_info_handle_freeinode(struct inode *inode, if (!oifi) { status = -ENOMEM; mlog_errno(status); - goto bail; + goto out_err; } if (o2info_from_user(*oifi, req)) @@ -431,7 +431,7 @@ bail: o2info_set_request_error(&oifi->ifi_req, req); kfree(oifi); - +out_err: return status; } @@ -666,7 +666,7 @@ int ocfs2_info_handle_freefrag(struct inode *inode, if (!oiff) { status = -ENOMEM; mlog_errno(status); - goto bail; + goto out_err; } if (o2info_from_user(*oiff, req)) @@ -716,7 +716,7 @@ bail: o2info_set_request_error(&oiff->iff_req, req); kfree(oiff); - +out_err: return status; } -- cgit v1.2.3-70-g09d2 From 730e663bd82c1a10a85ff00728d34152a5a67ec8 Mon Sep 17 00:00:00 2001 From: Akinobu Mita Date: Mon, 30 May 2011 21:58:05 +0900 Subject: ocfs2: use proper little-endian bitops Using __test_and_{set,clear}_bit_le() with ignoring its return value can be replaced with __{set,clear}_bit_le(). Signed-off-by: Akinobu Mita Cc: Joel Becker Cc: Mark Fasheh Cc: ocfs2-devel@oss.oracle.com Signed-off-by: Joel Becker --- fs/ocfs2/ocfs2.h | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'fs') diff --git a/fs/ocfs2/ocfs2.h b/fs/ocfs2/ocfs2.h index 409285854f6..bcde467acca 100644 --- a/fs/ocfs2/ocfs2.h +++ b/fs/ocfs2/ocfs2.h @@ -836,13 +836,13 @@ static inline unsigned int ocfs2_clusters_to_megabytes(struct super_block *sb, static inline void _ocfs2_set_bit(unsigned int bit, unsigned long *bitmap) { - __test_and_set_bit_le(bit, bitmap); + __set_bit_le(bit, bitmap); } #define ocfs2_set_bit(bit, addr) _ocfs2_set_bit((bit), (unsigned long *)(addr)) static inline void _ocfs2_clear_bit(unsigned int bit, unsigned long *bitmap) { - __test_and_clear_bit_le(bit, bitmap); + __clear_bit_le(bit, bitmap); } #define ocfs2_clear_bit(bit, addr) _ocfs2_clear_bit((bit), (unsigned long *)(addr)) -- cgit v1.2.3-70-g09d2 From d2eece376648d2f7ba0a7d78f3c4d0421e608ac2 Mon Sep 17 00:00:00 2001 From: Sunil Mushran Date: Sun, 24 Jul 2011 10:21:54 -0700 Subject: ocfs2/cluster: Abort heartbeat start on hard-ro devices Currently if the heartbeat device is hard-ro, the o2hb thread keeps chugging along and dumping errors along the way. The user needs to manually stop the heartbeat. The patch addresses this shortcoming by adding a limit to the number of times the hb thread will iterate in an unsteady state. If the hb thread does not ready steady state in that many interation, the start is aborted. Signed-off-by: Sunil Mushran --- fs/ocfs2/cluster/heartbeat.c | 185 +++++++++++++++++++++++++++---------------- 1 file changed, 116 insertions(+), 69 deletions(-) (limited to 'fs') diff --git a/fs/ocfs2/cluster/heartbeat.c b/fs/ocfs2/cluster/heartbeat.c index 9a3e6bbff27..4728a746281 100644 --- a/fs/ocfs2/cluster/heartbeat.c +++ b/fs/ocfs2/cluster/heartbeat.c @@ -216,6 +216,7 @@ struct o2hb_region { struct list_head hr_all_item; unsigned hr_unclean_stop:1, + hr_aborted_start:1, hr_item_pinned:1, hr_item_dropped:1; @@ -254,6 +255,10 @@ struct o2hb_region { * a more complete api that doesn't lead to this sort of fragility. */ atomic_t hr_steady_iterations; + /* terminate o2hb thread if it does not reach steady state + * (hr_steady_iterations == 0) within hr_unsteady_iterations */ + atomic_t hr_unsteady_iterations; + char hr_dev_name[BDEVNAME_SIZE]; unsigned int hr_timeout_ms; @@ -324,6 +329,10 @@ static void o2hb_write_timeout(struct work_struct *work) static void o2hb_arm_write_timeout(struct o2hb_region *reg) { + /* Arm writeout only after thread reaches steady state */ + if (atomic_read(®->hr_steady_iterations) != 0) + return; + mlog(ML_HEARTBEAT, "Queue write timeout for %u ms\n", O2HB_MAX_WRITE_TIMEOUT_MS); @@ -537,9 +546,14 @@ static int o2hb_verify_crc(struct o2hb_region *reg, return read == computed; } -/* We want to make sure that nobody is heartbeating on top of us -- - * this will help detect an invalid configuration. */ -static void o2hb_check_last_timestamp(struct o2hb_region *reg) +/* + * Compare the slot data with what we wrote in the last iteration. + * If the match fails, print an appropriate error message. This is to + * detect errors like... another node hearting on the same slot, + * flaky device that is losing writes, etc. + * Returns 1 if check succeeds, 0 otherwise. + */ +static int o2hb_check_own_slot(struct o2hb_region *reg) { struct o2hb_disk_slot *slot; struct o2hb_disk_heartbeat_block *hb_block; @@ -548,13 +562,13 @@ static void o2hb_check_last_timestamp(struct o2hb_region *reg) slot = ®->hr_slots[o2nm_this_node()]; /* Don't check on our 1st timestamp */ if (!slot->ds_last_time) - return; + return 0; hb_block = slot->ds_raw_block; if (le64_to_cpu(hb_block->hb_seq) == slot->ds_last_time && le64_to_cpu(hb_block->hb_generation) == slot->ds_last_generation && hb_block->hb_node == slot->ds_node_num) - return; + return 1; #define ERRSTR1 "Another node is heartbeating on device" #define ERRSTR2 "Heartbeat generation mismatch on device" @@ -574,6 +588,8 @@ static void o2hb_check_last_timestamp(struct o2hb_region *reg) (unsigned long long)slot->ds_last_time, hb_block->hb_node, (unsigned long long)le64_to_cpu(hb_block->hb_generation), (unsigned long long)le64_to_cpu(hb_block->hb_seq)); + + return 0; } static inline void o2hb_prepare_block(struct o2hb_region *reg, @@ -719,17 +735,24 @@ static void o2hb_shutdown_slot(struct o2hb_disk_slot *slot) o2nm_node_put(node); } -static void o2hb_set_quorum_device(struct o2hb_region *reg, - struct o2hb_disk_slot *slot) +static void o2hb_set_quorum_device(struct o2hb_region *reg) { - assert_spin_locked(&o2hb_live_lock); - if (!o2hb_global_heartbeat_active()) return; - if (test_bit(reg->hr_region_num, o2hb_quorum_region_bitmap)) + /* Prevent race with o2hb_heartbeat_group_drop_item() */ + if (kthread_should_stop()) + return; + + /* Tag region as quorum only after thread reaches steady state */ + if (atomic_read(®->hr_steady_iterations) != 0) return; + spin_lock(&o2hb_live_lock); + + if (test_bit(reg->hr_region_num, o2hb_quorum_region_bitmap)) + goto unlock; + /* * A region can be added to the quorum only when it sees all * live nodes heartbeat on it. In other words, the region has been @@ -737,13 +760,10 @@ static void o2hb_set_quorum_device(struct o2hb_region *reg, */ if (memcmp(reg->hr_live_node_bitmap, o2hb_live_node_bitmap, sizeof(o2hb_live_node_bitmap))) - return; - - if (slot->ds_changed_samples < O2HB_LIVE_THRESHOLD) - return; + goto unlock; - printk(KERN_NOTICE "o2hb: Region %s is now a quorum device\n", - config_item_name(®->hr_item)); + printk(KERN_NOTICE "o2hb: Region %s (%s) is now a quorum device\n", + config_item_name(®->hr_item), reg->hr_dev_name); set_bit(reg->hr_region_num, o2hb_quorum_region_bitmap); @@ -754,6 +774,8 @@ static void o2hb_set_quorum_device(struct o2hb_region *reg, if (o2hb_pop_count(&o2hb_quorum_region_bitmap, O2NM_MAX_REGIONS) > O2HB_PIN_CUT_OFF) o2hb_region_unpin(NULL); +unlock: + spin_unlock(&o2hb_live_lock); } static int o2hb_check_slot(struct o2hb_region *reg, @@ -925,8 +947,6 @@ fire_callbacks: slot->ds_equal_samples = 0; } out: - o2hb_set_quorum_device(reg, slot); - spin_unlock(&o2hb_live_lock); o2hb_run_event_list(&event); @@ -957,7 +977,8 @@ static int o2hb_highest_node(unsigned long *nodes, static int o2hb_do_disk_heartbeat(struct o2hb_region *reg) { - int i, ret, highest_node, change = 0; + int i, ret, highest_node; + int membership_change = 0, own_slot_ok = 0; unsigned long configured_nodes[BITS_TO_LONGS(O2NM_MAX_NODES)]; unsigned long live_node_bitmap[BITS_TO_LONGS(O2NM_MAX_NODES)]; struct o2hb_bio_wait_ctxt write_wc; @@ -966,7 +987,7 @@ static int o2hb_do_disk_heartbeat(struct o2hb_region *reg) sizeof(configured_nodes)); if (ret) { mlog_errno(ret); - return ret; + goto bail; } /* @@ -982,8 +1003,9 @@ static int o2hb_do_disk_heartbeat(struct o2hb_region *reg) highest_node = o2hb_highest_node(configured_nodes, O2NM_MAX_NODES); if (highest_node >= O2NM_MAX_NODES) { - mlog(ML_NOTICE, "ocfs2_heartbeat: no configured nodes found!\n"); - return -EINVAL; + mlog(ML_NOTICE, "o2hb: No configured nodes found!\n"); + ret = -EINVAL; + goto bail; } /* No sense in reading the slots of nodes that don't exist @@ -993,29 +1015,27 @@ static int o2hb_do_disk_heartbeat(struct o2hb_region *reg) ret = o2hb_read_slots(reg, highest_node + 1); if (ret < 0) { mlog_errno(ret); - return ret; + goto bail; } /* With an up to date view of the slots, we can check that no * other node has been improperly configured to heartbeat in * our slot. */ - o2hb_check_last_timestamp(reg); + own_slot_ok = o2hb_check_own_slot(reg); /* fill in the proper info for our next heartbeat */ o2hb_prepare_block(reg, reg->hr_generation); - /* And fire off the write. Note that we don't wait on this I/O - * until later. */ ret = o2hb_issue_node_write(reg, &write_wc); if (ret < 0) { mlog_errno(ret); - return ret; + goto bail; } i = -1; while((i = find_next_bit(configured_nodes, O2NM_MAX_NODES, i + 1)) < O2NM_MAX_NODES) { - change |= o2hb_check_slot(reg, ®->hr_slots[i]); + membership_change |= o2hb_check_slot(reg, ®->hr_slots[i]); } /* @@ -1030,18 +1050,39 @@ static int o2hb_do_disk_heartbeat(struct o2hb_region *reg) * disk */ mlog(ML_ERROR, "Write error %d on device \"%s\"\n", write_wc.wc_error, reg->hr_dev_name); - return write_wc.wc_error; + ret = write_wc.wc_error; + goto bail; } - o2hb_arm_write_timeout(reg); + /* Skip disarming the timeout if own slot has stale/bad data */ + if (own_slot_ok) { + o2hb_set_quorum_device(reg); + o2hb_arm_write_timeout(reg); + } +bail: /* let the person who launched us know when things are steady */ - if (!change && (atomic_read(®->hr_steady_iterations) != 0)) { - if (atomic_dec_and_test(®->hr_steady_iterations)) + if (atomic_read(®->hr_steady_iterations) != 0) { + if (!ret && own_slot_ok && !membership_change) { + if (atomic_dec_and_test(®->hr_steady_iterations)) + wake_up(&o2hb_steady_queue); + } + } + + if (atomic_read(®->hr_steady_iterations) != 0) { + if (atomic_dec_and_test(®->hr_unsteady_iterations)) { + printk(KERN_NOTICE "o2hb: Unable to stabilize " + "heartbeart on region %s (%s)\n", + config_item_name(®->hr_item), + reg->hr_dev_name); + atomic_set(®->hr_steady_iterations, 0); + reg->hr_aborted_start = 1; wake_up(&o2hb_steady_queue); + ret = -EIO; + } } - return 0; + return ret; } /* Subtract b from a, storing the result in a. a *must* have a larger @@ -1095,7 +1136,8 @@ static int o2hb_thread(void *data) /* Pin node */ o2nm_depend_this_node(); - while (!kthread_should_stop() && !reg->hr_unclean_stop) { + while (!kthread_should_stop() && + !reg->hr_unclean_stop && !reg->hr_aborted_start) { /* We track the time spent inside * o2hb_do_disk_heartbeat so that we avoid more than * hr_timeout_ms between disk writes. On busy systems @@ -1103,10 +1145,7 @@ static int o2hb_thread(void *data) * likely to time itself out. */ do_gettimeofday(&before_hb); - i = 0; - do { - ret = o2hb_do_disk_heartbeat(reg); - } while (ret && ++i < 2); + ret = o2hb_do_disk_heartbeat(reg); do_gettimeofday(&after_hb); elapsed_msec = o2hb_elapsed_msecs(&before_hb, &after_hb); @@ -1117,7 +1156,8 @@ static int o2hb_thread(void *data) after_hb.tv_sec, (unsigned long) after_hb.tv_usec, elapsed_msec); - if (elapsed_msec < reg->hr_timeout_ms) { + if (!kthread_should_stop() && + elapsed_msec < reg->hr_timeout_ms) { /* the kthread api has blocked signals for us so no * need to record the return value. */ msleep_interruptible(reg->hr_timeout_ms - elapsed_msec); @@ -1134,20 +1174,20 @@ static int o2hb_thread(void *data) * to timeout on this region when we could just as easily * write a clear generation - thus indicating to them that * this node has left this region. - * - * XXX: Should we skip this on unclean_stop? */ - o2hb_prepare_block(reg, 0); - ret = o2hb_issue_node_write(reg, &write_wc); - if (ret == 0) { - o2hb_wait_on_io(reg, &write_wc); - } else { - mlog_errno(ret); + */ + if (!reg->hr_unclean_stop && !reg->hr_aborted_start) { + o2hb_prepare_block(reg, 0); + ret = o2hb_issue_node_write(reg, &write_wc); + if (ret == 0) + o2hb_wait_on_io(reg, &write_wc); + else + mlog_errno(ret); } /* Unpin node */ o2nm_undepend_this_node(); - mlog(ML_HEARTBEAT|ML_KTHREAD, "hb thread exiting\n"); + mlog(ML_HEARTBEAT|ML_KTHREAD, "o2hb thread exiting\n"); return 0; } @@ -1426,6 +1466,8 @@ static void o2hb_region_release(struct config_item *item) struct page *page; struct o2hb_region *reg = to_o2hb_region(item); + mlog(ML_HEARTBEAT, "hb region release (%s)\n", reg->hr_dev_name); + if (reg->hr_tmp_block) kfree(reg->hr_tmp_block); @@ -1792,7 +1834,10 @@ static ssize_t o2hb_region_dev_write(struct o2hb_region *reg, live_threshold <<= 1; spin_unlock(&o2hb_live_lock); } - atomic_set(®->hr_steady_iterations, live_threshold + 1); + ++live_threshold; + atomic_set(®->hr_steady_iterations, live_threshold); + /* unsteady_iterations is double the steady_iterations */ + atomic_set(®->hr_unsteady_iterations, (live_threshold << 1)); hb_task = kthread_run(o2hb_thread, reg, "o2hb-%s", reg->hr_item.ci_name); @@ -1809,14 +1854,12 @@ static ssize_t o2hb_region_dev_write(struct o2hb_region *reg, ret = wait_event_interruptible(o2hb_steady_queue, atomic_read(®->hr_steady_iterations) == 0); if (ret) { - /* We got interrupted (hello ptrace!). Clean up */ - spin_lock(&o2hb_live_lock); - hb_task = reg->hr_task; - reg->hr_task = NULL; - spin_unlock(&o2hb_live_lock); + atomic_set(®->hr_steady_iterations, 0); + reg->hr_aborted_start = 1; + } - if (hb_task) - kthread_stop(hb_task); + if (reg->hr_aborted_start) { + ret = -EIO; goto out; } @@ -1833,8 +1876,8 @@ static ssize_t o2hb_region_dev_write(struct o2hb_region *reg, ret = -EIO; if (hb_task && o2hb_global_heartbeat_active()) - printk(KERN_NOTICE "o2hb: Heartbeat started on region %s\n", - config_item_name(®->hr_item)); + printk(KERN_NOTICE "o2hb: Heartbeat started on region %s (%s)\n", + config_item_name(®->hr_item), reg->hr_dev_name); out: if (filp) @@ -2092,13 +2135,6 @@ static void o2hb_heartbeat_group_drop_item(struct config_group *group, /* stop the thread when the user removes the region dir */ spin_lock(&o2hb_live_lock); - if (o2hb_global_heartbeat_active()) { - clear_bit(reg->hr_region_num, o2hb_region_bitmap); - clear_bit(reg->hr_region_num, o2hb_live_region_bitmap); - if (test_bit(reg->hr_region_num, o2hb_quorum_region_bitmap)) - quorum_region = 1; - clear_bit(reg->hr_region_num, o2hb_quorum_region_bitmap); - } hb_task = reg->hr_task; reg->hr_task = NULL; reg->hr_item_dropped = 1; @@ -2107,19 +2143,30 @@ static void o2hb_heartbeat_group_drop_item(struct config_group *group, if (hb_task) kthread_stop(hb_task); + if (o2hb_global_heartbeat_active()) { + spin_lock(&o2hb_live_lock); + clear_bit(reg->hr_region_num, o2hb_region_bitmap); + clear_bit(reg->hr_region_num, o2hb_live_region_bitmap); + if (test_bit(reg->hr_region_num, o2hb_quorum_region_bitmap)) + quorum_region = 1; + clear_bit(reg->hr_region_num, o2hb_quorum_region_bitmap); + spin_unlock(&o2hb_live_lock); + printk(KERN_NOTICE "o2hb: Heartbeat %s on region %s (%s)\n", + ((atomic_read(®->hr_steady_iterations) == 0) ? + "stopped" : "start aborted"), config_item_name(item), + reg->hr_dev_name); + } + /* * If we're racing a dev_write(), we need to wake them. They will * check reg->hr_task */ if (atomic_read(®->hr_steady_iterations) != 0) { + reg->hr_aborted_start = 1; atomic_set(®->hr_steady_iterations, 0); wake_up(&o2hb_steady_queue); } - if (o2hb_global_heartbeat_active()) - printk(KERN_NOTICE "o2hb: Heartbeat stopped on region %s\n", - config_item_name(®->hr_item)); - config_item_put(item); if (!o2hb_global_heartbeat_active() || !quorum_region) -- cgit v1.2.3-70-g09d2 From 1dfecf810e0eacb35987905082f23e5c2cd26e91 Mon Sep 17 00:00:00 2001 From: Sunil Mushran Date: Sun, 24 Jul 2011 10:22:54 -0700 Subject: ocfs2/cluster: Clean up messages in o2net o2net messages needed a facelift. Signed-off-by: Sunil Mushran --- fs/ocfs2/cluster/tcp.c | 119 ++++++++++++++++++++++--------------------------- 1 file changed, 53 insertions(+), 66 deletions(-) (limited to 'fs') diff --git a/fs/ocfs2/cluster/tcp.c b/fs/ocfs2/cluster/tcp.c index db5ee4b4f47..6e97895ecf2 100644 --- a/fs/ocfs2/cluster/tcp.c +++ b/fs/ocfs2/cluster/tcp.c @@ -545,7 +545,7 @@ static void o2net_set_nn_state(struct o2net_node *nn, } if (was_valid && !valid) { - printk(KERN_NOTICE "o2net: no longer connected to " + printk(KERN_NOTICE "o2net: No longer connected to " SC_NODEF_FMT "\n", SC_NODEF_ARGS(old_sc)); o2net_complete_nodes_nsw(nn); } @@ -555,7 +555,7 @@ static void o2net_set_nn_state(struct o2net_node *nn, cancel_delayed_work(&nn->nn_connect_expired); printk(KERN_NOTICE "o2net: %s " SC_NODEF_FMT "\n", o2nm_this_node() > sc->sc_node->nd_num ? - "connected to" : "accepted connection from", + "Connected to" : "Accepted connection from", SC_NODEF_ARGS(sc)); } @@ -643,7 +643,7 @@ static void o2net_state_change(struct sock *sk) o2net_sc_queue_work(sc, &sc->sc_connect_work); break; default: - printk(KERN_INFO "o2net: connection to " SC_NODEF_FMT + printk(KERN_INFO "o2net: Connection to " SC_NODEF_FMT " shutdown, state %d\n", SC_NODEF_ARGS(sc), sk->sk_state); o2net_sc_queue_work(sc, &sc->sc_shutdown_work); @@ -1284,11 +1284,11 @@ static int o2net_check_handshake(struct o2net_sock_container *sc) struct o2net_node *nn = o2net_nn_from_num(sc->sc_node->nd_num); if (hand->protocol_version != cpu_to_be64(O2NET_PROTOCOL_VERSION)) { - mlog(ML_NOTICE, SC_NODEF_FMT " advertised net protocol " - "version %llu but %llu is required, disconnecting\n", - SC_NODEF_ARGS(sc), - (unsigned long long)be64_to_cpu(hand->protocol_version), - O2NET_PROTOCOL_VERSION); + printk(KERN_NOTICE "o2net: " SC_NODEF_FMT " Advertised net " + "protocol version %llu but %llu is required. " + "Disconnecting.\n", SC_NODEF_ARGS(sc), + (unsigned long long)be64_to_cpu(hand->protocol_version), + O2NET_PROTOCOL_VERSION); /* don't bother reconnecting if its the wrong version. */ o2net_ensure_shutdown(nn, sc, -ENOTCONN); @@ -1302,33 +1302,33 @@ static int o2net_check_handshake(struct o2net_sock_container *sc) */ if (be32_to_cpu(hand->o2net_idle_timeout_ms) != o2net_idle_timeout()) { - mlog(ML_NOTICE, SC_NODEF_FMT " uses a network idle timeout of " - "%u ms, but we use %u ms locally. disconnecting\n", - SC_NODEF_ARGS(sc), - be32_to_cpu(hand->o2net_idle_timeout_ms), - o2net_idle_timeout()); + printk(KERN_NOTICE "o2net: " SC_NODEF_FMT " uses a network " + "idle timeout of %u ms, but we use %u ms locally. " + "Disconnecting.\n", SC_NODEF_ARGS(sc), + be32_to_cpu(hand->o2net_idle_timeout_ms), + o2net_idle_timeout()); o2net_ensure_shutdown(nn, sc, -ENOTCONN); return -1; } if (be32_to_cpu(hand->o2net_keepalive_delay_ms) != o2net_keepalive_delay()) { - mlog(ML_NOTICE, SC_NODEF_FMT " uses a keepalive delay of " - "%u ms, but we use %u ms locally. disconnecting\n", - SC_NODEF_ARGS(sc), - be32_to_cpu(hand->o2net_keepalive_delay_ms), - o2net_keepalive_delay()); + printk(KERN_NOTICE "o2net: " SC_NODEF_FMT " uses a keepalive " + "delay of %u ms, but we use %u ms locally. " + "Disconnecting.\n", SC_NODEF_ARGS(sc), + be32_to_cpu(hand->o2net_keepalive_delay_ms), + o2net_keepalive_delay()); o2net_ensure_shutdown(nn, sc, -ENOTCONN); return -1; } if (be32_to_cpu(hand->o2hb_heartbeat_timeout_ms) != O2HB_MAX_WRITE_TIMEOUT_MS) { - mlog(ML_NOTICE, SC_NODEF_FMT " uses a heartbeat timeout of " - "%u ms, but we use %u ms locally. disconnecting\n", - SC_NODEF_ARGS(sc), - be32_to_cpu(hand->o2hb_heartbeat_timeout_ms), - O2HB_MAX_WRITE_TIMEOUT_MS); + printk(KERN_NOTICE "o2net: " SC_NODEF_FMT " uses a heartbeat " + "timeout of %u ms, but we use %u ms locally. " + "Disconnecting.\n", SC_NODEF_ARGS(sc), + be32_to_cpu(hand->o2hb_heartbeat_timeout_ms), + O2HB_MAX_WRITE_TIMEOUT_MS); o2net_ensure_shutdown(nn, sc, -ENOTCONN); return -1; } @@ -1539,28 +1539,16 @@ static void o2net_idle_timer(unsigned long data) { struct o2net_sock_container *sc = (struct o2net_sock_container *)data; struct o2net_node *nn = o2net_nn_from_num(sc->sc_node->nd_num); - #ifdef CONFIG_DEBUG_FS - ktime_t now = ktime_get(); + unsigned long msecs = ktime_to_ms(ktime_get()) - + ktime_to_ms(sc->sc_tv_timer); +#else + unsigned long msecs = o2net_idle_timeout(); #endif - printk(KERN_NOTICE "o2net: connection to " SC_NODEF_FMT " has been idle for %u.%u " - "seconds, shutting it down.\n", SC_NODEF_ARGS(sc), - o2net_idle_timeout() / 1000, - o2net_idle_timeout() % 1000); - -#ifdef CONFIG_DEBUG_FS - mlog(ML_NOTICE, "Here are some times that might help debug the " - "situation: (Timer: %lld, Now %lld, DataReady %lld, Advance %lld-%lld, " - "Key 0x%08x, Func %u, FuncTime %lld-%lld)\n", - (long long)ktime_to_us(sc->sc_tv_timer), (long long)ktime_to_us(now), - (long long)ktime_to_us(sc->sc_tv_data_ready), - (long long)ktime_to_us(sc->sc_tv_advance_start), - (long long)ktime_to_us(sc->sc_tv_advance_stop), - sc->sc_msg_key, sc->sc_msg_type, - (long long)ktime_to_us(sc->sc_tv_func_start), - (long long)ktime_to_us(sc->sc_tv_func_stop)); -#endif + printk(KERN_NOTICE "o2net: Connection to " SC_NODEF_FMT " has been " + "idle for %lu.%lu secs, shutting it down.\n", SC_NODEF_ARGS(sc), + msecs / 1000, msecs % 1000); /* * Initialize the nn_timeout so that the next connection attempt @@ -1693,8 +1681,8 @@ static void o2net_start_connect(struct work_struct *work) out: if (ret) { - mlog(ML_NOTICE, "connect attempt to " SC_NODEF_FMT " failed " - "with errno %d\n", SC_NODEF_ARGS(sc), ret); + printk(KERN_NOTICE "o2net: Connect attempt to " SC_NODEF_FMT + " failed with errno %d\n", SC_NODEF_ARGS(sc), ret); /* 0 err so that another will be queued and attempted * from set_nn_state */ if (sc) @@ -1717,8 +1705,8 @@ static void o2net_connect_expired(struct work_struct *work) spin_lock(&nn->nn_lock); if (!nn->nn_sc_valid) { - mlog(ML_ERROR, "no connection established with node %u after " - "%u.%u seconds, giving up and returning errors.\n", + printk(KERN_NOTICE "o2net: No connection established with " + "node %u after %u.%u seconds, giving up.\n", o2net_num_from_nn(nn), o2net_idle_timeout() / 1000, o2net_idle_timeout() % 1000); @@ -1861,21 +1849,21 @@ static int o2net_accept_one(struct socket *sock) node = o2nm_get_node_by_ip(sin.sin_addr.s_addr); if (node == NULL) { - mlog(ML_NOTICE, "attempt to connect from unknown node at %pI4:%d\n", - &sin.sin_addr.s_addr, ntohs(sin.sin_port)); + printk(KERN_NOTICE "o2net: Attempt to connect from unknown " + "node at %pI4:%d\n", &sin.sin_addr.s_addr, + ntohs(sin.sin_port)); ret = -EINVAL; goto out; } if (o2nm_this_node() >= node->nd_num) { local_node = o2nm_get_node_by_num(o2nm_this_node()); - mlog(ML_NOTICE, "unexpected connect attempt seen at node '%s' (" - "%u, %pI4:%d) from node '%s' (%u, %pI4:%d)\n", - local_node->nd_name, local_node->nd_num, - &(local_node->nd_ipv4_address), - ntohs(local_node->nd_ipv4_port), - node->nd_name, node->nd_num, &sin.sin_addr.s_addr, - ntohs(sin.sin_port)); + printk(KERN_NOTICE "o2net: Unexpected connect attempt seen " + "at node '%s' (%u, %pI4:%d) from node '%s' (%u, " + "%pI4:%d)\n", local_node->nd_name, local_node->nd_num, + &(local_node->nd_ipv4_address), + ntohs(local_node->nd_ipv4_port), node->nd_name, + node->nd_num, &sin.sin_addr.s_addr, ntohs(sin.sin_port)); ret = -EINVAL; goto out; } @@ -1900,10 +1888,10 @@ static int o2net_accept_one(struct socket *sock) ret = 0; spin_unlock(&nn->nn_lock); if (ret) { - mlog(ML_NOTICE, "attempt to connect from node '%s' at " - "%pI4:%d but it already has an open connection\n", - node->nd_name, &sin.sin_addr.s_addr, - ntohs(sin.sin_port)); + printk(KERN_NOTICE "o2net: Attempt to connect from node '%s' " + "at %pI4:%d but it already has an open connection\n", + node->nd_name, &sin.sin_addr.s_addr, + ntohs(sin.sin_port)); goto out; } @@ -1983,7 +1971,7 @@ static int o2net_open_listening_sock(__be32 addr, __be16 port) ret = sock_create(PF_INET, SOCK_STREAM, IPPROTO_TCP, &sock); if (ret < 0) { - mlog(ML_ERROR, "unable to create socket, ret=%d\n", ret); + printk(KERN_ERR "o2net: Error %d while creating socket\n", ret); goto out; } @@ -2000,16 +1988,15 @@ static int o2net_open_listening_sock(__be32 addr, __be16 port) sock->sk->sk_reuse = 1; ret = sock->ops->bind(sock, (struct sockaddr *)&sin, sizeof(sin)); if (ret < 0) { - mlog(ML_ERROR, "unable to bind socket at %pI4:%u, " - "ret=%d\n", &addr, ntohs(port), ret); + printk(KERN_ERR "o2net: Error %d while binding socket at " + "%pI4:%u\n", ret, &addr, ntohs(port)); goto out; } ret = sock->ops->listen(sock, 64); - if (ret < 0) { - mlog(ML_ERROR, "unable to listen on %pI4:%u, ret=%d\n", - &addr, ntohs(port), ret); - } + if (ret < 0) + printk(KERN_ERR "o2net: Error %d while listening on %pI4:%u\n", + ret, &addr, ntohs(port)); out: if (ret) { -- cgit v1.2.3-70-g09d2 From 8decab3c8dadcdf4f54ffb30df6e6f67b398b6e0 Mon Sep 17 00:00:00 2001 From: Sunil Mushran Date: Sun, 24 Jul 2011 10:23:54 -0700 Subject: ocfs2/dlm: Clean up messages in o2dlm o2dlm messages needed a facelift. Signed-off-by: Sunil Mushran --- fs/ocfs2/dlm/dlmdomain.c | 18 +++++++++------- fs/ocfs2/dlm/dlmlock.c | 22 ++++++++----------- fs/ocfs2/dlm/dlmmaster.c | 46 ++++++++++++++++++---------------------- fs/ocfs2/dlm/dlmrecovery.c | 53 +++++++++++++++++++++++++++------------------- fs/ocfs2/dlm/dlmthread.c | 2 -- 5 files changed, 71 insertions(+), 70 deletions(-) (limited to 'fs') diff --git a/fs/ocfs2/dlm/dlmdomain.c b/fs/ocfs2/dlm/dlmdomain.c index 6ed6b95dcf9..ce225fb0cea 100644 --- a/fs/ocfs2/dlm/dlmdomain.c +++ b/fs/ocfs2/dlm/dlmdomain.c @@ -539,17 +539,17 @@ again: static void __dlm_print_nodes(struct dlm_ctxt *dlm) { - int node = -1; + int node = -1, num = 0; assert_spin_locked(&dlm->spinlock); - printk(KERN_NOTICE "o2dlm: Nodes in domain %s: ", dlm->name); - + printk("( "); while ((node = find_next_bit(dlm->domain_map, O2NM_MAX_NODES, node + 1)) < O2NM_MAX_NODES) { printk("%d ", node); + ++num; } - printk("\n"); + printk(") %u nodes\n", num); } static int dlm_exit_domain_handler(struct o2net_msg *msg, u32 len, void *data, @@ -566,11 +566,10 @@ static int dlm_exit_domain_handler(struct o2net_msg *msg, u32 len, void *data, node = exit_msg->node_idx; - printk(KERN_NOTICE "o2dlm: Node %u leaves domain %s\n", node, dlm->name); - spin_lock(&dlm->spinlock); clear_bit(node, dlm->domain_map); clear_bit(node, dlm->exit_domain_map); + printk(KERN_NOTICE "o2dlm: Node %u leaves domain %s ", node, dlm->name); __dlm_print_nodes(dlm); /* notify anything attached to the heartbeat events */ @@ -755,6 +754,7 @@ void dlm_unregister_domain(struct dlm_ctxt *dlm) dlm_mark_domain_leaving(dlm); dlm_leave_domain(dlm); + printk(KERN_NOTICE "o2dlm: Leaving domain %s\n", dlm->name); dlm_force_free_mles(dlm); dlm_complete_dlm_shutdown(dlm); } @@ -970,7 +970,7 @@ static int dlm_assert_joined_handler(struct o2net_msg *msg, u32 len, void *data, clear_bit(assert->node_idx, dlm->exit_domain_map); __dlm_set_joining_node(dlm, DLM_LOCK_RES_OWNER_UNKNOWN); - printk(KERN_NOTICE "o2dlm: Node %u joins domain %s\n", + printk(KERN_NOTICE "o2dlm: Node %u joins domain %s ", assert->node_idx, dlm->name); __dlm_print_nodes(dlm); @@ -1701,8 +1701,10 @@ static int dlm_try_to_join_domain(struct dlm_ctxt *dlm) bail: spin_lock(&dlm->spinlock); __dlm_set_joining_node(dlm, DLM_LOCK_RES_OWNER_UNKNOWN); - if (!status) + if (!status) { + printk(KERN_NOTICE "o2dlm: Joining domain %s ", dlm->name); __dlm_print_nodes(dlm); + } spin_unlock(&dlm->spinlock); if (ctxt) { diff --git a/fs/ocfs2/dlm/dlmlock.c b/fs/ocfs2/dlm/dlmlock.c index 8d39e0fd66f..c7f3e22bda1 100644 --- a/fs/ocfs2/dlm/dlmlock.c +++ b/fs/ocfs2/dlm/dlmlock.c @@ -319,27 +319,23 @@ static enum dlm_status dlm_send_remote_lock_request(struct dlm_ctxt *dlm, tmpret = o2net_send_message(DLM_CREATE_LOCK_MSG, dlm->key, &create, sizeof(create), res->owner, &status); if (tmpret >= 0) { - // successfully sent and received - ret = status; // this is already a dlm_status + ret = status; if (ret == DLM_REJECTED) { - mlog(ML_ERROR, "%s:%.*s: BUG. this is a stale lockres " - "no longer owned by %u. that node is coming back " - "up currently.\n", dlm->name, create.namelen, + mlog(ML_ERROR, "%s: res %.*s, Stale lockres no longer " + "owned by node %u. That node is coming back up " + "currently.\n", dlm->name, create.namelen, create.name, res->owner); dlm_print_one_lock_resource(res); BUG(); } } else { - mlog(ML_ERROR, "Error %d when sending message %u (key 0x%x) to " - "node %u\n", tmpret, DLM_CREATE_LOCK_MSG, dlm->key, - res->owner); - if (dlm_is_host_down(tmpret)) { + mlog(ML_ERROR, "%s: res %.*s, Error %d send CREATE LOCK to " + "node %u\n", dlm->name, create.namelen, create.name, + tmpret, res->owner); + if (dlm_is_host_down(tmpret)) ret = DLM_RECOVERING; - mlog(0, "node %u died so returning DLM_RECOVERING " - "from lock message!\n", res->owner); - } else { + else ret = dlm_err_to_dlm_status(tmpret); - } } return ret; diff --git a/fs/ocfs2/dlm/dlmmaster.c b/fs/ocfs2/dlm/dlmmaster.c index 11eefb8c12e..9f3b093b629 100644 --- a/fs/ocfs2/dlm/dlmmaster.c +++ b/fs/ocfs2/dlm/dlmmaster.c @@ -829,8 +829,8 @@ lookup: * but they might own this lockres. wait on them. */ bit = find_next_bit(dlm->recovery_map, O2NM_MAX_NODES, 0); if (bit < O2NM_MAX_NODES) { - mlog(ML_NOTICE, "%s:%.*s: at least one node (%d) to " - "recover before lock mastery can begin\n", + mlog(0, "%s: res %.*s, At least one node (%d) " + "to recover before lock mastery can begin\n", dlm->name, namelen, (char *)lockid, bit); wait_on_recovery = 1; } @@ -864,8 +864,8 @@ redo_request: * dlm spinlock would be detectable be a change on the mle, * so we only need to clear out the recovery map once. */ if (dlm_is_recovery_lock(lockid, namelen)) { - mlog(ML_NOTICE, "%s: recovery map is not empty, but " - "must master $RECOVERY lock now\n", dlm->name); + mlog(0, "%s: Recovery map is not empty, but must " + "master $RECOVERY lock now\n", dlm->name); if (!dlm_pre_master_reco_lockres(dlm, res)) wait_on_recovery = 0; else { @@ -883,8 +883,8 @@ redo_request: spin_lock(&dlm->spinlock); bit = find_next_bit(dlm->recovery_map, O2NM_MAX_NODES, 0); if (bit < O2NM_MAX_NODES) { - mlog(ML_NOTICE, "%s:%.*s: at least one node (%d) to " - "recover before lock mastery can begin\n", + mlog(0, "%s: res %.*s, At least one node (%d) " + "to recover before lock mastery can begin\n", dlm->name, namelen, (char *)lockid, bit); wait_on_recovery = 1; } else @@ -913,8 +913,8 @@ redo_request: * yet, keep going until it does. this is how the * master will know that asserts are needed back to * the lower nodes. */ - mlog(0, "%s:%.*s: requests only up to %u but master " - "is %u, keep going\n", dlm->name, namelen, + mlog(0, "%s: res %.*s, Requests only up to %u but " + "master is %u, keep going\n", dlm->name, namelen, lockid, nodenum, mle->master); } } @@ -924,13 +924,12 @@ wait: ret = dlm_wait_for_lock_mastery(dlm, res, mle, &blocked); if (ret < 0) { wait_on_recovery = 1; - mlog(0, "%s:%.*s: node map changed, redo the " - "master request now, blocked=%d\n", - dlm->name, res->lockname.len, + mlog(0, "%s: res %.*s, Node map changed, redo the master " + "request now, blocked=%d\n", dlm->name, res->lockname.len, res->lockname.name, blocked); if (++tries > 20) { - mlog(ML_ERROR, "%s:%.*s: spinning on " - "dlm_wait_for_lock_mastery, blocked=%d\n", + mlog(ML_ERROR, "%s: res %.*s, Spinning on " + "dlm_wait_for_lock_mastery, blocked = %d\n", dlm->name, res->lockname.len, res->lockname.name, blocked); dlm_print_one_lock_resource(res); @@ -940,7 +939,8 @@ wait: goto redo_request; } - mlog(0, "lockres mastered by %u\n", res->owner); + mlog(0, "%s: res %.*s, Mastered by %u\n", dlm->name, res->lockname.len, + res->lockname.name, res->owner); /* make sure we never continue without this */ BUG_ON(res->owner == O2NM_MAX_NODES); @@ -2187,8 +2187,6 @@ int dlm_drop_lockres_ref(struct dlm_ctxt *dlm, struct dlm_lock_resource *res) namelen = res->lockname.len; BUG_ON(namelen > O2NM_MAX_NAME_LEN); - mlog(0, "%s:%.*s: sending deref to %d\n", - dlm->name, namelen, lockname, res->owner); memset(&deref, 0, sizeof(deref)); deref.node_idx = dlm->node_num; deref.namelen = namelen; @@ -2197,14 +2195,12 @@ int dlm_drop_lockres_ref(struct dlm_ctxt *dlm, struct dlm_lock_resource *res) ret = o2net_send_message(DLM_DEREF_LOCKRES_MSG, dlm->key, &deref, sizeof(deref), res->owner, &r); if (ret < 0) - mlog(ML_ERROR, "Error %d when sending message %u (key 0x%x) to " - "node %u\n", ret, DLM_DEREF_LOCKRES_MSG, dlm->key, - res->owner); + mlog(ML_ERROR, "%s: res %.*s, error %d send DEREF to node %u\n", + dlm->name, namelen, lockname, ret, res->owner); else if (r < 0) { /* BAD. other node says I did not have a ref. */ - mlog(ML_ERROR,"while dropping ref on %s:%.*s " - "(master=%u) got %d.\n", dlm->name, namelen, - lockname, res->owner, r); + mlog(ML_ERROR, "%s: res %.*s, DEREF to node %u got %d\n", + dlm->name, namelen, lockname, res->owner, r); dlm_print_one_lock_resource(res); BUG(); } @@ -2916,9 +2912,9 @@ static int dlm_do_migrate_request(struct dlm_ctxt *dlm, &migrate, sizeof(migrate), nodenum, &status); if (ret < 0) { - mlog(ML_ERROR, "Error %d when sending message %u (key " - "0x%x) to node %u\n", ret, DLM_MIGRATE_REQUEST_MSG, - dlm->key, nodenum); + mlog(ML_ERROR, "%s: res %.*s, Error %d send " + "MIGRATE_REQUEST to node %u\n", dlm->name, + migrate.namelen, migrate.name, ret, nodenum); if (!dlm_is_host_down(ret)) { mlog(ML_ERROR, "unhandled error=%d!\n", ret); BUG(); diff --git a/fs/ocfs2/dlm/dlmrecovery.c b/fs/ocfs2/dlm/dlmrecovery.c index 7efab6d28a2..a3c312c43b9 100644 --- a/fs/ocfs2/dlm/dlmrecovery.c +++ b/fs/ocfs2/dlm/dlmrecovery.c @@ -430,6 +430,8 @@ static void dlm_begin_recovery(struct dlm_ctxt *dlm) { spin_lock(&dlm->spinlock); BUG_ON(dlm->reco.state & DLM_RECO_STATE_ACTIVE); + printk(KERN_NOTICE "o2dlm: Begin recovery on domain %s for node %u\n", + dlm->name, dlm->reco.dead_node); dlm->reco.state |= DLM_RECO_STATE_ACTIVE; spin_unlock(&dlm->spinlock); } @@ -440,9 +442,18 @@ static void dlm_end_recovery(struct dlm_ctxt *dlm) BUG_ON(!(dlm->reco.state & DLM_RECO_STATE_ACTIVE)); dlm->reco.state &= ~DLM_RECO_STATE_ACTIVE; spin_unlock(&dlm->spinlock); + printk(KERN_NOTICE "o2dlm: End recovery on domain %s\n", dlm->name); wake_up(&dlm->reco.event); } +static void dlm_print_recovery_master(struct dlm_ctxt *dlm) +{ + printk(KERN_NOTICE "o2dlm: Node %u (%s) is the Recovery Master for the " + "dead node %u in domain %s\n", dlm->reco.new_master, + (dlm->node_num == dlm->reco.new_master ? "me" : "he"), + dlm->reco.dead_node, dlm->name); +} + static int dlm_do_recovery(struct dlm_ctxt *dlm) { int status = 0; @@ -505,9 +516,8 @@ static int dlm_do_recovery(struct dlm_ctxt *dlm) } mlog(0, "another node will master this recovery session.\n"); } - mlog(0, "dlm=%s (%d), new_master=%u, this node=%u, dead_node=%u\n", - dlm->name, task_pid_nr(dlm->dlm_reco_thread_task), dlm->reco.new_master, - dlm->node_num, dlm->reco.dead_node); + + dlm_print_recovery_master(dlm); /* it is safe to start everything back up here * because all of the dead node's lock resources @@ -518,15 +528,13 @@ static int dlm_do_recovery(struct dlm_ctxt *dlm) return 0; master_here: - mlog(ML_NOTICE, "(%d) Node %u is the Recovery Master for the Dead Node " - "%u for Domain %s\n", task_pid_nr(dlm->dlm_reco_thread_task), - dlm->node_num, dlm->reco.dead_node, dlm->name); + dlm_print_recovery_master(dlm); status = dlm_remaster_locks(dlm, dlm->reco.dead_node); if (status < 0) { /* we should never hit this anymore */ - mlog(ML_ERROR, "error %d remastering locks for node %u, " - "retrying.\n", status, dlm->reco.dead_node); + mlog(ML_ERROR, "%s: Error %d remastering locks for node %u, " + "retrying.\n", dlm->name, status, dlm->reco.dead_node); /* yield a bit to allow any final network messages * to get handled on remaining nodes */ msleep(100); @@ -567,7 +575,7 @@ static int dlm_remaster_locks(struct dlm_ctxt *dlm, u8 dead_node) BUG_ON(ndata->state != DLM_RECO_NODE_DATA_INIT); ndata->state = DLM_RECO_NODE_DATA_REQUESTING; - mlog(0, "requesting lock info from node %u\n", + mlog(0, "%s: Requesting lock info from node %u\n", dlm->name, ndata->node_num); if (ndata->node_num == dlm->node_num) { @@ -640,7 +648,7 @@ static int dlm_remaster_locks(struct dlm_ctxt *dlm, u8 dead_node) spin_unlock(&dlm_reco_state_lock); } - mlog(0, "done requesting all lock info\n"); + mlog(0, "%s: Done requesting all lock info\n", dlm->name); /* nodes should be sending reco data now * just need to wait */ @@ -802,10 +810,9 @@ static int dlm_request_all_locks(struct dlm_ctxt *dlm, u8 request_from, /* negative status is handled by caller */ if (ret < 0) - mlog(ML_ERROR, "Error %d when sending message %u (key " - "0x%x) to node %u\n", ret, DLM_LOCK_REQUEST_MSG, - dlm->key, request_from); - + mlog(ML_ERROR, "%s: Error %d send LOCK_REQUEST to node %u " + "to recover dead node %u\n", dlm->name, ret, + request_from, dead_node); // return from here, then // sleep until all received or error return ret; @@ -956,9 +963,9 @@ static int dlm_send_all_done_msg(struct dlm_ctxt *dlm, u8 dead_node, u8 send_to) ret = o2net_send_message(DLM_RECO_DATA_DONE_MSG, dlm->key, &done_msg, sizeof(done_msg), send_to, &tmpret); if (ret < 0) { - mlog(ML_ERROR, "Error %d when sending message %u (key " - "0x%x) to node %u\n", ret, DLM_RECO_DATA_DONE_MSG, - dlm->key, send_to); + mlog(ML_ERROR, "%s: Error %d send RECO_DATA_DONE to node %u " + "to recover dead node %u\n", dlm->name, ret, send_to, + dead_node); if (!dlm_is_host_down(ret)) { BUG(); } @@ -1127,9 +1134,11 @@ static int dlm_send_mig_lockres_msg(struct dlm_ctxt *dlm, if (ret < 0) { /* XXX: negative status is not handled. * this will end up killing this node. */ - mlog(ML_ERROR, "Error %d when sending message %u (key " - "0x%x) to node %u\n", ret, DLM_MIG_LOCKRES_MSG, - dlm->key, send_to); + mlog(ML_ERROR, "%s: res %.*s, Error %d send MIG_LOCKRES to " + "node %u (%s)\n", dlm->name, mres->lockname_len, + mres->lockname, ret, send_to, + (orig_flags & DLM_MRES_MIGRATION ? + "migration" : "recovery")); } else { /* might get an -ENOMEM back here */ ret = status; @@ -2324,9 +2333,9 @@ static void dlm_do_local_recovery_cleanup(struct dlm_ctxt *dlm, u8 dead_node) dlm_revalidate_lvb(dlm, res, dead_node); if (res->owner == dead_node) { if (res->state & DLM_LOCK_RES_DROPPING_REF) { - mlog(ML_NOTICE, "Ignore %.*s for " + mlog(ML_NOTICE, "%s: res %.*s, Skip " "recovery as it is being freed\n", - res->lockname.len, + dlm->name, res->lockname.len, res->lockname.name); } else dlm_move_lockres_to_recovery_list(dlm, diff --git a/fs/ocfs2/dlm/dlmthread.c b/fs/ocfs2/dlm/dlmthread.c index 1d6d1d22c47..46faac20f16 100644 --- a/fs/ocfs2/dlm/dlmthread.c +++ b/fs/ocfs2/dlm/dlmthread.c @@ -185,8 +185,6 @@ static void dlm_purge_lockres(struct dlm_ctxt *dlm, /* clear our bit from the master's refmap, ignore errors */ ret = dlm_drop_lockres_ref(dlm, res); if (ret < 0) { - mlog(ML_ERROR, "%s: deref %.*s failed %d\n", dlm->name, - res->lockname.len, res->lockname.name, ret); if (!dlm_is_host_down(ret)) BUG(); } -- cgit v1.2.3-70-g09d2 From 394eb3d38a3ecc549cc34a3040103a9164be516b Mon Sep 17 00:00:00 2001 From: Sunil Mushran Date: Sun, 24 Jul 2011 10:24:54 -0700 Subject: ocfs2: Clean up messages in stack_o2cb.c o2cb messages needed a facelift. Signed-off-by: Sunil Mushran --- fs/ocfs2/stack_o2cb.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'fs') diff --git a/fs/ocfs2/stack_o2cb.c b/fs/ocfs2/stack_o2cb.c index 19965b00c43..be935d89a4e 100644 --- a/fs/ocfs2/stack_o2cb.c +++ b/fs/ocfs2/stack_o2cb.c @@ -263,8 +263,8 @@ static void o2dlm_eviction_cb(int node_num, void *data) { struct ocfs2_cluster_connection *conn = data; - mlog(ML_NOTICE, "o2dlm has evicted node %d from group %.*s\n", - node_num, conn->cc_namelen, conn->cc_name); + printk(KERN_NOTICE "o2cb: o2dlm has evicted node %d from domain %.*s\n", + node_num, conn->cc_namelen, conn->cc_name); conn->cc_recovery_handler(node_num, conn->cc_recovery_data); } -- cgit v1.2.3-70-g09d2 From 0afbba13226fcdbd4327f6b13a42f6efbb8c9caf Mon Sep 17 00:00:00 2001 From: Sunil Mushran Date: Sun, 24 Jul 2011 10:25:54 -0700 Subject: ocfs2/dlm: Cleanup up dlm_finish_local_lockres_recovery() dlm_finish_local_lockres_recovery() needed a facelift. Signed-off-by: Sunil Mushran --- fs/ocfs2/dlm/dlmrecovery.c | 57 ++++++++++++++++++++-------------------------- 1 file changed, 25 insertions(+), 32 deletions(-) (limited to 'fs') diff --git a/fs/ocfs2/dlm/dlmrecovery.c b/fs/ocfs2/dlm/dlmrecovery.c index a3c312c43b9..41e74f0e2d3 100644 --- a/fs/ocfs2/dlm/dlmrecovery.c +++ b/fs/ocfs2/dlm/dlmrecovery.c @@ -2093,6 +2093,9 @@ static void dlm_finish_local_lockres_recovery(struct dlm_ctxt *dlm, list_for_each_entry_safe(res, next, &dlm->reco.resources, recovering) { if (res->owner == dead_node) { + mlog(0, "%s: res %.*s, Changing owner from %u to %u\n", + dlm->name, res->lockname.len, res->lockname.name, + res->owner, new_master); list_del_init(&res->recovering); spin_lock(&res->spinlock); /* new_master has our reference from @@ -2114,40 +2117,30 @@ static void dlm_finish_local_lockres_recovery(struct dlm_ctxt *dlm, for (i = 0; i < DLM_HASH_BUCKETS; i++) { bucket = dlm_lockres_hash(dlm, i); hlist_for_each_entry(res, hash_iter, bucket, hash_node) { - if (res->state & DLM_LOCK_RES_RECOVERING) { - if (res->owner == dead_node) { - mlog(0, "(this=%u) res %.*s owner=%u " - "was not on recovering list, but " - "clearing state anyway\n", - dlm->node_num, res->lockname.len, - res->lockname.name, new_master); - } else if (res->owner == dlm->node_num) { - mlog(0, "(this=%u) res %.*s owner=%u " - "was not on recovering list, " - "owner is THIS node, clearing\n", - dlm->node_num, res->lockname.len, - res->lockname.name, new_master); - } else - continue; + if (!(res->state & DLM_LOCK_RES_RECOVERING)) + continue; - if (!list_empty(&res->recovering)) { - mlog(0, "%s:%.*s: lockres was " - "marked RECOVERING, owner=%u\n", - dlm->name, res->lockname.len, - res->lockname.name, res->owner); - list_del_init(&res->recovering); - dlm_lockres_put(res); - } - spin_lock(&res->spinlock); - /* new_master has our reference from - * the lock state sent during recovery */ - dlm_change_lockres_owner(dlm, res, new_master); - res->state &= ~DLM_LOCK_RES_RECOVERING; - if (__dlm_lockres_has_locks(res)) - __dlm_dirty_lockres(dlm, res); - spin_unlock(&res->spinlock); - wake_up(&res->wq); + if (res->owner != dead_node && + res->owner != dlm->node_num) + continue; + + if (!list_empty(&res->recovering)) { + list_del_init(&res->recovering); + dlm_lockres_put(res); } + + /* new_master has our reference from + * the lock state sent during recovery */ + mlog(0, "%s: res %.*s, Changing owner from %u to %u\n", + dlm->name, res->lockname.len, res->lockname.name, + res->owner, new_master); + spin_lock(&res->spinlock); + dlm_change_lockres_owner(dlm, res, new_master); + res->state &= ~DLM_LOCK_RES_RECOVERING; + if (__dlm_lockres_has_locks(res)) + __dlm_dirty_lockres(dlm, res); + spin_unlock(&res->spinlock); + wake_up(&res->wq); } } } -- cgit v1.2.3-70-g09d2 From 8d400b81cc83b171ff872587723a37eb7fae9abe Mon Sep 17 00:00:00 2001 From: Sunil Mushran Date: Sun, 24 Jul 2011 10:26:54 -0700 Subject: ocfs2/dlm: Clean up refmap helpers Patch cleans up helpers that set/clear refmap bits and grab/drop inflight lock ref counts. Signed-off-by: Sunil Mushran --- fs/ocfs2/dlm/dlmcommon.h | 49 +++++--------------------- fs/ocfs2/dlm/dlmmaster.c | 88 ++++++++++++++++++++++++++++------------------ fs/ocfs2/dlm/dlmrecovery.c | 8 ++--- 3 files changed, 66 insertions(+), 79 deletions(-) (limited to 'fs') diff --git a/fs/ocfs2/dlm/dlmcommon.h b/fs/ocfs2/dlm/dlmcommon.h index d602abb51b6..4a5d5804b13 100644 --- a/fs/ocfs2/dlm/dlmcommon.h +++ b/fs/ocfs2/dlm/dlmcommon.h @@ -902,46 +902,15 @@ struct dlm_lock_resource *dlm_new_lockres(struct dlm_ctxt *dlm, const char *name, unsigned int namelen); -#define dlm_lockres_set_refmap_bit(bit,res) \ - __dlm_lockres_set_refmap_bit(bit,res,__FILE__,__LINE__) -#define dlm_lockres_clear_refmap_bit(bit,res) \ - __dlm_lockres_clear_refmap_bit(bit,res,__FILE__,__LINE__) - -static inline void __dlm_lockres_set_refmap_bit(int bit, - struct dlm_lock_resource *res, - const char *file, - int line) -{ - //printk("%s:%d:%.*s: setting bit %d\n", file, line, - // res->lockname.len, res->lockname.name, bit); - set_bit(bit, res->refmap); -} - -static inline void __dlm_lockres_clear_refmap_bit(int bit, - struct dlm_lock_resource *res, - const char *file, - int line) -{ - //printk("%s:%d:%.*s: clearing bit %d\n", file, line, - // res->lockname.len, res->lockname.name, bit); - clear_bit(bit, res->refmap); -} - -void __dlm_lockres_drop_inflight_ref(struct dlm_ctxt *dlm, - struct dlm_lock_resource *res, - const char *file, - int line); -void __dlm_lockres_grab_inflight_ref(struct dlm_ctxt *dlm, - struct dlm_lock_resource *res, - int new_lockres, - const char *file, - int line); -#define dlm_lockres_drop_inflight_ref(d,r) \ - __dlm_lockres_drop_inflight_ref(d,r,__FILE__,__LINE__) -#define dlm_lockres_grab_inflight_ref(d,r) \ - __dlm_lockres_grab_inflight_ref(d,r,0,__FILE__,__LINE__) -#define dlm_lockres_grab_inflight_ref_new(d,r) \ - __dlm_lockres_grab_inflight_ref(d,r,1,__FILE__,__LINE__) +void dlm_lockres_set_refmap_bit(struct dlm_ctxt *dlm, + struct dlm_lock_resource *res, int bit); +void dlm_lockres_clear_refmap_bit(struct dlm_ctxt *dlm, + struct dlm_lock_resource *res, int bit); + +void dlm_lockres_drop_inflight_ref(struct dlm_ctxt *dlm, + struct dlm_lock_resource *res); +void dlm_lockres_grab_inflight_ref(struct dlm_ctxt *dlm, + struct dlm_lock_resource *res); void dlm_queue_ast(struct dlm_ctxt *dlm, struct dlm_lock *lock); void dlm_queue_bast(struct dlm_ctxt *dlm, struct dlm_lock *lock); diff --git a/fs/ocfs2/dlm/dlmmaster.c b/fs/ocfs2/dlm/dlmmaster.c index 9f3b093b629..11e446ffb67 100644 --- a/fs/ocfs2/dlm/dlmmaster.c +++ b/fs/ocfs2/dlm/dlmmaster.c @@ -631,39 +631,58 @@ error: return NULL; } -void __dlm_lockres_grab_inflight_ref(struct dlm_ctxt *dlm, - struct dlm_lock_resource *res, - int new_lockres, - const char *file, - int line) +void dlm_lockres_set_refmap_bit(struct dlm_ctxt *dlm, + struct dlm_lock_resource *res, int bit) { - if (!new_lockres) - assert_spin_locked(&res->spinlock); + assert_spin_locked(&res->spinlock); + + mlog(0, "res %.*s, set node %u, %ps()\n", res->lockname.len, + res->lockname.name, bit, __builtin_return_address(0)); + + set_bit(bit, res->refmap); +} + +void dlm_lockres_clear_refmap_bit(struct dlm_ctxt *dlm, + struct dlm_lock_resource *res, int bit) +{ + assert_spin_locked(&res->spinlock); + + mlog(0, "res %.*s, clr node %u, %ps()\n", res->lockname.len, + res->lockname.name, bit, __builtin_return_address(0)); + + clear_bit(bit, res->refmap); +} + + +void dlm_lockres_grab_inflight_ref(struct dlm_ctxt *dlm, + struct dlm_lock_resource *res) +{ + assert_spin_locked(&res->spinlock); if (!test_bit(dlm->node_num, res->refmap)) { BUG_ON(res->inflight_locks != 0); - dlm_lockres_set_refmap_bit(dlm->node_num, res); + dlm_lockres_set_refmap_bit(dlm, res, dlm->node_num); } res->inflight_locks++; - mlog(0, "%s:%.*s: inflight++: now %u\n", - dlm->name, res->lockname.len, res->lockname.name, - res->inflight_locks); + mlog(0, "%s: res %.*s, inflight++: now %u, %ps()\n", dlm->name, + res->lockname.len, res->lockname.name, res->inflight_locks, + __builtin_return_address(0)); } -void __dlm_lockres_drop_inflight_ref(struct dlm_ctxt *dlm, - struct dlm_lock_resource *res, - const char *file, - int line) +void dlm_lockres_drop_inflight_ref(struct dlm_ctxt *dlm, + struct dlm_lock_resource *res) { assert_spin_locked(&res->spinlock); BUG_ON(res->inflight_locks == 0); + res->inflight_locks--; - mlog(0, "%s:%.*s: inflight--: now %u\n", - dlm->name, res->lockname.len, res->lockname.name, - res->inflight_locks); + mlog(0, "%s: res %.*s, inflight--: now %u, %ps()\n", dlm->name, + res->lockname.len, res->lockname.name, res->inflight_locks, + __builtin_return_address(0)); + if (res->inflight_locks == 0) - dlm_lockres_clear_refmap_bit(dlm->node_num, res); + dlm_lockres_clear_refmap_bit(dlm, res, dlm->node_num); wake_up(&res->wq); } @@ -843,8 +862,10 @@ lookup: /* finally add the lockres to its hash bucket */ __dlm_insert_lockres(dlm, res); - /* since this lockres is new it doesn't not require the spinlock */ - dlm_lockres_grab_inflight_ref_new(dlm, res); + + spin_lock(&res->spinlock); + dlm_lockres_grab_inflight_ref(dlm, res); + spin_unlock(&res->spinlock); /* if this node does not become the master make sure to drop * this inflight reference below */ @@ -1426,9 +1447,7 @@ way_up_top: } if (res->owner == dlm->node_num) { - mlog(0, "%s:%.*s: setting bit %u in refmap\n", - dlm->name, namelen, name, request->node_idx); - dlm_lockres_set_refmap_bit(request->node_idx, res); + dlm_lockres_set_refmap_bit(dlm, res, request->node_idx); spin_unlock(&res->spinlock); response = DLM_MASTER_RESP_YES; if (mle) @@ -1493,10 +1512,8 @@ way_up_top: * go back and clean the mles on any * other nodes */ dispatch_assert = 1; - dlm_lockres_set_refmap_bit(request->node_idx, res); - mlog(0, "%s:%.*s: setting bit %u in refmap\n", - dlm->name, namelen, name, - request->node_idx); + dlm_lockres_set_refmap_bit(dlm, res, + request->node_idx); } else response = DLM_MASTER_RESP_NO; } else { @@ -1702,7 +1719,7 @@ again: "lockres, set the bit in the refmap\n", namelen, lockname, to); spin_lock(&res->spinlock); - dlm_lockres_set_refmap_bit(to, res); + dlm_lockres_set_refmap_bit(dlm, res, to); spin_unlock(&res->spinlock); } } @@ -2256,7 +2273,7 @@ int dlm_deref_lockres_handler(struct o2net_msg *msg, u32 len, void *data, else { BUG_ON(res->state & DLM_LOCK_RES_DROPPING_REF); if (test_bit(node, res->refmap)) { - dlm_lockres_clear_refmap_bit(node, res); + dlm_lockres_clear_refmap_bit(dlm, res, node); cleared = 1; } } @@ -2316,7 +2333,7 @@ static void dlm_deref_lockres_worker(struct dlm_work_item *item, void *data) BUG_ON(res->state & DLM_LOCK_RES_DROPPING_REF); if (test_bit(node, res->refmap)) { __dlm_wait_on_lockres_flags(res, DLM_LOCK_RES_SETREF_INPROG); - dlm_lockres_clear_refmap_bit(node, res); + dlm_lockres_clear_refmap_bit(dlm, res, node); cleared = 1; } spin_unlock(&res->spinlock); @@ -2798,7 +2815,8 @@ static void dlm_remove_nonlocal_locks(struct dlm_ctxt *dlm, BUG_ON(!list_empty(&lock->bast_list)); BUG_ON(lock->ast_pending); BUG_ON(lock->bast_pending); - dlm_lockres_clear_refmap_bit(lock->ml.node, res); + dlm_lockres_clear_refmap_bit(dlm, res, + lock->ml.node); list_del_init(&lock->list); dlm_lock_put(lock); /* In a normal unlock, we would have added a @@ -2819,7 +2837,7 @@ static void dlm_remove_nonlocal_locks(struct dlm_ctxt *dlm, mlog(0, "%s:%.*s: node %u had a ref to this " "migrating lockres, clearing\n", dlm->name, res->lockname.len, res->lockname.name, bit); - dlm_lockres_clear_refmap_bit(bit, res); + dlm_lockres_clear_refmap_bit(dlm, res, bit); } bit++; } @@ -2933,7 +2951,7 @@ static int dlm_do_migrate_request(struct dlm_ctxt *dlm, dlm->name, res->lockname.len, res->lockname.name, nodenum); spin_lock(&res->spinlock); - dlm_lockres_set_refmap_bit(nodenum, res); + dlm_lockres_set_refmap_bit(dlm, res, nodenum); spin_unlock(&res->spinlock); } } @@ -3267,7 +3285,7 @@ int dlm_finish_migration(struct dlm_ctxt *dlm, struct dlm_lock_resource *res, * mastery reference here since old_master will briefly have * a reference after the migration completes */ spin_lock(&res->spinlock); - dlm_lockres_set_refmap_bit(old_master, res); + dlm_lockres_set_refmap_bit(dlm, res, old_master); spin_unlock(&res->spinlock); mlog(0, "now time to do a migrate request to other nodes\n"); diff --git a/fs/ocfs2/dlm/dlmrecovery.c b/fs/ocfs2/dlm/dlmrecovery.c index 41e74f0e2d3..f1a3aa84ad8 100644 --- a/fs/ocfs2/dlm/dlmrecovery.c +++ b/fs/ocfs2/dlm/dlmrecovery.c @@ -1776,7 +1776,7 @@ static int dlm_process_recovery_data(struct dlm_ctxt *dlm, dlm->name, mres->lockname_len, mres->lockname, from); spin_lock(&res->spinlock); - dlm_lockres_set_refmap_bit(from, res); + dlm_lockres_set_refmap_bit(dlm, res, from); spin_unlock(&res->spinlock); added++; break; @@ -1974,7 +1974,7 @@ skip_lvb: mlog(0, "%s:%.*s: added lock for node %u, " "setting refmap bit\n", dlm->name, res->lockname.len, res->lockname.name, ml->node); - dlm_lockres_set_refmap_bit(ml->node, res); + dlm_lockres_set_refmap_bit(dlm, res, ml->node); added++; } spin_unlock(&res->spinlock); @@ -2254,12 +2254,12 @@ static void dlm_free_dead_locks(struct dlm_ctxt *dlm, res->lockname.len, res->lockname.name, freed, dead_node); __dlm_print_one_lock_resource(res); } - dlm_lockres_clear_refmap_bit(dead_node, res); + dlm_lockres_clear_refmap_bit(dlm, res, dead_node); } else if (test_bit(dead_node, res->refmap)) { mlog(0, "%s:%.*s: dead node %u had a ref, but had " "no locks and had not purged before dying\n", dlm->name, res->lockname.len, res->lockname.name, dead_node); - dlm_lockres_clear_refmap_bit(dead_node, res); + dlm_lockres_clear_refmap_bit(dlm, res, dead_node); } /* do not kick thread yet */ -- cgit v1.2.3-70-g09d2 From e9f0b6a6233105c188064328c3f15e21942d0241 Mon Sep 17 00:00:00 2001 From: Sunil Mushran Date: Sun, 24 Jul 2011 10:27:54 -0700 Subject: ocfs2/dlm: Trace insert/remove of resource to/from hash Add mlog to trace adding and removing the resource from/to the hash table. Signed-off-by: Sunil Mushran --- fs/ocfs2/dlm/dlmcommon.h | 5 ++--- fs/ocfs2/dlm/dlmdomain.c | 19 ++++++++++++------- fs/ocfs2/dlm/dlmthread.c | 2 +- 3 files changed, 15 insertions(+), 11 deletions(-) (limited to 'fs') diff --git a/fs/ocfs2/dlm/dlmcommon.h b/fs/ocfs2/dlm/dlmcommon.h index 4a5d5804b13..b1748ac7475 100644 --- a/fs/ocfs2/dlm/dlmcommon.h +++ b/fs/ocfs2/dlm/dlmcommon.h @@ -877,9 +877,8 @@ static inline void dlm_lockres_get(struct dlm_lock_resource *res) kref_get(&res->refs); } void dlm_lockres_put(struct dlm_lock_resource *res); -void __dlm_unhash_lockres(struct dlm_lock_resource *res); -void __dlm_insert_lockres(struct dlm_ctxt *dlm, - struct dlm_lock_resource *res); +void __dlm_unhash_lockres(struct dlm_ctxt *dlm, struct dlm_lock_resource *res); +void __dlm_insert_lockres(struct dlm_ctxt *dlm, struct dlm_lock_resource *res); struct dlm_lock_resource * __dlm_lookup_lockres_full(struct dlm_ctxt *dlm, const char *name, unsigned int len, diff --git a/fs/ocfs2/dlm/dlmdomain.c b/fs/ocfs2/dlm/dlmdomain.c index ce225fb0cea..200d9888ffe 100644 --- a/fs/ocfs2/dlm/dlmdomain.c +++ b/fs/ocfs2/dlm/dlmdomain.c @@ -157,16 +157,18 @@ static int dlm_protocol_compare(struct dlm_protocol_version *existing, static void dlm_unregister_domain_handlers(struct dlm_ctxt *dlm); -void __dlm_unhash_lockres(struct dlm_lock_resource *lockres) +void __dlm_unhash_lockres(struct dlm_ctxt *dlm, struct dlm_lock_resource *res) { - if (!hlist_unhashed(&lockres->hash_node)) { - hlist_del_init(&lockres->hash_node); - dlm_lockres_put(lockres); - } + if (hlist_unhashed(&res->hash_node)) + return; + + mlog(0, "%s: Unhash res %.*s\n", dlm->name, res->lockname.len, + res->lockname.name); + hlist_del_init(&res->hash_node); + dlm_lockres_put(res); } -void __dlm_insert_lockres(struct dlm_ctxt *dlm, - struct dlm_lock_resource *res) +void __dlm_insert_lockres(struct dlm_ctxt *dlm, struct dlm_lock_resource *res) { struct hlist_head *bucket; struct qstr *q; @@ -180,6 +182,9 @@ void __dlm_insert_lockres(struct dlm_ctxt *dlm, dlm_lockres_get(res); hlist_add_head(&res->hash_node, bucket); + + mlog(0, "%s: Hash res %.*s\n", dlm->name, res->lockname.len, + res->lockname.name); } struct dlm_lock_resource * __dlm_lookup_lockres_full(struct dlm_ctxt *dlm, diff --git a/fs/ocfs2/dlm/dlmthread.c b/fs/ocfs2/dlm/dlmthread.c index 46faac20f16..4eff65e7e59 100644 --- a/fs/ocfs2/dlm/dlmthread.c +++ b/fs/ocfs2/dlm/dlmthread.c @@ -207,7 +207,7 @@ static void dlm_purge_lockres(struct dlm_ctxt *dlm, BUG(); } - __dlm_unhash_lockres(res); + __dlm_unhash_lockres(dlm, res); /* lockres is not in the hash now. drop the flag and wake up * any processes waiting in dlm_get_lock_resource. */ -- cgit v1.2.3-70-g09d2 From ed8625c6fb93d750ed022db571a8a7b7a6724b3b Mon Sep 17 00:00:00 2001 From: Sunil Mushran Date: Sun, 24 Jul 2011 10:28:54 -0700 Subject: ocfs2/dlm: Cleanup dlm_wait_for_node_death() and dlm_wait_for_node_recovery() dlm_wait_for_node_death() and dlm_wait_for_node_recovery() needed a facelift. Signed-off-by: Sunil Mushran --- fs/ocfs2/dlm/dlmcommon.h | 4 ++-- fs/ocfs2/dlm/dlmrecovery.c | 46 ++++++++++++++++++++++------------------------ 2 files changed, 24 insertions(+), 26 deletions(-) (limited to 'fs') diff --git a/fs/ocfs2/dlm/dlmcommon.h b/fs/ocfs2/dlm/dlmcommon.h index b1748ac7475..a5952ceecba 100644 --- a/fs/ocfs2/dlm/dlmcommon.h +++ b/fs/ocfs2/dlm/dlmcommon.h @@ -859,8 +859,8 @@ void dlm_complete_recovery_thread(struct dlm_ctxt *dlm); void dlm_wait_for_recovery(struct dlm_ctxt *dlm); void dlm_kick_recovery_thread(struct dlm_ctxt *dlm); int dlm_is_node_dead(struct dlm_ctxt *dlm, u8 node); -int dlm_wait_for_node_death(struct dlm_ctxt *dlm, u8 node, int timeout); -int dlm_wait_for_node_recovery(struct dlm_ctxt *dlm, u8 node, int timeout); +void dlm_wait_for_node_death(struct dlm_ctxt *dlm, u8 node, int timeout); +void dlm_wait_for_node_recovery(struct dlm_ctxt *dlm, u8 node, int timeout); void dlm_put(struct dlm_ctxt *dlm); struct dlm_ctxt *dlm_grab(struct dlm_ctxt *dlm); diff --git a/fs/ocfs2/dlm/dlmrecovery.c b/fs/ocfs2/dlm/dlmrecovery.c index f1a3aa84ad8..01ebfd0bdad 100644 --- a/fs/ocfs2/dlm/dlmrecovery.c +++ b/fs/ocfs2/dlm/dlmrecovery.c @@ -362,40 +362,38 @@ static int dlm_is_node_recovered(struct dlm_ctxt *dlm, u8 node) } -int dlm_wait_for_node_death(struct dlm_ctxt *dlm, u8 node, int timeout) +void dlm_wait_for_node_death(struct dlm_ctxt *dlm, u8 node, int timeout) { - if (timeout) { - mlog(ML_NOTICE, "%s: waiting %dms for notification of " - "death of node %u\n", dlm->name, timeout, node); + if (dlm_is_node_dead(dlm, node)) + return; + + printk(KERN_NOTICE "o2dlm: Waiting on the death of node %u in " + "domain %s\n", node, dlm->name); + + if (timeout) wait_event_timeout(dlm->dlm_reco_thread_wq, - dlm_is_node_dead(dlm, node), - msecs_to_jiffies(timeout)); - } else { - mlog(ML_NOTICE, "%s: waiting indefinitely for notification " - "of death of node %u\n", dlm->name, node); + dlm_is_node_dead(dlm, node), + msecs_to_jiffies(timeout)); + else wait_event(dlm->dlm_reco_thread_wq, dlm_is_node_dead(dlm, node)); - } - /* for now, return 0 */ - return 0; } -int dlm_wait_for_node_recovery(struct dlm_ctxt *dlm, u8 node, int timeout) +void dlm_wait_for_node_recovery(struct dlm_ctxt *dlm, u8 node, int timeout) { - if (timeout) { - mlog(0, "%s: waiting %dms for notification of " - "recovery of node %u\n", dlm->name, timeout, node); + if (dlm_is_node_recovered(dlm, node)) + return; + + printk(KERN_NOTICE "o2dlm: Waiting on the recovery of node %u in " + "domain %s\n", node, dlm->name); + + if (timeout) wait_event_timeout(dlm->dlm_reco_thread_wq, - dlm_is_node_recovered(dlm, node), - msecs_to_jiffies(timeout)); - } else { - mlog(0, "%s: waiting indefinitely for notification " - "of recovery of node %u\n", dlm->name, node); + dlm_is_node_recovered(dlm, node), + msecs_to_jiffies(timeout)); + else wait_event(dlm->dlm_reco_thread_wq, dlm_is_node_recovered(dlm, node)); - } - /* for now, return 0 */ - return 0; } /* callers of the top-level api calls (dlmlock/dlmunlock) should -- cgit v1.2.3-70-g09d2 From ff0a522e7db79625aa27a433467eb94c5e255718 Mon Sep 17 00:00:00 2001 From: Sunil Mushran Date: Sun, 24 Jul 2011 10:29:54 -0700 Subject: ocfs2/dlm: Take inflight reference count for remotely mastered resources too The inflight reference count, in the lock resource, is taken to pin the resource in memory. We take it when a new resource is created and release it after a lock is attached to it. We do this to prevent the resource from getting purged prematurely. Earlier this reference count was being taken for locally mastered resources only. This patch extends the same functionality for remotely mastered ones. We are doing this because the same premature purging could occur for remotely mastered resources if the remote node were to die before completion of the create lock. Fix for Oracle bug#12405575. Signed-off-by: Sunil Mushran --- fs/ocfs2/dlm/dlmlock.c | 12 ++++++++---- fs/ocfs2/dlm/dlmmaster.c | 47 +++++++++++++++++------------------------------ fs/ocfs2/dlm/dlmthread.c | 12 +++++++----- 3 files changed, 32 insertions(+), 39 deletions(-) (limited to 'fs') diff --git a/fs/ocfs2/dlm/dlmlock.c b/fs/ocfs2/dlm/dlmlock.c index c7f3e22bda1..3ef2c1adfb8 100644 --- a/fs/ocfs2/dlm/dlmlock.c +++ b/fs/ocfs2/dlm/dlmlock.c @@ -183,10 +183,6 @@ static enum dlm_status dlmlock_master(struct dlm_ctxt *dlm, kick_thread = 1; } } - /* reduce the inflight count, this may result in the lockres - * being purged below during calc_usage */ - if (lock->ml.node == dlm->node_num) - dlm_lockres_drop_inflight_ref(dlm, res); spin_unlock(&res->spinlock); wake_up(&res->wq); @@ -737,6 +733,14 @@ retry_lock: } } + /* Inflight taken in dlm_get_lock_resource() is dropped here */ + spin_lock(&res->spinlock); + dlm_lockres_drop_inflight_ref(dlm, res); + spin_unlock(&res->spinlock); + + dlm_lockres_calc_usage(dlm, res); + dlm_kick_thread(dlm, res); + if (status != DLM_NORMAL) { lock->lksb->flags &= ~DLM_LKSB_GET_LVB; if (status != DLM_NOTQUEUED) diff --git a/fs/ocfs2/dlm/dlmmaster.c b/fs/ocfs2/dlm/dlmmaster.c index 11e446ffb67..005261c333b 100644 --- a/fs/ocfs2/dlm/dlmmaster.c +++ b/fs/ocfs2/dlm/dlmmaster.c @@ -659,11 +659,8 @@ void dlm_lockres_grab_inflight_ref(struct dlm_ctxt *dlm, { assert_spin_locked(&res->spinlock); - if (!test_bit(dlm->node_num, res->refmap)) { - BUG_ON(res->inflight_locks != 0); - dlm_lockres_set_refmap_bit(dlm, res, dlm->node_num); - } res->inflight_locks++; + mlog(0, "%s: res %.*s, inflight++: now %u, %ps()\n", dlm->name, res->lockname.len, res->lockname.name, res->inflight_locks, __builtin_return_address(0)); @@ -677,12 +674,11 @@ void dlm_lockres_drop_inflight_ref(struct dlm_ctxt *dlm, BUG_ON(res->inflight_locks == 0); res->inflight_locks--; + mlog(0, "%s: res %.*s, inflight--: now %u, %ps()\n", dlm->name, res->lockname.len, res->lockname.name, res->inflight_locks, __builtin_return_address(0)); - if (res->inflight_locks == 0) - dlm_lockres_clear_refmap_bit(dlm, res, dlm->node_num); wake_up(&res->wq); } @@ -716,7 +712,6 @@ struct dlm_lock_resource * dlm_get_lock_resource(struct dlm_ctxt *dlm, unsigned int hash; int tries = 0; int bit, wait_on_recovery = 0; - int drop_inflight_if_nonlocal = 0; BUG_ON(!lockid); @@ -728,36 +723,33 @@ lookup: spin_lock(&dlm->spinlock); tmpres = __dlm_lookup_lockres_full(dlm, lockid, namelen, hash); if (tmpres) { - int dropping_ref = 0; - spin_unlock(&dlm->spinlock); - spin_lock(&tmpres->spinlock); - /* We wait for the other thread that is mastering the resource */ + /* Wait on the thread that is mastering the resource */ if (tmpres->owner == DLM_LOCK_RES_OWNER_UNKNOWN) { __dlm_wait_on_lockres(tmpres); BUG_ON(tmpres->owner == DLM_LOCK_RES_OWNER_UNKNOWN); + spin_unlock(&tmpres->spinlock); + dlm_lockres_put(tmpres); + tmpres = NULL; + goto lookup; } - if (tmpres->owner == dlm->node_num) { - BUG_ON(tmpres->state & DLM_LOCK_RES_DROPPING_REF); - dlm_lockres_grab_inflight_ref(dlm, tmpres); - } else if (tmpres->state & DLM_LOCK_RES_DROPPING_REF) - dropping_ref = 1; - spin_unlock(&tmpres->spinlock); - - /* wait until done messaging the master, drop our ref to allow - * the lockres to be purged, start over. */ - if (dropping_ref) { - spin_lock(&tmpres->spinlock); - __dlm_wait_on_lockres_flags(tmpres, DLM_LOCK_RES_DROPPING_REF); + /* Wait on the resource purge to complete before continuing */ + if (tmpres->state & DLM_LOCK_RES_DROPPING_REF) { + BUG_ON(tmpres->owner == dlm->node_num); + __dlm_wait_on_lockres_flags(tmpres, + DLM_LOCK_RES_DROPPING_REF); spin_unlock(&tmpres->spinlock); dlm_lockres_put(tmpres); tmpres = NULL; goto lookup; } - mlog(0, "found in hash!\n"); + /* Grab inflight ref to pin the resource */ + dlm_lockres_grab_inflight_ref(dlm, tmpres); + + spin_unlock(&tmpres->spinlock); if (res) dlm_lockres_put(res); res = tmpres; @@ -863,14 +855,11 @@ lookup: /* finally add the lockres to its hash bucket */ __dlm_insert_lockres(dlm, res); + /* Grab inflight ref to pin the resource */ spin_lock(&res->spinlock); dlm_lockres_grab_inflight_ref(dlm, res); spin_unlock(&res->spinlock); - /* if this node does not become the master make sure to drop - * this inflight reference below */ - drop_inflight_if_nonlocal = 1; - /* get an extra ref on the mle in case this is a BLOCK * if so, the creator of the BLOCK may try to put the last * ref at this time in the assert master handler, so we @@ -973,8 +962,6 @@ wait: wake_waiters: spin_lock(&res->spinlock); - if (res->owner != dlm->node_num && drop_inflight_if_nonlocal) - dlm_lockres_drop_inflight_ref(dlm, res); res->state &= ~DLM_LOCK_RES_IN_PROGRESS; spin_unlock(&res->spinlock); wake_up(&res->wq); diff --git a/fs/ocfs2/dlm/dlmthread.c b/fs/ocfs2/dlm/dlmthread.c index 4eff65e7e59..e73c833fc2a 100644 --- a/fs/ocfs2/dlm/dlmthread.c +++ b/fs/ocfs2/dlm/dlmthread.c @@ -94,24 +94,26 @@ int __dlm_lockres_unused(struct dlm_lock_resource *res) { int bit; + assert_spin_locked(&res->spinlock); + if (__dlm_lockres_has_locks(res)) return 0; + /* Locks are in the process of being created */ + if (res->inflight_locks) + return 0; + if (!list_empty(&res->dirty) || res->state & DLM_LOCK_RES_DIRTY) return 0; if (res->state & DLM_LOCK_RES_RECOVERING) return 0; + /* Another node has this resource with this node as the master */ bit = find_next_bit(res->refmap, O2NM_MAX_NODES, 0); if (bit < O2NM_MAX_NODES) return 0; - /* - * since the bit for dlm->node_num is not set, inflight_locks better - * be zero - */ - BUG_ON(res->inflight_locks != 0); return 1; } -- cgit v1.2.3-70-g09d2 From a2c0cc1579176bd0808ef7deea456767dfa80217 Mon Sep 17 00:00:00 2001 From: Sunil Mushran Date: Sun, 24 Jul 2011 10:30:54 -0700 Subject: ocfs2/dlm: dlmlock_remote() needs to account for remastery In dlmlock_remote(), we wait for the resource to stop being active before setting the inprogress flag. Active includes recovery, migration, etc. The problem here is that if the resource was being recovered or migrated, the new owner could very well be that node itself (and thus not a remote node). This problem was observed in Oracle bug#12583620. The error messages observed were as follows: dlm_send_remote_lock_request:337 ERROR: Error -40 (ELOOP) when sending message 503 (key 0xd6d8c7) to node 2 dlmlock_remote:271 ERROR: dlm status = DLM_BADARGS dlmlock:751 ERROR: dlm status = DLM_BADARGS Signed-off-by: Sunil Mushran --- fs/ocfs2/dlm/dlmlock.c | 18 ++++++++---------- 1 file changed, 8 insertions(+), 10 deletions(-) (limited to 'fs') diff --git a/fs/ocfs2/dlm/dlmlock.c b/fs/ocfs2/dlm/dlmlock.c index 3ef2c1adfb8..f32fcba0492 100644 --- a/fs/ocfs2/dlm/dlmlock.c +++ b/fs/ocfs2/dlm/dlmlock.c @@ -227,10 +227,16 @@ static enum dlm_status dlmlock_remote(struct dlm_ctxt *dlm, lock->ml.type, res->lockname.len, res->lockname.name, flags); + /* + * Wait if resource is getting recovered, remastered, etc. + * If the resource was remastered and new owner is self, then exit. + */ spin_lock(&res->spinlock); - - /* will exit this call with spinlock held */ __dlm_wait_on_lockres(res); + if (res->owner == dlm->node_num) { + spin_unlock(&res->spinlock); + return DLM_RECOVERING; + } res->state |= DLM_LOCK_RES_IN_PROGRESS; /* add lock to local (secondary) queue */ @@ -710,18 +716,10 @@ retry_lock: if (status == DLM_RECOVERING || status == DLM_MIGRATING || status == DLM_FORWARD) { - mlog(0, "retrying lock with migration/" - "recovery/in progress\n"); msleep(100); - /* no waiting for dlm_reco_thread */ if (recovery) { if (status != DLM_RECOVERING) goto retry_lock; - - mlog(0, "%s: got RECOVERING " - "for $RECOVERY lock, master " - "was %u\n", dlm->name, - res->owner); /* wait to see the node go down, then * drop down and allow the lockres to * get cleaned up. need to remaster. */ -- cgit v1.2.3-70-g09d2 From bb570a5d9e74f71d32751823052db4a97d6a5e7c Mon Sep 17 00:00:00 2001 From: Sunil Mushran Date: Sun, 24 Jul 2011 10:31:54 -0700 Subject: ocfs2/cluster: Fix output in file elapsed_time_in_ms The o2hb debugfs file, elapsed_time_in_ms, should return values only after the timer is armed atleast once. Signed-off-by: Sunil Mushran --- fs/ocfs2/cluster/heartbeat.c | 9 ++++++--- 1 file changed, 6 insertions(+), 3 deletions(-) (limited to 'fs') diff --git a/fs/ocfs2/cluster/heartbeat.c b/fs/ocfs2/cluster/heartbeat.c index 4728a746281..a4e855e3690 100644 --- a/fs/ocfs2/cluster/heartbeat.c +++ b/fs/ocfs2/cluster/heartbeat.c @@ -1198,6 +1198,7 @@ static int o2hb_debug_open(struct inode *inode, struct file *file) struct o2hb_debug_buf *db = inode->i_private; struct o2hb_region *reg; unsigned long map[BITS_TO_LONGS(O2NM_MAX_NODES)]; + unsigned long lts; char *buf = NULL; int i = -1; int out = 0; @@ -1234,9 +1235,11 @@ static int o2hb_debug_open(struct inode *inode, struct file *file) case O2HB_DB_TYPE_REGION_ELAPSED_TIME: reg = (struct o2hb_region *)db->db_data; - out += snprintf(buf + out, PAGE_SIZE - out, "%u\n", - jiffies_to_msecs(jiffies - - reg->hr_last_timeout_start)); + lts = reg->hr_last_timeout_start; + /* If 0, it has never been set before */ + if (lts) + lts = jiffies_to_msecs(jiffies - lts); + out += snprintf(buf + out, PAGE_SIZE - out, "%lu\n", lts); goto done; case O2HB_DB_TYPE_REGION_PINNED: -- cgit v1.2.3-70-g09d2 From 3ba169ccec1c5ad0f678e04fd29b990197fdfe79 Mon Sep 17 00:00:00 2001 From: Sunil Mushran Date: Sun, 24 Jul 2011 10:32:54 -0700 Subject: ocfs2/cluster: Add new function o2net_fill_node_map() Patch adds function o2net_fill_node_map() to return the bitmap of nodes that it is connected to. This bitmap is also accessible by the user via the debugfs file, /sys/kernel/debug/o2net/connected_nodes. Signed-off-by: Sunil Mushran --- fs/ocfs2/cluster/netdebug.c | 102 ++++++++++++++++++++++++++++++-------------- fs/ocfs2/cluster/tcp.c | 19 +++++++++ fs/ocfs2/cluster/tcp.h | 2 + 3 files changed, 90 insertions(+), 33 deletions(-) (limited to 'fs') diff --git a/fs/ocfs2/cluster/netdebug.c b/fs/ocfs2/cluster/netdebug.c index 3a5835904b3..dc45deb19e6 100644 --- a/fs/ocfs2/cluster/netdebug.c +++ b/fs/ocfs2/cluster/netdebug.c @@ -47,6 +47,7 @@ #define SC_DEBUG_NAME "sock_containers" #define NST_DEBUG_NAME "send_tracking" #define STATS_DEBUG_NAME "stats" +#define NODES_DEBUG_NAME "connected_nodes" #define SHOW_SOCK_CONTAINERS 0 #define SHOW_SOCK_STATS 1 @@ -55,6 +56,7 @@ static struct dentry *o2net_dentry; static struct dentry *sc_dentry; static struct dentry *nst_dentry; static struct dentry *stats_dentry; +static struct dentry *nodes_dentry; static DEFINE_SPINLOCK(o2net_debug_lock); @@ -491,53 +493,87 @@ static const struct file_operations sc_seq_fops = { .release = sc_fop_release, }; -int o2net_debugfs_init(void) +static int o2net_fill_bitmap(char *buf, int len) { - o2net_dentry = debugfs_create_dir(O2NET_DEBUG_DIR, NULL); - if (!o2net_dentry) { - mlog_errno(-ENOMEM); - goto bail; - } + unsigned long map[BITS_TO_LONGS(O2NM_MAX_NODES)]; + int i = -1, out = 0; - nst_dentry = debugfs_create_file(NST_DEBUG_NAME, S_IFREG|S_IRUSR, - o2net_dentry, NULL, - &nst_seq_fops); - if (!nst_dentry) { - mlog_errno(-ENOMEM); - goto bail; - } + o2net_fill_node_map(map, sizeof(map)); - sc_dentry = debugfs_create_file(SC_DEBUG_NAME, S_IFREG|S_IRUSR, - o2net_dentry, NULL, - &sc_seq_fops); - if (!sc_dentry) { - mlog_errno(-ENOMEM); - goto bail; - } + while ((i = find_next_bit(map, O2NM_MAX_NODES, i + 1)) < O2NM_MAX_NODES) + out += snprintf(buf + out, PAGE_SIZE - out, "%d ", i); + out += snprintf(buf + out, PAGE_SIZE - out, "\n"); - stats_dentry = debugfs_create_file(STATS_DEBUG_NAME, S_IFREG|S_IRUSR, - o2net_dentry, NULL, - &stats_seq_fops); - if (!stats_dentry) { - mlog_errno(-ENOMEM); - goto bail; - } + return out; +} + +static int nodes_fop_open(struct inode *inode, struct file *file) +{ + char *buf; + + buf = kmalloc(PAGE_SIZE, GFP_KERNEL); + if (!buf) + return -ENOMEM; + + i_size_write(inode, o2net_fill_bitmap(buf, PAGE_SIZE)); + + file->private_data = buf; return 0; -bail: - debugfs_remove(stats_dentry); - debugfs_remove(sc_dentry); - debugfs_remove(nst_dentry); - debugfs_remove(o2net_dentry); - return -ENOMEM; } +static int o2net_debug_release(struct inode *inode, struct file *file) +{ + kfree(file->private_data); + return 0; +} + +static ssize_t o2net_debug_read(struct file *file, char __user *buf, + size_t nbytes, loff_t *ppos) +{ + return simple_read_from_buffer(buf, nbytes, ppos, file->private_data, + i_size_read(file->f_mapping->host)); +} + +static const struct file_operations nodes_fops = { + .open = nodes_fop_open, + .release = o2net_debug_release, + .read = o2net_debug_read, + .llseek = generic_file_llseek, +}; + void o2net_debugfs_exit(void) { + debugfs_remove(nodes_dentry); debugfs_remove(stats_dentry); debugfs_remove(sc_dentry); debugfs_remove(nst_dentry); debugfs_remove(o2net_dentry); } +int o2net_debugfs_init(void) +{ + mode_t mode = S_IFREG|S_IRUSR; + + o2net_dentry = debugfs_create_dir(O2NET_DEBUG_DIR, NULL); + if (o2net_dentry) + nst_dentry = debugfs_create_file(NST_DEBUG_NAME, mode, + o2net_dentry, NULL, &nst_seq_fops); + if (nst_dentry) + sc_dentry = debugfs_create_file(SC_DEBUG_NAME, mode, + o2net_dentry, NULL, &sc_seq_fops); + if (sc_dentry) + stats_dentry = debugfs_create_file(STATS_DEBUG_NAME, mode, + o2net_dentry, NULL, &stats_seq_fops); + if (stats_dentry) + nodes_dentry = debugfs_create_file(NODES_DEBUG_NAME, mode, + o2net_dentry, NULL, &nodes_fops); + if (nodes_dentry) + return 0; + + o2net_debugfs_exit(); + mlog_errno(-ENOMEM); + return -ENOMEM; +} + #endif /* CONFIG_DEBUG_FS */ diff --git a/fs/ocfs2/cluster/tcp.c b/fs/ocfs2/cluster/tcp.c index 6e97895ecf2..ae13d5ca790 100644 --- a/fs/ocfs2/cluster/tcp.c +++ b/fs/ocfs2/cluster/tcp.c @@ -1034,6 +1034,25 @@ static int o2net_tx_can_proceed(struct o2net_node *nn, return ret; } +/* Get a map of all nodes to which this node is currently connected to */ +void o2net_fill_node_map(unsigned long *map, unsigned bytes) +{ + struct o2net_sock_container *sc; + int node, ret; + + BUG_ON(bytes < (BITS_TO_LONGS(O2NM_MAX_NODES) * sizeof(unsigned long))); + + memset(map, 0, bytes); + for (node = 0; node < O2NM_MAX_NODES; ++node) { + o2net_tx_can_proceed(o2net_nn_from_num(node), &sc, &ret); + if (!ret) { + set_bit(node, map); + sc_put(sc); + } + } +} +EXPORT_SYMBOL_GPL(o2net_fill_node_map); + int o2net_send_message_vec(u32 msg_type, u32 key, struct kvec *caller_vec, size_t caller_veclen, u8 target_node, int *status) { diff --git a/fs/ocfs2/cluster/tcp.h b/fs/ocfs2/cluster/tcp.h index fd6179eb26d..5bada2a69b5 100644 --- a/fs/ocfs2/cluster/tcp.h +++ b/fs/ocfs2/cluster/tcp.h @@ -106,6 +106,8 @@ int o2net_register_handler(u32 msg_type, u32 key, u32 max_len, struct list_head *unreg_list); void o2net_unregister_handler_list(struct list_head *list); +void o2net_fill_node_map(unsigned long *map, unsigned bytes); + struct o2nm_node; int o2net_register_hb_callbacks(void); void o2net_unregister_hb_callbacks(void); -- cgit v1.2.3-70-g09d2 From 6b27f62fc750d85bc6fc3718b3b38ec60edc2d74 Mon Sep 17 00:00:00 2001 From: Sunil Mushran Date: Sun, 24 Jul 2011 10:33:54 -0700 Subject: ocfs2/cluster: Cluster up now includes network connections too The cluster up check only checks to see if the node is heartbeating or not. If yes it continues assuming that the node is connected to all the nodes. But if that is not the case, the cluster join aborts with a stack of errors that are not easy to comprehend. This patch adds the network connect check upfront and prints the nodes that the node is not yet connected to, before aborting. Signed-off-by: Sunil Mushran --- fs/ocfs2/dlm/dlmdomain.c | 7 ----- fs/ocfs2/stack_o2cb.c | 67 +++++++++++++++++++++++++++++++++++++++++++----- 2 files changed, 61 insertions(+), 13 deletions(-) (limited to 'fs') diff --git a/fs/ocfs2/dlm/dlmdomain.c b/fs/ocfs2/dlm/dlmdomain.c index 200d9888ffe..92f2ead0fab 100644 --- a/fs/ocfs2/dlm/dlmdomain.c +++ b/fs/ocfs2/dlm/dlmdomain.c @@ -2138,13 +2138,6 @@ struct dlm_ctxt * dlm_register_domain(const char *domain, goto leave; } - if (!o2hb_check_local_node_heartbeating()) { - mlog(ML_ERROR, "the local node has not been configured, or is " - "not heartbeating\n"); - ret = -EPROTO; - goto leave; - } - mlog(0, "register called for domain \"%s\"\n", domain); retry: diff --git a/fs/ocfs2/stack_o2cb.c b/fs/ocfs2/stack_o2cb.c index be935d89a4e..94368017edb 100644 --- a/fs/ocfs2/stack_o2cb.c +++ b/fs/ocfs2/stack_o2cb.c @@ -28,6 +28,7 @@ #include "cluster/masklog.h" #include "cluster/nodemanager.h" #include "cluster/heartbeat.h" +#include "cluster/tcp.h" #include "stackglue.h" @@ -255,6 +256,61 @@ static void o2cb_dump_lksb(struct ocfs2_dlm_lksb *lksb) dlm_print_one_lock(lksb->lksb_o2dlm.lockid); } +/* + * Check if this node is heartbeating and is connected to all other + * heartbeating nodes. + */ +static int o2cb_cluster_check(void) +{ + u8 node_num; + int i; + unsigned long hbmap[BITS_TO_LONGS(O2NM_MAX_NODES)]; + unsigned long netmap[BITS_TO_LONGS(O2NM_MAX_NODES)]; + + node_num = o2nm_this_node(); + if (node_num == O2NM_MAX_NODES) { + printk(KERN_ERR "o2cb: This node has not been configured.\n"); + return -EINVAL; + } + + /* + * o2dlm expects o2net sockets to be created. If not, then + * dlm_join_domain() fails with a stack of errors which are both cryptic + * and incomplete. The idea here is to detect upfront whether we have + * managed to connect to all nodes or not. If not, then list the nodes + * to allow the user to check the configuration (incorrect IP, firewall, + * etc.) Yes, this is racy. But its not the end of the world. + */ +#define O2CB_MAP_STABILIZE_COUNT 60 + for (i = 0; i < O2CB_MAP_STABILIZE_COUNT; ++i) { + o2hb_fill_node_map(hbmap, sizeof(hbmap)); + if (!test_bit(node_num, hbmap)) { + printk(KERN_ERR "o2cb: %s heartbeat has not been " + "started.\n", (o2hb_global_heartbeat_active() ? + "Global" : "Local")); + return -EINVAL; + } + o2net_fill_node_map(netmap, sizeof(netmap)); + /* Force set the current node to allow easy compare */ + set_bit(node_num, netmap); + if (!memcmp(hbmap, netmap, sizeof(hbmap))) + return 0; + if (i < O2CB_MAP_STABILIZE_COUNT) + msleep(1000); + } + + printk(KERN_ERR "o2cb: This node could not connect to nodes:"); + i = -1; + while ((i = find_next_bit(hbmap, O2NM_MAX_NODES, + i + 1)) < O2NM_MAX_NODES) { + if (!test_bit(i, netmap)) + printk(" %u", i); + } + printk(".\n"); + + return -ENOTCONN; +} + /* * Called from the dlm when it's about to evict a node. This is how the * classic stack signals node death. @@ -280,12 +336,11 @@ static int o2cb_cluster_connect(struct ocfs2_cluster_connection *conn) BUG_ON(conn == NULL); BUG_ON(conn->cc_proto == NULL); - /* for now we only have one cluster/node, make sure we see it - * in the heartbeat universe */ - if (!o2hb_check_local_node_heartbeating()) { - if (o2hb_global_heartbeat_active()) - mlog(ML_ERROR, "Global heartbeat not started\n"); - rc = -EINVAL; + /* Ensure cluster stack is up and all nodes are connected */ + rc = o2cb_cluster_check(); + if (rc) { + printk(KERN_ERR "o2cb: Cluster check failed. Fix errors " + "before retrying.\n"); goto out; } -- cgit v1.2.3-70-g09d2 From 619c200de144b44f5e405305241bcd7edbb8c6cf Mon Sep 17 00:00:00 2001 From: Sunil Mushran Date: Sun, 24 Jul 2011 10:34:54 -0700 Subject: ocfs2: Clean up messages in the fs Convert useful messages from ML_NOTICE to KERN_NOTICE to improve readability. Signed-off-by: Sunil Mushran --- fs/ocfs2/journal.c | 9 ++++++--- fs/ocfs2/quota_local.c | 13 +++++++++---- fs/ocfs2/slot_map.c | 4 ++-- fs/ocfs2/super.c | 10 +++++----- 4 files changed, 22 insertions(+), 14 deletions(-) (limited to 'fs') diff --git a/fs/ocfs2/journal.c b/fs/ocfs2/journal.c index 295d56454e8..8eaaa785440 100644 --- a/fs/ocfs2/journal.c +++ b/fs/ocfs2/journal.c @@ -1544,9 +1544,9 @@ static int ocfs2_replay_journal(struct ocfs2_super *osb, /* we need to run complete recovery for offline orphan slots */ ocfs2_replay_map_set_state(osb, REPLAY_NEEDED); - mlog(ML_NOTICE, "Recovering node %d from slot %d on device (%u,%u)\n", - node_num, slot_num, - MAJOR(osb->sb->s_dev), MINOR(osb->sb->s_dev)); + printk(KERN_NOTICE "ocfs2: Begin replay journal (node %d, slot %d) on "\ + "device (%u,%u)\n", node_num, slot_num, MAJOR(osb->sb->s_dev), + MINOR(osb->sb->s_dev)); OCFS2_I(inode)->ip_clusters = le32_to_cpu(fe->i_clusters); @@ -1601,6 +1601,9 @@ static int ocfs2_replay_journal(struct ocfs2_super *osb, jbd2_journal_destroy(journal); + printk(KERN_NOTICE "ocfs2: End replay journal (node %d, slot %d) on "\ + "device (%u,%u)\n", node_num, slot_num, MAJOR(osb->sb->s_dev), + MINOR(osb->sb->s_dev)); done: /* drop the lock on this nodes journal */ if (got_lock) diff --git a/fs/ocfs2/quota_local.c b/fs/ocfs2/quota_local.c index dc8007fc924..942fd65bdad 100644 --- a/fs/ocfs2/quota_local.c +++ b/fs/ocfs2/quota_local.c @@ -404,7 +404,9 @@ struct ocfs2_quota_recovery *ocfs2_begin_quota_recovery( int status = 0; struct ocfs2_quota_recovery *rec; - mlog(ML_NOTICE, "Beginning quota recovery in slot %u\n", slot_num); + printk(KERN_NOTICE "ocfs2: Beginning quota recovery on device (%s) for " + "slot %u\n", osb->dev_str, slot_num); + rec = ocfs2_alloc_quota_recovery(); if (!rec) return ERR_PTR(-ENOMEM); @@ -596,7 +598,9 @@ int ocfs2_finish_quota_recovery(struct ocfs2_super *osb, struct inode *lqinode; unsigned int flags; - mlog(ML_NOTICE, "Finishing quota recovery in slot %u\n", slot_num); + printk(KERN_NOTICE "ocfs2: Finishing quota recovery on device (%s) for " + "slot %u\n", osb->dev_str, slot_num); + mutex_lock(&sb_dqopt(sb)->dqonoff_mutex); for (type = 0; type < MAXQUOTAS; type++) { if (list_empty(&(rec->r_list[type]))) @@ -612,8 +616,9 @@ int ocfs2_finish_quota_recovery(struct ocfs2_super *osb, /* Someone else is holding the lock? Then he must be * doing the recovery. Just skip the file... */ if (status == -EAGAIN) { - mlog(ML_NOTICE, "skipping quota recovery for slot %d " - "because quota file is locked.\n", slot_num); + printk(KERN_NOTICE "ocfs2: Skipping quota recovery on " + "device (%s) for slot %d because quota file is " + "locked.\n", osb->dev_str, slot_num); status = 0; goto out_put; } else if (status < 0) { diff --git a/fs/ocfs2/slot_map.c b/fs/ocfs2/slot_map.c index 26fc0014d50..1424c151ccc 100644 --- a/fs/ocfs2/slot_map.c +++ b/fs/ocfs2/slot_map.c @@ -493,8 +493,8 @@ int ocfs2_find_slot(struct ocfs2_super *osb) goto bail; } } else - mlog(ML_NOTICE, "slot %d is already allocated to this node!\n", - slot); + printk(KERN_INFO "ocfs2: Slot %d on device (%s) was already " + "allocated to this node!\n", slot, osb->dev_str); ocfs2_set_slot(si, slot, osb->node_num); osb->slot_num = slot; diff --git a/fs/ocfs2/super.c b/fs/ocfs2/super.c index 56f61027236..3e7850380d2 100644 --- a/fs/ocfs2/super.c +++ b/fs/ocfs2/super.c @@ -1107,9 +1107,9 @@ static int ocfs2_fill_super(struct super_block *sb, void *data, int silent) ocfs2_set_ro_flag(osb, 1); - printk(KERN_NOTICE "Readonly device detected. No cluster " - "services will be utilized for this mount. Recovery " - "will be skipped.\n"); + printk(KERN_NOTICE "ocfs2: Readonly device (%s) detected. " + "Cluster services will not be used for this mount. " + "Recovery will be skipped.\n", osb->dev_str); } if (!ocfs2_is_hard_readonly(osb)) { @@ -2462,8 +2462,8 @@ static int ocfs2_check_volume(struct ocfs2_super *osb) goto finally; } } else { - mlog(ML_NOTICE, "File system was not unmounted cleanly, " - "recovering volume.\n"); + printk(KERN_NOTICE "ocfs2: File system on device (%s) was not " + "unmounted cleanly, recovering it.\n", osb->dev_str); } local = ocfs2_mount_local(osb); -- cgit v1.2.3-70-g09d2 From a035bff6b82aca89c1223e2c614adc2d17ec8aa2 Mon Sep 17 00:00:00 2001 From: Sunil Mushran Date: Sun, 24 Jul 2011 10:35:54 -0700 Subject: ocfs2: Add comment about orphan scanning Add a comment that explains the reason as to why orphan scan scans all the slots. Signed-off-by: Sunil Mushran --- fs/ocfs2/journal.c | 14 ++++++++++++++ 1 file changed, 14 insertions(+) (limited to 'fs') diff --git a/fs/ocfs2/journal.c b/fs/ocfs2/journal.c index 8eaaa785440..0a42ae96dca 100644 --- a/fs/ocfs2/journal.c +++ b/fs/ocfs2/journal.c @@ -1811,6 +1811,20 @@ static inline unsigned long ocfs2_orphan_scan_timeout(void) * every slot, queuing a recovery of the slot on the ocfs2_wq thread. This * is done to catch any orphans that are left over in orphan directories. * + * It scans all slots, even ones that are in use. It does so to handle the + * case described below: + * + * Node 1 has an inode it was using. The dentry went away due to memory + * pressure. Node 1 closes the inode, but it's on the free list. The node + * has the open lock. + * Node 2 unlinks the inode. It grabs the dentry lock to notify others, + * but node 1 has no dentry and doesn't get the message. It trylocks the + * open lock, sees that another node has a PR, and does nothing. + * Later node 2 runs its orphan dir. It igets the inode, trylocks the + * open lock, sees the PR still, and does nothing. + * Basically, we have to trigger an orphan iput on node 1. The only way + * for this to happen is if node 1 runs node 2's orphan dir. + * * ocfs2_queue_orphan_scan gets called every ORPHAN_SCAN_SCHEDULE_TIMEOUT * seconds. It gets an EX lock on os_lockres and checks sequence number * stored in LVB. If the sequence number has changed, it means some other -- cgit v1.2.3-70-g09d2 From 5cffff9e29866a3de98c2c25135b3199491f93b0 Mon Sep 17 00:00:00 2001 From: Wengang Wang Date: Sun, 24 Jul 2011 10:36:54 -0700 Subject: ocfs2: Fix ocfs2_page_mkwrite() This patch address two shortcomings in ocfs2_page_mkwrite(): 1. Makes the function return better VM_FAULT_* errors. 2. It handles a error that is triggered when a page is dropped from the mapping due to memory pressure. This patch locks the page to prevent that. [Patch was cleaned up by Sunil Mushran.] Signed-off-by: Wengang Wang Signed-off-by: Sunil Mushran --- fs/ocfs2/aops.c | 51 +++++++++++++++++++++++++++++++++++++++++++-------- fs/ocfs2/mmap.c | 53 ++++++++++++++++++++++++----------------------------- 2 files changed, 67 insertions(+), 37 deletions(-) (limited to 'fs') diff --git a/fs/ocfs2/aops.c b/fs/ocfs2/aops.c index c1efe939c77..ff98c169b63 100644 --- a/fs/ocfs2/aops.c +++ b/fs/ocfs2/aops.c @@ -862,6 +862,12 @@ struct ocfs2_write_ctxt { struct page *w_pages[OCFS2_MAX_CTXT_PAGES]; struct page *w_target_page; + /* + * w_target_locked is used for page_mkwrite path indicating no unlocking + * against w_target_page in ocfs2_write_end_nolock. + */ + unsigned int w_target_locked:1; + /* * ocfs2_write_end() uses this to know what the real range to * write in the target should be. @@ -895,6 +901,24 @@ void ocfs2_unlock_and_free_pages(struct page **pages, int num_pages) static void ocfs2_free_write_ctxt(struct ocfs2_write_ctxt *wc) { + int i; + + /* + * w_target_locked is only set to true in the page_mkwrite() case. + * The intent is to allow us to lock the target page from write_begin() + * to write_end(). The caller must hold a ref on w_target_page. + */ + if (wc->w_target_locked) { + BUG_ON(!wc->w_target_page); + for (i = 0; i < wc->w_num_pages; i++) { + if (wc->w_target_page == wc->w_pages[i]) { + wc->w_pages[i] = NULL; + break; + } + } + mark_page_accessed(wc->w_target_page); + page_cache_release(wc->w_target_page); + } ocfs2_unlock_and_free_pages(wc->w_pages, wc->w_num_pages); brelse(wc->w_di_bh); @@ -1132,20 +1156,17 @@ static int ocfs2_grab_pages_for_write(struct address_space *mapping, */ lock_page(mmap_page); + /* Exit and let the caller retry */ if (mmap_page->mapping != mapping) { + WARN_ON(mmap_page->mapping); unlock_page(mmap_page); - /* - * Sanity check - the locking in - * ocfs2_pagemkwrite() should ensure - * that this code doesn't trigger. - */ - ret = -EINVAL; - mlog_errno(ret); + ret = -EAGAIN; goto out; } page_cache_get(mmap_page); wc->w_pages[i] = mmap_page; + wc->w_target_locked = true; } else { wc->w_pages[i] = find_or_create_page(mapping, index, GFP_NOFS); @@ -1160,6 +1181,8 @@ static int ocfs2_grab_pages_for_write(struct address_space *mapping, wc->w_target_page = wc->w_pages[i]; } out: + if (ret) + wc->w_target_locked = false; return ret; } @@ -1817,11 +1840,23 @@ try_again: */ ret = ocfs2_grab_pages_for_write(mapping, wc, wc->w_cpos, pos, len, cluster_of_pages, mmap_page); - if (ret) { + if (ret && ret != -EAGAIN) { mlog_errno(ret); goto out_quota; } + /* + * ocfs2_grab_pages_for_write() returns -EAGAIN if it could not lock + * the target page. In this case, we exit with no error and no target + * page. This will trigger the caller, page_mkwrite(), to re-try + * the operation. + */ + if (ret == -EAGAIN) { + BUG_ON(wc->w_target_page); + ret = 0; + goto out_quota; + } + ret = ocfs2_write_cluster_by_desc(mapping, data_ac, meta_ac, wc, pos, len); if (ret) { diff --git a/fs/ocfs2/mmap.c b/fs/ocfs2/mmap.c index 3e9393ca39e..9cd41083e99 100644 --- a/fs/ocfs2/mmap.c +++ b/fs/ocfs2/mmap.c @@ -61,7 +61,7 @@ static int ocfs2_fault(struct vm_area_struct *area, struct vm_fault *vmf) static int __ocfs2_page_mkwrite(struct file *file, struct buffer_head *di_bh, struct page *page) { - int ret; + int ret = VM_FAULT_NOPAGE; struct inode *inode = file->f_path.dentry->d_inode; struct address_space *mapping = inode->i_mapping; loff_t pos = page_offset(page); @@ -71,32 +71,25 @@ static int __ocfs2_page_mkwrite(struct file *file, struct buffer_head *di_bh, void *fsdata; loff_t size = i_size_read(inode); - /* - * Another node might have truncated while we were waiting on - * cluster locks. - * We don't check size == 0 before the shift. This is borrowed - * from do_generic_file_read. - */ last_index = (size - 1) >> PAGE_CACHE_SHIFT; - if (unlikely(!size || page->index > last_index)) { - ret = -EINVAL; - goto out; - } /* - * The i_size check above doesn't catch the case where nodes - * truncated and then re-extended the file. We'll re-check the - * page mapping after taking the page lock inside of - * ocfs2_write_begin_nolock(). + * There are cases that lead to the page no longer bebongs to the + * mapping. + * 1) pagecache truncates locally due to memory pressure. + * 2) pagecache truncates when another is taking EX lock against + * inode lock. see ocfs2_data_convert_worker. + * + * The i_size check doesn't catch the case where nodes truncated and + * then re-extended the file. We'll re-check the page mapping after + * taking the page lock inside of ocfs2_write_begin_nolock(). + * + * Let VM retry with these cases. */ - if (!PageUptodate(page) || page->mapping != inode->i_mapping) { - /* - * the page has been umapped in ocfs2_data_downconvert_worker. - * So return 0 here and let VFS retry. - */ - ret = 0; + if ((page->mapping != inode->i_mapping) || + (!PageUptodate(page)) || + (page_offset(page) >= size)) goto out; - } /* * Call ocfs2_write_begin() and ocfs2_write_end() to take @@ -116,17 +109,21 @@ static int __ocfs2_page_mkwrite(struct file *file, struct buffer_head *di_bh, if (ret) { if (ret != -ENOSPC) mlog_errno(ret); + if (ret == -ENOMEM) + ret = VM_FAULT_OOM; + else + ret = VM_FAULT_SIGBUS; goto out; } - ret = ocfs2_write_end_nolock(mapping, pos, len, len, locked_page, - fsdata); - if (ret < 0) { - mlog_errno(ret); + if (!locked_page) { + ret = VM_FAULT_NOPAGE; goto out; } + ret = ocfs2_write_end_nolock(mapping, pos, len, len, locked_page, + fsdata); BUG_ON(ret != len); - ret = 0; + ret = VM_FAULT_LOCKED; out: return ret; } @@ -168,8 +165,6 @@ static int ocfs2_page_mkwrite(struct vm_area_struct *vma, struct vm_fault *vmf) out: ocfs2_unblock_signals(&oldset); - if (ret) - ret = VM_FAULT_SIGBUS; return ret; } -- cgit v1.2.3-70-g09d2 From 93862d5e1ab875664c6cc95254fc365028a48bb1 Mon Sep 17 00:00:00 2001 From: Sunil Mushran Date: Mon, 25 Jul 2011 14:58:15 -0700 Subject: ocfs2: Implement llseek() ocfs2 implements its own llseek() to provide the SEEK_HOLE/SEEK_DATA functionality. SEEK_HOLE sets the file pointer to the start of either a hole or an unwritten (preallocated) extent, that is greater than or equal to the supplied offset. SEEK_DATA sets the file pointer to the start of an allocated extent (not unwritten) that is greater than or equal to the supplied offset. If the supplied offset is on a desired region, then the file pointer is set to it. Offsets greater than or equal to the file size return -ENXIO. Unwritten (preallocated) extents are considered holes because the file system treats reads to such regions in the same way as it does to holes. Signed-off-by: Sunil Mushran --- fs/ocfs2/extent_map.c | 96 +++++++++++++++++++++++++++++++++++++++++++++++++++ fs/ocfs2/extent_map.h | 2 ++ fs/ocfs2/file.c | 55 +++++++++++++++++++++++++++-- 3 files changed, 151 insertions(+), 2 deletions(-) (limited to 'fs') diff --git a/fs/ocfs2/extent_map.c b/fs/ocfs2/extent_map.c index 23457b491e8..2f5b92ef0e5 100644 --- a/fs/ocfs2/extent_map.c +++ b/fs/ocfs2/extent_map.c @@ -832,6 +832,102 @@ out: return ret; } +int ocfs2_seek_data_hole_offset(struct file *file, loff_t *offset, int origin) +{ + struct inode *inode = file->f_mapping->host; + int ret; + unsigned int is_last = 0, is_data = 0; + u16 cs_bits = OCFS2_SB(inode->i_sb)->s_clustersize_bits; + u32 cpos, cend, clen, hole_size; + u64 extoff, extlen; + struct buffer_head *di_bh = NULL; + struct ocfs2_extent_rec rec; + + BUG_ON(origin != SEEK_DATA && origin != SEEK_HOLE); + + ret = ocfs2_inode_lock(inode, &di_bh, 0); + if (ret) { + mlog_errno(ret); + goto out; + } + + down_read(&OCFS2_I(inode)->ip_alloc_sem); + + if (*offset >= inode->i_size) { + ret = -ENXIO; + goto out_unlock; + } + + if (OCFS2_I(inode)->ip_dyn_features & OCFS2_INLINE_DATA_FL) { + if (origin == SEEK_HOLE) + *offset = inode->i_size; + goto out_unlock; + } + + clen = 0; + cpos = *offset >> cs_bits; + cend = ocfs2_clusters_for_bytes(inode->i_sb, inode->i_size); + + while (cpos < cend && !is_last) { + ret = ocfs2_get_clusters_nocache(inode, di_bh, cpos, &hole_size, + &rec, &is_last); + if (ret) { + mlog_errno(ret); + goto out_unlock; + } + + extoff = cpos; + extoff <<= cs_bits; + + if (rec.e_blkno == 0ULL) { + clen = hole_size; + is_data = 0; + } else { + clen = le16_to_cpu(rec.e_leaf_clusters) - + (cpos - le32_to_cpu(rec.e_cpos)); + is_data = (rec.e_flags & OCFS2_EXT_UNWRITTEN) ? 0 : 1; + } + + if ((!is_data && origin == SEEK_HOLE) || + (is_data && origin == SEEK_DATA)) { + if (extoff > *offset) + *offset = extoff; + goto out_unlock; + } + + if (!is_last) + cpos += clen; + } + + if (origin == SEEK_HOLE) { + extoff = cpos; + extoff <<= cs_bits; + extlen = clen; + extlen <<= cs_bits; + + if ((extoff + extlen) > inode->i_size) + extlen = inode->i_size - extoff; + extoff += extlen; + if (extoff > *offset) + *offset = extoff; + goto out_unlock; + } + + ret = -ENXIO; + +out_unlock: + + brelse(di_bh); + + up_read(&OCFS2_I(inode)->ip_alloc_sem); + + ocfs2_inode_unlock(inode, 0); +out: + if (ret && ret != -ENXIO) + ret = -ENXIO; + return ret; +} + int ocfs2_read_virt_blocks(struct inode *inode, u64 v_block, int nr, struct buffer_head *bhs[], int flags, int (*validate)(struct super_block *sb, diff --git a/fs/ocfs2/extent_map.h b/fs/ocfs2/extent_map.h index e79d41c2c90..67ea57d2fd5 100644 --- a/fs/ocfs2/extent_map.h +++ b/fs/ocfs2/extent_map.h @@ -53,6 +53,8 @@ int ocfs2_extent_map_get_blocks(struct inode *inode, u64 v_blkno, u64 *p_blkno, int ocfs2_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, u64 map_start, u64 map_len); +int ocfs2_seek_data_hole_offset(struct file *file, loff_t *offset, int origin); + int ocfs2_xattr_get_clusters(struct inode *inode, u32 v_cluster, u32 *p_cluster, u32 *num_clusters, struct ocfs2_extent_list *el, diff --git a/fs/ocfs2/file.c b/fs/ocfs2/file.c index 0fc2bd34039..c0f015e11c2 100644 --- a/fs/ocfs2/file.c +++ b/fs/ocfs2/file.c @@ -2591,6 +2591,57 @@ bail: return ret; } +/* Refer generic_file_llseek_unlocked() */ +static loff_t ocfs2_file_llseek(struct file *file, loff_t offset, int origin) +{ + struct inode *inode = file->f_mapping->host; + int ret = 0; + + mutex_lock(&inode->i_mutex); + + switch (origin) { + case SEEK_SET: + break; + case SEEK_END: + offset += inode->i_size; + break; + case SEEK_CUR: + if (offset == 0) { + offset = file->f_pos; + goto out; + } + offset += file->f_pos; + break; + case SEEK_DATA: + case SEEK_HOLE: + ret = ocfs2_seek_data_hole_offset(file, &offset, origin); + if (ret) + goto out; + break; + default: + ret = -EINVAL; + goto out; + } + + if (offset < 0 && !(file->f_mode & FMODE_UNSIGNED_OFFSET)) + ret = -EINVAL; + if (!ret && offset > inode->i_sb->s_maxbytes) + ret = -EINVAL; + if (ret) + goto out; + + if (offset != file->f_pos) { + file->f_pos = offset; + file->f_version = 0; + } + +out: + mutex_unlock(&inode->i_mutex); + if (ret) + return ret; + return offset; +} + const struct inode_operations ocfs2_file_iops = { .setattr = ocfs2_setattr, .getattr = ocfs2_getattr, @@ -2615,7 +2666,7 @@ const struct inode_operations ocfs2_special_file_iops = { * ocfs2_fops_no_plocks and ocfs2_dops_no_plocks! */ const struct file_operations ocfs2_fops = { - .llseek = generic_file_llseek, + .llseek = ocfs2_file_llseek, .read = do_sync_read, .write = do_sync_write, .mmap = ocfs2_mmap, @@ -2663,7 +2714,7 @@ const struct file_operations ocfs2_dops = { * the cluster. */ const struct file_operations ocfs2_fops_no_plocks = { - .llseek = generic_file_llseek, + .llseek = ocfs2_file_llseek, .read = do_sync_read, .write = do_sync_write, .mmap = ocfs2_mmap, -- cgit v1.2.3-70-g09d2 From a11f7e63c59810a81494d4c4b028af707d4c7ca4 Mon Sep 17 00:00:00 2001 From: Mark Fasheh Date: Wed, 22 Jun 2011 14:23:38 -0700 Subject: ocfs2: serialize unaligned aio Fix a corruption that can happen when we have (two or more) outstanding aio's to an overlapping unaligned region. Ext4 (e9e3bcecf44c04b9e6b505fd8e2eb9cea58fb94d) and xfs recently had to fix similar issues. In our case what happens is that we can have an outstanding aio on a region and if a write comes in with some bytes overlapping the original aio we may decide to read that region into a page before continuing (typically because of buffered-io fallback). Since we have no ordering guarantees with the aio, we can read stale or bad data into the page and then write it back out. If the i/o is page and block aligned, then we avoid this issue as there won't be any need to read data from disk. I took the same approach as Eric in the ext4 patch and introduced some serialization of unaligned async direct i/o. I don't expect this to have an effect on the most common cases of AIO. Unaligned aio will be slower though, but that's far more acceptable than data corruption. Signed-off-by: Mark Fasheh Signed-off-by: Joel Becker --- fs/ocfs2/aops.c | 10 ++++++++++ fs/ocfs2/aops.h | 14 ++++++++++++++ fs/ocfs2/file.c | 38 ++++++++++++++++++++++++++++++++++++++ fs/ocfs2/inode.h | 3 +++ fs/ocfs2/super.c | 10 ++++++++-- 5 files changed, 73 insertions(+), 2 deletions(-) (limited to 'fs') diff --git a/fs/ocfs2/aops.c b/fs/ocfs2/aops.c index ac97bca282d..4c1ec8f2d8c 100644 --- a/fs/ocfs2/aops.c +++ b/fs/ocfs2/aops.c @@ -564,6 +564,7 @@ static void ocfs2_dio_end_io(struct kiocb *iocb, { struct inode *inode = iocb->ki_filp->f_path.dentry->d_inode; int level; + wait_queue_head_t *wq = ocfs2_ioend_wq(inode); /* this io's submitter should not have unlocked this before we could */ BUG_ON(!ocfs2_iocb_is_rw_locked(iocb)); @@ -573,6 +574,15 @@ static void ocfs2_dio_end_io(struct kiocb *iocb, ocfs2_iocb_clear_sem_locked(iocb); } + if (ocfs2_iocb_is_unaligned_aio(iocb)) { + ocfs2_iocb_clear_unaligned_aio(iocb); + + if (atomic_dec_and_test(&OCFS2_I(inode)->ip_unaligned_aio) && + waitqueue_active(wq)) { + wake_up_all(wq); + } + } + ocfs2_iocb_clear_rw_locked(iocb); level = ocfs2_iocb_rw_locked_level(iocb); diff --git a/fs/ocfs2/aops.h b/fs/ocfs2/aops.h index 75cf3ad987a..ffb2da370a9 100644 --- a/fs/ocfs2/aops.h +++ b/fs/ocfs2/aops.h @@ -78,6 +78,7 @@ enum ocfs2_iocb_lock_bits { OCFS2_IOCB_RW_LOCK = 0, OCFS2_IOCB_RW_LOCK_LEVEL, OCFS2_IOCB_SEM, + OCFS2_IOCB_UNALIGNED_IO, OCFS2_IOCB_NUM_LOCKS }; @@ -91,4 +92,17 @@ enum ocfs2_iocb_lock_bits { clear_bit(OCFS2_IOCB_SEM, (unsigned long *)&iocb->private) #define ocfs2_iocb_is_sem_locked(iocb) \ test_bit(OCFS2_IOCB_SEM, (unsigned long *)&iocb->private) + +#define ocfs2_iocb_set_unaligned_aio(iocb) \ + set_bit(OCFS2_IOCB_UNALIGNED_IO, (unsigned long *)&iocb->private) +#define ocfs2_iocb_clear_unaligned_aio(iocb) \ + clear_bit(OCFS2_IOCB_UNALIGNED_IO, (unsigned long *)&iocb->private) +#define ocfs2_iocb_is_unaligned_aio(iocb) \ + test_bit(OCFS2_IOCB_UNALIGNED_IO, (unsigned long *)&iocb->private) + +#define OCFS2_IOEND_WQ_HASH_SZ 37 +#define ocfs2_ioend_wq(v) (&ocfs2__ioend_wq[((unsigned long)(v)) %\ + OCFS2_IOEND_WQ_HASH_SZ]) +extern wait_queue_head_t ocfs2__ioend_wq[OCFS2_IOEND_WQ_HASH_SZ]; + #endif /* OCFS2_FILE_H */ diff --git a/fs/ocfs2/file.c b/fs/ocfs2/file.c index b1e35a392ca..145f4533a93 100644 --- a/fs/ocfs2/file.c +++ b/fs/ocfs2/file.c @@ -2038,6 +2038,23 @@ out: return ret; } +static void ocfs2_aiodio_wait(struct inode *inode) +{ + wait_queue_head_t *wq = ocfs2_ioend_wq(inode); + + wait_event(*wq, (atomic_read(&OCFS2_I(inode)->ip_unaligned_aio) == 0)); +} + +static int ocfs2_is_io_unaligned(struct inode *inode, size_t count, loff_t pos) +{ + int blockmask = inode->i_sb->s_blocksize - 1; + loff_t final_size = pos + count; + + if ((pos & blockmask) || (final_size & blockmask)) + return 1; + return 0; +} + static int ocfs2_prepare_inode_for_refcount(struct inode *inode, struct file *file, loff_t pos, size_t count, @@ -2216,6 +2233,7 @@ static ssize_t ocfs2_file_aio_write(struct kiocb *iocb, struct ocfs2_super *osb = OCFS2_SB(inode->i_sb); int full_coherency = !(osb->s_mount_opt & OCFS2_MOUNT_COHERENCY_BUFFERED); + int unaligned_dio = 0; trace_ocfs2_file_aio_write(inode, file, file->f_path.dentry, (unsigned long long)OCFS2_I(inode)->ip_blkno, @@ -2284,6 +2302,10 @@ relock: goto out; } + if (direct_io && !is_sync_kiocb(iocb)) + unaligned_dio = ocfs2_is_io_unaligned(inode, iocb->ki_left, + *ppos); + /* * We can't complete the direct I/O as requested, fall back to * buffered I/O. @@ -2299,6 +2321,18 @@ relock: goto relock; } + if (unaligned_dio) { + /* + * Wait on previous unaligned aio to complete before + * proceeding. + */ + ocfs2_aiodio_wait(inode); + + /* Mark the iocb as needing a decrement in ocfs2_dio_end_io */ + atomic_inc(&OCFS2_I(inode)->ip_unaligned_aio); + ocfs2_iocb_set_unaligned_aio(iocb); + } + /* * To later detect whether a journal commit for sync writes is * necessary, we sample i_size, and cluster count here. @@ -2371,8 +2405,12 @@ out_dio: if ((ret == -EIOCBQUEUED) || (!ocfs2_iocb_is_rw_locked(iocb))) { rw_level = -1; have_alloc_sem = 0; + unaligned_dio = 0; } + if (unaligned_dio) + atomic_dec(&OCFS2_I(inode)->ip_unaligned_aio); + out: if (rw_level != -1) ocfs2_rw_unlock(inode, rw_level); diff --git a/fs/ocfs2/inode.h b/fs/ocfs2/inode.h index 1c508b149b3..88924a3133f 100644 --- a/fs/ocfs2/inode.h +++ b/fs/ocfs2/inode.h @@ -43,6 +43,9 @@ struct ocfs2_inode_info /* protects extended attribute changes on this inode */ struct rw_semaphore ip_xattr_sem; + /* Number of outstanding AIO's which are not page aligned */ + atomic_t ip_unaligned_aio; + /* These fields are protected by ip_lock */ spinlock_t ip_lock; u32 ip_open_count; diff --git a/fs/ocfs2/super.c b/fs/ocfs2/super.c index 029c4cd8a69..603f5fe9f81 100644 --- a/fs/ocfs2/super.c +++ b/fs/ocfs2/super.c @@ -54,6 +54,7 @@ #include "ocfs1_fs_compat.h" #include "alloc.h" +#include "aops.h" #include "blockcheck.h" #include "dlmglue.h" #include "export.h" @@ -1616,12 +1617,17 @@ static int ocfs2_show_options(struct seq_file *s, struct vfsmount *mnt) return 0; } +wait_queue_head_t ocfs2__ioend_wq[OCFS2_IOEND_WQ_HASH_SZ]; + static int __init ocfs2_init(void) { - int status; + int status, i; ocfs2_print_version(); + for (i = 0; i < OCFS2_IOEND_WQ_HASH_SZ; i++) + init_waitqueue_head(&ocfs2__ioend_wq[i]); + status = init_ocfs2_uptodate_cache(); if (status < 0) { mlog_errno(status); @@ -1760,7 +1766,7 @@ static void ocfs2_inode_init_once(void *data) ocfs2_extent_map_init(&oi->vfs_inode); INIT_LIST_HEAD(&oi->ip_io_markers); oi->ip_dir_start_lookup = 0; - + atomic_set(&oi->ip_unaligned_aio, 0); init_rwsem(&oi->ip_alloc_sem); init_rwsem(&oi->ip_xattr_sem); mutex_init(&oi->ip_io_mutex); -- cgit v1.2.3-70-g09d2 From c7e25e6e0b0486492c5faaf6312b37413642c48e Mon Sep 17 00:00:00 2001 From: Jan Kara Date: Thu, 23 Jun 2011 22:51:47 +0200 Subject: ocfs2: Avoid livelock in ocfs2_readpage() When someone writes to an inode, readers accessing the same inode via ocfs2_readpage() just busyloop trying to get ip_alloc_sem because do_generic_file_read() looks up the page again and retries ->readpage() when previous attempt failed with AOP_TRUNCATED_PAGE. When there are enough readers, they can occupy all CPUs and in non-preempt kernel the system is deadlocked because writer holding ip_alloc_sem is never run to release the semaphore. Fix the problem by making reader block on ip_alloc_sem to break the busy loop. Signed-off-by: Jan Kara Signed-off-by: Joel Becker --- fs/ocfs2/aops.c | 8 ++++++++ 1 file changed, 8 insertions(+) (limited to 'fs') diff --git a/fs/ocfs2/aops.c b/fs/ocfs2/aops.c index 4c1ec8f2d8c..ba3ca1e63b5 100644 --- a/fs/ocfs2/aops.c +++ b/fs/ocfs2/aops.c @@ -290,7 +290,15 @@ static int ocfs2_readpage(struct file *file, struct page *page) } if (down_read_trylock(&oi->ip_alloc_sem) == 0) { + /* + * Unlock the page and cycle ip_alloc_sem so that we don't + * busyloop waiting for ip_alloc_sem to unlock + */ ret = AOP_TRUNCATED_PAGE; + unlock_page(page); + unlock = 0; + down_read(&oi->ip_alloc_sem); + up_read(&oi->ip_alloc_sem); goto out_inode_unlock; } -- cgit v1.2.3-70-g09d2 From 1788ea6e3b2a58cf4fb00206e362d9caff8d86a7 Mon Sep 17 00:00:00 2001 From: Jeff Layton Date: Fri, 4 Nov 2011 13:31:21 -0400 Subject: nfs: when attempting to open a directory, fall back on normal lookup (try #5) commit d953126 changed how nfs_atomic_lookup handles an -EISDIR return from an OPEN call. Prior to that patch, that caused the client to fall back to doing a normal lookup. When that patch went in, the code began returning that error to userspace. The d_revalidate codepath however never had the corresponding change, so it was still possible to end up with a NULL ctx->state pointer after that. That patch caused a regression. When we attempt to open a directory that does not have a cached dentry, that open now errors out with EISDIR. If you attempt the same open with a cached dentry, it will succeed. Fix this by reverting the change in nfs_atomic_lookup and allowing attempts to open directories to fall back to a normal lookup Also, add a NFSv4-specific f_ops->open routine that just returns -ENOTDIR. This should never be called if things are working properly, but if it ever is, then the dprintk may help in debugging. To facilitate this, a new file_operations field is also added to the nfs_rpc_ops struct. Cc: stable@kernel.org Signed-off-by: Jeff Layton Signed-off-by: Trond Myklebust --- fs/nfs/dir.c | 2 +- fs/nfs/file.c | 32 ++++++++++++++++++++++++++++++++ fs/nfs/inode.c | 2 +- fs/nfs/nfs3proc.c | 1 + fs/nfs/nfs4proc.c | 1 + fs/nfs/proc.c | 1 + include/linux/nfs_fs.h | 3 +++ include/linux/nfs_xdr.h | 1 + 8 files changed, 41 insertions(+), 2 deletions(-) (limited to 'fs') diff --git a/fs/nfs/dir.c b/fs/nfs/dir.c index b238d95ac48..ac289909814 100644 --- a/fs/nfs/dir.c +++ b/fs/nfs/dir.c @@ -1468,12 +1468,12 @@ static struct dentry *nfs_atomic_lookup(struct inode *dir, struct dentry *dentry res = NULL; goto out; /* This turned out not to be a regular file */ + case -EISDIR: case -ENOTDIR: goto no_open; case -ELOOP: if (!(nd->intent.open.flags & O_NOFOLLOW)) goto no_open; - /* case -EISDIR: */ /* case -EINVAL: */ default: res = ERR_CAST(inode); diff --git a/fs/nfs/file.c b/fs/nfs/file.c index 0a1f8312b4d..6d93e0754b5 100644 --- a/fs/nfs/file.c +++ b/fs/nfs/file.c @@ -886,3 +886,35 @@ static int nfs_setlease(struct file *file, long arg, struct file_lock **fl) file->f_path.dentry->d_name.name, arg); return -EINVAL; } + +#ifdef CONFIG_NFS_V4 +static int +nfs4_file_open(struct inode *inode, struct file *filp) +{ + /* + * NFSv4 opens are handled in d_lookup and d_revalidate. If we get to + * this point, then something is very wrong + */ + dprintk("NFS: %s called! inode=%p filp=%p\n", __func__, inode, filp); + return -ENOTDIR; +} + +const struct file_operations nfs4_file_operations = { + .llseek = nfs_file_llseek, + .read = do_sync_read, + .write = do_sync_write, + .aio_read = nfs_file_read, + .aio_write = nfs_file_write, + .mmap = nfs_file_mmap, + .open = nfs4_file_open, + .flush = nfs_file_flush, + .release = nfs_file_release, + .fsync = nfs_file_fsync, + .lock = nfs_lock, + .flock = nfs_flock, + .splice_read = nfs_file_splice_read, + .splice_write = nfs_file_splice_write, + .check_flags = nfs_check_flags, + .setlease = nfs_setlease, +}; +#endif /* CONFIG_NFS_V4 */ diff --git a/fs/nfs/inode.c b/fs/nfs/inode.c index c07a55aec83..50a15fa8cf9 100644 --- a/fs/nfs/inode.c +++ b/fs/nfs/inode.c @@ -291,7 +291,7 @@ nfs_fhget(struct super_block *sb, struct nfs_fh *fh, struct nfs_fattr *fattr) */ inode->i_op = NFS_SB(sb)->nfs_client->rpc_ops->file_inode_ops; if (S_ISREG(inode->i_mode)) { - inode->i_fop = &nfs_file_operations; + inode->i_fop = NFS_SB(sb)->nfs_client->rpc_ops->file_ops; inode->i_data.a_ops = &nfs_file_aops; inode->i_data.backing_dev_info = &NFS_SB(sb)->backing_dev_info; } else if (S_ISDIR(inode->i_mode)) { diff --git a/fs/nfs/nfs3proc.c b/fs/nfs/nfs3proc.c index 85f1690ca08..d4bc9ed9174 100644 --- a/fs/nfs/nfs3proc.c +++ b/fs/nfs/nfs3proc.c @@ -853,6 +853,7 @@ const struct nfs_rpc_ops nfs_v3_clientops = { .dentry_ops = &nfs_dentry_operations, .dir_inode_ops = &nfs3_dir_inode_operations, .file_inode_ops = &nfs3_file_inode_operations, + .file_ops = &nfs_file_operations, .getroot = nfs3_proc_get_root, .getattr = nfs3_proc_getattr, .setattr = nfs3_proc_setattr, diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c index b60fddf606f..069cb8094d4 100644 --- a/fs/nfs/nfs4proc.c +++ b/fs/nfs/nfs4proc.c @@ -6253,6 +6253,7 @@ const struct nfs_rpc_ops nfs_v4_clientops = { .dentry_ops = &nfs4_dentry_operations, .dir_inode_ops = &nfs4_dir_inode_operations, .file_inode_ops = &nfs4_file_inode_operations, + .file_ops = &nfs4_file_operations, .getroot = nfs4_proc_get_root, .getattr = nfs4_proc_getattr, .setattr = nfs4_proc_setattr, diff --git a/fs/nfs/proc.c b/fs/nfs/proc.c index ac40b8535d7..f48125da198 100644 --- a/fs/nfs/proc.c +++ b/fs/nfs/proc.c @@ -710,6 +710,7 @@ const struct nfs_rpc_ops nfs_v2_clientops = { .dentry_ops = &nfs_dentry_operations, .dir_inode_ops = &nfs_dir_inode_operations, .file_inode_ops = &nfs_file_inode_operations, + .file_ops = &nfs_file_operations, .getroot = nfs_proc_get_root, .getattr = nfs_proc_getattr, .setattr = nfs_proc_setattr, diff --git a/include/linux/nfs_fs.h b/include/linux/nfs_fs.h index ab2c6343361..92ecf5585fa 100644 --- a/include/linux/nfs_fs.h +++ b/include/linux/nfs_fs.h @@ -410,6 +410,9 @@ extern const struct inode_operations nfs_file_inode_operations; extern const struct inode_operations nfs3_file_inode_operations; #endif /* CONFIG_NFS_V3 */ extern const struct file_operations nfs_file_operations; +#ifdef CONFIG_NFS_V4 +extern const struct file_operations nfs4_file_operations; +#endif /* CONFIG_NFS_V4 */ extern const struct address_space_operations nfs_file_aops; extern const struct address_space_operations nfs_dir_aops; diff --git a/include/linux/nfs_xdr.h b/include/linux/nfs_xdr.h index c74595ba709..2a7c533be5d 100644 --- a/include/linux/nfs_xdr.h +++ b/include/linux/nfs_xdr.h @@ -1192,6 +1192,7 @@ struct nfs_rpc_ops { const struct dentry_operations *dentry_ops; const struct inode_operations *dir_inode_ops; const struct inode_operations *file_inode_ops; + const struct file_operations *file_ops; int (*getroot) (struct nfs_server *, struct nfs_fh *, struct nfs_fsinfo *); -- cgit v1.2.3-70-g09d2 From 0486958f57a496212e3c1e3d9194deebba3dc3d4 Mon Sep 17 00:00:00 2001 From: Jeff Layton Date: Fri, 4 Nov 2011 13:31:22 -0400 Subject: nfs: move nfs_file_operations declaration to bottom of file.c (try #2) ...a remove a set of forward declarations. Signed-off-by: Jeff Layton Signed-off-by: Trond Myklebust --- fs/nfs/file.c | 59 +++++++++++++++++++---------------------------------------- 1 file changed, 19 insertions(+), 40 deletions(-) (limited to 'fs') diff --git a/fs/nfs/file.c b/fs/nfs/file.c index 6d93e0754b5..eca56d4b39c 100644 --- a/fs/nfs/file.c +++ b/fs/nfs/file.c @@ -40,48 +40,8 @@ #define NFSDBG_FACILITY NFSDBG_FILE -static int nfs_file_open(struct inode *, struct file *); -static int nfs_file_release(struct inode *, struct file *); -static loff_t nfs_file_llseek(struct file *file, loff_t offset, int origin); -static int nfs_file_mmap(struct file *, struct vm_area_struct *); -static ssize_t nfs_file_splice_read(struct file *filp, loff_t *ppos, - struct pipe_inode_info *pipe, - size_t count, unsigned int flags); -static ssize_t nfs_file_read(struct kiocb *, const struct iovec *iov, - unsigned long nr_segs, loff_t pos); -static ssize_t nfs_file_splice_write(struct pipe_inode_info *pipe, - struct file *filp, loff_t *ppos, - size_t count, unsigned int flags); -static ssize_t nfs_file_write(struct kiocb *, const struct iovec *iov, - unsigned long nr_segs, loff_t pos); -static int nfs_file_flush(struct file *, fl_owner_t id); -static int nfs_file_fsync(struct file *, loff_t, loff_t, int datasync); -static int nfs_check_flags(int flags); -static int nfs_lock(struct file *filp, int cmd, struct file_lock *fl); -static int nfs_flock(struct file *filp, int cmd, struct file_lock *fl); -static int nfs_setlease(struct file *file, long arg, struct file_lock **fl); - static const struct vm_operations_struct nfs_file_vm_ops; -const struct file_operations nfs_file_operations = { - .llseek = nfs_file_llseek, - .read = do_sync_read, - .write = do_sync_write, - .aio_read = nfs_file_read, - .aio_write = nfs_file_write, - .mmap = nfs_file_mmap, - .open = nfs_file_open, - .flush = nfs_file_flush, - .release = nfs_file_release, - .fsync = nfs_file_fsync, - .lock = nfs_lock, - .flock = nfs_flock, - .splice_read = nfs_file_splice_read, - .splice_write = nfs_file_splice_write, - .check_flags = nfs_check_flags, - .setlease = nfs_setlease, -}; - const struct inode_operations nfs_file_inode_operations = { .permission = nfs_permission, .getattr = nfs_getattr, @@ -887,6 +847,25 @@ static int nfs_setlease(struct file *file, long arg, struct file_lock **fl) return -EINVAL; } +const struct file_operations nfs_file_operations = { + .llseek = nfs_file_llseek, + .read = do_sync_read, + .write = do_sync_write, + .aio_read = nfs_file_read, + .aio_write = nfs_file_write, + .mmap = nfs_file_mmap, + .open = nfs_file_open, + .flush = nfs_file_flush, + .release = nfs_file_release, + .fsync = nfs_file_fsync, + .lock = nfs_lock, + .flock = nfs_flock, + .splice_read = nfs_file_splice_read, + .splice_write = nfs_file_splice_write, + .check_flags = nfs_check_flags, + .setlease = nfs_setlease, +}; + #ifdef CONFIG_NFS_V4 static int nfs4_file_open(struct inode *inode, struct file *filp) -- cgit v1.2.3-70-g09d2 From 15a2015fbc692e1c97d7ce12d96e077f5ae7ea6d Mon Sep 17 00:00:00 2001 From: Sage Weil Date: Sat, 5 Nov 2011 22:06:31 -0700 Subject: ceph: fix iput race when queueing inode work If we queue a work item that calls iput(), make sure we ihold() before attempting to queue work. Otherwise our queued work might miraculously run before we notice the queue_work() succeeded and call ihold(), allowing the inode to be destroyed. That is, instead of if (queue_work(...)) ihold(); we need to do ihold(); if (!queue_work(...)) iput(); Reported-by: Amon Ott Signed-off-by: Sage Weil --- fs/ceph/inode.c | 9 ++++++--- 1 file changed, 6 insertions(+), 3 deletions(-) (limited to 'fs') diff --git a/fs/ceph/inode.c b/fs/ceph/inode.c index e392bfce84a..116f36502f1 100644 --- a/fs/ceph/inode.c +++ b/fs/ceph/inode.c @@ -1328,12 +1328,13 @@ int ceph_inode_set_size(struct inode *inode, loff_t size) */ void ceph_queue_writeback(struct inode *inode) { + ihold(inode); if (queue_work(ceph_inode_to_client(inode)->wb_wq, &ceph_inode(inode)->i_wb_work)) { dout("ceph_queue_writeback %p\n", inode); - ihold(inode); } else { dout("ceph_queue_writeback %p failed\n", inode); + iput(inode); } } @@ -1353,12 +1354,13 @@ static void ceph_writeback_work(struct work_struct *work) */ void ceph_queue_invalidate(struct inode *inode) { + ihold(inode); if (queue_work(ceph_inode_to_client(inode)->pg_inv_wq, &ceph_inode(inode)->i_pg_inv_work)) { dout("ceph_queue_invalidate %p\n", inode); - ihold(inode); } else { dout("ceph_queue_invalidate %p failed\n", inode); + iput(inode); } } @@ -1434,13 +1436,14 @@ void ceph_queue_vmtruncate(struct inode *inode) { struct ceph_inode_info *ci = ceph_inode(inode); + ihold(inode); if (queue_work(ceph_sb_to_client(inode->i_sb)->trunc_wq, &ci->i_vmtruncate_work)) { dout("ceph_queue_vmtruncate %p\n", inode); - ihold(inode); } else { dout("ceph_queue_vmtruncate %p failed, pending=%d\n", inode, ci->i_truncate_pending); + iput(inode); } } -- cgit v1.2.3-70-g09d2 From eb513689c97e3e73bb9b4459d490a8e894b4a546 Mon Sep 17 00:00:00 2001 From: Theodore Ts'o Date: Mon, 7 Nov 2011 10:47:42 -0500 Subject: ext4: ignore journalled data options on remount if fs has no journal This avoids a confusing failure in the init scripts when the /etc/fstab has data=writeback or data=journal but the file system does not have a journal. So check for this case explicitly, and warn the user that we are ignoring the (pointless, since they have no journal) data=* mount option. Signed-off-by: "Theodore Ts'o" --- fs/ext4/super.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/ext4/super.c b/fs/ext4/super.c index 9953d80145a..0435013dbd4 100644 --- a/fs/ext4/super.c +++ b/fs/ext4/super.c @@ -1683,7 +1683,9 @@ static int parse_options(char *options, struct super_block *sb, data_opt = EXT4_MOUNT_WRITEBACK_DATA; datacheck: if (is_remount) { - if (test_opt(sb, DATA_FLAGS) != data_opt) { + if (!sbi->s_journal) + ext4_msg(sb, KERN_WARNING, "Remounting file system with no journal so ignoring journalled data option"); + else if (test_opt(sb, DATA_FLAGS) != data_opt) { ext4_msg(sb, KERN_ERR, "Cannot change data mode on remount"); return 0; -- cgit v1.2.3-70-g09d2 From 2397256d6218e7bf5147a3b01dcc6aec20fd3916 Mon Sep 17 00:00:00 2001 From: Richard Weinberger Date: Mon, 7 Nov 2011 10:50:09 -0500 Subject: ext4: Remove kernel_lock annotations The BKL is gone, these annotations are useless. Signed-off-by: Richard Weinberger Signed-off-by: "Theodore Ts'o" --- fs/ext4/super.c | 2 -- 1 file changed, 2 deletions(-) (limited to 'fs') diff --git a/fs/ext4/super.c b/fs/ext4/super.c index 0435013dbd4..3858767ec67 100644 --- a/fs/ext4/super.c +++ b/fs/ext4/super.c @@ -3101,8 +3101,6 @@ static void ext4_destroy_lazyinit_thread(void) } static int ext4_fill_super(struct super_block *sb, void *data, int silent) - __releases(kernel_lock) - __acquires(kernel_lock) { char *orig_data = kstrdup(data, GFP_KERNEL); struct buffer_head *bh; -- cgit v1.2.3-70-g09d2 From 3c1fcb2c24519febd7ca9cf292fa5bdf513b601f Mon Sep 17 00:00:00 2001 From: Namjae Jeon Date: Mon, 7 Nov 2011 11:01:13 -0500 Subject: ext4: add blk_finish_plug in error case of writepages. blk_finish_plug is needed in error case of writepages. Signed-off-by: Namjae Jeon Signed-off-by: "Theodore Ts'o" --- fs/ext4/inode.c | 1 + 1 file changed, 1 insertion(+) (limited to 'fs') diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c index cc5a6da030a..18b262bb6a4 100644 --- a/fs/ext4/inode.c +++ b/fs/ext4/inode.c @@ -2270,6 +2270,7 @@ retry: ext4_msg(inode->i_sb, KERN_CRIT, "%s: jbd2_start: " "%ld pages, ino %lu; err %d", __func__, wbc->nr_to_write, inode->i_ino, ret); + blk_finish_plug(&plug); goto out_writepages; } -- cgit v1.2.3-70-g09d2 From a6f498a891c730327645a7afa10c5ae977de6fd8 Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Mon, 7 Nov 2011 15:50:32 -0500 Subject: NFS: Fix a regression in the referral code Fix a regression that was introduced by commit 0c2e53f11a6dae9e3af5f50f5ad0382e7c3e0cfa (NFS: Remove the unused "lookupfh()" version of nfs4_proc_lookup()). In the case where the lookup gets an NFS4ERR_MOVED, we want to return the result of nfs4_get_referral(). Instead, that value is getting clobbered by the call to nfs4_handle_exception()... Reported-by: Tigran Mkrtchyan Tested-by: Tigran Mkrtchyan Signed-off-by: Trond Myklebust --- fs/nfs/nfs4proc.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) (limited to 'fs') diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c index 069cb8094d4..be2bbac1381 100644 --- a/fs/nfs/nfs4proc.c +++ b/fs/nfs/nfs4proc.c @@ -2464,8 +2464,7 @@ static int nfs4_proc_lookup(struct rpc_clnt *clnt, struct inode *dir, struct qst case -NFS4ERR_BADNAME: return -ENOENT; case -NFS4ERR_MOVED: - err = nfs4_get_referral(dir, name, fattr, fhandle); - break; + return nfs4_get_referral(dir, name, fattr, fhandle); case -NFS4ERR_WRONGSEC: nfs_fixup_secinfo_attributes(fattr, fhandle); } -- cgit v1.2.3-70-g09d2 From 62e4a76987eab2b7fa952546614bc83e5bfc9d3e Mon Sep 17 00:00:00 2001 From: Trond Myklebust Date: Thu, 10 Nov 2011 14:30:37 -0500 Subject: NFS: Revert pnfs ugliness from the generic NFS read code path pNFS-specific code belongs in the pnfs layer. It should not be hijacking generic NFS read or write code paths. Signed-off-by: Trond Myklebust --- fs/nfs/internal.h | 2 ++ fs/nfs/pnfs.c | 26 +++++++++++++++++++++----- fs/nfs/read.c | 14 ++------------ 3 files changed, 25 insertions(+), 17 deletions(-) (limited to 'fs') diff --git a/fs/nfs/internal.h b/fs/nfs/internal.h index c1a1bd8ddf1..3f4d95751d5 100644 --- a/fs/nfs/internal.h +++ b/fs/nfs/internal.h @@ -299,6 +299,8 @@ extern void nfs_read_prepare(struct rpc_task *task, void *calldata); extern int nfs_generic_pagein(struct nfs_pageio_descriptor *desc, struct list_head *head); +extern void nfs_pageio_init_read_mds(struct nfs_pageio_descriptor *pgio, + struct inode *inode); extern void nfs_pageio_reset_read_mds(struct nfs_pageio_descriptor *pgio); extern void nfs_readdata_release(struct nfs_read_data *rdata); diff --git a/fs/nfs/pnfs.c b/fs/nfs/pnfs.c index a2478bc7444..e40a3ca22ea 100644 --- a/fs/nfs/pnfs.c +++ b/fs/nfs/pnfs.c @@ -1259,6 +1259,25 @@ pnfs_generic_pg_writepages(struct nfs_pageio_descriptor *desc) } EXPORT_SYMBOL_GPL(pnfs_generic_pg_writepages); +static void pnfs_ld_handle_read_error(struct nfs_read_data *data) +{ + struct nfs_pageio_descriptor pgio; + + put_lseg(data->lseg); + data->lseg = NULL; + dprintk("pnfs write error = %d\n", data->pnfs_error); + + nfs_pageio_init_read_mds(&pgio, data->inode); + + while (!list_empty(&data->pages)) { + struct nfs_page *req = nfs_list_entry(data->pages.next); + + nfs_list_remove_request(req); + nfs_pageio_add_request(&pgio, req); + } + nfs_pageio_complete(&pgio); +} + /* * Called by non rpc-based layout drivers */ @@ -1267,11 +1286,8 @@ void pnfs_ld_read_done(struct nfs_read_data *data) if (likely(!data->pnfs_error)) { __nfs4_read_done_cb(data); data->mds_ops->rpc_call_done(&data->task, data); - } else { - put_lseg(data->lseg); - data->lseg = NULL; - dprintk("pnfs write error = %d\n", data->pnfs_error); - } + } else + pnfs_ld_handle_read_error(data); data->mds_ops->rpc_release(data); } EXPORT_SYMBOL_GPL(pnfs_ld_read_done); diff --git a/fs/nfs/read.c b/fs/nfs/read.c index 8b48ec63f72..cfa175c223d 100644 --- a/fs/nfs/read.c +++ b/fs/nfs/read.c @@ -109,7 +109,7 @@ static void nfs_readpage_truncate_uninitialised_page(struct nfs_read_data *data) } } -static void nfs_pageio_init_read_mds(struct nfs_pageio_descriptor *pgio, +void nfs_pageio_init_read_mds(struct nfs_pageio_descriptor *pgio, struct inode *inode) { nfs_pageio_init(pgio, inode, &nfs_pageio_read_ops, @@ -534,23 +534,13 @@ static void nfs_readpage_result_full(struct rpc_task *task, void *calldata) static void nfs_readpage_release_full(void *calldata) { struct nfs_read_data *data = calldata; - struct nfs_pageio_descriptor pgio; - if (data->pnfs_error) { - nfs_pageio_init_read_mds(&pgio, data->inode); - pgio.pg_recoalesce = 1; - } while (!list_empty(&data->pages)) { struct nfs_page *req = nfs_list_entry(data->pages.next); nfs_list_remove_request(req); - if (!data->pnfs_error) - nfs_readpage_release(req); - else - nfs_pageio_add_request(&pgio, req); + nfs_readpage_release(req); } - if (data->pnfs_error) - nfs_pageio_complete(&pgio); nfs_readdata_release(calldata); } -- cgit v1.2.3-70-g09d2 From 774ac21da76f5c3018428725074e27a3fd40b128 Mon Sep 17 00:00:00 2001 From: Sage Weil Date: Fri, 11 Nov 2011 09:48:08 -0800 Subject: ceph: initialize root dentry Set up d_fsdata on the root dentry. This fixes a NULL pointer dereference in ceph_d_prune on umount. It also means we can eventually strip out all of the conditional checks on d_fsdata because it is now set unconditionally (prior to setting up the d_ops). Fix the ceph_d_prune debug print while we're here. Signed-off-by: Sage Weil --- fs/ceph/dir.c | 2 +- fs/ceph/super.c | 6 ++++-- 2 files changed, 5 insertions(+), 3 deletions(-) (limited to 'fs') diff --git a/fs/ceph/dir.c b/fs/ceph/dir.c index 2abd0dfad7f..bca3948e9db 100644 --- a/fs/ceph/dir.c +++ b/fs/ceph/dir.c @@ -1143,7 +1143,7 @@ static void ceph_d_prune(struct dentry *dentry) { struct ceph_dentry_info *di; - dout("d_release %p\n", dentry); + dout("ceph_d_prune %p\n", dentry); /* do we have a valid parent? */ if (!dentry->d_parent || IS_ROOT(dentry)) diff --git a/fs/ceph/super.c b/fs/ceph/super.c index a90846fac75..8dc73a594a9 100644 --- a/fs/ceph/super.c +++ b/fs/ceph/super.c @@ -638,10 +638,12 @@ static struct dentry *open_root_dentry(struct ceph_fs_client *fsc, if (err == 0) { dout("open_root_inode success\n"); if (ceph_ino(req->r_target_inode) == CEPH_INO_ROOT && - fsc->sb->s_root == NULL) + fsc->sb->s_root == NULL) { root = d_alloc_root(req->r_target_inode); - else + ceph_init_dentry(root); + } else { root = d_obtain_alias(req->r_target_inode); + } req->r_target_inode = NULL; dout("open_root_inode success, root dentry is %p\n", root); } else { -- cgit v1.2.3-70-g09d2 From f1ebcc74d5b2159f44c96b479b6eb8afc7829095 Mon Sep 17 00:00:00 2001 From: Liu Bo Date: Mon, 14 Nov 2011 20:48:06 -0500 Subject: Btrfs: fix tree corruption after multi-thread snapshots and inode_cache flush The btrfs snapshotting code requires that once a root has been snapshotted, we don't change it during a commit. But there are two cases to lead to tree corruptions: 1) multi-thread snapshots can commit serveral snapshots in a transaction, and this may change the src root when processing the following pending snapshots, which lead to the former snapshots corruptions; 2) the free inode cache was changing the roots when it root the cache, which lead to corruptions. This fixes things by making sure we force COW the block after we create a snapshot during commiting a transaction, then any changes to the roots will result in COW, and we get all the fs roots and snapshot roots to be consistent. Signed-off-by: Liu Bo Signed-off-by: Miao Xie Signed-off-by: Chris Mason --- fs/btrfs/ctree.c | 17 ++++++++++++++++- fs/btrfs/ctree.h | 2 ++ fs/btrfs/transaction.c | 8 ++++++++ 3 files changed, 26 insertions(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/btrfs/ctree.c b/fs/btrfs/ctree.c index 0fe615e4ea3..dede441bdee 100644 --- a/fs/btrfs/ctree.c +++ b/fs/btrfs/ctree.c @@ -514,10 +514,25 @@ static inline int should_cow_block(struct btrfs_trans_handle *trans, struct btrfs_root *root, struct extent_buffer *buf) { + /* ensure we can see the force_cow */ + smp_rmb(); + + /* + * We do not need to cow a block if + * 1) this block is not created or changed in this transaction; + * 2) this block does not belong to TREE_RELOC tree; + * 3) the root is not forced COW. + * + * What is forced COW: + * when we create snapshot during commiting the transaction, + * after we've finished coping src root, we must COW the shared + * block to ensure the metadata consistency. + */ if (btrfs_header_generation(buf) == trans->transid && !btrfs_header_flag(buf, BTRFS_HEADER_FLAG_WRITTEN) && !(root->root_key.objectid != BTRFS_TREE_RELOC_OBJECTID && - btrfs_header_flag(buf, BTRFS_HEADER_FLAG_RELOC))) + btrfs_header_flag(buf, BTRFS_HEADER_FLAG_RELOC)) && + !root->force_cow) return 0; return 1; } diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h index b9ba59ff929..b1cb3c05248 100644 --- a/fs/btrfs/ctree.h +++ b/fs/btrfs/ctree.h @@ -1271,6 +1271,8 @@ struct btrfs_root { * for stat. It may be used for more later */ dev_t anon_dev; + + int force_cow; }; struct btrfs_ioctl_defrag_range_args { diff --git a/fs/btrfs/transaction.c b/fs/btrfs/transaction.c index 6a0574e923b..81376d94cd3 100644 --- a/fs/btrfs/transaction.c +++ b/fs/btrfs/transaction.c @@ -785,6 +785,10 @@ static noinline int commit_fs_roots(struct btrfs_trans_handle *trans, btrfs_save_ino_cache(root, trans); + /* see comments in should_cow_block() */ + root->force_cow = 0; + smp_wmb(); + if (root->commit_root != root->node) { mutex_lock(&root->fs_commit_mutex); switch_commit_root(root); @@ -947,6 +951,10 @@ static noinline int create_pending_snapshot(struct btrfs_trans_handle *trans, btrfs_tree_unlock(old); free_extent_buffer(old); + /* see comments in should_cow_block() */ + root->force_cow = 1; + smp_wmb(); + btrfs_set_root_node(new_root_item, tmp); /* record when the snapshot was created in key.offset */ key.offset = trans->transid; -- cgit v1.2.3-70-g09d2 From bc5b8a9003132ae44559edd63a1623b7b99dfb68 Mon Sep 17 00:00:00 2001 From: Dan Carpenter Date: Mon, 14 Nov 2011 17:52:08 +0300 Subject: hfs: add sanity check for file name length On a corrupted file system the ->len field could be wrong leading to a buffer overflow. Reported-and-acked-by: Clement LECIGNE Signed-off-by: Dan Carpenter Cc: stable@kernel.org Signed-off-by: Linus Torvalds --- fs/hfs/trans.c | 2 ++ 1 file changed, 2 insertions(+) (limited to 'fs') diff --git a/fs/hfs/trans.c b/fs/hfs/trans.c index e673a88b8ae..b1ce4c7ad3f 100644 --- a/fs/hfs/trans.c +++ b/fs/hfs/trans.c @@ -40,6 +40,8 @@ int hfs_mac2asc(struct super_block *sb, char *out, const struct hfs_name *in) src = in->name; srclen = in->len; + if (srclen > HFS_NAMELEN) + srclen = HFS_NAMELEN; dst = out; dstlen = HFS_MAX_NAMELEN; if (nls_io) { -- cgit v1.2.3-70-g09d2 From db3e74b582915d66e10b0c73a62763418f54c340 Mon Sep 17 00:00:00 2001 From: Mitsuo Hayasaka Date: Thu, 10 Nov 2011 01:33:10 +0000 Subject: xfs: use doalloc flag in xfs_qm_dqattach_one() The doalloc arg in xfs_qm_dqattach_one() is a flag that indicates whether a new area to handle quota information will be allocated if needed. Originally, it was passed to xfs_qm_dqget(), but has been removed by the following commit (probably by mistake): commit 8e9b6e7fa4544ea8a0e030c8987b918509c8ff47 Author: Christoph Hellwig Date: Sun Feb 8 21:51:42 2009 +0100 xfs: remove the unused XFS_QMOPT_DQLOCK flag As the result, xfs_qm_dqget() called from xfs_qm_dqattach_one() never allocates the new area even if it is needed. This patch gives the doalloc arg to xfs_qm_dqget() in xfs_qm_dqattach_one() to fix this problem. Signed-off-by: Mitsuo Hayasaka Cc: Alex Elder Cc: Christoph Hellwig Reviewed-by: Christoph Hellwig Signed-off-by: Ben Myers --- fs/xfs/xfs_qm.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/xfs/xfs_qm.c b/fs/xfs/xfs_qm.c index 5cff443f6cd..0bbb1a41998 100644 --- a/fs/xfs/xfs_qm.c +++ b/fs/xfs/xfs_qm.c @@ -674,7 +674,8 @@ xfs_qm_dqattach_one( * disk and we didn't ask it to allocate; * ESRCH if quotas got turned off suddenly. */ - error = xfs_qm_dqget(ip->i_mount, ip, id, type, XFS_QMOPT_DOWARN, &dqp); + error = xfs_qm_dqget(ip->i_mount, ip, id, type, + doalloc | XFS_QMOPT_DOWARN, &dqp); if (error) return error; -- cgit v1.2.3-70-g09d2 From 121f099412bd6576dfb3d94222e89d9341362177 Mon Sep 17 00:00:00 2001 From: Dan Carpenter Date: Wed, 16 Nov 2011 09:21:50 +0100 Subject: bio: change some signed vars to unsigned This is just a cleanup patch to silence a static checker warning. The problem is that we cap "nr_iovecs" so it can't be larger than "UIO_MAXIOV" but we don't check for negative values. It turns out this is prevented at other layers, but logically it doesn't make sense to have negative nr_iovecs so making it unsigned is nicer. Signed-off-by: Dan Carpenter Signed-off-by: Andrew Morton Signed-off-by: Jens Axboe --- fs/bio.c | 7 ++++--- include/linux/bio.h | 4 ++-- 2 files changed, 6 insertions(+), 5 deletions(-) (limited to 'fs') diff --git a/fs/bio.c b/fs/bio.c index 41c93c72224..b1fe82cf88c 100644 --- a/fs/bio.c +++ b/fs/bio.c @@ -337,7 +337,7 @@ static void bio_fs_destructor(struct bio *bio) * RETURNS: * Pointer to new bio on success, NULL on failure. */ -struct bio *bio_alloc(gfp_t gfp_mask, int nr_iovecs) +struct bio *bio_alloc(gfp_t gfp_mask, unsigned int nr_iovecs) { struct bio *bio = bio_alloc_bioset(gfp_mask, nr_iovecs, fs_bio_set); @@ -365,7 +365,7 @@ static void bio_kmalloc_destructor(struct bio *bio) * %__GFP_WAIT, the allocation is guaranteed to succeed. * **/ -struct bio *bio_kmalloc(gfp_t gfp_mask, int nr_iovecs) +struct bio *bio_kmalloc(gfp_t gfp_mask, unsigned int nr_iovecs) { struct bio *bio; @@ -696,7 +696,8 @@ static void bio_free_map_data(struct bio_map_data *bmd) kfree(bmd); } -static struct bio_map_data *bio_alloc_map_data(int nr_segs, int iov_count, +static struct bio_map_data *bio_alloc_map_data(int nr_segs, + unsigned int iov_count, gfp_t gfp_mask) { struct bio_map_data *bmd; diff --git a/include/linux/bio.h b/include/linux/bio.h index d2a3cc23d82..847994aef0e 100644 --- a/include/linux/bio.h +++ b/include/linux/bio.h @@ -211,8 +211,8 @@ extern void bio_pair_release(struct bio_pair *dbio); extern struct bio_set *bioset_create(unsigned int, unsigned int); extern void bioset_free(struct bio_set *); -extern struct bio *bio_alloc(gfp_t, int); -extern struct bio *bio_kmalloc(gfp_t, int); +extern struct bio *bio_alloc(gfp_t, unsigned int); +extern struct bio *bio_kmalloc(gfp_t, unsigned int); extern struct bio *bio_alloc_bioset(gfp_t, int, struct bio_set *); extern void bio_put(struct bio *); extern void bio_free(struct bio *, struct bio_set *); -- cgit v1.2.3-70-g09d2 From 8d514bbf37eecf0a3e309284728637816a36764b Mon Sep 17 00:00:00 2001 From: Al Viro Date: Wed, 16 Nov 2011 16:06:09 -0500 Subject: btrfs: fix double mntput() in mount_subvol() Signed-off-by: Al Viro --- fs/btrfs/super.c | 1 - 1 file changed, 1 deletion(-) (limited to 'fs') diff --git a/fs/btrfs/super.c b/fs/btrfs/super.c index 8bd9d6d0e07..969a7747e88 100644 --- a/fs/btrfs/super.c +++ b/fs/btrfs/super.c @@ -861,7 +861,6 @@ static struct dentry *mount_subvol(const char *subvol_name, int flags, if (!is_subvolume_inode(path.dentry->d_inode)) { path_put(&path); - mntput(mnt); error = -EINVAL; printk(KERN_ERR "btrfs: '%s' is not a valid subvolume\n", subvol_name); -- cgit v1.2.3-70-g09d2 From c13344958780b4046305ee6235d686c846535529 Mon Sep 17 00:00:00 2001 From: Al Viro Date: Wed, 16 Nov 2011 16:12:14 -0500 Subject: switch create_mnt_ns() to saner calling conventions, fix double mntput() in nfs Life is much saner if create_mnt_ns(mnt) drops mnt in case of error... Switch it to such calling conventions, switch callers, fix double mntput() in fs/nfs/super.c one. Signed-off-by: Al Viro --- fs/btrfs/super.c | 4 +--- fs/namespace.c | 2 ++ fs/nfs/super.c | 23 ++++++++--------------- 3 files changed, 11 insertions(+), 18 deletions(-) (limited to 'fs') diff --git a/fs/btrfs/super.c b/fs/btrfs/super.c index 969a7747e88..cfbedd7755b 100644 --- a/fs/btrfs/super.c +++ b/fs/btrfs/super.c @@ -843,10 +843,8 @@ static struct dentry *mount_subvol(const char *subvol_name, int flags, return ERR_CAST(mnt); ns_private = create_mnt_ns(mnt); - if (IS_ERR(ns_private)) { - mntput(mnt); + if (IS_ERR(ns_private)) return ERR_CAST(ns_private); - } /* * This will trigger the automount of the subvol so we can just diff --git a/fs/namespace.c b/fs/namespace.c index e5e1c7d1839..aea4b768984 100644 --- a/fs/namespace.c +++ b/fs/namespace.c @@ -2483,6 +2483,8 @@ struct mnt_namespace *create_mnt_ns(struct vfsmount *mnt) __mnt_make_longterm(mnt); new_ns->root = mnt; list_add(&new_ns->list, &new_ns->root->mnt_list); + } else { + mntput(mnt); } return new_ns; } diff --git a/fs/nfs/super.c b/fs/nfs/super.c index 480b3b6bf71..46d69f38fd5 100644 --- a/fs/nfs/super.c +++ b/fs/nfs/super.c @@ -2794,22 +2794,21 @@ static struct dentry *nfs_follow_remote_path(struct vfsmount *root_mnt, int ret; ns_private = create_mnt_ns(root_mnt); - ret = PTR_ERR(ns_private); if (IS_ERR(ns_private)) - goto out_mntput; + return ERR_CAST(ns_private); ret = nfs_referral_loop_protect(); - if (ret != 0) - goto out_put_mnt_ns; - - ret = vfs_path_lookup(root_mnt->mnt_root, root_mnt, - export_path, LOOKUP_FOLLOW|LOOKUP_AUTOMOUNT, &path); + if (ret == 0) { + ret = vfs_path_lookup(root_mnt->mnt_root, root_mnt, + export_path, LOOKUP_FOLLOW|LOOKUP_AUTOMOUNT, + &path); + nfs_referral_loop_unprotect(); + } - nfs_referral_loop_unprotect(); put_mnt_ns(ns_private); if (ret != 0) - goto out_err; + return ERR_PTR(ret); s = path.mnt->mnt_sb; atomic_inc(&s->s_active); @@ -2818,12 +2817,6 @@ static struct dentry *nfs_follow_remote_path(struct vfsmount *root_mnt, path_put(&path); down_write(&s->s_umount); return dentry; -out_put_mnt_ns: - put_mnt_ns(ns_private); -out_mntput: - mntput(root_mnt); -out_err: - return ERR_PTR(ret); } static struct dentry *nfs4_try_mount(int flags, const char *dev_name, -- cgit v1.2.3-70-g09d2 From ea441d1104cf1efb471fa81bc91e9fd1e6ae29fd Mon Sep 17 00:00:00 2001 From: Al Viro Date: Wed, 16 Nov 2011 21:43:59 -0500 Subject: new helper: mount_subtree() takes vfsmount and relative path, does lookup within that vfsmount (possibly triggering automounts) and returns the result as root of subtree suitable for return by ->mount() (i.e. a reference to dentry and an active reference to its superblock grabbed, superblock locked exclusive). btrfs and nfs switched to it instead of open-coding the sucker. Signed-off-by: Al Viro --- fs/btrfs/super.c | 35 ++++++----------------------------- fs/namespace.c | 28 ++++++++++++++++++++++++++++ fs/nfs/super.c | 30 ++++++------------------------ include/linux/fs.h | 1 + 4 files changed, 41 insertions(+), 53 deletions(-) (limited to 'fs') diff --git a/fs/btrfs/super.c b/fs/btrfs/super.c index cfbedd7755b..17ee7fc5e64 100644 --- a/fs/btrfs/super.c +++ b/fs/btrfs/super.c @@ -825,13 +825,9 @@ static char *setup_root_args(char *args) static struct dentry *mount_subvol(const char *subvol_name, int flags, const char *device_name, char *data) { - struct super_block *s; struct dentry *root; struct vfsmount *mnt; - struct mnt_namespace *ns_private; char *newargs; - struct path path; - int error; newargs = setup_root_args(data); if (!newargs) @@ -842,36 +838,17 @@ static struct dentry *mount_subvol(const char *subvol_name, int flags, if (IS_ERR(mnt)) return ERR_CAST(mnt); - ns_private = create_mnt_ns(mnt); - if (IS_ERR(ns_private)) - return ERR_CAST(ns_private); - - /* - * This will trigger the automount of the subvol so we can just - * drop the mnt we have here and return the dentry that we - * found. - */ - error = vfs_path_lookup(mnt->mnt_root, mnt, subvol_name, - LOOKUP_FOLLOW, &path); - put_mnt_ns(ns_private); - if (error) - return ERR_PTR(error); + root = mount_subtree(mnt, subvol_name); - if (!is_subvolume_inode(path.dentry->d_inode)) { - path_put(&path); - error = -EINVAL; + if (!IS_ERR(root) && !is_subvolume_inode(root->d_inode)) { + struct super_block *s = root->d_sb; + dput(root); + root = ERR_PTR(-EINVAL); + deactivate_locked_super(s); printk(KERN_ERR "btrfs: '%s' is not a valid subvolume\n", subvol_name); - return ERR_PTR(-EINVAL); } - /* Get a ref to the sb and the dentry we found and return it */ - s = path.mnt->mnt_sb; - atomic_inc(&s->s_active); - root = dget(path.dentry); - path_put(&path); - down_write(&s->s_umount); - return root; } diff --git a/fs/namespace.c b/fs/namespace.c index aea4b768984..50ee30345b4 100644 --- a/fs/namespace.c +++ b/fs/namespace.c @@ -2490,6 +2490,34 @@ struct mnt_namespace *create_mnt_ns(struct vfsmount *mnt) } EXPORT_SYMBOL(create_mnt_ns); +struct dentry *mount_subtree(struct vfsmount *mnt, const char *name) +{ + struct mnt_namespace *ns; + struct path path; + int err; + + ns = create_mnt_ns(mnt); + if (IS_ERR(ns)) + return ERR_CAST(ns); + + err = vfs_path_lookup(mnt->mnt_root, mnt, + name, LOOKUP_FOLLOW|LOOKUP_AUTOMOUNT, &path); + + put_mnt_ns(ns); + + if (err) + return ERR_PTR(err); + + /* trade a vfsmount reference for active sb one */ + atomic_inc(&path.mnt->mnt_sb->s_active); + mntput(path.mnt); + /* lock the sucker */ + down_write(&path.mnt->mnt_sb->s_umount); + /* ... and return the root of (sub)tree on it */ + return path.dentry; +} +EXPORT_SYMBOL(mount_subtree); + SYSCALL_DEFINE5(mount, char __user *, dev_name, char __user *, dir_name, char __user *, type, unsigned long, flags, void __user *, data) { diff --git a/fs/nfs/super.c b/fs/nfs/super.c index 46d69f38fd5..134777406ee 100644 --- a/fs/nfs/super.c +++ b/fs/nfs/super.c @@ -2787,35 +2787,17 @@ static void nfs_referral_loop_unprotect(void) static struct dentry *nfs_follow_remote_path(struct vfsmount *root_mnt, const char *export_path) { - struct mnt_namespace *ns_private; - struct super_block *s; struct dentry *dentry; - struct path path; - int ret; + int ret = nfs_referral_loop_protect(); - ns_private = create_mnt_ns(root_mnt); - if (IS_ERR(ns_private)) - return ERR_CAST(ns_private); - - ret = nfs_referral_loop_protect(); - if (ret == 0) { - ret = vfs_path_lookup(root_mnt->mnt_root, root_mnt, - export_path, LOOKUP_FOLLOW|LOOKUP_AUTOMOUNT, - &path); - nfs_referral_loop_unprotect(); - } - - put_mnt_ns(ns_private); - - if (ret != 0) + if (ret) { + mntput(root_mnt); return ERR_PTR(ret); + } - s = path.mnt->mnt_sb; - atomic_inc(&s->s_active); - dentry = dget(path.dentry); + dentry = mount_subtree(root_mnt, export_path); + nfs_referral_loop_unprotect(); - path_put(&path); - down_write(&s->s_umount); return dentry; } diff --git a/include/linux/fs.h b/include/linux/fs.h index 0c4df261af7..e3130220ce3 100644 --- a/include/linux/fs.h +++ b/include/linux/fs.h @@ -1886,6 +1886,7 @@ extern struct dentry *mount_single(struct file_system_type *fs_type, extern struct dentry *mount_nodev(struct file_system_type *fs_type, int flags, void *data, int (*fill_super)(struct super_block *, void *, int)); +extern struct dentry *mount_subtree(struct vfsmount *mnt, const char *path); void generic_shutdown_super(struct super_block *sb); void kill_block_super(struct super_block *sb); void kill_anon_super(struct super_block *sb); -- cgit v1.2.3-70-g09d2 From fc9f899483435935c1cd7005df29681929d1c99b Mon Sep 17 00:00:00 2001 From: Julia Lawall Date: Sat, 9 Jul 2011 18:04:39 +0200 Subject: fs/ocfs2/dlm/dlmlock.c: free kmem_cache_zalloc'd data using kmem_cache_free Memory allocated using kmem_cache_zalloc should be freed using kmem_cache_free, not kfree. The semantic patch that fixes this problem is as follows: (http://coccinelle.lip6.fr/) // @@ expression x,e,e1,e2; @@ x = kmem_cache_zalloc(e1,e2) ... when != x = e ?-kfree(x) +kmem_cache_free(e1,x) // Signed-off-by: Julia Lawall Signed-off-by: Joel Becker --- fs/ocfs2/dlm/dlmlock.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/ocfs2/dlm/dlmlock.c b/fs/ocfs2/dlm/dlmlock.c index f32fcba0492..975810b9849 100644 --- a/fs/ocfs2/dlm/dlmlock.c +++ b/fs/ocfs2/dlm/dlmlock.c @@ -438,7 +438,7 @@ struct dlm_lock * dlm_new_lock(int type, u8 node, u64 cookie, /* zero memory only if kernel-allocated */ lksb = kzalloc(sizeof(*lksb), GFP_NOFS); if (!lksb) { - kfree(lock); + kmem_cache_free(dlm_lock_cache, lock); return NULL; } kernel_allocated = 1; -- cgit v1.2.3-70-g09d2 From 8298524803339a9a8df053ebdfebc2975ec55be9 Mon Sep 17 00:00:00 2001 From: Wengang Wang Date: Tue, 12 Jul 2011 16:43:14 +0800 Subject: ocfs2: make direntry invalid when deleting it When we deleting a direntry from a directory, if it's the first in a block we invalid it by setting inode to 0; otherwise, we merge the deleted one to the prior and contiguous direntry. And we don't truncate directories. There is a problem for the later case since inode is not set to 0. This problem happens when the caller passes a file position as parameter to ocfs2_dir_foreach_blk(). If the position happens to point to a stale(not the first, deleted in betweens of ocfs2_dir_foreach_blk()s) direntry, we are not able to recognize its staleness. So that we treat it as a live one wrongly. The fix is to set inode to 0 in both cases indicating the direntry is stale. This won't introduce additional IOs. Signed-off-by: Wengang Wang Signed-off-by: Joel Becker --- fs/ocfs2/dir.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) (limited to 'fs') diff --git a/fs/ocfs2/dir.c b/fs/ocfs2/dir.c index 8582e3f4f12..3302088e1f0 100644 --- a/fs/ocfs2/dir.c +++ b/fs/ocfs2/dir.c @@ -1184,8 +1184,7 @@ static int __ocfs2_delete_entry(handle_t *handle, struct inode *dir, if (pde) le16_add_cpu(&pde->rec_len, le16_to_cpu(de->rec_len)); - else - de->inode = 0; + de->inode = 0; dir->i_version++; ocfs2_journal_dirty(handle, bh); goto bail; -- cgit v1.2.3-70-g09d2 From b8a0ae579fb8d9b21008ac386be08b9428902455 Mon Sep 17 00:00:00 2001 From: Wengang Wang Date: Wed, 12 Oct 2011 15:22:15 +0800 Subject: ocfs2: Commit transactions in error cases -v2 There are three cases found that in error cases, journal transactions are not committed nor aborted. We should take care of these case by committing the transactions. Otherwise, there would left a journal handle which will lead to , in same process context, the comming ocfs2_start_trans() gets wrong credits. Signed-off-by: Wengang Wang Signed-off-by: Joel Becker --- fs/ocfs2/alloc.c | 2 +- fs/ocfs2/ioctl.c | 3 ++- fs/ocfs2/xattr.c | 10 ++++++---- 3 files changed, 9 insertions(+), 6 deletions(-) (limited to 'fs') diff --git a/fs/ocfs2/alloc.c b/fs/ocfs2/alloc.c index ed553c60de8..3165aebb43c 100644 --- a/fs/ocfs2/alloc.c +++ b/fs/ocfs2/alloc.c @@ -5699,7 +5699,7 @@ int ocfs2_remove_btree_range(struct inode *inode, OCFS2_JOURNAL_ACCESS_WRITE); if (ret) { mlog_errno(ret); - goto out; + goto out_commit; } dquot_free_space_nodirty(inode, diff --git a/fs/ocfs2/ioctl.c b/fs/ocfs2/ioctl.c index d9a6ce77926..726ff265b29 100644 --- a/fs/ocfs2/ioctl.c +++ b/fs/ocfs2/ioctl.c @@ -122,7 +122,7 @@ static int ocfs2_set_inode_attr(struct inode *inode, unsigned flags, if ((oldflags & OCFS2_IMMUTABLE_FL) || ((flags ^ oldflags) & (OCFS2_APPEND_FL | OCFS2_IMMUTABLE_FL))) { if (!capable(CAP_LINUX_IMMUTABLE)) - goto bail_unlock; + goto bail_commit; } ocfs2_inode->ip_attr = flags; @@ -132,6 +132,7 @@ static int ocfs2_set_inode_attr(struct inode *inode, unsigned flags, if (status < 0) mlog_errno(status); +bail_commit: ocfs2_commit_trans(osb, handle); bail_unlock: ocfs2_inode_unlock(inode, 1); diff --git a/fs/ocfs2/xattr.c b/fs/ocfs2/xattr.c index 81ecf9c0bf0..93d6c80b392 100644 --- a/fs/ocfs2/xattr.c +++ b/fs/ocfs2/xattr.c @@ -2376,16 +2376,18 @@ static int ocfs2_remove_value_outside(struct inode*inode, } ret = ocfs2_xattr_value_truncate(inode, vb, 0, &ctxt); - if (ret < 0) { - mlog_errno(ret); - break; - } ocfs2_commit_trans(osb, ctxt.handle); if (ctxt.meta_ac) { ocfs2_free_alloc_context(ctxt.meta_ac); ctxt.meta_ac = NULL; } + + if (ret < 0) { + mlog_errno(ret); + break; + } + } if (ctxt.meta_ac) -- cgit v1.2.3-70-g09d2 From e41d33af7c037a6f7350f8410b45ce83b50d1b80 Mon Sep 17 00:00:00 2001 From: Dan Magenheimer Date: Mon, 17 Oct 2011 16:47:17 -0700 Subject: ocfs2: send correct UUID to cleancache initialization ocfs2: Fix cleancache initialization call to correctly pass uuid As reported by Steven Whitehouse in https://lkml.org/lkml/2011/5/27/221 the ocfs2 volume UUID is incorrectly passed to cleancache. As a result, shared-ephemeral tmem pools will not actually be created; instead they will be private (unshared) which misses out on a major benefit of tmem. Reported-by: Steven Whitehouse Signed-off-by: Dan Magenheimer Signed-off-by: Joel Becker --- fs/ocfs2/super.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/ocfs2/super.c b/fs/ocfs2/super.c index 938e2b2b0c9..4994f8b0e60 100644 --- a/fs/ocfs2/super.c +++ b/fs/ocfs2/super.c @@ -2360,7 +2360,7 @@ static int ocfs2_initialize_super(struct super_block *sb, mlog_errno(status); goto bail; } - cleancache_init_shared_fs((char *)&uuid_net_key, sb); + cleancache_init_shared_fs((char *)&di->id2.i_super.s_uuid, sb); bail: return status; -- cgit v1.2.3-70-g09d2 From 0393afea31874947b1d149b82d17b7dccac4f210 Mon Sep 17 00:00:00 2001 From: "Xiaowei.Hu" Date: Wed, 19 Oct 2011 09:34:19 +0800 Subject: ocfs2: Add a missing journal credit in ocfs2_link_credits() -v2 With indexed_dir enabled, ocfs2 maintains a list of dirblocks having space. The credit calculation in ocfs2_link_credits() did not correctly account for adding an entry that exactly fills a dirblock that triggers removing that dirblock by changing the pointer in the previous block in the list. The credit calculation did not account for that previous block. To expose, do: mkfs.ocfs2 -b 512 -M local /dev/sdX mount /dev/sdX /ocfs2 mkdir /ocfs2/linkdir touch /ocfs2/linkdir/file1 for i in `seq 1 29` ; do link /ocfs2/linkdir/file1 /ocfs2/linkdir/linklinklinklinklinklink$i; done rm -f /ocfs2/linkdir/linklinklinklinklinklink10 sleep 8 link /ocfs2/linkdir/file1 /ocfs2/linkdir/linklinklinklinklinklinkaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa Note: The link names have been crafted for a 512 byte blocksize. Reproducing with a larger blocksize will require longer (or more) links. The sleep is important. We want jbd2 to commit the transaction so that the missing block does not piggy back on account of the previous transaction. Signed-off-by: XiaoweiHu Reviewed-by: WengangWang Reviewed-by: Sunil.Mushran Signed-off-by: Joel Becker --- fs/ocfs2/journal.h | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) (limited to 'fs') diff --git a/fs/ocfs2/journal.h b/fs/ocfs2/journal.h index 68cf2f6d3c6..a3385b63ff5 100644 --- a/fs/ocfs2/journal.h +++ b/fs/ocfs2/journal.h @@ -441,10 +441,11 @@ static inline int ocfs2_mknod_credits(struct super_block *sb, int is_dir, #define OCFS2_SIMPLE_DIR_EXTEND_CREDITS (2) /* file update (nlink, etc) + directory mtime/ctime + dir entry block + quota - * update on dir + index leaf + dx root update for free list */ + * update on dir + index leaf + dx root update for free list + + * previous dirblock update in the free list */ static inline int ocfs2_link_credits(struct super_block *sb) { - return 2*OCFS2_INODE_UPDATE_CREDITS + 3 + + return 2*OCFS2_INODE_UPDATE_CREDITS + 4 + ocfs2_quota_trans_credits(sb); } -- cgit v1.2.3-70-g09d2 From df295d4a4b3c98af1a2445a82aef169e7e5d96b8 Mon Sep 17 00:00:00 2001 From: Mark Fasheh Date: Wed, 16 Nov 2011 12:03:10 -0800 Subject: ocfs2: honor O_(D)SYNC flag in fallocate We need to sync the transaction which updates i_size if the file is marked as needing sync semantics. Signed-off-by: Mark Fasheh Signed-off-by: Joel Becker --- fs/ocfs2/file.c | 3 +++ 1 file changed, 3 insertions(+) (limited to 'fs') diff --git a/fs/ocfs2/file.c b/fs/ocfs2/file.c index 5c4a74e04ab..09e3de57cde 100644 --- a/fs/ocfs2/file.c +++ b/fs/ocfs2/file.c @@ -1950,6 +1950,9 @@ static int __ocfs2_change_file_space(struct file *file, struct inode *inode, if (ret < 0) mlog_errno(ret); + if (file->f_flags & O_SYNC) + handle->h_sync = 1; + ocfs2_commit_trans(osb, handle); out_inode_unlock: -- cgit v1.2.3-70-g09d2 From 249ec93c01db8898058899a80ffb537c8d27f86f Mon Sep 17 00:00:00 2001 From: Jan Kara Date: Tue, 8 Nov 2011 00:20:39 +0100 Subject: ocfs2: Use filemap_write_and_wait() instead of write_inode_now() Since ocfs2 has no ->write_inode method, there's no point in calling write_inode_now() from ocfs2_cleanup_delete_inode(). Use filemap_write_and_wait() instead. This helps us to cleanup inode writing interfaces... Signed-off-by: Jan Kara Signed-off-by: Joel Becker --- fs/ocfs2/inode.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/ocfs2/inode.c b/fs/ocfs2/inode.c index b4c8bb6b8d2..c4bf6ac4a0b 100644 --- a/fs/ocfs2/inode.c +++ b/fs/ocfs2/inode.c @@ -951,7 +951,7 @@ static void ocfs2_cleanup_delete_inode(struct inode *inode, trace_ocfs2_cleanup_delete_inode( (unsigned long long)OCFS2_I(inode)->ip_blkno, sync_data); if (sync_data) - write_inode_now(inode, 1); + filemap_write_and_wait(inode->i_mapping); truncate_inode_pages(&inode->i_data, 0); } -- cgit v1.2.3-70-g09d2 From f6f8285132907757ef84ef8dae0a1244b8cde6ac Mon Sep 17 00:00:00 2001 From: Kees Cook Date: Thu, 17 Nov 2011 12:58:07 -0800 Subject: pstore: pass allocated memory region back to caller The buf_lock cannot be held while populating the inodes, so make the backend pass forward an allocated and filled buffer instead. This solves the following backtrace. The effect is that "buf" is only ever used to notify the backends that something was written to it, and shouldn't be used in the read path. To replace the buf_lock during the read path, isolate the open/read/close loop with a separate mutex to maintain serialized access to the backend. Note that is is up to the pstore backend to cope if the (*write)() path is called in the middle of the read path. [ 59.691019] BUG: sleeping function called from invalid context at .../mm/slub.c:847 [ 59.691019] in_atomic(): 0, irqs_disabled(): 1, pid: 1819, name: mount [ 59.691019] Pid: 1819, comm: mount Not tainted 3.0.8 #1 [ 59.691019] Call Trace: [ 59.691019] [<810252d5>] __might_sleep+0xc3/0xca [ 59.691019] [<810a26e6>] kmem_cache_alloc+0x32/0xf3 [ 59.691019] [<810b53ac>] ? __d_lookup_rcu+0x6f/0xf4 [ 59.691019] [<810b68b1>] alloc_inode+0x2a/0x64 [ 59.691019] [<810b6903>] new_inode+0x18/0x43 [ 59.691019] [<81142447>] pstore_get_inode.isra.1+0x11/0x98 [ 59.691019] [<81142623>] pstore_mkfile+0xae/0x26f [ 59.691019] [<810a2a66>] ? kmem_cache_free+0x19/0xb1 [ 59.691019] [<8116c821>] ? ida_get_new_above+0x140/0x158 [ 59.691019] [<811708ea>] ? __init_rwsem+0x1e/0x2c [ 59.691019] [<810b67e8>] ? inode_init_always+0x111/0x1b0 [ 59.691019] [<8102127e>] ? should_resched+0xd/0x27 [ 59.691019] [<8137977f>] ? _cond_resched+0xd/0x21 [ 59.691019] [<81142abf>] pstore_get_records+0x52/0xa7 [ 59.691019] [<8114254b>] pstore_fill_super+0x7d/0x91 [ 59.691019] [<810a7ff5>] mount_single+0x46/0x82 [ 59.691019] [<8114231a>] pstore_mount+0x15/0x17 [ 59.691019] [<811424ce>] ? pstore_get_inode.isra.1+0x98/0x98 [ 59.691019] [<810a8199>] mount_fs+0x5a/0x12d [ 59.691019] [<810b9174>] ? alloc_vfsmnt+0xa4/0x14a [ 59.691019] [<810b9474>] vfs_kern_mount+0x4f/0x7d [ 59.691019] [<810b9d7e>] do_kern_mount+0x34/0xb2 [ 59.691019] [<810bb15f>] do_mount+0x5fc/0x64a [ 59.691019] [<810912fb>] ? strndup_user+0x2e/0x3f [ 59.691019] [<810bb3cb>] sys_mount+0x66/0x99 [ 59.691019] [<8137b537>] sysenter_do_call+0x12/0x26 Signed-off-by: Kees Cook Signed-off-by: Tony Luck --- drivers/acpi/apei/erst.c | 31 ++++++++++++++++++++++--------- drivers/firmware/efivars.c | 9 +++++++-- fs/pstore/platform.c | 13 ++++++++----- include/linux/pstore.h | 4 +++- 4 files changed, 40 insertions(+), 17 deletions(-) (limited to 'fs') diff --git a/drivers/acpi/apei/erst.c b/drivers/acpi/apei/erst.c index 127408069ca..631b9477b99 100644 --- a/drivers/acpi/apei/erst.c +++ b/drivers/acpi/apei/erst.c @@ -932,7 +932,8 @@ static int erst_check_table(struct acpi_table_erst *erst_tab) static int erst_open_pstore(struct pstore_info *psi); static int erst_close_pstore(struct pstore_info *psi); static ssize_t erst_reader(u64 *id, enum pstore_type_id *type, - struct timespec *time, struct pstore_info *psi); + struct timespec *time, char **buf, + struct pstore_info *psi); static int erst_writer(enum pstore_type_id type, u64 *id, unsigned int part, size_t size, struct pstore_info *psi); static int erst_clearer(enum pstore_type_id type, u64 id, @@ -986,17 +987,23 @@ static int erst_close_pstore(struct pstore_info *psi) } static ssize_t erst_reader(u64 *id, enum pstore_type_id *type, - struct timespec *time, struct pstore_info *psi) + struct timespec *time, char **buf, + struct pstore_info *psi) { int rc; ssize_t len = 0; u64 record_id; - struct cper_pstore_record *rcd = (struct cper_pstore_record *) - (erst_info.buf - sizeof(*rcd)); + struct cper_pstore_record *rcd; + size_t rcd_len = sizeof(*rcd) + erst_info.bufsize; if (erst_disable) return -ENODEV; + rcd = kmalloc(rcd_len, GFP_KERNEL); + if (!rcd) { + rc = -ENOMEM; + goto out; + } skip: rc = erst_get_record_id_next(&reader_pos, &record_id); if (rc) @@ -1004,22 +1011,27 @@ skip: /* no more record */ if (record_id == APEI_ERST_INVALID_RECORD_ID) { - rc = -1; + rc = -EINVAL; goto out; } - len = erst_read(record_id, &rcd->hdr, sizeof(*rcd) + - erst_info.bufsize); + len = erst_read(record_id, &rcd->hdr, rcd_len); /* The record may be cleared by others, try read next record */ if (len == -ENOENT) goto skip; - else if (len < 0) { - rc = -1; + else if (len < sizeof(*rcd)) { + rc = -EIO; goto out; } if (uuid_le_cmp(rcd->hdr.creator_id, CPER_CREATOR_PSTORE) != 0) goto skip; + *buf = kmalloc(len, GFP_KERNEL); + if (*buf == NULL) { + rc = -ENOMEM; + goto out; + } + memcpy(*buf, rcd->data, len - sizeof(*rcd)); *id = record_id; if (uuid_le_cmp(rcd->sec_hdr.section_type, CPER_SECTION_TYPE_DMESG) == 0) @@ -1037,6 +1049,7 @@ skip: time->tv_nsec = 0; out: + kfree(rcd); return (rc < 0) ? rc : (len - sizeof(*rcd)); } diff --git a/drivers/firmware/efivars.c b/drivers/firmware/efivars.c index 8370f72d87f..a54a6b972ce 100644 --- a/drivers/firmware/efivars.c +++ b/drivers/firmware/efivars.c @@ -457,7 +457,8 @@ static int efi_pstore_close(struct pstore_info *psi) } static ssize_t efi_pstore_read(u64 *id, enum pstore_type_id *type, - struct timespec *timespec, struct pstore_info *psi) + struct timespec *timespec, + char **buf, struct pstore_info *psi) { efi_guid_t vendor = LINUX_EFI_CRASH_GUID; struct efivars *efivars = psi->data; @@ -478,7 +479,11 @@ static ssize_t efi_pstore_read(u64 *id, enum pstore_type_id *type, timespec->tv_nsec = 0; get_var_data_locked(efivars, &efivars->walk_entry->var); size = efivars->walk_entry->var.DataSize; - memcpy(psi->buf, efivars->walk_entry->var.Data, size); + *buf = kmalloc(size, GFP_KERNEL); + if (*buf == NULL) + return -ENOMEM; + memcpy(*buf, efivars->walk_entry->var.Data, + size); efivars->walk_entry = list_entry(efivars->walk_entry->list.next, struct efivar_entry, list); return size; diff --git a/fs/pstore/platform.c b/fs/pstore/platform.c index 2bd620f0d79..57bbf9078ac 100644 --- a/fs/pstore/platform.c +++ b/fs/pstore/platform.c @@ -167,6 +167,7 @@ int pstore_register(struct pstore_info *psi) } psinfo = psi; + mutex_init(&psinfo->read_mutex); spin_unlock(&pstore_lock); if (owner && !try_module_get(owner)) { @@ -195,30 +196,32 @@ EXPORT_SYMBOL_GPL(pstore_register); void pstore_get_records(int quiet) { struct pstore_info *psi = psinfo; + char *buf = NULL; ssize_t size; u64 id; enum pstore_type_id type; struct timespec time; int failed = 0, rc; - unsigned long flags; if (!psi) return; - spin_lock_irqsave(&psinfo->buf_lock, flags); + mutex_lock(&psi->read_mutex); rc = psi->open(psi); if (rc) goto out; - while ((size = psi->read(&id, &type, &time, psi)) > 0) { - rc = pstore_mkfile(type, psi->name, id, psi->buf, (size_t)size, + while ((size = psi->read(&id, &type, &time, &buf, psi)) > 0) { + rc = pstore_mkfile(type, psi->name, id, buf, (size_t)size, time, psi); + kfree(buf); + buf = NULL; if (rc && (rc != -EEXIST || !quiet)) failed++; } psi->close(psi); out: - spin_unlock_irqrestore(&psinfo->buf_lock, flags); + mutex_unlock(&psi->read_mutex); if (failed) printk(KERN_WARNING "pstore: failed to load %d record(s) from '%s'\n", diff --git a/include/linux/pstore.h b/include/linux/pstore.h index ea567321ae3..2ca8cde5459 100644 --- a/include/linux/pstore.h +++ b/include/linux/pstore.h @@ -35,10 +35,12 @@ struct pstore_info { spinlock_t buf_lock; /* serialize access to 'buf' */ char *buf; size_t bufsize; + struct mutex read_mutex; /* serialize open/read/close */ int (*open)(struct pstore_info *psi); int (*close)(struct pstore_info *psi); ssize_t (*read)(u64 *id, enum pstore_type_id *type, - struct timespec *time, struct pstore_info *psi); + struct timespec *time, char **buf, + struct pstore_info *psi); int (*write)(enum pstore_type_id type, u64 *id, unsigned int part, size_t size, struct pstore_info *psi); int (*erase)(enum pstore_type_id type, u64 id, -- cgit v1.2.3-70-g09d2 From 016e8d44bc06dd3322f26712bdd3f3a6973592d0 Mon Sep 17 00:00:00 2001 From: Josh Boyer Date: Fri, 19 Aug 2011 14:50:26 -0400 Subject: fs/minix: Verify bitmap block counts before mounting Newer versions of MINIX can create filesystems that allocate an extra bitmap block. Mounting of this succeeds, but doing a statfs call will result in an oops in count_free because of a negative number being used for the bh index. Avoid this by verifying the number of allocated blocks at mount time, erroring out if there are not enough and make statfs ignore the extras if there are too many. This fixes https://bugzilla.kernel.org/show_bug.cgi?id=18792 Signed-off-by: Josh Boyer Signed-off-by: Al Viro --- fs/minix/bitmap.c | 18 ++++++++++++------ fs/minix/inode.c | 25 +++++++++++++++++++++++-- fs/minix/minix.h | 9 +++++++-- 3 files changed, 42 insertions(+), 10 deletions(-) (limited to 'fs') diff --git a/fs/minix/bitmap.c b/fs/minix/bitmap.c index 3f32bcb0d9b..7c82c29429d 100644 --- a/fs/minix/bitmap.c +++ b/fs/minix/bitmap.c @@ -20,10 +20,11 @@ static const int nibblemap[] = { 4,3,3,2,3,2,2,1,3,2,2,1,2,1,1,0 }; static DEFINE_SPINLOCK(bitmap_lock); -static unsigned long count_free(struct buffer_head *map[], unsigned numblocks, __u32 numbits) +static unsigned long count_free(struct buffer_head *map[], unsigned blocksize, __u32 numbits) { unsigned i, j, sum = 0; struct buffer_head *bh; + unsigned numblocks = minix_blocks_needed(numbits, blocksize); for (i=0; is_zmap, sbi->s_zmap_blocks, - sbi->s_nzones - sbi->s_firstdatazone + 1) + struct minix_sb_info *sbi = minix_sb(sb); + u32 bits = sbi->s_nzones - (sbi->s_firstdatazone + 1); + + return (count_free(sbi->s_zmap, sb->s_blocksize, bits) << sbi->s_log_zone_size); } @@ -273,7 +276,10 @@ struct inode *minix_new_inode(const struct inode *dir, int mode, int *error) return inode; } -unsigned long minix_count_free_inodes(struct minix_sb_info *sbi) +unsigned long minix_count_free_inodes(struct super_block *sb) { - return count_free(sbi->s_imap, sbi->s_imap_blocks, sbi->s_ninodes + 1); + struct minix_sb_info *sbi = minix_sb(sb); + u32 bits = sbi->s_ninodes + 1; + + return count_free(sbi->s_imap, sb->s_blocksize, bits); } diff --git a/fs/minix/inode.c b/fs/minix/inode.c index 64cdcd662ff..1d9e33966db 100644 --- a/fs/minix/inode.c +++ b/fs/minix/inode.c @@ -279,6 +279,27 @@ static int minix_fill_super(struct super_block *s, void *data, int silent) else if (sbi->s_mount_state & MINIX_ERROR_FS) printk("MINIX-fs: mounting file system with errors, " "running fsck is recommended\n"); + + /* Apparently minix can create filesystems that allocate more blocks for + * the bitmaps than needed. We simply ignore that, but verify it didn't + * create one with not enough blocks and bail out if so. + */ + block = minix_blocks_needed(sbi->s_ninodes, s->s_blocksize); + if (sbi->s_imap_blocks < block) { + printk("MINIX-fs: file system does not have enough " + "imap blocks allocated. Refusing to mount\n"); + goto out_iput; + } + + block = minix_blocks_needed( + (sbi->s_nzones - (sbi->s_firstdatazone + 1)), + s->s_blocksize); + if (sbi->s_zmap_blocks < block) { + printk("MINIX-fs: file system does not have enough " + "zmap blocks allocated. Refusing to mount.\n"); + goto out_iput; + } + return 0; out_iput: @@ -339,10 +360,10 @@ static int minix_statfs(struct dentry *dentry, struct kstatfs *buf) buf->f_type = sb->s_magic; buf->f_bsize = sb->s_blocksize; buf->f_blocks = (sbi->s_nzones - sbi->s_firstdatazone) << sbi->s_log_zone_size; - buf->f_bfree = minix_count_free_blocks(sbi); + buf->f_bfree = minix_count_free_blocks(sb); buf->f_bavail = buf->f_bfree; buf->f_files = sbi->s_ninodes; - buf->f_ffree = minix_count_free_inodes(sbi); + buf->f_ffree = minix_count_free_inodes(sb); buf->f_namelen = sbi->s_namelen; buf->f_fsid.val[0] = (u32)id; buf->f_fsid.val[1] = (u32)(id >> 32); diff --git a/fs/minix/minix.h b/fs/minix/minix.h index 341e2122879..6415fe0d238 100644 --- a/fs/minix/minix.h +++ b/fs/minix/minix.h @@ -48,10 +48,10 @@ extern struct minix_inode * minix_V1_raw_inode(struct super_block *, ino_t, stru extern struct minix2_inode * minix_V2_raw_inode(struct super_block *, ino_t, struct buffer_head **); extern struct inode * minix_new_inode(const struct inode *, int, int *); extern void minix_free_inode(struct inode * inode); -extern unsigned long minix_count_free_inodes(struct minix_sb_info *sbi); +extern unsigned long minix_count_free_inodes(struct super_block *sb); extern int minix_new_block(struct inode * inode); extern void minix_free_block(struct inode *inode, unsigned long block); -extern unsigned long minix_count_free_blocks(struct minix_sb_info *sbi); +extern unsigned long minix_count_free_blocks(struct super_block *sb); extern int minix_getattr(struct vfsmount *, struct dentry *, struct kstat *); extern int minix_prepare_chunk(struct page *page, loff_t pos, unsigned len); @@ -88,6 +88,11 @@ static inline struct minix_inode_info *minix_i(struct inode *inode) return list_entry(inode, struct minix_inode_info, vfs_inode); } +static inline unsigned minix_blocks_needed(unsigned bits, unsigned blocksize) +{ + return DIV_ROUND_UP(bits, blocksize * 8); +} + #if defined(CONFIG_MINIX_FS_NATIVE_ENDIAN) && \ defined(CONFIG_MINIX_FS_BIG_ENDIAN_16BIT_INDEXED) -- cgit v1.2.3-70-g09d2 From f1fd306a91f875e65af0e04855b23adda6831ac9 Mon Sep 17 00:00:00 2001 From: Al Viro Date: Fri, 26 Aug 2011 22:38:50 -0400 Subject: minixfs: kill manual hweight(), simplify Signed-off-by: Al Viro --- fs/minix/bitmap.c | 41 ++++++++++++++--------------------------- fs/minix/minix.h | 2 +- 2 files changed, 15 insertions(+), 28 deletions(-) (limited to 'fs') diff --git a/fs/minix/bitmap.c b/fs/minix/bitmap.c index 7c82c29429d..ef175cb8cfd 100644 --- a/fs/minix/bitmap.c +++ b/fs/minix/bitmap.c @@ -16,39 +16,26 @@ #include #include -static const int nibblemap[] = { 4,3,3,2,3,2,2,1,3,2,2,1,2,1,1,0 }; - static DEFINE_SPINLOCK(bitmap_lock); -static unsigned long count_free(struct buffer_head *map[], unsigned blocksize, __u32 numbits) +/* + * bitmap consists of blocks filled with 16bit words + * bit set == busy, bit clear == free + * endianness is a mess, but for counting zero bits it really doesn't matter... + */ +static __u32 count_free(struct buffer_head *map[], unsigned blocksize, __u32 numbits) { - unsigned i, j, sum = 0; - struct buffer_head *bh; - unsigned numblocks = minix_blocks_needed(numbits, blocksize); - - for (i=0; ib_size; j++) - sum += nibblemap[bh->b_data[j] & 0xf] - + nibblemap[(bh->b_data[j]>>4) & 0xf]; - } + __u32 sum = 0; + unsigned blocks = DIV_ROUND_UP(numbits, blocksize * 8); - if (numblocks==0 || !(bh=map[numblocks-1])) - return(0); - i = ((numbits - (numblocks-1) * bh->b_size * 8) / 16) * 2; - for (j=0; jb_data[j] & 0xf] - + nibblemap[(bh->b_data[j]>>4) & 0xf]; + while (blocks--) { + unsigned words = blocksize / 2; + __u16 *p = (__u16 *)(*map++)->b_data; + while (words--) + sum += 16 - hweight16(*p++); } - i = numbits%16; - if (i!=0) { - i = *(__u16 *)(&bh->b_data[j]) | ~((1<>4) & 0xf]; - sum += nibblemap[(i>>8) & 0xf] + nibblemap[(i>>12) & 0xf]; - } - return(sum); + return sum; } void minix_free_block(struct inode *inode, unsigned long block) diff --git a/fs/minix/minix.h b/fs/minix/minix.h index 6415fe0d238..26bbd55e82e 100644 --- a/fs/minix/minix.h +++ b/fs/minix/minix.h @@ -130,7 +130,7 @@ static inline int minix_find_first_zero_bit(const void *vaddr, unsigned size) if (!size) return 0; - size = (size >> 4) + ((size & 15) > 0); + size >>= 4; while (*p++ == 0xffff) { if (--size == 0) return (p - addr) << 4; -- cgit v1.2.3-70-g09d2 From 387125fc722a8ed432066b85a552917343bdafca Mon Sep 17 00:00:00 2001 From: Chris Mason Date: Fri, 18 Nov 2011 15:07:51 -0500 Subject: Btrfs: fix barrier flushes When btrfs is writing the super blocks, it send barrier flushes to make sure writeback caching drives get all the metadata on disk in the right order. But, we have two bugs in the way these are sent down. When doing full commits (not via the tree log), we are sending the barrier down before the last super when it should be going down before the first. In multi-device setups, we should be waiting for the barriers to complete on all devices before writing any of the supers. Both of these bugs can cause corruptions on power failures. We fix it with some new code to send down empty barriers to all devices before writing the first super. Alexandre Oliva found the multi-device bug. Arne Jansen did the async barrier loop. Signed-off-by: Chris Mason Reported-by: Alexandre Oliva --- fs/btrfs/disk-io.c | 145 ++++++++++++++++++++++++++++++++++++++++++++++------- fs/btrfs/volumes.h | 6 +++ 2 files changed, 134 insertions(+), 17 deletions(-) (limited to 'fs') diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index b6a5c0dd0dd..48d30138237 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -2573,22 +2573,10 @@ static int write_dev_supers(struct btrfs_device *device, int errors = 0; u32 crc; u64 bytenr; - int last_barrier = 0; if (max_mirrors == 0) max_mirrors = BTRFS_SUPER_MIRROR_MAX; - /* make sure only the last submit_bh does a barrier */ - if (do_barriers) { - for (i = 0; i < max_mirrors; i++) { - bytenr = btrfs_sb_offset(i); - if (bytenr + BTRFS_SUPER_INFO_SIZE >= - device->total_bytes) - break; - last_barrier = i; - } - } - for (i = 0; i < max_mirrors; i++) { bytenr = btrfs_sb_offset(i); if (bytenr + BTRFS_SUPER_INFO_SIZE >= device->total_bytes) @@ -2634,17 +2622,136 @@ static int write_dev_supers(struct btrfs_device *device, bh->b_end_io = btrfs_end_buffer_write_sync; } - if (i == last_barrier && do_barriers) - ret = submit_bh(WRITE_FLUSH_FUA, bh); - else - ret = submit_bh(WRITE_SYNC, bh); - + /* + * we fua the first super. The others we allow + * to go down lazy. + */ + ret = submit_bh(WRITE_FUA, bh); if (ret) errors++; } return errors < i ? 0 : -1; } +/* + * endio for the write_dev_flush, this will wake anyone waiting + * for the barrier when it is done + */ +static void btrfs_end_empty_barrier(struct bio *bio, int err) +{ + if (err) { + if (err == -EOPNOTSUPP) + set_bit(BIO_EOPNOTSUPP, &bio->bi_flags); + clear_bit(BIO_UPTODATE, &bio->bi_flags); + } + if (bio->bi_private) + complete(bio->bi_private); + bio_put(bio); +} + +/* + * trigger flushes for one the devices. If you pass wait == 0, the flushes are + * sent down. With wait == 1, it waits for the previous flush. + * + * any device where the flush fails with eopnotsupp are flagged as not-barrier + * capable + */ +static int write_dev_flush(struct btrfs_device *device, int wait) +{ + struct bio *bio; + int ret = 0; + + if (device->nobarriers) + return 0; + + if (wait) { + bio = device->flush_bio; + if (!bio) + return 0; + + wait_for_completion(&device->flush_wait); + + if (bio_flagged(bio, BIO_EOPNOTSUPP)) { + printk("btrfs: disabling barriers on dev %s\n", + device->name); + device->nobarriers = 1; + } + if (!bio_flagged(bio, BIO_UPTODATE)) { + ret = -EIO; + } + + /* drop the reference from the wait == 0 run */ + bio_put(bio); + device->flush_bio = NULL; + + return ret; + } + + /* + * one reference for us, and we leave it for the + * caller + */ + device->flush_bio = NULL;; + bio = bio_alloc(GFP_NOFS, 0); + if (!bio) + return -ENOMEM; + + bio->bi_end_io = btrfs_end_empty_barrier; + bio->bi_bdev = device->bdev; + init_completion(&device->flush_wait); + bio->bi_private = &device->flush_wait; + device->flush_bio = bio; + + bio_get(bio); + submit_bio(WRITE_FLUSH, bio); + + return 0; +} + +/* + * send an empty flush down to each device in parallel, + * then wait for them + */ +static int barrier_all_devices(struct btrfs_fs_info *info) +{ + struct list_head *head; + struct btrfs_device *dev; + int errors = 0; + int ret; + + /* send down all the barriers */ + head = &info->fs_devices->devices; + list_for_each_entry_rcu(dev, head, dev_list) { + if (!dev->bdev) { + errors++; + continue; + } + if (!dev->in_fs_metadata || !dev->writeable) + continue; + + ret = write_dev_flush(dev, 0); + if (ret) + errors++; + } + + /* wait for all the barriers */ + list_for_each_entry_rcu(dev, head, dev_list) { + if (!dev->bdev) { + errors++; + continue; + } + if (!dev->in_fs_metadata || !dev->writeable) + continue; + + ret = write_dev_flush(dev, 1); + if (ret) + errors++; + } + if (errors) + return -EIO; + return 0; +} + int write_all_supers(struct btrfs_root *root, int max_mirrors) { struct list_head *head; @@ -2666,6 +2773,10 @@ int write_all_supers(struct btrfs_root *root, int max_mirrors) mutex_lock(&root->fs_info->fs_devices->device_list_mutex); head = &root->fs_info->fs_devices->devices; + + if (do_barriers) + barrier_all_devices(root->fs_info); + list_for_each_entry_rcu(dev, head, dev_list) { if (!dev->bdev) { total_errors++; diff --git a/fs/btrfs/volumes.h b/fs/btrfs/volumes.h index ab5b1c49f35..78f2d4d4f37 100644 --- a/fs/btrfs/volumes.h +++ b/fs/btrfs/volumes.h @@ -100,6 +100,12 @@ struct btrfs_device { struct reada_zone *reada_curr_zone; struct radix_tree_root reada_zones; struct radix_tree_root reada_extents; + + /* for sending down flush barriers */ + struct bio *flush_bio; + struct completion flush_wait; + int nobarriers; + }; struct btrfs_fs_devices { -- cgit v1.2.3-70-g09d2 From 745c4d8e160afaf6c75e887c27ea4b75c8142b26 Mon Sep 17 00:00:00 2001 From: Jeff Mahoney Date: Sun, 20 Nov 2011 07:31:57 -0500 Subject: btrfs: Fix up 32/64-bit compatibility for new ioctls This patch casts to unsigned long before casting to a pointer and fixes the following warnings: fs/btrfs/extent_io.c:2289:20: warning: cast from pointer to integer of different size [-Wpointer-to-int-cast] fs/btrfs/ioctl.c:2933:37: warning: cast from pointer to integer of different size [-Wpointer-to-int-cast] fs/btrfs/ioctl.c:2937:21: warning: cast to pointer from integer of different size [-Wint-to-pointer-cast] fs/btrfs/ioctl.c:3020:21: warning: cast to pointer from integer of different size [-Wint-to-pointer-cast] fs/btrfs/scrub.c:275:4: warning: cast to pointer from integer of different size [-Wint-to-pointer-cast] fs/btrfs/backref.c:686:27: warning: cast from pointer to integer of different size [-Wpointer-to-int-cast] Signed-off-by: Jeff Mahoney Signed-off-by: Chris Mason --- fs/btrfs/backref.c | 2 +- fs/btrfs/extent_io.c | 2 +- fs/btrfs/ioctl.c | 9 ++++++--- fs/btrfs/scrub.c | 2 +- 4 files changed, 9 insertions(+), 6 deletions(-) (limited to 'fs') diff --git a/fs/btrfs/backref.c b/fs/btrfs/backref.c index 8855aad3929..22c64fff1bd 100644 --- a/fs/btrfs/backref.c +++ b/fs/btrfs/backref.c @@ -683,7 +683,7 @@ static int inode_to_path(u64 inum, struct btrfs_inode_ref *iref, return PTR_ERR(fspath); if (fspath > fspath_min) { - ipath->fspath->val[i] = (u64)fspath; + ipath->fspath->val[i] = (u64)(unsigned long)fspath; ++ipath->fspath->elem_cnt; ipath->fspath->bytes_left = fspath - fspath_min; } else { diff --git a/fs/btrfs/extent_io.c b/fs/btrfs/extent_io.c index 1f87c4d0e7a..47fdba7853c 100644 --- a/fs/btrfs/extent_io.c +++ b/fs/btrfs/extent_io.c @@ -2286,7 +2286,7 @@ static void end_bio_extent_readpage(struct bio *bio, int err) } if (!uptodate) { u64 failed_mirror; - failed_mirror = (u64)bio->bi_bdev; + failed_mirror = (unsigned long)bio->bi_bdev; if (tree->ops && tree->ops->readpage_io_failed_hook) ret = tree->ops->readpage_io_failed_hook( bio, page, start, end, diff --git a/fs/btrfs/ioctl.c b/fs/btrfs/ioctl.c index 4a34c472f12..bda53fcd1a0 100644 --- a/fs/btrfs/ioctl.c +++ b/fs/btrfs/ioctl.c @@ -2930,11 +2930,13 @@ static long btrfs_ioctl_ino_to_path(struct btrfs_root *root, void __user *arg) goto out; for (i = 0; i < ipath->fspath->elem_cnt; ++i) { - rel_ptr = ipath->fspath->val[i] - (u64)ipath->fspath->val; + rel_ptr = ipath->fspath->val[i] - + (u64)(unsigned long)ipath->fspath->val; ipath->fspath->val[i] = rel_ptr; } - ret = copy_to_user((void *)ipa->fspath, (void *)ipath->fspath, size); + ret = copy_to_user((void *)(unsigned long)ipa->fspath, + (void *)(unsigned long)ipath->fspath, size); if (ret) { ret = -EFAULT; goto out; @@ -3017,7 +3019,8 @@ static long btrfs_ioctl_logical_to_ino(struct btrfs_root *root, if (ret < 0) goto out; - ret = copy_to_user((void *)loi->inodes, (void *)inodes, size); + ret = copy_to_user((void *)(unsigned long)loi->inodes, + (void *)(unsigned long)inodes, size); if (ret) ret = -EFAULT; diff --git a/fs/btrfs/scrub.c b/fs/btrfs/scrub.c index f4190f22edf..fab420db512 100644 --- a/fs/btrfs/scrub.c +++ b/fs/btrfs/scrub.c @@ -272,7 +272,7 @@ static int scrub_print_warning_inode(u64 inum, u64 offset, u64 root, void *ctx) swarn->logical, swarn->dev->name, (unsigned long long)swarn->sector, root, inum, offset, min(isize - offset, (u64)PAGE_SIZE), nlink, - (char *)ipath->fspath->val[i]); + (char *)(unsigned long)ipath->fspath->val[i]); free_ipath(ipath); return 0; -- cgit v1.2.3-70-g09d2 From 32240a913d9f3a5aad42175d7696590ea1bfdb08 Mon Sep 17 00:00:00 2001 From: Jan Schmidt Date: Sun, 20 Nov 2011 07:33:38 -0500 Subject: btrfs: mirror_num should be int, not u64 My previous patch introduced some u64 for failed_mirror variables, this one makes it consistent again. Signed-off-by: Jan Schmidt Signed-off-by: Chris Mason --- fs/btrfs/disk-io.c | 2 +- fs/btrfs/extent_io.c | 4 ++-- fs/btrfs/extent_io.h | 2 +- 3 files changed, 4 insertions(+), 4 deletions(-) (limited to 'fs') diff --git a/fs/btrfs/disk-io.c b/fs/btrfs/disk-io.c index 48d30138237..94abc25392f 100644 --- a/fs/btrfs/disk-io.c +++ b/fs/btrfs/disk-io.c @@ -620,7 +620,7 @@ out: static int btree_io_failed_hook(struct bio *failed_bio, struct page *page, u64 start, u64 end, - u64 mirror_num, struct extent_state *state) + int mirror_num, struct extent_state *state) { struct extent_io_tree *tree; unsigned long len; diff --git a/fs/btrfs/extent_io.c b/fs/btrfs/extent_io.c index 47fdba7853c..a877b8b212e 100644 --- a/fs/btrfs/extent_io.c +++ b/fs/btrfs/extent_io.c @@ -2285,8 +2285,8 @@ static void end_bio_extent_readpage(struct bio *bio, int err) clean_io_failure(start, page); } if (!uptodate) { - u64 failed_mirror; - failed_mirror = (unsigned long)bio->bi_bdev; + int failed_mirror; + failed_mirror = (int)(unsigned long)bio->bi_bdev; if (tree->ops && tree->ops->readpage_io_failed_hook) ret = tree->ops->readpage_io_failed_hook( bio, page, start, end, diff --git a/fs/btrfs/extent_io.h b/fs/btrfs/extent_io.h index feb9be0e23b..7604c300132 100644 --- a/fs/btrfs/extent_io.h +++ b/fs/btrfs/extent_io.h @@ -70,7 +70,7 @@ struct extent_io_ops { unsigned long bio_flags); int (*readpage_io_hook)(struct page *page, u64 start, u64 end); int (*readpage_io_failed_hook)(struct bio *bio, struct page *page, - u64 start, u64 end, u64 failed_mirror, + u64 start, u64 end, int failed_mirror, struct extent_state *state); int (*writepage_io_failed_hook)(struct bio *bio, struct page *page, u64 start, u64 end, -- cgit v1.2.3-70-g09d2 From 0f0fbf1d0e188d129756e9508090af4bdbfde00b Mon Sep 17 00:00:00 2001 From: Li Zefan Date: Sun, 20 Nov 2011 07:33:38 -0500 Subject: Btrfs: fix to search one more bitmap for cluster setup Suppose there are two bitmaps [0, 256], [256, 512] and one extent [100, 120] in the free space cache, and we want to setup a cluster with offset=100, bytes=50. In this case, there will be only one bitmap [256, 512] in the temporary bitmaps list, and then setup_cluster_bitmap() won't search bitmap [0, 256]. The cause is, the list is constructed in setup_cluster_no_bitmap(), and only bitmaps with bitmap_entry->offset >= offset will be added into the list, and the very bitmap that convers offset has bitmap_entry->offset <= offset. Signed-off-by: Li Zefan Signed-off-by: Chris Mason --- fs/btrfs/free-space-cache.c | 12 ++++++++++++ 1 file changed, 12 insertions(+) (limited to 'fs') diff --git a/fs/btrfs/free-space-cache.c b/fs/btrfs/free-space-cache.c index 181760f9d2a..8f792f41fea 100644 --- a/fs/btrfs/free-space-cache.c +++ b/fs/btrfs/free-space-cache.c @@ -2453,10 +2453,22 @@ setup_cluster_bitmap(struct btrfs_block_group_cache *block_group, struct btrfs_free_space *entry; struct rb_node *node; int ret = -ENOSPC; + u64 bitmap_offset = offset_to_bitmap(ctl, offset); if (ctl->total_bitmaps == 0) return -ENOSPC; + /* + * The bitmap that covers offset won't be in the list unless offset + * is just its start offset. + */ + entry = list_first_entry(bitmaps, struct btrfs_free_space, list); + if (entry->offset != bitmap_offset) { + entry = tree_search_offset(ctl, bitmap_offset, 1, 0); + if (entry && list_empty(&entry->list)) + list_add(&entry->list, bitmaps); + } + /* * First check our cached list of bitmaps and see if there is an entry * here that will work. -- cgit v1.2.3-70-g09d2 From 52621cb6ed0e0e14358bb317bda7cd5fbd5c2a27 Mon Sep 17 00:00:00 2001 From: Li Zefan Date: Sun, 20 Nov 2011 07:33:38 -0500 Subject: Btrfs: avoid unnecessary bitmap search for cluster setup setup_cluster_no_bitmap() searches all the extents and bitmaps starting from offset. Therefore if it returns -ENOSPC, all the bitmaps starting from offset are in the bitmaps list, so it's sufficient to search from this list in setup_cluser_bitmap(). Signed-off-by: Li Zefan Signed-off-by: Chris Mason --- fs/btrfs/free-space-cache.c | 42 ++++-------------------------------------- 1 file changed, 4 insertions(+), 38 deletions(-) (limited to 'fs') diff --git a/fs/btrfs/free-space-cache.c b/fs/btrfs/free-space-cache.c index 8f792f41fea..8c32434da2c 100644 --- a/fs/btrfs/free-space-cache.c +++ b/fs/btrfs/free-space-cache.c @@ -2451,7 +2451,6 @@ setup_cluster_bitmap(struct btrfs_block_group_cache *block_group, { struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl; struct btrfs_free_space *entry; - struct rb_node *node; int ret = -ENOSPC; u64 bitmap_offset = offset_to_bitmap(ctl, offset); @@ -2469,10 +2468,6 @@ setup_cluster_bitmap(struct btrfs_block_group_cache *block_group, list_add(&entry->list, bitmaps); } - /* - * First check our cached list of bitmaps and see if there is an entry - * here that will work. - */ list_for_each_entry(entry, bitmaps, list) { if (entry->bytes < min_bytes) continue; @@ -2483,38 +2478,10 @@ setup_cluster_bitmap(struct btrfs_block_group_cache *block_group, } /* - * If we do have entries on our list and we are here then we didn't find - * anything, so go ahead and get the next entry after the last entry in - * this list and start the search from there. + * The bitmaps list has all the bitmaps that record free space + * starting after offset, so no more search is required. */ - if (!list_empty(bitmaps)) { - entry = list_entry(bitmaps->prev, struct btrfs_free_space, - list); - node = rb_next(&entry->offset_index); - if (!node) - return -ENOSPC; - entry = rb_entry(node, struct btrfs_free_space, offset_index); - goto search; - } - - entry = tree_search_offset(ctl, offset_to_bitmap(ctl, offset), 0, 1); - if (!entry) - return -ENOSPC; - -search: - node = &entry->offset_index; - do { - entry = rb_entry(node, struct btrfs_free_space, offset_index); - node = rb_next(&entry->offset_index); - if (!entry->bitmap) - continue; - if (entry->bytes < min_bytes) - continue; - ret = btrfs_bitmap_cluster(block_group, entry, cluster, offset, - bytes, min_bytes); - } while (ret && node); - - return ret; + return -ENOSPC; } /* @@ -2532,8 +2499,8 @@ int btrfs_find_space_cluster(struct btrfs_trans_handle *trans, u64 offset, u64 bytes, u64 empty_size) { struct btrfs_free_space_ctl *ctl = block_group->free_space_ctl; - struct list_head bitmaps; struct btrfs_free_space *entry, *tmp; + LIST_HEAD(bitmaps); u64 min_bytes; int ret; @@ -2572,7 +2539,6 @@ int btrfs_find_space_cluster(struct btrfs_trans_handle *trans, goto out; } - INIT_LIST_HEAD(&bitmaps); ret = setup_cluster_no_bitmap(block_group, cluster, &bitmaps, offset, bytes, min_bytes); if (ret) -- cgit v1.2.3-70-g09d2 From fadc0d8be4dfca80f6c568bc5874931893c6709b Mon Sep 17 00:00:00 2001 From: David Sterba Date: Sun, 20 Nov 2011 07:33:38 -0500 Subject: btrfs: fix stat blocks accounting Round inode bytes and delalloc bytes up to real blocksize before converting to sector size. Otherwise eg. files smaller than 512 are reported with zero blocks due to incorrect rounding. Signed-off-by: David Sterba Signed-off-by: Chris Mason --- fs/btrfs/inode.c | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-) (limited to 'fs') diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c index e16215f480d..8ad26b135a1 100644 --- a/fs/btrfs/inode.c +++ b/fs/btrfs/inode.c @@ -6794,11 +6794,13 @@ static int btrfs_getattr(struct vfsmount *mnt, struct dentry *dentry, struct kstat *stat) { struct inode *inode = dentry->d_inode; + u32 blocksize = inode->i_sb->s_blocksize; + generic_fillattr(inode, stat); stat->dev = BTRFS_I(inode)->root->anon_dev; stat->blksize = PAGE_CACHE_SIZE; - stat->blocks = (inode_get_bytes(inode) + - BTRFS_I(inode)->delalloc_bytes) >> 9; + stat->blocks = (ALIGN(inode_get_bytes(inode), blocksize) + + ALIGN(BTRFS_I(inode)->delalloc_bytes, blocksize)) >> 9; return 0; } -- cgit v1.2.3-70-g09d2 From 5bb1468238e20b15921909e9f9601e945f03bac7 Mon Sep 17 00:00:00 2001 From: Arnd Hannemann Date: Sun, 20 Nov 2011 07:33:38 -0500 Subject: Btrfs: prefix resize related printks with btrfs: For the user it is confusing to find something like: [10197.627710] new size for /dev/mapper/vg0-usr_share is 3221225472 in kernel log, because it doesn't point directly to btrfs. This patch prefixes those messages with "btrfs:" like other btrfs related printks. Signed-off-by: Arnd Hannemann Signed-off-by: Chris Mason --- fs/btrfs/ioctl.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) (limited to 'fs') diff --git a/fs/btrfs/ioctl.c b/fs/btrfs/ioctl.c index bda53fcd1a0..a90e749ed6d 100644 --- a/fs/btrfs/ioctl.c +++ b/fs/btrfs/ioctl.c @@ -1216,12 +1216,12 @@ static noinline int btrfs_ioctl_resize(struct btrfs_root *root, *devstr = '\0'; devstr = vol_args->name; devid = simple_strtoull(devstr, &end, 10); - printk(KERN_INFO "resizing devid %llu\n", + printk(KERN_INFO "btrfs: resizing devid %llu\n", (unsigned long long)devid); } device = btrfs_find_device(root, devid, NULL, NULL); if (!device) { - printk(KERN_INFO "resizer unable to find device %llu\n", + printk(KERN_INFO "btrfs: resizer unable to find device %llu\n", (unsigned long long)devid); ret = -EINVAL; goto out_unlock; @@ -1267,7 +1267,7 @@ static noinline int btrfs_ioctl_resize(struct btrfs_root *root, do_div(new_size, root->sectorsize); new_size *= root->sectorsize; - printk(KERN_INFO "new size for %s is %llu\n", + printk(KERN_INFO "btrfs: new size for %s is %llu\n", device->name, (unsigned long long)new_size); if (new_size > old_size) { -- cgit v1.2.3-70-g09d2 From 291c7d2f577428f896daa5002e784959328a80aa Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Mon, 14 Nov 2011 13:52:14 -0500 Subject: Btrfs: wait on caching if we're loading the free space cache We've been hitting panics when running xfstest 13 in a loop for long periods of time. And actually this problem has always existed so we've been hitting these things randomly for a while. Basically what happens is we get a thread coming into the allocator and reading the space cache off of disk and adding the entries to the free space cache as we go. Then we get another thread that comes in and tries to allocate from that block group. Since block_group->cached != BTRFS_CACHE_NO it goes ahead and tries to do the allocation. We do this because if we're doing the old slow way of caching we don't want to hold people up and wait for everything to finish. The problem with this is we could end up discarding the space cache at some arbitrary point in the future, which means we could very well end up allocating space that is either bad, or when the real caching happens it could end up thinking the space isn't in use when it really is and cause all sorts of other problems. The solution is to add a new flag to indicate we are loading the free space cache from disk, and always try to cache the block group if cache->cached != BTRFS_CACHE_FINISHED. That way if we are loading the space cache anybody else who tries to allocate from the block group will have to wait until it's finished to make sure it completes successfully. Thanks, Signed-off-by: Josef Bacik --- fs/btrfs/ctree.h | 3 +- fs/btrfs/extent-tree.c | 119 ++++++++++++++++++++++++++++++++----------------- 2 files changed, 81 insertions(+), 41 deletions(-) (limited to 'fs') diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h index b1cb3c05248..04a5dfcee5a 100644 --- a/fs/btrfs/ctree.h +++ b/fs/btrfs/ctree.h @@ -848,7 +848,8 @@ struct btrfs_free_cluster { enum btrfs_caching_type { BTRFS_CACHE_NO = 0, BTRFS_CACHE_STARTED = 1, - BTRFS_CACHE_FINISHED = 2, + BTRFS_CACHE_FAST = 2, + BTRFS_CACHE_FINISHED = 3, }; enum btrfs_disk_cache_state { diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 0f47b3e2010..5d86877f10e 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -467,13 +467,59 @@ static int cache_block_group(struct btrfs_block_group_cache *cache, struct btrfs_root *root, int load_cache_only) { + DEFINE_WAIT(wait); struct btrfs_fs_info *fs_info = cache->fs_info; struct btrfs_caching_control *caching_ctl; int ret = 0; - smp_mb(); - if (cache->cached != BTRFS_CACHE_NO) + caching_ctl = kzalloc(sizeof(*caching_ctl), GFP_NOFS); + BUG_ON(!caching_ctl); + + INIT_LIST_HEAD(&caching_ctl->list); + mutex_init(&caching_ctl->mutex); + init_waitqueue_head(&caching_ctl->wait); + caching_ctl->block_group = cache; + caching_ctl->progress = cache->key.objectid; + atomic_set(&caching_ctl->count, 1); + caching_ctl->work.func = caching_thread; + + spin_lock(&cache->lock); + /* + * This should be a rare occasion, but this could happen I think in the + * case where one thread starts to load the space cache info, and then + * some other thread starts a transaction commit which tries to do an + * allocation while the other thread is still loading the space cache + * info. The previous loop should have kept us from choosing this block + * group, but if we've moved to the state where we will wait on caching + * block groups we need to first check if we're doing a fast load here, + * so we can wait for it to finish, otherwise we could end up allocating + * from a block group who's cache gets evicted for one reason or + * another. + */ + while (cache->cached == BTRFS_CACHE_FAST) { + struct btrfs_caching_control *ctl; + + ctl = cache->caching_ctl; + atomic_inc(&ctl->count); + prepare_to_wait(&ctl->wait, &wait, TASK_UNINTERRUPTIBLE); + spin_unlock(&cache->lock); + + schedule(); + + finish_wait(&ctl->wait, &wait); + put_caching_control(ctl); + spin_lock(&cache->lock); + } + + if (cache->cached != BTRFS_CACHE_NO) { + spin_unlock(&cache->lock); + kfree(caching_ctl); return 0; + } + WARN_ON(cache->caching_ctl); + cache->caching_ctl = caching_ctl; + cache->cached = BTRFS_CACHE_FAST; + spin_unlock(&cache->lock); /* * We can't do the read from on-disk cache during a commit since we need @@ -484,56 +530,51 @@ static int cache_block_group(struct btrfs_block_group_cache *cache, if (trans && (!trans->transaction->in_commit) && (root && root != root->fs_info->tree_root) && btrfs_test_opt(root, SPACE_CACHE)) { - spin_lock(&cache->lock); - if (cache->cached != BTRFS_CACHE_NO) { - spin_unlock(&cache->lock); - return 0; - } - cache->cached = BTRFS_CACHE_STARTED; - spin_unlock(&cache->lock); - ret = load_free_space_cache(fs_info, cache); spin_lock(&cache->lock); if (ret == 1) { + cache->caching_ctl = NULL; cache->cached = BTRFS_CACHE_FINISHED; cache->last_byte_to_unpin = (u64)-1; } else { - cache->cached = BTRFS_CACHE_NO; + if (load_cache_only) { + cache->caching_ctl = NULL; + cache->cached = BTRFS_CACHE_NO; + } else { + cache->cached = BTRFS_CACHE_STARTED; + } } spin_unlock(&cache->lock); + wake_up(&caching_ctl->wait); if (ret == 1) { + put_caching_control(caching_ctl); free_excluded_extents(fs_info->extent_root, cache); return 0; } + } else { + /* + * We are not going to do the fast caching, set cached to the + * appropriate value and wakeup any waiters. + */ + spin_lock(&cache->lock); + if (load_cache_only) { + cache->caching_ctl = NULL; + cache->cached = BTRFS_CACHE_NO; + } else { + cache->cached = BTRFS_CACHE_STARTED; + } + spin_unlock(&cache->lock); + wake_up(&caching_ctl->wait); } - if (load_cache_only) - return 0; - - caching_ctl = kzalloc(sizeof(*caching_ctl), GFP_NOFS); - BUG_ON(!caching_ctl); - - INIT_LIST_HEAD(&caching_ctl->list); - mutex_init(&caching_ctl->mutex); - init_waitqueue_head(&caching_ctl->wait); - caching_ctl->block_group = cache; - caching_ctl->progress = cache->key.objectid; - /* one for caching kthread, one for caching block group list */ - atomic_set(&caching_ctl->count, 2); - caching_ctl->work.func = caching_thread; - - spin_lock(&cache->lock); - if (cache->cached != BTRFS_CACHE_NO) { - spin_unlock(&cache->lock); - kfree(caching_ctl); + if (load_cache_only) { + put_caching_control(caching_ctl); return 0; } - cache->caching_ctl = caching_ctl; - cache->cached = BTRFS_CACHE_STARTED; - spin_unlock(&cache->lock); down_write(&fs_info->extent_commit_sem); + atomic_inc(&caching_ctl->count); list_add_tail(&caching_ctl->list, &fs_info->caching_block_groups); up_write(&fs_info->extent_commit_sem); @@ -5177,13 +5218,15 @@ search: } have_block_group: - if (unlikely(block_group->cached == BTRFS_CACHE_NO)) { + cached = block_group_cache_done(block_group); + if (unlikely(!cached)) { u64 free_percent; + found_uncached_bg = true; ret = cache_block_group(block_group, trans, orig_root, 1); if (block_group->cached == BTRFS_CACHE_FINISHED) - goto have_block_group; + goto alloc; free_percent = btrfs_block_group_used(&block_group->item); free_percent *= 100; @@ -5205,7 +5248,6 @@ have_block_group: orig_root, 0); BUG_ON(ret); } - found_uncached_bg = true; /* * If loop is set for cached only, try the next block @@ -5215,10 +5257,7 @@ have_block_group: goto loop; } - cached = block_group_cache_done(block_group); - if (unlikely(!cached)) - found_uncached_bg = true; - +alloc: if (unlikely(block_group->ro)) goto loop; -- cgit v1.2.3-70-g09d2 From f7d61dcd6873c49bcc42be2caa2af1c2511aa915 Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Tue, 15 Nov 2011 09:31:24 -0500 Subject: Btrfs: clear pages dirty for io and set them extent mapped When doing the io_ctl helpers to clean up the free space cache stuff I stopped using our normal prepare_pages stuff, which means I of course forgot to do things like set the pages extent mapped, which will cause us all sorts of wonderful propblems. Thanks, Signed-off-by: Josef Bacik --- fs/btrfs/free-space-cache.c | 5 +++++ 1 file changed, 5 insertions(+) (limited to 'fs') diff --git a/fs/btrfs/free-space-cache.c b/fs/btrfs/free-space-cache.c index 8c32434da2c..aedacdbf77e 100644 --- a/fs/btrfs/free-space-cache.c +++ b/fs/btrfs/free-space-cache.c @@ -351,6 +351,11 @@ static int io_ctl_prepare_pages(struct io_ctl *io_ctl, struct inode *inode, } } + for (i = 0; i < io_ctl->num_pages; i++) { + clear_page_dirty_for_io(io_ctl->pages[i]); + set_page_extent_mapped(io_ctl->pages[i]); + } + return 0; } -- cgit v1.2.3-70-g09d2 From 4d479cf010d56ec9c54f3099992d039918f1296b Mon Sep 17 00:00:00 2001 From: Josef Bacik Date: Thu, 17 Nov 2011 11:34:31 -0500 Subject: Btrfs: sectorsize align offsets in fiemap We've been hitting BUG()'s in btrfs_cont_expand and btrfs_fallocate and anywhere else that calls btrfs_get_extent while running xfstests 13 in a loop. This is because fiemap is calling btrfs_get_extent with non-sectorsize aligned offsets, which will end up adding mappings that are not sectorsize aligned, which will cause problems in some cases for subsequent calls to btrfs_get_extent for similar areas that are sectorsize aligned. With this patch I ran xfstests 13 in a loop for a couple of hours and didn't hit the problem that I could previously hit in at most 20 minutes. Thanks, Signed-off-by: Josef Bacik --- fs/btrfs/extent_io.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/btrfs/extent_io.c b/fs/btrfs/extent_io.c index a877b8b212e..9472d3de5e5 100644 --- a/fs/btrfs/extent_io.c +++ b/fs/btrfs/extent_io.c @@ -3366,6 +3366,9 @@ int extent_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, return -ENOMEM; path->leave_spinning = 1; + start = ALIGN(start, BTRFS_I(inode)->root->sectorsize); + len = ALIGN(len, BTRFS_I(inode)->root->sectorsize); + /* * lookup the last file extent. We're not using i_size here * because there might be preallocation past i_size @@ -3413,7 +3416,7 @@ int extent_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, lock_extent_bits(&BTRFS_I(inode)->io_tree, start, start + len, 0, &cached_state, GFP_NOFS); - em = get_extent_skip_holes(inode, off, last_for_get_extent, + em = get_extent_skip_holes(inode, start, last_for_get_extent, get_extent); if (!em) goto out; -- cgit v1.2.3-70-g09d2 From dd179946db2493646955efc112d73c85b3cafcb1 Mon Sep 17 00:00:00 2001 From: David Howells Date: Tue, 16 Aug 2011 15:31:30 +0100 Subject: VFS: Log the fact that we've given ELOOP rather than creating a loop To prevent an NFS server from being used to create a directory loop in an NFS superblock on the client, the following patch was committed: commit 1836750115f20b774e55c032a3893e8c5bdf41ed Author: Al Viro Date: Tue Jul 12 21:42:24 2011 -0400 Subject: fix loop checks in d_materialise_unique() This causes ELOOP to be reported to anyone trying to access the dentry that would otherwise cause the kernel to complete the loop. However, no indication is given to the caller as to why an operation that ought to work doesn't. The fault is with the kernel, which doesn't want to try and solve the problem as it gets horrendously messy if there's another mountpoint somewhere in the trees being spliced that can't be moved[*]. [*] The real problem is that we don't handle the excision of a subtree that gets moved _out_ of what we can see. This can happen on the server where a directory is merely moved between two other dirs on the same filesystem, but where destination dir is not accessible by the client. So, given the choice to return ELOOP rather than trying to reconfigure the dentry tree, we should give the caller some indication of why they aren't being allowed to make what should be a legitimate request and log a message. Signed-off-by: David Howells Acked-by: Sachin Prabhu Signed-off-by: Al Viro --- fs/dcache.c | 11 ++++++++++- 1 file changed, 10 insertions(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/dcache.c b/fs/dcache.c index a901c6901bc..10ba92def3f 100644 --- a/fs/dcache.c +++ b/fs/dcache.c @@ -36,6 +36,7 @@ #include #include #include +#include #include "internal.h" /* @@ -2383,8 +2384,16 @@ struct dentry *d_materialise_unique(struct dentry *dentry, struct inode *inode) actual = __d_unalias(inode, dentry, alias); } write_sequnlock(&rename_lock); - if (IS_ERR(actual)) + if (IS_ERR(actual)) { + if (PTR_ERR(actual) == -ELOOP) + pr_warn_ratelimited( + "VFS: Lookup of '%s' in %s %s" + " would have caused loop\n", + dentry->d_name.name, + inode->i_sb->s_type->name, + inode->i_sb->s_id); dput(alias); + } goto out_nolock; } } -- cgit v1.2.3-70-g09d2 From 6e58ad69efe9f4c91eb15f6bc365293414c397ce Mon Sep 17 00:00:00 2001 From: Yongqiang Yang Date: Mon, 21 Nov 2011 12:09:19 -0500 Subject: ext4: fix up a undefined error in ext4_free_blocks in debugging code sbi is not defined, so let ext4_free_blocks use EXT4_SB(sb) instead when EXT4FS_DEBUG is defined. Signed-off-by: Yongqiang Yang --- fs/ext4/balloc.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/ext4/balloc.c b/fs/ext4/balloc.c index f6dba4505f1..12ccacda44e 100644 --- a/fs/ext4/balloc.c +++ b/fs/ext4/balloc.c @@ -565,7 +565,7 @@ ext4_fsblk_t ext4_count_free_clusters(struct super_block *sb) brelse(bitmap_bh); printk(KERN_DEBUG "ext4_count_free_clusters: stored = %llu" ", computed = %llu, %llu\n", - EXT4_B2C(sbi, ext4_free_blocks_count(es)), + EXT4_B2C(EXT4_SB(sb), ext4_free_blocks_count(es)), desc_count, bitmap_count); return bitmap_count; #else -- cgit v1.2.3-70-g09d2 From 24a70313969fc3fc440216b40babdb42564acff3 Mon Sep 17 00:00:00 2001 From: Chris Mason Date: Mon, 21 Nov 2011 09:39:11 -0500 Subject: Btrfs: remove free-space-cache.c WARN during log replay The log replay code only partially loads block groups, since the block group caching code is able to detect and deal with extents the logging code has pinned down. While the logging code is pinning down block groups, there is a bogus WARN_ON we're hitting if the code wasn't able to find an extent in the cache. This commit removes the warning because it can happen any time there isn't a valid free space cache for that block group. Signed-off-by: Chris Mason --- fs/btrfs/free-space-cache.c | 8 +++++++- 1 file changed, 7 insertions(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/btrfs/free-space-cache.c b/fs/btrfs/free-space-cache.c index aedacdbf77e..6e5b7e46369 100644 --- a/fs/btrfs/free-space-cache.c +++ b/fs/btrfs/free-space-cache.c @@ -1849,7 +1849,13 @@ again: info = tree_search_offset(ctl, offset_to_bitmap(ctl, offset), 1, 0); if (!info) { - WARN_ON(1); + /* the tree logging code might be calling us before we + * have fully loaded the free space rbtree for this + * block group. So it is possible the entry won't + * be in the rbtree yet at all. The caching code + * will make sure not to put it in the rbtree if + * the logging code has pinned it. + */ goto out_lock; } } -- cgit v1.2.3-70-g09d2 From d31da0f0ba3bc0a827a63879310818c22d9a95be Mon Sep 17 00:00:00 2001 From: Al Viro Date: Tue, 22 Nov 2011 12:31:21 -0500 Subject: mount_subtree() pointless use-after-free d'oh... we'd carefully pinned mnt->mnt_sb down, dropped mnt and attempt to grab s_umount on mnt->mnt_sb. The trouble is, *mnt might've been overwritten by now... Signed-off-by: Al Viro --- fs/namespace.c | 6 ++++-- 1 file changed, 4 insertions(+), 2 deletions(-) (limited to 'fs') diff --git a/fs/namespace.c b/fs/namespace.c index 50ee30345b4..6d3a1963879 100644 --- a/fs/namespace.c +++ b/fs/namespace.c @@ -2493,6 +2493,7 @@ EXPORT_SYMBOL(create_mnt_ns); struct dentry *mount_subtree(struct vfsmount *mnt, const char *name) { struct mnt_namespace *ns; + struct super_block *s; struct path path; int err; @@ -2509,10 +2510,11 @@ struct dentry *mount_subtree(struct vfsmount *mnt, const char *name) return ERR_PTR(err); /* trade a vfsmount reference for active sb one */ - atomic_inc(&path.mnt->mnt_sb->s_active); + s = path.mnt->mnt_sb; + atomic_inc(&s->s_active); mntput(path.mnt); /* lock the sucker */ - down_write(&path.mnt->mnt_sb->s_umount); + down_write(&s->s_umount); /* ... and return the root of (sub)tree on it */ return path.dentry; } -- cgit v1.2.3-70-g09d2 From b59db43ad4434519feb338eacb01d77eb50825c5 Mon Sep 17 00:00:00 2001 From: Tyler Hicks Date: Mon, 21 Nov 2011 17:31:02 -0600 Subject: eCryptfs: Prevent file create race condition The file creation path prematurely called d_instantiate() and unlock_new_inode() before the eCryptfs inode info was fully allocated and initialized and before the eCryptfs metadata was written to the lower file. This could result in race conditions in subsequent file and inode operations leading to unexpected error conditions or a null pointer dereference while attempting to use the unallocated memory. https://launchpad.net/bugs/813146 Signed-off-by: Tyler Hicks Cc: stable@kernel.org --- fs/ecryptfs/crypto.c | 22 +++++++++--------- fs/ecryptfs/ecryptfs_kernel.h | 5 +++-- fs/ecryptfs/inode.c | 52 ++++++++++++++++++++++++++----------------- 3 files changed, 46 insertions(+), 33 deletions(-) (limited to 'fs') diff --git a/fs/ecryptfs/crypto.c b/fs/ecryptfs/crypto.c index 58609bde3b9..203a1fdff66 100644 --- a/fs/ecryptfs/crypto.c +++ b/fs/ecryptfs/crypto.c @@ -967,7 +967,7 @@ static void ecryptfs_set_default_crypt_stat_vals( /** * ecryptfs_new_file_context - * @ecryptfs_dentry: The eCryptfs dentry + * @ecryptfs_inode: The eCryptfs inode * * If the crypto context for the file has not yet been established, * this is where we do that. Establishing a new crypto context @@ -984,13 +984,13 @@ static void ecryptfs_set_default_crypt_stat_vals( * * Returns zero on success; non-zero otherwise */ -int ecryptfs_new_file_context(struct dentry *ecryptfs_dentry) +int ecryptfs_new_file_context(struct inode *ecryptfs_inode) { struct ecryptfs_crypt_stat *crypt_stat = - &ecryptfs_inode_to_private(ecryptfs_dentry->d_inode)->crypt_stat; + &ecryptfs_inode_to_private(ecryptfs_inode)->crypt_stat; struct ecryptfs_mount_crypt_stat *mount_crypt_stat = &ecryptfs_superblock_to_private( - ecryptfs_dentry->d_sb)->mount_crypt_stat; + ecryptfs_inode->i_sb)->mount_crypt_stat; int cipher_name_len; int rc = 0; @@ -1299,12 +1299,12 @@ static int ecryptfs_write_headers_virt(char *page_virt, size_t max, } static int -ecryptfs_write_metadata_to_contents(struct dentry *ecryptfs_dentry, +ecryptfs_write_metadata_to_contents(struct inode *ecryptfs_inode, char *virt, size_t virt_len) { int rc; - rc = ecryptfs_write_lower(ecryptfs_dentry->d_inode, virt, + rc = ecryptfs_write_lower(ecryptfs_inode, virt, 0, virt_len); if (rc < 0) printk(KERN_ERR "%s: Error attempting to write header " @@ -1338,7 +1338,8 @@ static unsigned long ecryptfs_get_zeroed_pages(gfp_t gfp_mask, /** * ecryptfs_write_metadata - * @ecryptfs_dentry: The eCryptfs dentry + * @ecryptfs_dentry: The eCryptfs dentry, which should be negative + * @ecryptfs_inode: The newly created eCryptfs inode * * Write the file headers out. This will likely involve a userspace * callout, in which the session key is encrypted with one or more @@ -1348,10 +1349,11 @@ static unsigned long ecryptfs_get_zeroed_pages(gfp_t gfp_mask, * * Returns zero on success; non-zero on error */ -int ecryptfs_write_metadata(struct dentry *ecryptfs_dentry) +int ecryptfs_write_metadata(struct dentry *ecryptfs_dentry, + struct inode *ecryptfs_inode) { struct ecryptfs_crypt_stat *crypt_stat = - &ecryptfs_inode_to_private(ecryptfs_dentry->d_inode)->crypt_stat; + &ecryptfs_inode_to_private(ecryptfs_inode)->crypt_stat; unsigned int order; char *virt; size_t virt_len; @@ -1391,7 +1393,7 @@ int ecryptfs_write_metadata(struct dentry *ecryptfs_dentry) rc = ecryptfs_write_metadata_to_xattr(ecryptfs_dentry, virt, size); else - rc = ecryptfs_write_metadata_to_contents(ecryptfs_dentry, virt, + rc = ecryptfs_write_metadata_to_contents(ecryptfs_inode, virt, virt_len); if (rc) { printk(KERN_ERR "%s: Error writing metadata out to lower file; " diff --git a/fs/ecryptfs/ecryptfs_kernel.h b/fs/ecryptfs/ecryptfs_kernel.h index 54481a3b2c7..a9f29b12fbf 100644 --- a/fs/ecryptfs/ecryptfs_kernel.h +++ b/fs/ecryptfs/ecryptfs_kernel.h @@ -584,9 +584,10 @@ int ecryptfs_init_crypt_ctx(struct ecryptfs_crypt_stat *crypt_stat); int ecryptfs_write_inode_size_to_metadata(struct inode *ecryptfs_inode); int ecryptfs_encrypt_page(struct page *page); int ecryptfs_decrypt_page(struct page *page); -int ecryptfs_write_metadata(struct dentry *ecryptfs_dentry); +int ecryptfs_write_metadata(struct dentry *ecryptfs_dentry, + struct inode *ecryptfs_inode); int ecryptfs_read_metadata(struct dentry *ecryptfs_dentry); -int ecryptfs_new_file_context(struct dentry *ecryptfs_dentry); +int ecryptfs_new_file_context(struct inode *ecryptfs_inode); void ecryptfs_write_crypt_stat_flags(char *page_virt, struct ecryptfs_crypt_stat *crypt_stat, size_t *written); diff --git a/fs/ecryptfs/inode.c b/fs/ecryptfs/inode.c index a36d327f152..32f90a3ae63 100644 --- a/fs/ecryptfs/inode.c +++ b/fs/ecryptfs/inode.c @@ -172,22 +172,23 @@ ecryptfs_create_underlying_file(struct inode *lower_dir_inode, * it. It will also update the eCryptfs directory inode to mimic the * stat of the lower directory inode. * - * Returns zero on success; non-zero on error condition + * Returns the new eCryptfs inode on success; an ERR_PTR on error condition */ -static int +static struct inode * ecryptfs_do_create(struct inode *directory_inode, struct dentry *ecryptfs_dentry, int mode) { int rc; struct dentry *lower_dentry; struct dentry *lower_dir_dentry; + struct inode *inode; lower_dentry = ecryptfs_dentry_to_lower(ecryptfs_dentry); lower_dir_dentry = lock_parent(lower_dentry); if (IS_ERR(lower_dir_dentry)) { ecryptfs_printk(KERN_ERR, "Error locking directory of " "dentry\n"); - rc = PTR_ERR(lower_dir_dentry); + inode = ERR_CAST(lower_dir_dentry); goto out; } rc = ecryptfs_create_underlying_file(lower_dir_dentry->d_inode, @@ -195,20 +196,19 @@ ecryptfs_do_create(struct inode *directory_inode, if (rc) { printk(KERN_ERR "%s: Failure to create dentry in lower fs; " "rc = [%d]\n", __func__, rc); + inode = ERR_PTR(rc); goto out_lock; } - rc = ecryptfs_interpose(lower_dentry, ecryptfs_dentry, - directory_inode->i_sb); - if (rc) { - ecryptfs_printk(KERN_ERR, "Failure in ecryptfs_interpose\n"); + inode = __ecryptfs_get_inode(lower_dentry->d_inode, + directory_inode->i_sb); + if (IS_ERR(inode)) goto out_lock; - } fsstack_copy_attr_times(directory_inode, lower_dir_dentry->d_inode); fsstack_copy_inode_size(directory_inode, lower_dir_dentry->d_inode); out_lock: unlock_dir(lower_dir_dentry); out: - return rc; + return inode; } /** @@ -219,26 +219,26 @@ out: * * Returns zero on success */ -static int ecryptfs_initialize_file(struct dentry *ecryptfs_dentry) +static int ecryptfs_initialize_file(struct dentry *ecryptfs_dentry, + struct inode *ecryptfs_inode) { struct ecryptfs_crypt_stat *crypt_stat = - &ecryptfs_inode_to_private(ecryptfs_dentry->d_inode)->crypt_stat; + &ecryptfs_inode_to_private(ecryptfs_inode)->crypt_stat; int rc = 0; - if (S_ISDIR(ecryptfs_dentry->d_inode->i_mode)) { + if (S_ISDIR(ecryptfs_inode->i_mode)) { ecryptfs_printk(KERN_DEBUG, "This is a directory\n"); crypt_stat->flags &= ~(ECRYPTFS_ENCRYPTED); goto out; } ecryptfs_printk(KERN_DEBUG, "Initializing crypto context\n"); - rc = ecryptfs_new_file_context(ecryptfs_dentry); + rc = ecryptfs_new_file_context(ecryptfs_inode); if (rc) { ecryptfs_printk(KERN_ERR, "Error creating new file " "context; rc = [%d]\n", rc); goto out; } - rc = ecryptfs_get_lower_file(ecryptfs_dentry, - ecryptfs_dentry->d_inode); + rc = ecryptfs_get_lower_file(ecryptfs_dentry, ecryptfs_inode); if (rc) { printk(KERN_ERR "%s: Error attempting to initialize " "the lower file for the dentry with name " @@ -246,10 +246,10 @@ static int ecryptfs_initialize_file(struct dentry *ecryptfs_dentry) ecryptfs_dentry->d_name.name, rc); goto out; } - rc = ecryptfs_write_metadata(ecryptfs_dentry); + rc = ecryptfs_write_metadata(ecryptfs_dentry, ecryptfs_inode); if (rc) printk(KERN_ERR "Error writing headers; rc = [%d]\n", rc); - ecryptfs_put_lower_file(ecryptfs_dentry->d_inode); + ecryptfs_put_lower_file(ecryptfs_inode); out: return rc; } @@ -269,18 +269,28 @@ static int ecryptfs_create(struct inode *directory_inode, struct dentry *ecryptfs_dentry, int mode, struct nameidata *nd) { + struct inode *ecryptfs_inode; int rc; - /* ecryptfs_do_create() calls ecryptfs_interpose() */ - rc = ecryptfs_do_create(directory_inode, ecryptfs_dentry, mode); - if (unlikely(rc)) { + ecryptfs_inode = ecryptfs_do_create(directory_inode, ecryptfs_dentry, + mode); + if (unlikely(IS_ERR(ecryptfs_inode))) { ecryptfs_printk(KERN_WARNING, "Failed to create file in" "lower filesystem\n"); + rc = PTR_ERR(ecryptfs_inode); goto out; } /* At this point, a file exists on "disk"; we need to make sure * that this on disk file is prepared to be an ecryptfs file */ - rc = ecryptfs_initialize_file(ecryptfs_dentry); + rc = ecryptfs_initialize_file(ecryptfs_dentry, ecryptfs_inode); + if (rc) { + drop_nlink(ecryptfs_inode); + unlock_new_inode(ecryptfs_inode); + iput(ecryptfs_inode); + goto out; + } + d_instantiate(ecryptfs_dentry, ecryptfs_inode); + unlock_new_inode(ecryptfs_inode); out: return rc; } -- cgit v1.2.3-70-g09d2 From 32001d6fe9ac6b0423e674a3093aa56740849f3b Mon Sep 17 00:00:00 2001 From: Tyler Hicks Date: Mon, 21 Nov 2011 17:31:29 -0600 Subject: eCryptfs: Flush file in vma close Dirty pages weren't being written back when an mmap'ed eCryptfs file was closed before the mapping was unmapped. Since f_ops->flush() is not called by the munmap() path, the lower file was simply being released. This patch flushes the eCryptfs file in the vm_ops->close() path. https://launchpad.net/bugs/870326 Signed-off-by: Tyler Hicks Cc: stable@kernel.org [2.6.39+] --- fs/ecryptfs/file.c | 23 ++++++++++++++++++++++- 1 file changed, 22 insertions(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/ecryptfs/file.c b/fs/ecryptfs/file.c index c6ac98cf9ba..d3f95f941c4 100644 --- a/fs/ecryptfs/file.c +++ b/fs/ecryptfs/file.c @@ -139,6 +139,27 @@ out: return rc; } +static void ecryptfs_vma_close(struct vm_area_struct *vma) +{ + filemap_write_and_wait(vma->vm_file->f_mapping); +} + +static const struct vm_operations_struct ecryptfs_file_vm_ops = { + .close = ecryptfs_vma_close, + .fault = filemap_fault, +}; + +static int ecryptfs_file_mmap(struct file *file, struct vm_area_struct *vma) +{ + int rc; + + rc = generic_file_mmap(file, vma); + if (!rc) + vma->vm_ops = &ecryptfs_file_vm_ops; + + return rc; +} + struct kmem_cache *ecryptfs_file_info_cache; /** @@ -349,7 +370,7 @@ const struct file_operations ecryptfs_main_fops = { #ifdef CONFIG_COMPAT .compat_ioctl = ecryptfs_compat_ioctl, #endif - .mmap = generic_file_mmap, + .mmap = ecryptfs_file_mmap, .open = ecryptfs_open, .flush = ecryptfs_flush, .release = ecryptfs_release, -- cgit v1.2.3-70-g09d2 From 0f751e641a71157aa584c2a2e22fda52b52b8a56 Mon Sep 17 00:00:00 2001 From: Tyler Hicks Date: Wed, 23 Nov 2011 11:31:24 -0600 Subject: eCryptfs: Extend array bounds for all filename chars From mhalcrow's original commit message: Characters with ASCII values greater than the size of filename_rev_map[] are valid filename characters. ecryptfs_decode_from_filename() will access kernel memory beyond that array, and ecryptfs_parse_tag_70_packet() will then decrypt those characters. The attacker, using the FNEK of the crafted file, can then re-encrypt the characters to reveal the kernel memory past the end of the filename_rev_map[] array. I expect low security impact since this array is statically allocated in the text area, and the amount of memory past the array that is accessible is limited by the largest possible ASCII filename character. This patch solves the issue reported by mhalcrow but with an implementation suggested by Linus to simply extend the length of filename_rev_map[] to 256. Characters greater than 0x7A are mapped to 0x00, which is how invalid characters less than 0x7A were previously being handled. Signed-off-by: Tyler Hicks Reported-by: Michael Halcrow Cc: stable@kernel.org --- fs/ecryptfs/crypto.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'fs') diff --git a/fs/ecryptfs/crypto.c b/fs/ecryptfs/crypto.c index 203a1fdff66..2a834255c75 100644 --- a/fs/ecryptfs/crypto.c +++ b/fs/ecryptfs/crypto.c @@ -1945,7 +1945,7 @@ static unsigned char *portable_filename_chars = ("-.0123456789ABCD" /* We could either offset on every reverse map or just pad some 0x00's * at the front here */ -static const unsigned char filename_rev_map[] = { +static const unsigned char filename_rev_map[256] = { 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 7 */ 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 15 */ 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, /* 23 */ @@ -1961,7 +1961,7 @@ static const unsigned char filename_rev_map[] = { 0x00, 0x26, 0x27, 0x28, 0x29, 0x2A, 0x2B, 0x2C, /* 103 */ 0x2D, 0x2E, 0x2F, 0x30, 0x31, 0x32, 0x33, 0x34, /* 111 */ 0x35, 0x36, 0x37, 0x38, 0x39, 0x3A, 0x3B, 0x3C, /* 119 */ - 0x3D, 0x3E, 0x3F + 0x3D, 0x3E, 0x3F /* 123 - 255 initialized to 0x00 */ }; /** -- cgit v1.2.3-70-g09d2 From 4c81f045c0bd2cbb78cc6446a4cd98038fe11a2e Mon Sep 17 00:00:00 2001 From: Tejun Heo Date: Thu, 24 Nov 2011 19:22:24 -0500 Subject: ext4: fix racy use-after-free in ext4_end_io_dio() ext4_end_io_dio() queues io_end->work and then clears iocb->private; however, io_end->work calls aio_complete() which frees the iocb object. If that slab object gets reallocated, then ext4_end_io_dio() can end up clearing someone else's iocb->private, this use-after-free can cause a leak of a struct ext4_io_end_t structure. Detected and tested with slab poisoning. [ Note: Can also reproduce using 12 fio's against 12 file systems with the following configuration file: [global] direct=1 ioengine=libaio iodepth=1 bs=4k ba=4k size=128m [create] filename=${TESTDIR} rw=write -- tytso ] Google-Bug-Id: 5354697 Signed-off-by: Tejun Heo Signed-off-by: "Theodore Ts'o" Reported-by: Kent Overstreet Tested-by: Kent Overstreet Cc: stable@kernel.org --- fs/ext4/inode.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/ext4/inode.c b/fs/ext4/inode.c index fffec40d599..848f436df29 100644 --- a/fs/ext4/inode.c +++ b/fs/ext4/inode.c @@ -2807,8 +2807,8 @@ out: spin_unlock_irqrestore(&ei->i_completed_io_lock, flags); /* queue the work to convert unwritten extents to written */ - queue_work(wq, &io_end->work); iocb->private = NULL; + queue_work(wq, &io_end->work); /* XXX: probably should move into the real I/O completion handler */ inode_dio_done(inode); -- cgit v1.2.3-70-g09d2 From fa8b18edd752a8b4e9d1ee2cd615b82c93cf8bba Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Sun, 20 Nov 2011 15:35:32 +0000 Subject: xfs: validate acl count This prevents in-memory corruption and possible panics if the on-disk ACL is badly corrupted. Signed-off-by: Christoph Hellwig Signed-off-by: Ben Myers --- fs/xfs/xfs_acl.c | 2 ++ 1 file changed, 2 insertions(+) (limited to 'fs') diff --git a/fs/xfs/xfs_acl.c b/fs/xfs/xfs_acl.c index b6c4b3795c4..76e4266d2e7 100644 --- a/fs/xfs/xfs_acl.c +++ b/fs/xfs/xfs_acl.c @@ -42,6 +42,8 @@ xfs_acl_from_disk(struct xfs_acl *aclp) int count, i; count = be32_to_cpu(aclp->acl_cnt); + if (count > XFS_ACL_MAX_ENTRIES) + return ERR_PTR(-EFSCORRUPTED); acl = posix_acl_alloc(count, GFP_KERNEL); if (!acl) -- cgit v1.2.3-70-g09d2 From 4dd2cb4a28b7ab1f37163a4eba280926a13a8749 Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Tue, 29 Nov 2011 12:06:14 -0600 Subject: xfs: force buffer writeback before blocking on the ilock in inode reclaim If we are doing synchronous inode reclaim we block the VM from making progress in memory reclaim. So if we encouter a flush locked inode promote it in the delwri list and wake up xfsbufd to write it out now. Without this we can get hangs of up to 30 seconds during workloads hitting synchronous inode reclaim. The scheme is copied from what we do for dquot reclaims. Reported-by: Simon Kirby Signed-off-by: Christoph Hellwig Tested-by: Simon Kirby Signed-off-by: Ben Myers --- fs/xfs/xfs_inode.c | 21 +++++++++++++++++++++ fs/xfs/xfs_inode.h | 1 + fs/xfs/xfs_sync.c | 11 +++++++++++ 3 files changed, 33 insertions(+) (limited to 'fs') diff --git a/fs/xfs/xfs_inode.c b/fs/xfs/xfs_inode.c index c0237c602f1..755ee816488 100644 --- a/fs/xfs/xfs_inode.c +++ b/fs/xfs/xfs_inode.c @@ -2835,6 +2835,27 @@ corrupt_out: return XFS_ERROR(EFSCORRUPTED); } +void +xfs_promote_inode( + struct xfs_inode *ip) +{ + struct xfs_buf *bp; + + ASSERT(xfs_isilocked(ip, XFS_ILOCK_EXCL|XFS_ILOCK_SHARED)); + + bp = xfs_incore(ip->i_mount->m_ddev_targp, ip->i_imap.im_blkno, + ip->i_imap.im_len, XBF_TRYLOCK); + if (!bp) + return; + + if (XFS_BUF_ISDELAYWRITE(bp)) { + xfs_buf_delwri_promote(bp); + wake_up_process(ip->i_mount->m_ddev_targp->bt_task); + } + + xfs_buf_relse(bp); +} + /* * Return a pointer to the extent record at file index idx. */ diff --git a/fs/xfs/xfs_inode.h b/fs/xfs/xfs_inode.h index 760140d1dd6..b4cd4739f98 100644 --- a/fs/xfs/xfs_inode.h +++ b/fs/xfs/xfs_inode.h @@ -498,6 +498,7 @@ int xfs_iunlink(struct xfs_trans *, xfs_inode_t *); void xfs_iext_realloc(xfs_inode_t *, int, int); void xfs_iunpin_wait(xfs_inode_t *); int xfs_iflush(xfs_inode_t *, uint); +void xfs_promote_inode(struct xfs_inode *); void xfs_lock_inodes(xfs_inode_t **, int, uint); void xfs_lock_two_inodes(xfs_inode_t *, xfs_inode_t *, uint); diff --git a/fs/xfs/xfs_sync.c b/fs/xfs/xfs_sync.c index aa3dc1a4d53..be5c51d8f75 100644 --- a/fs/xfs/xfs_sync.c +++ b/fs/xfs/xfs_sync.c @@ -770,6 +770,17 @@ restart: if (!xfs_iflock_nowait(ip)) { if (!(sync_mode & SYNC_WAIT)) goto out; + + /* + * If we only have a single dirty inode in a cluster there is + * a fair chance that the AIL push may have pushed it into + * the buffer, but xfsbufd won't touch it until 30 seconds + * from now, and thus we will lock up here. + * + * Promote the inode buffer to the front of the delwri list + * and wake up xfsbufd now. + */ + xfs_promote_inode(ip); xfs_iflock(ip); } -- cgit v1.2.3-70-g09d2 From 4c393a6059f8442a70512a48ce4639b882b6f6ad Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Sat, 19 Nov 2011 17:44:30 +0000 Subject: xfs: fix attr2 vs large data fork assert With Dmitry fsstress updates I've seen very reproducible crashes in xfs_attr_shortform_remove because xfs_attr_shortform_bytesfit claims that the attributes would not fit inline into the inode after removing an attribute. It turns out that we were operating on an inode with lots of delalloc extents, and thus an if_bytes values for the data fork that is larger than biggest possible on-disk storage for it which utterly confuses the code near the end of xfs_attr_shortform_bytesfit. Fix this by always allowing the current attribute fork, like we already do for the attr1 format, given that delalloc conversion will take care for moving either the data or attribute area out of line if it doesn't fit at that point - or making the point moot by merging extents at this point. Also document the function better, and clean up some loose bits. Reviewed-by: Dave Chinner Signed-off-by: Christoph Hellwig Signed-off-by: Ben Myers --- fs/xfs/xfs_attr_leaf.c | 64 ++++++++++++++++++++++++++++++-------------------- 1 file changed, 39 insertions(+), 25 deletions(-) (limited to 'fs') diff --git a/fs/xfs/xfs_attr_leaf.c b/fs/xfs/xfs_attr_leaf.c index d4906e7c978..c1b55e59655 100644 --- a/fs/xfs/xfs_attr_leaf.c +++ b/fs/xfs/xfs_attr_leaf.c @@ -110,6 +110,7 @@ xfs_attr_namesp_match(int arg_flags, int ondisk_flags) /* * Query whether the requested number of additional bytes of extended * attribute space will be able to fit inline. + * * Returns zero if not, else the di_forkoff fork offset to be used in the * literal area for attribute data once the new bytes have been added. * @@ -122,7 +123,7 @@ xfs_attr_shortform_bytesfit(xfs_inode_t *dp, int bytes) int offset; int minforkoff; /* lower limit on valid forkoff locations */ int maxforkoff; /* upper limit on valid forkoff locations */ - int dsize; + int dsize; xfs_mount_t *mp = dp->i_mount; offset = (XFS_LITINO(mp) - bytes) >> 3; /* rounded down */ @@ -136,47 +137,60 @@ xfs_attr_shortform_bytesfit(xfs_inode_t *dp, int bytes) return (offset >= minforkoff) ? minforkoff : 0; } - if (!(mp->m_flags & XFS_MOUNT_ATTR2)) { - if (bytes <= XFS_IFORK_ASIZE(dp)) - return dp->i_d.di_forkoff; + /* + * If the requested numbers of bytes is smaller or equal to the + * current attribute fork size we can always proceed. + * + * Note that if_bytes in the data fork might actually be larger than + * the current data fork size is due to delalloc extents. In that + * case either the extent count will go down when they are converted + * to real extents, or the delalloc conversion will take care of the + * literal area rebalancing. + */ + if (bytes <= XFS_IFORK_ASIZE(dp)) + return dp->i_d.di_forkoff; + + /* + * For attr2 we can try to move the forkoff if there is space in the + * literal area, but for the old format we are done if there is no + * space in the fixed attribute fork. + */ + if (!(mp->m_flags & XFS_MOUNT_ATTR2)) return 0; - } dsize = dp->i_df.if_bytes; - + switch (dp->i_d.di_format) { case XFS_DINODE_FMT_EXTENTS: - /* + /* * If there is no attr fork and the data fork is extents, - * determine if creating the default attr fork will result - * in the extents form migrating to btree. If so, the - * minimum offset only needs to be the space required for + * determine if creating the default attr fork will result + * in the extents form migrating to btree. If so, the + * minimum offset only needs to be the space required for * the btree root. - */ + */ if (!dp->i_d.di_forkoff && dp->i_df.if_bytes > xfs_default_attroffset(dp)) dsize = XFS_BMDR_SPACE_CALC(MINDBTPTRS); break; - case XFS_DINODE_FMT_BTREE: /* - * If have data btree then keep forkoff if we have one, - * otherwise we are adding a new attr, so then we set - * minforkoff to where the btree root can finish so we have + * If we have a data btree then keep forkoff if we have one, + * otherwise we are adding a new attr, so then we set + * minforkoff to where the btree root can finish so we have * plenty of room for attrs */ if (dp->i_d.di_forkoff) { - if (offset < dp->i_d.di_forkoff) + if (offset < dp->i_d.di_forkoff) return 0; - else - return dp->i_d.di_forkoff; - } else - dsize = XFS_BMAP_BROOT_SPACE(dp->i_df.if_broot); + return dp->i_d.di_forkoff; + } + dsize = XFS_BMAP_BROOT_SPACE(dp->i_df.if_broot); break; } - - /* - * A data fork btree root must have space for at least + + /* + * A data fork btree root must have space for at least * MINDBTPTRS key/ptr pairs if the data fork is small or empty. */ minforkoff = MAX(dsize, XFS_BMDR_SPACE_CALC(MINDBTPTRS)); @@ -186,10 +200,10 @@ xfs_attr_shortform_bytesfit(xfs_inode_t *dp, int bytes) maxforkoff = XFS_LITINO(mp) - XFS_BMDR_SPACE_CALC(MINABTPTRS); maxforkoff = maxforkoff >> 3; /* rounded down */ - if (offset >= minforkoff && offset < maxforkoff) - return offset; if (offset >= maxforkoff) return maxforkoff; + if (offset >= minforkoff) + return offset; return 0; } -- cgit v1.2.3-70-g09d2 From 26bdef541d26fd6a5ddffdf8949ace22f94f809f Mon Sep 17 00:00:00 2001 From: Dan Carpenter Date: Wed, 16 Nov 2011 11:28:01 +0300 Subject: btrfs scrub: handle -ENOMEM from init_ipath() init_ipath() can return an ERR_PTR(-ENOMEM). Signed-off-by: Dan Carpenter --- fs/btrfs/scrub.c | 5 +++++ 1 file changed, 5 insertions(+) (limited to 'fs') diff --git a/fs/btrfs/scrub.c b/fs/btrfs/scrub.c index fab420db512..c27bcb67f33 100644 --- a/fs/btrfs/scrub.c +++ b/fs/btrfs/scrub.c @@ -256,6 +256,11 @@ static int scrub_print_warning_inode(u64 inum, u64 offset, u64 root, void *ctx) btrfs_release_path(swarn->path); ipath = init_ipath(4096, local_root, swarn->path); + if (IS_ERR(ipath)) { + ret = PTR_ERR(ipath); + ipath = NULL; + goto err; + } ret = paths_from_inode(inum, ipath); if (ret < 0) -- cgit v1.2.3-70-g09d2 From aa38a711a893accf5b5192f3d705a120deaa81e0 Mon Sep 17 00:00:00 2001 From: Miao Xie Date: Fri, 18 Nov 2011 17:43:00 +0800 Subject: Btrfs: fix deadlock on metadata reservation when evicting a inode When I ran the xfstests, I found the test tasks was blocked on meta-data reservation. By debugging, I found the reason of this bug: start transaction | v reserve meta-data space | v flush delay allocation -> iput inode -> evict inode ^ | | v wait for delay allocation flush <- reserve meta-data space And besides that, the flush on evicting inode will block the thread, which is reclaiming the memory, and make oom happen easily. Fix this bug by skipping the flush step when evicting inode. Signed-off-by: Miao Xie --- fs/btrfs/ctree.h | 3 +++ fs/btrfs/extent-tree.c | 22 ++++++++++++++++++---- fs/btrfs/inode.c | 2 +- 3 files changed, 22 insertions(+), 5 deletions(-) (limited to 'fs') diff --git a/fs/btrfs/ctree.h b/fs/btrfs/ctree.h index 04a5dfcee5a..50634abef9b 100644 --- a/fs/btrfs/ctree.h +++ b/fs/btrfs/ctree.h @@ -2369,6 +2369,9 @@ int btrfs_block_rsv_check(struct btrfs_root *root, int btrfs_block_rsv_refill(struct btrfs_root *root, struct btrfs_block_rsv *block_rsv, u64 min_reserved); +int btrfs_block_rsv_refill_noflush(struct btrfs_root *root, + struct btrfs_block_rsv *block_rsv, + u64 min_reserved); int btrfs_block_rsv_migrate(struct btrfs_block_rsv *src_rsv, struct btrfs_block_rsv *dst_rsv, u64 num_bytes); diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 5d86877f10e..b7e5f6898d0 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -3887,9 +3887,9 @@ int btrfs_block_rsv_check(struct btrfs_root *root, return ret; } -int btrfs_block_rsv_refill(struct btrfs_root *root, - struct btrfs_block_rsv *block_rsv, - u64 min_reserved) +static inline int __btrfs_block_rsv_refill(struct btrfs_root *root, + struct btrfs_block_rsv *block_rsv, + u64 min_reserved, int flush) { u64 num_bytes = 0; int ret = -ENOSPC; @@ -3908,7 +3908,7 @@ int btrfs_block_rsv_refill(struct btrfs_root *root, if (!ret) return 0; - ret = reserve_metadata_bytes(root, block_rsv, num_bytes, 1); + ret = reserve_metadata_bytes(root, block_rsv, num_bytes, flush); if (!ret) { block_rsv_add_bytes(block_rsv, num_bytes, 0); return 0; @@ -3917,6 +3917,20 @@ int btrfs_block_rsv_refill(struct btrfs_root *root, return ret; } +int btrfs_block_rsv_refill(struct btrfs_root *root, + struct btrfs_block_rsv *block_rsv, + u64 min_reserved) +{ + return __btrfs_block_rsv_refill(root, block_rsv, min_reserved, 1); +} + +int btrfs_block_rsv_refill_noflush(struct btrfs_root *root, + struct btrfs_block_rsv *block_rsv, + u64 min_reserved) +{ + return __btrfs_block_rsv_refill(root, block_rsv, min_reserved, 0); +} + int btrfs_block_rsv_migrate(struct btrfs_block_rsv *src_rsv, struct btrfs_block_rsv *dst_rsv, u64 num_bytes) diff --git a/fs/btrfs/inode.c b/fs/btrfs/inode.c index 8ad26b135a1..c5ccec23984 100644 --- a/fs/btrfs/inode.c +++ b/fs/btrfs/inode.c @@ -3490,7 +3490,7 @@ void btrfs_evict_inode(struct inode *inode) * doing the truncate. */ while (1) { - ret = btrfs_block_rsv_refill(root, rsv, min_size); + ret = btrfs_block_rsv_refill_noflush(root, rsv, min_size); /* * Try and steal from the global reserve since we will -- cgit v1.2.3-70-g09d2 From ece7d20e8be6730fbb29f4550de6b19b1a3a9387 Mon Sep 17 00:00:00 2001 From: Mike Fleetwood Date: Fri, 18 Nov 2011 18:55:01 +0000 Subject: Btrfs: Don't error on resizing FS to same size It seems overly harsh to fail a resize of a btrfs file system to the same size when a shrink or grow would succeed. User app GParted trips over this error. Allow it by bypassing the shrink or grow operation. Signed-off-by: Mike Fleetwood --- fs/btrfs/ioctl.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/btrfs/ioctl.c b/fs/btrfs/ioctl.c index a90e749ed6d..72d461656f6 100644 --- a/fs/btrfs/ioctl.c +++ b/fs/btrfs/ioctl.c @@ -1278,7 +1278,7 @@ static noinline int btrfs_ioctl_resize(struct btrfs_root *root, } ret = btrfs_grow_device(trans, device, new_size); btrfs_commit_transaction(trans, root); - } else { + } else if (new_size < old_size) { ret = btrfs_shrink_device(device, new_size); } -- cgit v1.2.3-70-g09d2 From b772a86ea6d932ac29d5e50e67c977653c832f8a Mon Sep 17 00:00:00 2001 From: Li Zefan Date: Mon, 28 Nov 2011 16:43:00 +0800 Subject: Btrfs: fix oops when calling statfs on readonly device To reproduce this bug: # dd if=/dev/zero of=img bs=1M count=256 # mkfs.btrfs img # losetup -r /dev/loop1 img # mount /dev/loop1 /mnt OOPS!! It triggered BUG_ON(!nr_devices) in btrfs_calc_avail_data_space(). To fix this, instead of checking write-only devices, we check all open deivces: # df -h /dev/loop1 Filesystem Size Used Avail Use% Mounted on /dev/loop1 250M 28K 238M 1% /mnt Signed-off-by: Li Zefan --- fs/btrfs/super.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) (limited to 'fs') diff --git a/fs/btrfs/super.c b/fs/btrfs/super.c index 8bd9d6d0e07..1a3ce9e0b49 100644 --- a/fs/btrfs/super.c +++ b/fs/btrfs/super.c @@ -1083,7 +1083,7 @@ static int btrfs_calc_avail_data_space(struct btrfs_root *root, u64 *free_bytes) int i = 0, nr_devices; int ret; - nr_devices = fs_info->fs_devices->rw_devices; + nr_devices = fs_info->fs_devices->open_devices; BUG_ON(!nr_devices); devices_info = kmalloc(sizeof(*devices_info) * nr_devices, @@ -1105,8 +1105,8 @@ static int btrfs_calc_avail_data_space(struct btrfs_root *root, u64 *free_bytes) else min_stripe_size = BTRFS_STRIPE_LEN; - list_for_each_entry(device, &fs_devices->alloc_list, dev_alloc_list) { - if (!device->in_fs_metadata) + list_for_each_entry(device, &fs_devices->devices, dev_list) { + if (!device->in_fs_metadata || !device->bdev) continue; avail_space = device->total_bytes - device->bytes_used; -- cgit v1.2.3-70-g09d2 From f2d0f6765d6332f9be732965a0c6f3b8a55082b4 Mon Sep 17 00:00:00 2001 From: Alexandre Oliva Date: Mon, 28 Nov 2011 12:04:43 -0200 Subject: Btrfs: initialize new bitmaps' list We're failing to create clusters with bitmaps because setup_cluster_no_bitmap checks that the list is empty before inserting the bitmap entry in the list for setup_cluster_bitmap, but the list field is only initialized when it is restored from the on-disk free space cache, or when it is written out to disk. Besides a potential race condition due to the multiple use of the list field, filesystem performance severely degrades over time: as we use up all non-bitmap free extents, the try-to-set-up-cluster dance is done at every metadata block allocation. For every block group, we fail to set up a cluster, and after failing on them all up to twice, we fall back to the much slower unclustered allocation. To make matters worse, before the unclustered allocation, we try to create new block groups until we reach the 1% threshold, which introduces additional bitmaps and thus block groups that we'll iterate over at each metadata block request. --- fs/btrfs/free-space-cache.c | 1 + 1 file changed, 1 insertion(+) (limited to 'fs') diff --git a/fs/btrfs/free-space-cache.c b/fs/btrfs/free-space-cache.c index 6e5b7e46369..ff179b1e742 100644 --- a/fs/btrfs/free-space-cache.c +++ b/fs/btrfs/free-space-cache.c @@ -1470,6 +1470,7 @@ static void add_new_bitmap(struct btrfs_free_space_ctl *ctl, { info->offset = offset_to_bitmap(ctl, offset); info->bytes = 0; + INIT_LIST_HEAD(&info->list); link_free_space(ctl, info); ctl->total_bitmaps++; -- cgit v1.2.3-70-g09d2 From b78d09bceb524ee6481c21b77bda22d766b10e6a Mon Sep 17 00:00:00 2001 From: Alexandre Oliva Date: Wed, 30 Nov 2011 13:43:00 -0500 Subject: Btrfs: reset cluster's max_size when creating bitmap The field that indicates the size of the largest contiguous chunk of free space in the cluster is not initialized when setting up bitmaps, it's only increased when we find a larger contiguous chunk. We end up retaining a larger value than appropriate for highly-fragmented clusters, which may cause pointless searches for large contiguous groups, and even cause clusters that do not meet the density requirements to be set up. Signed-off-by: Alexandre Oliva Signed-off-by: Chris Mason --- fs/btrfs/free-space-cache.c | 1 + 1 file changed, 1 insertion(+) (limited to 'fs') diff --git a/fs/btrfs/free-space-cache.c b/fs/btrfs/free-space-cache.c index ff179b1e742..ec23d43d0c3 100644 --- a/fs/btrfs/free-space-cache.c +++ b/fs/btrfs/free-space-cache.c @@ -2320,6 +2320,7 @@ again: if (!found) { start = i; + cluster->max_size = 0; found = true; } -- cgit v1.2.3-70-g09d2 From 1b22bad779be7fe07242be04749ec969164528b8 Mon Sep 17 00:00:00 2001 From: Alexandre Oliva Date: Wed, 30 Nov 2011 13:43:00 -0500 Subject: Btrfs: start search for new cluster at the beginning Instead of starting at zero (offset is always zero), request a cluster starting at search_start, that denotes the beginning of the current block group. Signed-off-by: Alexandre Oliva Signed-off-by: Chris Mason --- fs/btrfs/extent-tree.c | 6 ++---- 1 file changed, 2 insertions(+), 4 deletions(-) (limited to 'fs') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index b7e5f6898d0..97c12067a4b 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -5301,10 +5301,8 @@ alloc: spin_lock(&last_ptr->refill_lock); if (last_ptr->block_group && (last_ptr->block_group->ro || - !block_group_bits(last_ptr->block_group, data))) { - offset = 0; + !block_group_bits(last_ptr->block_group, data))) goto refill_cluster; - } offset = btrfs_alloc_from_cluster(block_group, last_ptr, num_bytes, search_start); @@ -5355,7 +5353,7 @@ refill_cluster: /* allocate a cluster in this block group */ ret = btrfs_find_space_cluster(trans, root, block_group, last_ptr, - offset, num_bytes, + search_start, num_bytes, empty_cluster + empty_size); if (ret == 0) { /* -- cgit v1.2.3-70-g09d2 From 425d83156ca27f74e2cc3f370138038c3c8947f8 Mon Sep 17 00:00:00 2001 From: Alexandre Oliva Date: Wed, 30 Nov 2011 13:43:00 -0500 Subject: Btrfs: skip block groups without enough space for a cluster We test whether a block group has enough free space to hold the requested block, but when we're doing clustered allocation, we can save some cycles by testing whether it has enough room for the cluster upfront, otherwise we end up attempting to set up a cluster and failing. Only in the NO_EMPTY_SIZE loop do we attempt an unclustered allocation, and by then we'll have zeroed the cluster size, so this patch won't stop us from using the block group as a last resort. Signed-off-by: Alexandre Oliva Signed-off-by: Chris Mason --- fs/btrfs/extent-tree.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 97c12067a4b..71c8e7049d0 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -5278,7 +5278,7 @@ alloc: spin_lock(&block_group->free_space_ctl->tree_lock); if (cached && block_group->free_space_ctl->free_space < - num_bytes + empty_size) { + num_bytes + empty_cluster + empty_size) { spin_unlock(&block_group->free_space_ctl->tree_lock); goto loop; } -- cgit v1.2.3-70-g09d2 From be064d113906f04ea13088a8260e1e68ae0a4050 Mon Sep 17 00:00:00 2001 From: Alexandre Oliva Date: Wed, 30 Nov 2011 13:43:00 -0500 Subject: Btrfs: skip allocation attempt from empty cluster If we don't have a cluster, don't bother trying to allocate from it, jumping right away to the attempt to allocate a new cluster. Signed-off-by: Alexandre Oliva Signed-off-by: Chris Mason --- fs/btrfs/extent-tree.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) (limited to 'fs') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 71c8e7049d0..813c6bb96c9 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -5299,9 +5299,9 @@ alloc: * people trying to start a new cluster */ spin_lock(&last_ptr->refill_lock); - if (last_ptr->block_group && - (last_ptr->block_group->ro || - !block_group_bits(last_ptr->block_group, data))) + if (!last_ptr->block_group || + last_ptr->block_group->ro || + !block_group_bits(last_ptr->block_group, data)) goto refill_cluster; offset = btrfs_alloc_from_cluster(block_group, last_ptr, -- cgit v1.2.3-70-g09d2 From f4a8e6563ea5366f563cb741a27fe90c5fa7f0fc Mon Sep 17 00:00:00 2001 From: Jan Schmidt Date: Thu, 1 Dec 2011 09:30:36 -0500 Subject: Btrfs: fix meta data raid-repair merge problem Commit 4a54c8c16 introduced raid-repair, killing the individual readpage_io_failed_hook entries from inode.c and disk-io.c. Commit 4bb31e92 introduced new readahead code, adding a readpage_io_failed_hook to disk-io.c. The raid-repair commit had logic to disable raid-repair, if readpage_io_failed_hook is set. Thus, the readahead commit effectively disabled raid-repair for meta data. This commit changes the logic to always attempt raid-repair when needed and call the readpage_io_failed_hook in case raid-repair fails. This is much more straight forward and should have been like that from the beginning. Signed-off-by: Jan Schmidt Reported-by: Stefan Behrens Signed-off-by: Chris Mason --- fs/btrfs/extent_io.c | 27 ++++++++++++++++++++------- 1 file changed, 20 insertions(+), 7 deletions(-) (limited to 'fs') diff --git a/fs/btrfs/extent_io.c b/fs/btrfs/extent_io.c index 9472d3de5e5..be1bf627a14 100644 --- a/fs/btrfs/extent_io.c +++ b/fs/btrfs/extent_io.c @@ -2287,14 +2287,20 @@ static void end_bio_extent_readpage(struct bio *bio, int err) if (!uptodate) { int failed_mirror; failed_mirror = (int)(unsigned long)bio->bi_bdev; - if (tree->ops && tree->ops->readpage_io_failed_hook) - ret = tree->ops->readpage_io_failed_hook( - bio, page, start, end, - failed_mirror, state); - else - ret = bio_readpage_error(bio, page, start, end, - failed_mirror, NULL); + /* + * The generic bio_readpage_error handles errors the + * following way: If possible, new read requests are + * created and submitted and will end up in + * end_bio_extent_readpage as well (if we're lucky, not + * in the !uptodate case). In that case it returns 0 and + * we just go on with the next page in our bio. If it + * can't handle the error it will return -EIO and we + * remain responsible for that page. + */ + ret = bio_readpage_error(bio, page, start, end, + failed_mirror, NULL); if (ret == 0) { +error_handled: uptodate = test_bit(BIO_UPTODATE, &bio->bi_flags); if (err) @@ -2302,6 +2308,13 @@ static void end_bio_extent_readpage(struct bio *bio, int err) uncache_state(&cached); continue; } + if (tree->ops && tree->ops->readpage_io_failed_hook) { + ret = tree->ops->readpage_io_failed_hook( + bio, page, start, end, + failed_mirror, state); + if (ret == 0) + goto error_handled; + } } if (uptodate) { -- cgit v1.2.3-70-g09d2 From 939255798a468e1a92f03546de6e87be7b491e57 Mon Sep 17 00:00:00 2001 From: Akinobu Mita Date: Tue, 15 Nov 2011 14:56:34 -0800 Subject: ocfs2: avoid unaligned access to dqc_bitmap The dqc_bitmap field of struct ocfs2_local_disk_chunk is 32-bit aligned, but not 64-bit aligned. The dqc_bitmap is accessed by ocfs2_set_bit(), ocfs2_clear_bit(), ocfs2_test_bit(), or ocfs2_find_next_zero_bit(). These are wrapper macros for ext2_*_bit() which need to take an unsigned long aligned address (though some architectures are able to handle unaligned address correctly) So some 64bit architectures may not be able to access the dqc_bitmap correctly. This avoids such unaligned access by using another wrapper functions for ext2_*_bit(). The code is taken from fs/ext4/mballoc.c which also need to handle unaligned bitmap access. Signed-off-by: Akinobu Mita Acked-by: Joel Becker Cc: Mark Fasheh Signed-off-by: Andrew Morton Signed-off-by: Joel Becker --- fs/ocfs2/ocfs2.h | 47 +++++++++++++++++++++++++++++++++++++++++++++++ fs/ocfs2/quota_local.c | 10 +++++----- 2 files changed, 52 insertions(+), 5 deletions(-) (limited to 'fs') diff --git a/fs/ocfs2/ocfs2.h b/fs/ocfs2/ocfs2.h index bcde467acca..d355e6e36b3 100644 --- a/fs/ocfs2/ocfs2.h +++ b/fs/ocfs2/ocfs2.h @@ -849,5 +849,52 @@ static inline void _ocfs2_clear_bit(unsigned int bit, unsigned long *bitmap) #define ocfs2_test_bit test_bit_le #define ocfs2_find_next_zero_bit find_next_zero_bit_le #define ocfs2_find_next_bit find_next_bit_le + +static inline void *correct_addr_and_bit_unaligned(int *bit, void *addr) +{ +#if BITS_PER_LONG == 64 + *bit += ((unsigned long) addr & 7UL) << 3; + addr = (void *) ((unsigned long) addr & ~7UL); +#elif BITS_PER_LONG == 32 + *bit += ((unsigned long) addr & 3UL) << 3; + addr = (void *) ((unsigned long) addr & ~3UL); +#else +#error "how many bits you are?!" +#endif + return addr; +} + +static inline void ocfs2_set_bit_unaligned(int bit, void *bitmap) +{ + bitmap = correct_addr_and_bit_unaligned(&bit, bitmap); + ocfs2_set_bit(bit, bitmap); +} + +static inline void ocfs2_clear_bit_unaligned(int bit, void *bitmap) +{ + bitmap = correct_addr_and_bit_unaligned(&bit, bitmap); + ocfs2_clear_bit(bit, bitmap); +} + +static inline int ocfs2_test_bit_unaligned(int bit, void *bitmap) +{ + bitmap = correct_addr_and_bit_unaligned(&bit, bitmap); + return ocfs2_test_bit(bit, bitmap); +} + +static inline int ocfs2_find_next_zero_bit_unaligned(void *bitmap, int max, + int start) +{ + int fix = 0, ret, tmpmax; + bitmap = correct_addr_and_bit_unaligned(&fix, bitmap); + tmpmax = max + fix; + start += fix; + + ret = ocfs2_find_next_zero_bit(bitmap, tmpmax, start) - fix; + if (ret > max) + return max; + return ret; +} + #endif /* OCFS2_H */ diff --git a/fs/ocfs2/quota_local.c b/fs/ocfs2/quota_local.c index 942fd65bdad..f100bf70a90 100644 --- a/fs/ocfs2/quota_local.c +++ b/fs/ocfs2/quota_local.c @@ -551,8 +551,8 @@ static int ocfs2_recover_local_quota_file(struct inode *lqinode, goto out_commit; } lock_buffer(qbh); - WARN_ON(!ocfs2_test_bit(bit, dchunk->dqc_bitmap)); - ocfs2_clear_bit(bit, dchunk->dqc_bitmap); + WARN_ON(!ocfs2_test_bit_unaligned(bit, dchunk->dqc_bitmap)); + ocfs2_clear_bit_unaligned(bit, dchunk->dqc_bitmap); le32_add_cpu(&dchunk->dqc_free, 1); unlock_buffer(qbh); ocfs2_journal_dirty(handle, qbh); @@ -949,7 +949,7 @@ static struct ocfs2_quota_chunk *ocfs2_find_free_entry(struct super_block *sb, * ol_quota_entries_per_block(sb); } - found = ocfs2_find_next_zero_bit(dchunk->dqc_bitmap, len, 0); + found = ocfs2_find_next_zero_bit_unaligned(dchunk->dqc_bitmap, len, 0); /* We failed? */ if (found == len) { mlog(ML_ERROR, "Did not find empty entry in chunk %d with %u" @@ -1213,7 +1213,7 @@ static void olq_alloc_dquot(struct buffer_head *bh, void *private) struct ocfs2_local_disk_chunk *dchunk; dchunk = (struct ocfs2_local_disk_chunk *)bh->b_data; - ocfs2_set_bit(*offset, dchunk->dqc_bitmap); + ocfs2_set_bit_unaligned(*offset, dchunk->dqc_bitmap); le32_add_cpu(&dchunk->dqc_free, -1); } @@ -1294,7 +1294,7 @@ int ocfs2_local_release_dquot(handle_t *handle, struct dquot *dquot) (od->dq_chunk->qc_headerbh->b_data); /* Mark structure as freed */ lock_buffer(od->dq_chunk->qc_headerbh); - ocfs2_clear_bit(offset, dchunk->dqc_bitmap); + ocfs2_clear_bit_unaligned(offset, dchunk->dqc_bitmap); le32_add_cpu(&dchunk->dqc_free, 1); unlock_buffer(od->dq_chunk->qc_headerbh); ocfs2_journal_dirty(handle, od->dq_chunk->qc_headerbh); -- cgit v1.2.3-70-g09d2 From a99ebf43f49f4499ab0e2a8a9132ad6ed6ba2409 Mon Sep 17 00:00:00 2001 From: Dave Chinner Date: Thu, 1 Dec 2011 11:24:20 +0000 Subject: xfs: fix allocation length overflow in xfs_bmapi_write() When testing the new xfstests --large-fs option that does very large file preallocations, this assert was tripped deep in xfs_alloc_vextent(): XFS: Assertion failed: args->minlen <= args->maxlen, file: fs/xfs/xfs_alloc.c, line: 2239 The allocation was trying to allocate a zero length extent because the lower 32 bits of the allocation length was zero. The remaining length of the allocation to be done was an exact multiple of 2^32 - the first case I saw was at 496TB remaining to be allocated. This turns out to be an overflow when converting the allocation length (a 64 bit quantity) into the extent length to allocate (a 32 bit quantity), and it requires the length to be allocated an exact multiple of 2^32 blocks to trip the assert. Fix it by limiting the extent lenth to allocate to MAXEXTLEN. Signed-off-by: Dave Chinner Signed-off-by: Ben Myers Reviewed-by: Christoph Hellwig --- fs/xfs/xfs_bmap.c | 20 +++++++++++++++++++- 1 file changed, 19 insertions(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/xfs/xfs_bmap.c b/fs/xfs/xfs_bmap.c index c68baeb0974..d0ab7883705 100644 --- a/fs/xfs/xfs_bmap.c +++ b/fs/xfs/xfs_bmap.c @@ -2383,6 +2383,8 @@ xfs_bmap_btalloc( int tryagain; int error; + ASSERT(ap->length); + mp = ap->ip->i_mount; align = ap->userdata ? xfs_get_extsz_hint(ap->ip) : 0; if (unlikely(align)) { @@ -4629,6 +4631,8 @@ xfs_bmapi_allocate( int error; int rt; + ASSERT(bma->length > 0); + rt = (whichfork == XFS_DATA_FORK) && XFS_IS_REALTIME_INODE(bma->ip); /* @@ -4849,6 +4853,7 @@ xfs_bmapi_write( ASSERT(*nmap <= XFS_BMAP_MAX_NMAP); ASSERT(!(flags & XFS_BMAPI_IGSTATE)); ASSERT(tp != NULL); + ASSERT(len > 0); whichfork = (flags & XFS_BMAPI_ATTRFORK) ? XFS_ATTR_FORK : XFS_DATA_FORK; @@ -4918,9 +4923,22 @@ xfs_bmapi_write( bma.eof = eof; bma.conv = !!(flags & XFS_BMAPI_CONVERT); bma.wasdel = wasdelay; - bma.length = len; bma.offset = bno; + /* + * There's a 32/64 bit type mismatch between the + * allocation length request (which can be 64 bits in + * length) and the bma length request, which is + * xfs_extlen_t and therefore 32 bits. Hence we have to + * check for 32-bit overflows and handle them here. + */ + if (len > (xfs_filblks_t)MAXEXTLEN) + bma.length = MAXEXTLEN; + else + bma.length = len; + + ASSERT(len > 0); + ASSERT(bma.length > 0); error = xfs_bmapi_allocate(&bma, flags); if (error) goto error0; -- cgit v1.2.3-70-g09d2 From c29f7d457ac63311feb11928a866efd2fe153d74 Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Wed, 30 Nov 2011 08:58:18 +0000 Subject: xfs: fix nfs export of 64-bit inodes numbers on 32-bit kernels The i_ino field in the VFS inode is of type unsigned long and thus can't hold the full 64-bit inode number on 32-bit kernels. We have the full inode number in the XFS inode, so use that one for nfs exports. Note that I've also switched the 32-bit file handles types to it, just to make the code more consistent and copy & paste errors less likely to happen. Reported-by: Guoquan Yang Reported-by: Hank Peng Signed-off-by: Christoph Hellwig Signed-off-by: Ben Myers --- fs/xfs/xfs_export.c | 8 ++++---- 1 file changed, 4 insertions(+), 4 deletions(-) (limited to 'fs') diff --git a/fs/xfs/xfs_export.c b/fs/xfs/xfs_export.c index da108977b21..558910f5e3c 100644 --- a/fs/xfs/xfs_export.c +++ b/fs/xfs/xfs_export.c @@ -98,22 +98,22 @@ xfs_fs_encode_fh( switch (fileid_type) { case FILEID_INO32_GEN_PARENT: spin_lock(&dentry->d_lock); - fid->i32.parent_ino = dentry->d_parent->d_inode->i_ino; + fid->i32.parent_ino = XFS_I(dentry->d_parent->d_inode)->i_ino; fid->i32.parent_gen = dentry->d_parent->d_inode->i_generation; spin_unlock(&dentry->d_lock); /*FALLTHRU*/ case FILEID_INO32_GEN: - fid->i32.ino = inode->i_ino; + fid->i32.ino = XFS_I(inode)->i_ino; fid->i32.gen = inode->i_generation; break; case FILEID_INO32_GEN_PARENT | XFS_FILEID_TYPE_64FLAG: spin_lock(&dentry->d_lock); - fid64->parent_ino = dentry->d_parent->d_inode->i_ino; + fid64->parent_ino = XFS_I(dentry->d_parent->d_inode)->i_ino; fid64->parent_gen = dentry->d_parent->d_inode->i_generation; spin_unlock(&dentry->d_lock); /*FALLTHRU*/ case FILEID_INO32_GEN | XFS_FILEID_TYPE_64FLAG: - fid64->ino = inode->i_ino; + fid64->ino = XFS_I(inode)->i_ino; fid64->gen = inode->i_generation; break; } -- cgit v1.2.3-70-g09d2 From 9f9c19ec1a59422c7687b11847ed3408128aa0d6 Mon Sep 17 00:00:00 2001 From: Christoph Hellwig Date: Mon, 28 Nov 2011 08:17:36 +0000 Subject: xfs: fix the logspace waiting algorithm Apply the scheme used in log_regrant_write_log_space to wake up any other threads waiting for log space before the newly added one to log_regrant_write_log_space as well, and factor the code into readable helpers. For each of the queues we have add two helpers: - one to try to wake up all waiting threads. This helper will also be usable by xfs_log_move_tail once we remove the current opportunistic wakeups in it. - one to sleep on t_wait until enough log space is available, loosely modelled after Linux waitqueues. And use them to reimplement the guts of log_regrant_write_log_space and log_regrant_write_log_space. These two function now use one and the same algorithm for waiting on log space instead of subtly different ones before, with an option to completely unify them in the near future. Also move the filesystem shutdown handling to the common caller given that we had to touch it anyway. Based on hard debugging and an earlier patch from Chandra Seetharaman . Signed-off-by: Christoph Hellwig Reviewed-by: Chandra Seetharaman Tested-by: Chandra Seetharaman Signed-off-by: Ben Myers --- fs/xfs/xfs_log.c | 348 ++++++++++++++++++++++++++--------------------------- fs/xfs/xfs_trace.h | 12 +- 2 files changed, 177 insertions(+), 183 deletions(-) (limited to 'fs') diff --git a/fs/xfs/xfs_log.c b/fs/xfs/xfs_log.c index a14cd89fe46..34817adf4b9 100644 --- a/fs/xfs/xfs_log.c +++ b/fs/xfs/xfs_log.c @@ -150,6 +150,117 @@ xlog_grant_add_space( } while (head_val != old); } +STATIC bool +xlog_reserveq_wake( + struct log *log, + int *free_bytes) +{ + struct xlog_ticket *tic; + int need_bytes; + + list_for_each_entry(tic, &log->l_reserveq, t_queue) { + if (tic->t_flags & XLOG_TIC_PERM_RESERV) + need_bytes = tic->t_unit_res * tic->t_cnt; + else + need_bytes = tic->t_unit_res; + + if (*free_bytes < need_bytes) + return false; + *free_bytes -= need_bytes; + + trace_xfs_log_grant_wake_up(log, tic); + wake_up(&tic->t_wait); + } + + return true; +} + +STATIC bool +xlog_writeq_wake( + struct log *log, + int *free_bytes) +{ + struct xlog_ticket *tic; + int need_bytes; + + list_for_each_entry(tic, &log->l_writeq, t_queue) { + ASSERT(tic->t_flags & XLOG_TIC_PERM_RESERV); + + need_bytes = tic->t_unit_res; + + if (*free_bytes < need_bytes) + return false; + *free_bytes -= need_bytes; + + trace_xfs_log_regrant_write_wake_up(log, tic); + wake_up(&tic->t_wait); + } + + return true; +} + +STATIC int +xlog_reserveq_wait( + struct log *log, + struct xlog_ticket *tic, + int need_bytes) +{ + list_add_tail(&tic->t_queue, &log->l_reserveq); + + do { + if (XLOG_FORCED_SHUTDOWN(log)) + goto shutdown; + xlog_grant_push_ail(log, need_bytes); + + XFS_STATS_INC(xs_sleep_logspace); + trace_xfs_log_grant_sleep(log, tic); + + xlog_wait(&tic->t_wait, &log->l_grant_reserve_lock); + trace_xfs_log_grant_wake(log, tic); + + spin_lock(&log->l_grant_reserve_lock); + if (XLOG_FORCED_SHUTDOWN(log)) + goto shutdown; + } while (xlog_space_left(log, &log->l_grant_reserve_head) < need_bytes); + + list_del_init(&tic->t_queue); + return 0; +shutdown: + list_del_init(&tic->t_queue); + return XFS_ERROR(EIO); +} + +STATIC int +xlog_writeq_wait( + struct log *log, + struct xlog_ticket *tic, + int need_bytes) +{ + list_add_tail(&tic->t_queue, &log->l_writeq); + + do { + if (XLOG_FORCED_SHUTDOWN(log)) + goto shutdown; + xlog_grant_push_ail(log, need_bytes); + + XFS_STATS_INC(xs_sleep_logspace); + trace_xfs_log_regrant_write_sleep(log, tic); + + xlog_wait(&tic->t_wait, &log->l_grant_write_lock); + trace_xfs_log_regrant_write_wake(log, tic); + + spin_lock(&log->l_grant_write_lock); + if (XLOG_FORCED_SHUTDOWN(log)) + goto shutdown; + } while (xlog_space_left(log, &log->l_grant_write_head) < need_bytes); + + list_del_init(&tic->t_queue); + return 0; +shutdown: + list_del_init(&tic->t_queue); + return XFS_ERROR(EIO); +} + static void xlog_tic_reset_res(xlog_ticket_t *tic) { @@ -350,8 +461,19 @@ xfs_log_reserve( retval = xlog_grant_log_space(log, internal_ticket); } + if (unlikely(retval)) { + /* + * If we are failing, make sure the ticket doesn't have any + * current reservations. We don't want to add this back + * when the ticket/ transaction gets cancelled. + */ + internal_ticket->t_curr_res = 0; + /* ungrant will give back unit_res * t_cnt. */ + internal_ticket->t_cnt = 0; + } + return retval; -} /* xfs_log_reserve */ +} /* @@ -2481,8 +2603,8 @@ restart: /* * Atomically get the log space required for a log ticket. * - * Once a ticket gets put onto the reserveq, it will only return after - * the needed reservation is satisfied. + * Once a ticket gets put onto the reserveq, it will only return after the + * needed reservation is satisfied. * * This function is structured so that it has a lock free fast path. This is * necessary because every new transaction reservation will come through this @@ -2490,113 +2612,53 @@ restart: * every pass. * * As tickets are only ever moved on and off the reserveq under the - * l_grant_reserve_lock, we only need to take that lock if we are going - * to add the ticket to the queue and sleep. We can avoid taking the lock if the - * ticket was never added to the reserveq because the t_queue list head will be - * empty and we hold the only reference to it so it can safely be checked - * unlocked. + * l_grant_reserve_lock, we only need to take that lock if we are going to add + * the ticket to the queue and sleep. We can avoid taking the lock if the ticket + * was never added to the reserveq because the t_queue list head will be empty + * and we hold the only reference to it so it can safely be checked unlocked. */ STATIC int -xlog_grant_log_space(xlog_t *log, - xlog_ticket_t *tic) +xlog_grant_log_space( + struct log *log, + struct xlog_ticket *tic) { - int free_bytes; - int need_bytes; + int free_bytes, need_bytes; + int error = 0; -#ifdef DEBUG - if (log->l_flags & XLOG_ACTIVE_RECOVERY) - panic("grant Recovery problem"); -#endif + ASSERT(!(log->l_flags & XLOG_ACTIVE_RECOVERY)); trace_xfs_log_grant_enter(log, tic); + /* + * If there are other waiters on the queue then give them a chance at + * logspace before us. Wake up the first waiters, if we do not wake + * up all the waiters then go to sleep waiting for more free space, + * otherwise try to get some space for this transaction. + */ need_bytes = tic->t_unit_res; if (tic->t_flags & XFS_LOG_PERM_RESERV) need_bytes *= tic->t_ocnt; - - /* something is already sleeping; insert new transaction at end */ - if (!list_empty_careful(&log->l_reserveq)) { - spin_lock(&log->l_grant_reserve_lock); - /* recheck the queue now we are locked */ - if (list_empty(&log->l_reserveq)) { - spin_unlock(&log->l_grant_reserve_lock); - goto redo; - } - list_add_tail(&tic->t_queue, &log->l_reserveq); - - trace_xfs_log_grant_sleep1(log, tic); - - /* - * Gotta check this before going to sleep, while we're - * holding the grant lock. - */ - if (XLOG_FORCED_SHUTDOWN(log)) - goto error_return; - - XFS_STATS_INC(xs_sleep_logspace); - xlog_wait(&tic->t_wait, &log->l_grant_reserve_lock); - - /* - * If we got an error, and the filesystem is shutting down, - * we'll catch it down below. So just continue... - */ - trace_xfs_log_grant_wake1(log, tic); - } - -redo: - if (XLOG_FORCED_SHUTDOWN(log)) - goto error_return_unlocked; - free_bytes = xlog_space_left(log, &log->l_grant_reserve_head); - if (free_bytes < need_bytes) { + if (!list_empty_careful(&log->l_reserveq)) { spin_lock(&log->l_grant_reserve_lock); - if (list_empty(&tic->t_queue)) - list_add_tail(&tic->t_queue, &log->l_reserveq); - - trace_xfs_log_grant_sleep2(log, tic); - - if (XLOG_FORCED_SHUTDOWN(log)) - goto error_return; - - xlog_grant_push_ail(log, need_bytes); - - XFS_STATS_INC(xs_sleep_logspace); - xlog_wait(&tic->t_wait, &log->l_grant_reserve_lock); - - trace_xfs_log_grant_wake2(log, tic); - goto redo; - } - - if (!list_empty(&tic->t_queue)) { + if (!xlog_reserveq_wake(log, &free_bytes) || + free_bytes < need_bytes) + error = xlog_reserveq_wait(log, tic, need_bytes); + spin_unlock(&log->l_grant_reserve_lock); + } else if (free_bytes < need_bytes) { spin_lock(&log->l_grant_reserve_lock); - list_del_init(&tic->t_queue); + error = xlog_reserveq_wait(log, tic, need_bytes); spin_unlock(&log->l_grant_reserve_lock); } + if (error) + return error; - /* we've got enough space */ xlog_grant_add_space(log, &log->l_grant_reserve_head, need_bytes); xlog_grant_add_space(log, &log->l_grant_write_head, need_bytes); trace_xfs_log_grant_exit(log, tic); xlog_verify_grant_tail(log); return 0; - -error_return_unlocked: - spin_lock(&log->l_grant_reserve_lock); -error_return: - list_del_init(&tic->t_queue); - spin_unlock(&log->l_grant_reserve_lock); - trace_xfs_log_grant_error(log, tic); - - /* - * If we are failing, make sure the ticket doesn't have any - * current reservations. We don't want to add this back when - * the ticket/transaction gets cancelled. - */ - tic->t_curr_res = 0; - tic->t_cnt = 0; /* ungrant will give back unit_res * t_cnt. */ - return XFS_ERROR(EIO); -} /* xlog_grant_log_space */ - +} /* * Replenish the byte reservation required by moving the grant write head. @@ -2605,10 +2667,12 @@ error_return: * free fast path. */ STATIC int -xlog_regrant_write_log_space(xlog_t *log, - xlog_ticket_t *tic) +xlog_regrant_write_log_space( + struct log *log, + struct xlog_ticket *tic) { - int free_bytes, need_bytes; + int free_bytes, need_bytes; + int error = 0; tic->t_curr_res = tic->t_unit_res; xlog_tic_reset_res(tic); @@ -2616,104 +2680,38 @@ xlog_regrant_write_log_space(xlog_t *log, if (tic->t_cnt > 0) return 0; -#ifdef DEBUG - if (log->l_flags & XLOG_ACTIVE_RECOVERY) - panic("regrant Recovery problem"); -#endif + ASSERT(!(log->l_flags & XLOG_ACTIVE_RECOVERY)); trace_xfs_log_regrant_write_enter(log, tic); - if (XLOG_FORCED_SHUTDOWN(log)) - goto error_return_unlocked; - /* If there are other waiters on the queue then give them a - * chance at logspace before us. Wake up the first waiters, - * if we do not wake up all the waiters then go to sleep waiting - * for more free space, otherwise try to get some space for - * this transaction. + /* + * If there are other waiters on the queue then give them a chance at + * logspace before us. Wake up the first waiters, if we do not wake + * up all the waiters then go to sleep waiting for more free space, + * otherwise try to get some space for this transaction. */ need_bytes = tic->t_unit_res; - if (!list_empty_careful(&log->l_writeq)) { - struct xlog_ticket *ntic; - - spin_lock(&log->l_grant_write_lock); - free_bytes = xlog_space_left(log, &log->l_grant_write_head); - list_for_each_entry(ntic, &log->l_writeq, t_queue) { - ASSERT(ntic->t_flags & XLOG_TIC_PERM_RESERV); - - if (free_bytes < ntic->t_unit_res) - break; - free_bytes -= ntic->t_unit_res; - wake_up(&ntic->t_wait); - } - - if (ntic != list_first_entry(&log->l_writeq, - struct xlog_ticket, t_queue)) { - if (list_empty(&tic->t_queue)) - list_add_tail(&tic->t_queue, &log->l_writeq); - trace_xfs_log_regrant_write_sleep1(log, tic); - - xlog_grant_push_ail(log, need_bytes); - - XFS_STATS_INC(xs_sleep_logspace); - xlog_wait(&tic->t_wait, &log->l_grant_write_lock); - trace_xfs_log_regrant_write_wake1(log, tic); - } else - spin_unlock(&log->l_grant_write_lock); - } - -redo: - if (XLOG_FORCED_SHUTDOWN(log)) - goto error_return_unlocked; - free_bytes = xlog_space_left(log, &log->l_grant_write_head); - if (free_bytes < need_bytes) { + if (!list_empty_careful(&log->l_writeq)) { spin_lock(&log->l_grant_write_lock); - if (list_empty(&tic->t_queue)) - list_add_tail(&tic->t_queue, &log->l_writeq); - - if (XLOG_FORCED_SHUTDOWN(log)) - goto error_return; - - xlog_grant_push_ail(log, need_bytes); - - XFS_STATS_INC(xs_sleep_logspace); - trace_xfs_log_regrant_write_sleep2(log, tic); - xlog_wait(&tic->t_wait, &log->l_grant_write_lock); - - trace_xfs_log_regrant_write_wake2(log, tic); - goto redo; - } - - if (!list_empty(&tic->t_queue)) { + if (!xlog_writeq_wake(log, &free_bytes) || + free_bytes < need_bytes) + error = xlog_writeq_wait(log, tic, need_bytes); + spin_unlock(&log->l_grant_write_lock); + } else if (free_bytes < need_bytes) { spin_lock(&log->l_grant_write_lock); - list_del_init(&tic->t_queue); + error = xlog_writeq_wait(log, tic, need_bytes); spin_unlock(&log->l_grant_write_lock); } - /* we've got enough space */ + if (error) + return error; + xlog_grant_add_space(log, &log->l_grant_write_head, need_bytes); trace_xfs_log_regrant_write_exit(log, tic); xlog_verify_grant_tail(log); return 0; - - - error_return_unlocked: - spin_lock(&log->l_grant_write_lock); - error_return: - list_del_init(&tic->t_queue); - spin_unlock(&log->l_grant_write_lock); - trace_xfs_log_regrant_write_error(log, tic); - - /* - * If we are failing, make sure the ticket doesn't have any - * current reservations. We don't want to add this back when - * the ticket/transaction gets cancelled. - */ - tic->t_curr_res = 0; - tic->t_cnt = 0; /* ungrant will give back unit_res * t_cnt. */ - return XFS_ERROR(EIO); -} /* xlog_regrant_write_log_space */ - +} /* The first cnt-1 times through here we don't need to * move the grant write head because the permanent diff --git a/fs/xfs/xfs_trace.h b/fs/xfs/xfs_trace.h index f1d2802b2f0..49403579887 100644 --- a/fs/xfs/xfs_trace.h +++ b/fs/xfs/xfs_trace.h @@ -834,18 +834,14 @@ DEFINE_LOGGRANT_EVENT(xfs_log_umount_write); DEFINE_LOGGRANT_EVENT(xfs_log_grant_enter); DEFINE_LOGGRANT_EVENT(xfs_log_grant_exit); DEFINE_LOGGRANT_EVENT(xfs_log_grant_error); -DEFINE_LOGGRANT_EVENT(xfs_log_grant_sleep1); -DEFINE_LOGGRANT_EVENT(xfs_log_grant_wake1); -DEFINE_LOGGRANT_EVENT(xfs_log_grant_sleep2); -DEFINE_LOGGRANT_EVENT(xfs_log_grant_wake2); +DEFINE_LOGGRANT_EVENT(xfs_log_grant_sleep); +DEFINE_LOGGRANT_EVENT(xfs_log_grant_wake); DEFINE_LOGGRANT_EVENT(xfs_log_grant_wake_up); DEFINE_LOGGRANT_EVENT(xfs_log_regrant_write_enter); DEFINE_LOGGRANT_EVENT(xfs_log_regrant_write_exit); DEFINE_LOGGRANT_EVENT(xfs_log_regrant_write_error); -DEFINE_LOGGRANT_EVENT(xfs_log_regrant_write_sleep1); -DEFINE_LOGGRANT_EVENT(xfs_log_regrant_write_wake1); -DEFINE_LOGGRANT_EVENT(xfs_log_regrant_write_sleep2); -DEFINE_LOGGRANT_EVENT(xfs_log_regrant_write_wake2); +DEFINE_LOGGRANT_EVENT(xfs_log_regrant_write_sleep); +DEFINE_LOGGRANT_EVENT(xfs_log_regrant_write_wake); DEFINE_LOGGRANT_EVENT(xfs_log_regrant_write_wake_up); DEFINE_LOGGRANT_EVENT(xfs_log_regrant_reserve_enter); DEFINE_LOGGRANT_EVENT(xfs_log_regrant_reserve_exit); -- cgit v1.2.3-70-g09d2 From 02125a826459a6ad142f8d91c5b6357562f96615 Mon Sep 17 00:00:00 2001 From: Al Viro Date: Mon, 5 Dec 2011 08:43:34 -0500 Subject: fix apparmor dereferencing potentially freed dentry, sanitize __d_path() API __d_path() API is asking for trouble and in case of apparmor d_namespace_path() getting just that. The root cause is that when __d_path() misses the root it had been told to look for, it stores the location of the most remote ancestor in *root. Without grabbing references. Sure, at the moment of call it had been pinned down by what we have in *path. And if we raced with umount -l, we could have very well stopped at vfsmount/dentry that got freed as soon as prepend_path() dropped vfsmount_lock. It is safe to compare these pointers with pre-existing (and known to be still alive) vfsmount and dentry, as long as all we are asking is "is it the same address?". Dereferencing is not safe and apparmor ended up stepping into that. d_namespace_path() really wants to examine the place where we stopped, even if it's not connected to our namespace. As the result, it looked at ->d_sb->s_magic of a dentry that might've been already freed by that point. All other callers had been careful enough to avoid that, but it's really a bad interface - it invites that kind of trouble. The fix is fairly straightforward, even though it's bigger than I'd like: * prepend_path() root argument becomes const. * __d_path() is never called with NULL/NULL root. It was a kludge to start with. Instead, we have an explicit function - d_absolute_root(). Same as __d_path(), except that it doesn't get root passed and stops where it stops. apparmor and tomoyo are using it. * __d_path() returns NULL on path outside of root. The main caller is show_mountinfo() and that's precisely what we pass root for - to skip those outside chroot jail. Those who don't want that can (and do) use d_path(). * __d_path() root argument becomes const. Everyone agrees, I hope. * apparmor does *NOT* try to use __d_path() or any of its variants when it sees that path->mnt is an internal vfsmount. In that case it's definitely not mounted anywhere and dentry_path() is exactly what we want there. Handling of sysctl()-triggered weirdness is moved to that place. * if apparmor is asked to do pathname relative to chroot jail and __d_path() tells it we it's not in that jail, the sucker just calls d_absolute_path() instead. That's the other remaining caller of __d_path(), BTW. * seq_path_root() does _NOT_ return -ENAMETOOLONG (it's stupid anyway - the normal seq_file logics will take care of growing the buffer and redoing the call of ->show() just fine). However, if it gets path not reachable from root, it returns SEQ_SKIP. The only caller adjusted (i.e. stopped ignoring the return value as it used to do). Reviewed-by: John Johansen ACKed-by: John Johansen Signed-off-by: Al Viro Cc: stable@vger.kernel.org --- fs/dcache.c | 71 ++++++++++++++++++++++++++++------------------ fs/namespace.c | 20 +++++++------ fs/seq_file.c | 6 ++-- include/linux/dcache.h | 3 +- include/linux/fs.h | 1 + security/apparmor/path.c | 65 ++++++++++++++++++++++++------------------ security/tomoyo/realpath.c | 3 +- 7 files changed, 100 insertions(+), 69 deletions(-) (limited to 'fs') diff --git a/fs/dcache.c b/fs/dcache.c index 10ba92def3f..89509b5a090 100644 --- a/fs/dcache.c +++ b/fs/dcache.c @@ -2439,16 +2439,14 @@ static int prepend_name(char **buffer, int *buflen, struct qstr *name) /** * prepend_path - Prepend path string to a buffer * @path: the dentry/vfsmount to report - * @root: root vfsmnt/dentry (may be modified by this function) + * @root: root vfsmnt/dentry * @buffer: pointer to the end of the buffer * @buflen: pointer to buffer length * * Caller holds the rename_lock. - * - * If path is not reachable from the supplied root, then the value of - * root is changed (without modifying refcounts). */ -static int prepend_path(const struct path *path, struct path *root, +static int prepend_path(const struct path *path, + const struct path *root, char **buffer, int *buflen) { struct dentry *dentry = path->dentry; @@ -2483,10 +2481,10 @@ static int prepend_path(const struct path *path, struct path *root, dentry = parent; } -out: if (!error && !slash) error = prepend(buffer, buflen, "/", 1); +out: br_read_unlock(vfsmount_lock); return error; @@ -2500,15 +2498,17 @@ global_root: WARN(1, "Root dentry has weird name <%.*s>\n", (int) dentry->d_name.len, dentry->d_name.name); } - root->mnt = vfsmnt; - root->dentry = dentry; + if (!slash) + error = prepend(buffer, buflen, "/", 1); + if (!error) + error = vfsmnt->mnt_ns ? 1 : 2; goto out; } /** * __d_path - return the path of a dentry * @path: the dentry/vfsmount to report - * @root: root vfsmnt/dentry (may be modified by this function) + * @root: root vfsmnt/dentry * @buf: buffer to return value in * @buflen: buffer length * @@ -2519,10 +2519,10 @@ global_root: * * "buflen" should be positive. * - * If path is not reachable from the supplied root, then the value of - * root is changed (without modifying refcounts). + * If the path is not reachable from the supplied root, return %NULL. */ -char *__d_path(const struct path *path, struct path *root, +char *__d_path(const struct path *path, + const struct path *root, char *buf, int buflen) { char *res = buf + buflen; @@ -2533,7 +2533,28 @@ char *__d_path(const struct path *path, struct path *root, error = prepend_path(path, root, &res, &buflen); write_sequnlock(&rename_lock); - if (error) + if (error < 0) + return ERR_PTR(error); + if (error > 0) + return NULL; + return res; +} + +char *d_absolute_path(const struct path *path, + char *buf, int buflen) +{ + struct path root = {}; + char *res = buf + buflen; + int error; + + prepend(&res, &buflen, "\0", 1); + write_seqlock(&rename_lock); + error = prepend_path(path, &root, &res, &buflen); + write_sequnlock(&rename_lock); + + if (error > 1) + error = -EINVAL; + if (error < 0) return ERR_PTR(error); return res; } @@ -2541,8 +2562,9 @@ char *__d_path(const struct path *path, struct path *root, /* * same as __d_path but appends "(deleted)" for unlinked files. */ -static int path_with_deleted(const struct path *path, struct path *root, - char **buf, int *buflen) +static int path_with_deleted(const struct path *path, + const struct path *root, + char **buf, int *buflen) { prepend(buf, buflen, "\0", 1); if (d_unlinked(path->dentry)) { @@ -2579,7 +2601,6 @@ char *d_path(const struct path *path, char *buf, int buflen) { char *res = buf + buflen; struct path root; - struct path tmp; int error; /* @@ -2594,9 +2615,8 @@ char *d_path(const struct path *path, char *buf, int buflen) get_fs_root(current->fs, &root); write_seqlock(&rename_lock); - tmp = root; - error = path_with_deleted(path, &tmp, &res, &buflen); - if (error) + error = path_with_deleted(path, &root, &res, &buflen); + if (error < 0) res = ERR_PTR(error); write_sequnlock(&rename_lock); path_put(&root); @@ -2617,7 +2637,6 @@ char *d_path_with_unreachable(const struct path *path, char *buf, int buflen) { char *res = buf + buflen; struct path root; - struct path tmp; int error; if (path->dentry->d_op && path->dentry->d_op->d_dname) @@ -2625,9 +2644,8 @@ char *d_path_with_unreachable(const struct path *path, char *buf, int buflen) get_fs_root(current->fs, &root); write_seqlock(&rename_lock); - tmp = root; - error = path_with_deleted(path, &tmp, &res, &buflen); - if (!error && !path_equal(&tmp, &root)) + error = path_with_deleted(path, &root, &res, &buflen); + if (error > 0) error = prepend_unreachable(&res, &buflen); write_sequnlock(&rename_lock); path_put(&root); @@ -2758,19 +2776,18 @@ SYSCALL_DEFINE2(getcwd, char __user *, buf, unsigned long, size) write_seqlock(&rename_lock); if (!d_unlinked(pwd.dentry)) { unsigned long len; - struct path tmp = root; char *cwd = page + PAGE_SIZE; int buflen = PAGE_SIZE; prepend(&cwd, &buflen, "\0", 1); - error = prepend_path(&pwd, &tmp, &cwd, &buflen); + error = prepend_path(&pwd, &root, &cwd, &buflen); write_sequnlock(&rename_lock); - if (error) + if (error < 0) goto out; /* Unreachable from current root */ - if (!path_equal(&tmp, &root)) { + if (error > 0) { error = prepend_unreachable(&cwd, &buflen); if (error) goto out; diff --git a/fs/namespace.c b/fs/namespace.c index 6d3a1963879..cfc6d4448aa 100644 --- a/fs/namespace.c +++ b/fs/namespace.c @@ -1048,15 +1048,12 @@ static int show_mountinfo(struct seq_file *m, void *v) if (err) goto out; seq_putc(m, ' '); - seq_path_root(m, &mnt_path, &root, " \t\n\\"); - if (root.mnt != p->root.mnt || root.dentry != p->root.dentry) { - /* - * Mountpoint is outside root, discard that one. Ugly, - * but less so than trying to do that in iterator in a - * race-free way (due to renames). - */ - return SEQ_SKIP; - } + + /* mountpoints outside of chroot jail will give SEQ_SKIP on this */ + err = seq_path_root(m, &mnt_path, &root, " \t\n\\"); + if (err) + goto out; + seq_puts(m, mnt->mnt_flags & MNT_READONLY ? " ro" : " rw"); show_mnt_opts(m, mnt); @@ -2776,3 +2773,8 @@ void kern_unmount(struct vfsmount *mnt) } } EXPORT_SYMBOL(kern_unmount); + +bool our_mnt(struct vfsmount *mnt) +{ + return check_mnt(mnt); +} diff --git a/fs/seq_file.c b/fs/seq_file.c index 05d6b0e78c9..dba43c3ea3a 100644 --- a/fs/seq_file.c +++ b/fs/seq_file.c @@ -449,8 +449,6 @@ EXPORT_SYMBOL(seq_path); /* * Same as seq_path, but relative to supplied root. - * - * root may be changed, see __d_path(). */ int seq_path_root(struct seq_file *m, struct path *path, struct path *root, char *esc) @@ -463,6 +461,8 @@ int seq_path_root(struct seq_file *m, struct path *path, struct path *root, char *p; p = __d_path(path, root, buf, size); + if (!p) + return SEQ_SKIP; res = PTR_ERR(p); if (!IS_ERR(p)) { char *end = mangle_path(buf, p, esc); @@ -474,7 +474,7 @@ int seq_path_root(struct seq_file *m, struct path *path, struct path *root, } seq_commit(m, res); - return res < 0 ? res : 0; + return res < 0 && res != -ENAMETOOLONG ? res : 0; } /* diff --git a/include/linux/dcache.h b/include/linux/dcache.h index 4df92619936..ed9f74f6c51 100644 --- a/include/linux/dcache.h +++ b/include/linux/dcache.h @@ -339,7 +339,8 @@ extern int d_validate(struct dentry *, struct dentry *); */ extern char *dynamic_dname(struct dentry *, char *, int, const char *, ...); -extern char *__d_path(const struct path *path, struct path *root, char *, int); +extern char *__d_path(const struct path *, const struct path *, char *, int); +extern char *d_absolute_path(const struct path *, char *, int); extern char *d_path(const struct path *, char *, int); extern char *d_path_with_unreachable(const struct path *, char *, int); extern char *dentry_path_raw(struct dentry *, char *, int); diff --git a/include/linux/fs.h b/include/linux/fs.h index e3130220ce3..019dc558df1 100644 --- a/include/linux/fs.h +++ b/include/linux/fs.h @@ -1942,6 +1942,7 @@ extern int fd_statfs(int, struct kstatfs *); extern int statfs_by_dentry(struct dentry *, struct kstatfs *); extern int freeze_super(struct super_block *super); extern int thaw_super(struct super_block *super); +extern bool our_mnt(struct vfsmount *mnt); extern int current_umask(void); diff --git a/security/apparmor/path.c b/security/apparmor/path.c index 36cc0cc39e7..b566eba4a65 100644 --- a/security/apparmor/path.c +++ b/security/apparmor/path.c @@ -57,23 +57,44 @@ static int prepend(char **buffer, int buflen, const char *str, int namelen) static int d_namespace_path(struct path *path, char *buf, int buflen, char **name, int flags) { - struct path root, tmp; char *res; - int connected, error = 0; + int error = 0; + int connected = 1; + + if (path->mnt->mnt_flags & MNT_INTERNAL) { + /* it's not mounted anywhere */ + res = dentry_path(path->dentry, buf, buflen); + *name = res; + if (IS_ERR(res)) { + *name = buf; + return PTR_ERR(res); + } + if (path->dentry->d_sb->s_magic == PROC_SUPER_MAGIC && + strncmp(*name, "/sys/", 5) == 0) { + /* TODO: convert over to using a per namespace + * control instead of hard coded /proc + */ + return prepend(name, *name - buf, "/proc", 5); + } + return 0; + } - /* Get the root we want to resolve too, released below */ + /* resolve paths relative to chroot?*/ if (flags & PATH_CHROOT_REL) { - /* resolve paths relative to chroot */ + struct path root; get_fs_root(current->fs, &root); - } else { - /* resolve paths relative to namespace */ - root.mnt = current->nsproxy->mnt_ns->root; - root.dentry = root.mnt->mnt_root; - path_get(&root); + res = __d_path(path, &root, buf, buflen); + if (res && !IS_ERR(res)) { + /* everything's fine */ + *name = res; + path_put(&root); + goto ok; + } + path_put(&root); + connected = 0; } - tmp = root; - res = __d_path(path, &tmp, buf, buflen); + res = d_absolute_path(path, buf, buflen); *name = res; /* handle error conditions - and still allow a partial path to @@ -84,7 +105,10 @@ static int d_namespace_path(struct path *path, char *buf, int buflen, *name = buf; goto out; } + if (!our_mnt(path->mnt)) + connected = 0; +ok: /* Handle two cases: * 1. A deleted dentry && profile is not allowing mediation of deleted * 2. On some filesystems, newly allocated dentries appear to the @@ -97,10 +121,7 @@ static int d_namespace_path(struct path *path, char *buf, int buflen, goto out; } - /* Determine if the path is connected to the expected root */ - connected = tmp.dentry == root.dentry && tmp.mnt == root.mnt; - - /* If the path is not connected, + /* If the path is not connected to the expected root, * check if it is a sysctl and handle specially else remove any * leading / that __d_path may have returned. * Unless @@ -112,17 +133,9 @@ static int d_namespace_path(struct path *path, char *buf, int buflen, * namespace root. */ if (!connected) { - /* is the disconnect path a sysctl? */ - if (tmp.dentry->d_sb->s_magic == PROC_SUPER_MAGIC && - strncmp(*name, "/sys/", 5) == 0) { - /* TODO: convert over to using a per namespace - * control instead of hard coded /proc - */ - error = prepend(name, *name - buf, "/proc", 5); - } else if (!(flags & PATH_CONNECT_PATH) && + if (!(flags & PATH_CONNECT_PATH) && !(((flags & CHROOT_NSCONNECT) == CHROOT_NSCONNECT) && - (tmp.mnt == current->nsproxy->mnt_ns->root && - tmp.dentry == tmp.mnt->mnt_root))) { + our_mnt(path->mnt))) { /* disconnected path, don't return pathname starting * with '/' */ @@ -133,8 +146,6 @@ static int d_namespace_path(struct path *path, char *buf, int buflen, } out: - path_put(&root); - return error; } diff --git a/security/tomoyo/realpath.c b/security/tomoyo/realpath.c index 738bbdf8d4c..36fa7c9bedc 100644 --- a/security/tomoyo/realpath.c +++ b/security/tomoyo/realpath.c @@ -101,9 +101,8 @@ static char *tomoyo_get_absolute_path(struct path *path, char * const buffer, { char *pos = ERR_PTR(-ENOMEM); if (buflen >= 256) { - struct path ns_root = { }; /* go to whatever namespace root we are under */ - pos = __d_path(path, &ns_root, buffer, buflen - 1); + pos = d_absolute_path(path, buffer, buflen - 1); if (!IS_ERR(pos) && *pos == '/' && pos[1]) { struct inode *inode = path->dentry->d_inode; if (inode && S_ISDIR(inode->i_mode)) { -- cgit v1.2.3-70-g09d2 From 062c05c46bd4358aad7a0e0cb5ffeb98ab935286 Mon Sep 17 00:00:00 2001 From: Alexandre Oliva Date: Wed, 7 Dec 2011 19:50:42 -0500 Subject: Btrfs: try to allocate from cluster even at LOOP_NO_EMPTY_SIZE If we reach LOOP_NO_EMPTY_SIZE, we won't even try to use a cluster that others might have set up. Odds are that there won't be one, but if someone else succeeded in setting it up, we might as well use it, even if we don't try to set up a cluster again. Signed-off-by: Alexandre Oliva Signed-off-by: Chris Mason --- fs/btrfs/extent-tree.c | 26 ++++++++++++++++++-------- 1 file changed, 18 insertions(+), 8 deletions(-) (limited to 'fs') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index 813c6bb96c9..db0b23b14f2 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -5285,15 +5285,10 @@ alloc: spin_unlock(&block_group->free_space_ctl->tree_lock); /* - * Ok we want to try and use the cluster allocator, so lets look - * there, unless we are on LOOP_NO_EMPTY_SIZE, since we will - * have tried the cluster allocator plenty of times at this - * point and not have found anything, so we are likely way too - * fragmented for the clustering stuff to find anything, so lets - * just skip it and let the allocator find whatever block it can - * find + * Ok we want to try and use the cluster allocator, so + * lets look there */ - if (last_ptr && loop < LOOP_NO_EMPTY_SIZE) { + if (last_ptr) { /* * the refill lock keeps out other * people trying to start a new cluster @@ -5342,6 +5337,20 @@ alloc: } spin_unlock(&last_ptr->lock); refill_cluster: + /* If we are on LOOP_NO_EMPTY_SIZE, we can't + * set up a new clusters, so lets just skip it + * and let the allocator find whatever block + * it can find. If we reach this point, we + * will have tried the cluster allocator + * plenty of times and not have found + * anything, so we are likely way too + * fragmented for the clustering stuff to find + * anything. */ + if (loop >= LOOP_NO_EMPTY_SIZE) { + spin_unlock(&last_ptr->refill_lock); + goto unclustered_alloc; + } + /* * this cluster didn't work out, free it and * start over @@ -5389,6 +5398,7 @@ refill_cluster: goto loop; } +unclustered_alloc: offset = btrfs_find_space_for_alloc(block_group, search_start, num_bytes, empty_size); /* -- cgit v1.2.3-70-g09d2 From 274bd4fb3ed6b72c1d77ef8850511f09fc6b8e4d Mon Sep 17 00:00:00 2001 From: Alexandre Oliva Date: Wed, 7 Dec 2011 20:08:40 -0500 Subject: Btrfs: try cluster but don't advance in search list When we find an existing cluster, we switch to its block group as the current block group, possibly skipping multiple blocks in the process. Furthermore, under heavy contention, multiple threads may fail to allocate from a cluster and then release just-created clusters just to proceed to create new ones in a different block group. This patch tries to allocate from an existing cluster regardless of its block group, and doesn't switch to that group, instead proceeding to try to allocate a cluster from the group it was iterating before the attempt. Signed-off-by: Alexandre Oliva Signed-off-by: Chris Mason --- fs/btrfs/extent-tree.c | 74 +++++++++++++++++++++----------------------------- 1 file changed, 31 insertions(+), 43 deletions(-) (limited to 'fs') diff --git a/fs/btrfs/extent-tree.c b/fs/btrfs/extent-tree.c index db0b23b14f2..05e1386b8be 100644 --- a/fs/btrfs/extent-tree.c +++ b/fs/btrfs/extent-tree.c @@ -5106,11 +5106,11 @@ static noinline int find_free_extent(struct btrfs_trans_handle *trans, struct btrfs_root *root = orig_root->fs_info->extent_root; struct btrfs_free_cluster *last_ptr = NULL; struct btrfs_block_group_cache *block_group = NULL; + struct btrfs_block_group_cache *used_block_group; int empty_cluster = 2 * 1024 * 1024; int allowed_chunk_alloc = 0; int done_chunk_alloc = 0; struct btrfs_space_info *space_info; - int last_ptr_loop = 0; int loop = 0; int index = 0; int alloc_type = (data & BTRFS_BLOCK_GROUP_DATA) ? @@ -5172,6 +5172,7 @@ static noinline int find_free_extent(struct btrfs_trans_handle *trans, ideal_cache: block_group = btrfs_lookup_block_group(root->fs_info, search_start); + used_block_group = block_group; /* * we don't want to use the block group if it doesn't match our * allocation bits, or if its not cached. @@ -5209,6 +5210,7 @@ search: u64 offset; int cached; + used_block_group = block_group; btrfs_get_block_group(block_group); search_start = block_group->key.objectid; @@ -5294,49 +5296,33 @@ alloc: * people trying to start a new cluster */ spin_lock(&last_ptr->refill_lock); - if (!last_ptr->block_group || - last_ptr->block_group->ro || - !block_group_bits(last_ptr->block_group, data)) + used_block_group = last_ptr->block_group; + if (used_block_group != block_group && + (!used_block_group || + used_block_group->ro || + !block_group_bits(used_block_group, data))) { + used_block_group = block_group; goto refill_cluster; + } + + if (used_block_group != block_group) + btrfs_get_block_group(used_block_group); - offset = btrfs_alloc_from_cluster(block_group, last_ptr, - num_bytes, search_start); + offset = btrfs_alloc_from_cluster(used_block_group, + last_ptr, num_bytes, used_block_group->key.objectid); if (offset) { /* we have a block, we're done */ spin_unlock(&last_ptr->refill_lock); goto checks; } - spin_lock(&last_ptr->lock); - /* - * whoops, this cluster doesn't actually point to - * this block group. Get a ref on the block - * group is does point to and try again - */ - if (!last_ptr_loop && last_ptr->block_group && - last_ptr->block_group != block_group && - index <= - get_block_group_index(last_ptr->block_group)) { - - btrfs_put_block_group(block_group); - block_group = last_ptr->block_group; - btrfs_get_block_group(block_group); - spin_unlock(&last_ptr->lock); - spin_unlock(&last_ptr->refill_lock); - - last_ptr_loop = 1; - search_start = block_group->key.objectid; - /* - * we know this block group is properly - * in the list because - * btrfs_remove_block_group, drops the - * cluster before it removes the block - * group from the list - */ - goto have_block_group; + WARN_ON(last_ptr->block_group != used_block_group); + if (used_block_group != block_group) { + btrfs_put_block_group(used_block_group); + used_block_group = block_group; } - spin_unlock(&last_ptr->lock); refill_cluster: + BUG_ON(used_block_group != block_group); /* If we are on LOOP_NO_EMPTY_SIZE, we can't * set up a new clusters, so lets just skip it * and let the allocator find whatever block @@ -5357,8 +5343,6 @@ refill_cluster: */ btrfs_return_cluster_to_free_space(NULL, last_ptr); - last_ptr_loop = 0; - /* allocate a cluster in this block group */ ret = btrfs_find_space_cluster(trans, root, block_group, last_ptr, @@ -5425,14 +5409,14 @@ checks: search_start = stripe_align(root, offset); /* move on to the next group */ if (search_start + num_bytes >= search_end) { - btrfs_add_free_space(block_group, offset, num_bytes); + btrfs_add_free_space(used_block_group, offset, num_bytes); goto loop; } /* move on to the next group */ if (search_start + num_bytes > - block_group->key.objectid + block_group->key.offset) { - btrfs_add_free_space(block_group, offset, num_bytes); + used_block_group->key.objectid + used_block_group->key.offset) { + btrfs_add_free_space(used_block_group, offset, num_bytes); goto loop; } @@ -5440,14 +5424,14 @@ checks: ins->offset = num_bytes; if (offset < search_start) - btrfs_add_free_space(block_group, offset, + btrfs_add_free_space(used_block_group, offset, search_start - offset); BUG_ON(offset > search_start); - ret = btrfs_update_reserved_bytes(block_group, num_bytes, + ret = btrfs_update_reserved_bytes(used_block_group, num_bytes, alloc_type); if (ret == -EAGAIN) { - btrfs_add_free_space(block_group, offset, num_bytes); + btrfs_add_free_space(used_block_group, offset, num_bytes); goto loop; } @@ -5456,15 +5440,19 @@ checks: ins->offset = num_bytes; if (offset < search_start) - btrfs_add_free_space(block_group, offset, + btrfs_add_free_space(used_block_group, offset, search_start - offset); BUG_ON(offset > search_start); + if (used_block_group != block_group) + btrfs_put_block_group(used_block_group); btrfs_put_block_group(block_group); break; loop: failed_cluster_refill = false; failed_alloc = false; BUG_ON(index != get_block_group_index(block_group)); + if (used_block_group != block_group) + btrfs_put_block_group(used_block_group); btrfs_put_block_group(block_group); } up_read(&space_info->groups_sem); -- cgit v1.2.3-70-g09d2 From a5d16333612718569ffd26064270e535cb9c3928 Mon Sep 17 00:00:00 2001 From: Li Zefan Date: Wed, 7 Dec 2011 20:08:40 -0500 Subject: Btrfs: check if the to-be-added device is writable If we call ioctl(BTRFS_IOC_ADD_DEV) directly, we'll succeed in adding a readonly device to a btrfs filesystem, and btrfs will write to that device, emitting kernel errors: [ 3109.833692] lost page write due to I/O error on loop2 [ 3109.833720] lost page write due to I/O error on loop2 ... Signed-off-by: Li Zefan Signed-off-by: Chris Mason --- fs/btrfs/volumes.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'fs') diff --git a/fs/btrfs/volumes.c b/fs/btrfs/volumes.c index c37433d3cd8..0a8c8f8304b 100644 --- a/fs/btrfs/volumes.c +++ b/fs/btrfs/volumes.c @@ -1611,7 +1611,7 @@ int btrfs_init_new_device(struct btrfs_root *root, char *device_path) if ((sb->s_flags & MS_RDONLY) && !root->fs_info->fs_devices->seeding) return -EINVAL; - bdev = blkdev_get_by_path(device_path, FMODE_EXCL, + bdev = blkdev_get_by_path(device_path, FMODE_WRITE | FMODE_EXCL, root->fs_info->bdev_holder); if (IS_ERR(bdev)) return PTR_ERR(bdev); -- cgit v1.2.3-70-g09d2 From 1cf4ffdb3289624a6462c94f2ce05545b32ef736 Mon Sep 17 00:00:00 2001 From: Liu Bo Date: Wed, 7 Dec 2011 20:08:40 -0500 Subject: Btrfs: drop spin lock when memory alloc fails Drop spin lock in convert_extent_bit() when memory alloc fails, otherwise, it will be a deadlock. Signed-off-by: Liu Bo Signed-off-by: Chris Mason --- fs/btrfs/extent_io.c | 24 ++++++++++++++++-------- 1 file changed, 16 insertions(+), 8 deletions(-) (limited to 'fs') diff --git a/fs/btrfs/extent_io.c b/fs/btrfs/extent_io.c index be1bf627a14..49f3c9dc09f 100644 --- a/fs/btrfs/extent_io.c +++ b/fs/btrfs/extent_io.c @@ -935,8 +935,10 @@ again: node = tree_search(tree, start); if (!node) { prealloc = alloc_extent_state_atomic(prealloc); - if (!prealloc) - return -ENOMEM; + if (!prealloc) { + err = -ENOMEM; + goto out; + } err = insert_state(tree, prealloc, start, end, &bits); prealloc = NULL; BUG_ON(err == -EEXIST); @@ -992,8 +994,10 @@ hit_next: */ if (state->start < start) { prealloc = alloc_extent_state_atomic(prealloc); - if (!prealloc) - return -ENOMEM; + if (!prealloc) { + err = -ENOMEM; + goto out; + } err = split_state(tree, state, prealloc, start); BUG_ON(err == -EEXIST); prealloc = NULL; @@ -1024,8 +1028,10 @@ hit_next: this_end = last_start - 1; prealloc = alloc_extent_state_atomic(prealloc); - if (!prealloc) - return -ENOMEM; + if (!prealloc) { + err = -ENOMEM; + goto out; + } /* * Avoid to free 'prealloc' if it can be merged with @@ -1051,8 +1057,10 @@ hit_next: */ if (state->start <= end && state->end > end) { prealloc = alloc_extent_state_atomic(prealloc); - if (!prealloc) - return -ENOMEM; + if (!prealloc) { + err = -ENOMEM; + goto out; + } err = split_state(tree, state, prealloc, end + 1); BUG_ON(err == -EEXIST); -- cgit v1.2.3-70-g09d2 From 9a5101c8968c7b3e7de5564af6358d7332cf9a6b Mon Sep 17 00:00:00 2001 From: Pavel Shilovsky Date: Mon, 7 Nov 2011 16:11:24 +0300 Subject: CIFS: Add descriptions to the brlock cache functions Signed-off-by: Pavel Shilovsky Signed-off-by: Steve French --- fs/cifs/file.c | 26 ++++++++++++++++++++++++++ 1 file changed, 26 insertions(+) (limited to 'fs') diff --git a/fs/cifs/file.c b/fs/cifs/file.c index cf0b1539b32..4dd9283885e 100644 --- a/fs/cifs/file.c +++ b/fs/cifs/file.c @@ -702,6 +702,13 @@ cifs_find_lock_conflict(struct cifsInodeInfo *cinode, struct cifsLockInfo *lock, lock->type, lock->netfid, conf_lock); } +/* + * Check if there is another lock that prevents us to set the lock (mandatory + * style). If such a lock exists, update the flock structure with its + * properties. Otherwise, set the flock type to F_UNLCK if we can cache brlocks + * or leave it the same if we can't. Returns 0 if we don't need to request to + * the server or 1 otherwise. + */ static int cifs_lock_test(struct cifsInodeInfo *cinode, __u64 offset, __u64 length, __u8 type, __u16 netfid, struct file_lock *flock) @@ -739,6 +746,12 @@ cifs_lock_add(struct cifsInodeInfo *cinode, struct cifsLockInfo *lock) mutex_unlock(&cinode->lock_mutex); } +/* + * Set the byte-range lock (mandatory style). Returns: + * 1) 0, if we set the lock and don't need to request to the server; + * 2) 1, if no locks prevent us but we need to request to the server; + * 3) -EACCESS, if there is a lock that prevents us and wait is false. + */ static int cifs_lock_add_if(struct cifsInodeInfo *cinode, struct cifsLockInfo *lock, bool wait) @@ -778,6 +791,13 @@ try_again: return rc; } +/* + * Check if there is another lock that prevents us to set the lock (posix + * style). If such a lock exists, update the flock structure with its + * properties. Otherwise, set the flock type to F_UNLCK if we can cache brlocks + * or leave it the same if we can't. Returns 0 if we don't need to request to + * the server or 1 otherwise. + */ static int cifs_posix_lock_test(struct file *file, struct file_lock *flock) { @@ -800,6 +820,12 @@ cifs_posix_lock_test(struct file *file, struct file_lock *flock) return rc; } +/* + * Set the byte-range lock (posix style). Returns: + * 1) 0, if we set the lock and don't need to request to the server; + * 2) 1, if we need to request to the server; + * 3) <0, if the error occurs while setting the lock. + */ static int cifs_posix_lock_set(struct file *file, struct file_lock *flock) { -- cgit v1.2.3-70-g09d2 From 59edb63ad08e4f3a9477223c9aaf3fba753085bf Mon Sep 17 00:00:00 2001 From: Steve French Date: Thu, 10 Nov 2011 12:48:20 -0600 Subject: cifs: Fix sparse warning when calling cifs_strtoUCS Fix sparse endian check warning while calling cifs_strtoUCS CHECK fs/cifs/smbencrypt.c fs/cifs/smbencrypt.c:216:37: warning: incorrect type in argument 1 (different base types) fs/cifs/smbencrypt.c:216:37: expected restricted __le16 [usertype] * fs/cifs/smbencrypt.c:216:37: got unsigned short * Signed-off-by: Steve French Acked-by: Shirish Pargaonkar Date: Thu, 1 Dec 2011 20:22:41 -0500 Subject: cifs: attempt to freeze while looping on a receive attempt In the recent overhaul of the demultiplex thread receive path, I neglected to ensure that we attempt to freeze on each pass through the receive loop. Reported-and-Tested-by: Woody Suwalski Reported-and-Tested-by: Adam Williamson Signed-off-by: Jeff Layton Signed-off-by: Steve French --- fs/cifs/connect.c | 2 ++ 1 file changed, 2 insertions(+) (limited to 'fs') diff --git a/fs/cifs/connect.c b/fs/cifs/connect.c index d6a972df033..8cd4b52d421 100644 --- a/fs/cifs/connect.c +++ b/fs/cifs/connect.c @@ -441,6 +441,8 @@ cifs_readv_from_socket(struct TCP_Server_Info *server, struct kvec *iov_orig, smb_msg.msg_controllen = 0; for (total_read = 0; to_read; total_read += length, to_read -= length) { + try_to_freeze(); + if (server_unresponsive(server)) { total_read = -EAGAIN; break; -- cgit v1.2.3-70-g09d2 From 7023676f9ee851d94f0942e879243fc1f9081c47 Mon Sep 17 00:00:00 2001 From: Jeff Layton Date: Thu, 1 Dec 2011 20:23:34 -0500 Subject: cifs: check for NULL last_entry before calling cifs_save_resume_key Prior to commit eaf35b1, cifs_save_resume_key had some NULL pointer checks at the top. It turns out that at least one of those NULL pointer checks is needed after all. When the LastNameOffset in a FIND reply appears to be beyond the end of the buffer, CIFSFindFirst and CIFSFindNext will set srch_inf.last_entry to NULL. Since eaf35b1, the code will now oops in this situation. Fix this by having the callers check for a NULL last entry pointer before calling cifs_save_resume_key. No change is needed for the call site in cifs_readdir as it's not reachable with a NULL current_entry pointer. This should fix: https://bugzilla.redhat.com/show_bug.cgi?id=750247 Cc: stable@vger.kernel.org Cc: Christoph Hellwig Reported-by: Adam G. Metzler Signed-off-by: Jeff Layton Signed-off-by: Steve French --- fs/cifs/readdir.c | 10 ++++++++-- 1 file changed, 8 insertions(+), 2 deletions(-) (limited to 'fs') diff --git a/fs/cifs/readdir.c b/fs/cifs/readdir.c index 5de03ec2014..a090bbe6ee2 100644 --- a/fs/cifs/readdir.c +++ b/fs/cifs/readdir.c @@ -554,7 +554,10 @@ static int find_cifs_entry(const int xid, struct cifs_tcon *pTcon, rc); return rc; } - cifs_save_resume_key(cifsFile->srch_inf.last_entry, cifsFile); + /* FindFirst/Next set last_entry to NULL on malformed reply */ + if (cifsFile->srch_inf.last_entry) + cifs_save_resume_key(cifsFile->srch_inf.last_entry, + cifsFile); } while ((index_to_find >= cifsFile->srch_inf.index_of_last_entry) && @@ -562,7 +565,10 @@ static int find_cifs_entry(const int xid, struct cifs_tcon *pTcon, cFYI(1, "calling findnext2"); rc = CIFSFindNext(xid, pTcon, cifsFile->netfid, &cifsFile->srch_inf); - cifs_save_resume_key(cifsFile->srch_inf.last_entry, cifsFile); + /* FindFirst/Next set last_entry to NULL on malformed reply */ + if (cifsFile->srch_inf.last_entry) + cifs_save_resume_key(cifsFile->srch_inf.last_entry, + cifsFile); if (rc) return -ENOENT; } -- cgit v1.2.3-70-g09d2 From b53fc7c2974a50913f49e1d800fe904a28c338e3 Mon Sep 17 00:00:00 2001 From: Claudio Scordino Date: Thu, 8 Dec 2011 14:33:56 -0800 Subject: fs/proc/meminfo.c: fix compilation error Fix the error message "directives may not be used inside a macro argument" which appears when the kernel is compiled for the cris architecture. Signed-off-by: Claudio Scordino Cc: Andrea Arcangeli Cc: Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- fs/proc/meminfo.c | 7 ++++--- 1 file changed, 4 insertions(+), 3 deletions(-) (limited to 'fs') diff --git a/fs/proc/meminfo.c b/fs/proc/meminfo.c index 586174168e2..80e4645f799 100644 --- a/fs/proc/meminfo.c +++ b/fs/proc/meminfo.c @@ -131,12 +131,13 @@ static int meminfo_proc_show(struct seq_file *m, void *v) K(i.freeswap), K(global_page_state(NR_FILE_DIRTY)), K(global_page_state(NR_WRITEBACK)), - K(global_page_state(NR_ANON_PAGES) #ifdef CONFIG_TRANSPARENT_HUGEPAGE + K(global_page_state(NR_ANON_PAGES) + global_page_state(NR_ANON_TRANSPARENT_HUGEPAGES) * - HPAGE_PMD_NR + HPAGE_PMD_NR), +#else + K(global_page_state(NR_ANON_PAGES)), #endif - ), K(global_page_state(NR_FILE_MAPPED)), K(global_page_state(NR_SHMEM)), K(global_page_state(NR_SLAB_RECLAIMABLE) + -- cgit v1.2.3-70-g09d2 From 2a95ea6c0d129b4568fb64e1deda16ceb20e6636 Mon Sep 17 00:00:00 2001 From: Michal Hocko Date: Thu, 8 Dec 2011 14:34:32 -0800 Subject: procfs: do not overflow get_{idle,iowait}_time for nohz Since commit a25cac5198d4 ("proc: Consider NO_HZ when printing idle and iowait times") we are reporting idle/io_wait time also while a CPU is tickless. We rely on get_{idle,iowait}_time functions to retrieve proper data. These functions, however, use usecs_to_cputime to translate micro seconds time to cputime64_t. This is just an alias to usecs_to_jiffies which reduces the data type from u64 to unsigned int and also checks whether the given parameter overflows jiffies_to_usecs(MAX_JIFFY_OFFSET) and returns MAX_JIFFY_OFFSET in that case. When we overflow depends on CONFIG_HZ but especially for CONFIG_HZ_300 it is quite low (1431649781) so we are getting MAX_JIFFY_OFFSET for >3000s! until we overflow unsigned int. Just for reference CONFIG_HZ_100 has an overflow window around 20s, CONFIG_HZ_250 ~8s and CONFIG_HZ_1000 ~2s. This results in a bug when people saw [h]top going mad reporting 100% CPU usage even though there was basically no CPU load. The reason was simply that /proc/stat stopped reporting idle/io_wait changes (and reported MAX_JIFFY_OFFSET) and so the only change happening was for user system time. Let's use nsecs_to_jiffies64 instead which doesn't reduce the precision to 32b type and it is much more appropriate for cumulative time values (unlike usecs_to_jiffies which intended for timeout calculations). Signed-off-by: Michal Hocko Tested-by: Artem S. Tashkinov Cc: Dave Jones Cc: Arnd Bergmann Cc: Alexey Dobriyan Cc: Thomas Gleixner Signed-off-by: Andrew Morton Signed-off-by: Linus Torvalds --- fs/proc/stat.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'fs') diff --git a/fs/proc/stat.c b/fs/proc/stat.c index 42b274da92c..2a30d67dd6b 100644 --- a/fs/proc/stat.c +++ b/fs/proc/stat.c @@ -32,7 +32,7 @@ static cputime64_t get_idle_time(int cpu) idle = kstat_cpu(cpu).cpustat.idle; idle = cputime64_add(idle, arch_idle_time(cpu)); } else - idle = usecs_to_cputime(idle_time); + idle = nsecs_to_jiffies64(1000 * idle_time); return idle; } @@ -46,7 +46,7 @@ static cputime64_t get_iowait_time(int cpu) /* !NO_HZ so we can rely on cpustat.iowait */ iowait = kstat_cpu(cpu).cpustat.iowait; else - iowait = usecs_to_cputime(iowait_time); + iowait = nsecs_to_jiffies64(1000 * iowait_time); return iowait; } -- cgit v1.2.3-70-g09d2