aboutsummaryrefslogtreecommitdiff
path: root/fs/ocfs2/journal.c
diff options
context:
space:
mode:
Diffstat (limited to 'fs/ocfs2/journal.c')
-rw-r--r--fs/ocfs2/journal.c352
1 files changed, 176 insertions, 176 deletions
diff --git a/fs/ocfs2/journal.c b/fs/ocfs2/journal.c
index bf34c491ae9..4b0c68849b3 100644
--- a/fs/ocfs2/journal.c
+++ b/fs/ocfs2/journal.c
@@ -30,8 +30,8 @@
#include <linux/kthread.h>
#include <linux/time.h>
#include <linux/random.h>
+#include <linux/delay.h>
-#define MLOG_MASK_PREFIX ML_JOURNAL
#include <cluster/masklog.h>
#include "ocfs2.h"
@@ -52,6 +52,7 @@
#include "quota.h"
#include "buffer_head_io.h"
+#include "ocfs2_trace.h"
DEFINE_SPINLOCK(trans_inc_lock);
@@ -301,19 +302,17 @@ static int ocfs2_commit_cache(struct ocfs2_super *osb)
{
int status = 0;
unsigned int flushed;
- unsigned long old_id;
struct ocfs2_journal *journal = NULL;
- mlog_entry_void();
-
journal = osb->journal;
/* Flush all pending commits and checkpoint the journal. */
down_write(&journal->j_trans_barrier);
- if (atomic_read(&journal->j_num_trans) == 0) {
+ flushed = atomic_read(&journal->j_num_trans);
+ trace_ocfs2_commit_cache_begin(flushed);
+ if (flushed == 0) {
up_write(&journal->j_trans_barrier);
- mlog(0, "No transactions for me to flush!\n");
goto finally;
}
@@ -326,25 +325,20 @@ static int ocfs2_commit_cache(struct ocfs2_super *osb)
goto finally;
}
- old_id = ocfs2_inc_trans_id(journal);
+ ocfs2_inc_trans_id(journal);
flushed = atomic_read(&journal->j_num_trans);
atomic_set(&journal->j_num_trans, 0);
up_write(&journal->j_trans_barrier);
- mlog(0, "commit_thread: flushed transaction %lu (%u handles)\n",
- journal->j_trans_id, flushed);
+ trace_ocfs2_commit_cache_end(journal->j_trans_id, flushed);
ocfs2_wake_downconvert_thread(osb);
wake_up(&journal->j_checkpointed);
finally:
- mlog_exit(status);
return status;
}
-/* pass it NULL and it will allocate a new handle object for you. If
- * you pass it a handle however, it may still return error, in which
- * case it has free'd the passed handle for you. */
handle_t *ocfs2_start_trans(struct ocfs2_super *osb, int max_buffs)
{
journal_t *journal = osb->journal->j_journal;
@@ -362,11 +356,14 @@ handle_t *ocfs2_start_trans(struct ocfs2_super *osb, int max_buffs)
if (journal_current_handle())
return jbd2_journal_start(journal, max_buffs);
+ sb_start_intwrite(osb->sb);
+
down_read(&osb->journal->j_trans_barrier);
handle = jbd2_journal_start(journal, max_buffs);
if (IS_ERR(handle)) {
up_read(&osb->journal->j_trans_barrier);
+ sb_end_intwrite(osb->sb);
mlog_errno(PTR_ERR(handle));
@@ -395,16 +392,16 @@ int ocfs2_commit_trans(struct ocfs2_super *osb,
if (ret < 0)
mlog_errno(ret);
- if (!nested)
+ if (!nested) {
up_read(&journal->j_trans_barrier);
+ sb_end_intwrite(osb->sb);
+ }
return ret;
}
/*
- * 'nblocks' is what you want to add to the current
- * transaction. extend_trans will either extend the current handle by
- * nblocks, or commit it and start a new one with nblocks credits.
+ * 'nblocks' is what you want to add to the current transaction.
*
* This might call jbd2_journal_restart() which will commit dirty buffers
* and then restart the transaction. Before calling
@@ -422,14 +419,17 @@ int ocfs2_commit_trans(struct ocfs2_super *osb,
*/
int ocfs2_extend_trans(handle_t *handle, int nblocks)
{
- int status;
+ int status, old_nblocks;
BUG_ON(!handle);
- BUG_ON(!nblocks);
+ BUG_ON(nblocks < 0);
- mlog_entry_void();
+ if (!nblocks)
+ return 0;
+
+ old_nblocks = handle->h_buffer_credits;
- mlog(0, "Trying to extend transaction by %d blocks\n", nblocks);
+ trace_ocfs2_extend_trans(old_nblocks, nblocks);
#ifdef CONFIG_OCFS2_DEBUG_FS
status = 1;
@@ -442,10 +442,9 @@ int ocfs2_extend_trans(handle_t *handle, int nblocks)
#endif
if (status > 0) {
- mlog(0,
- "jbd2_journal_extend failed, trying "
- "jbd2_journal_restart\n");
- status = jbd2_journal_restart(handle, nblocks);
+ trace_ocfs2_extend_trans_restart(old_nblocks + nblocks);
+ status = jbd2_journal_restart(handle,
+ old_nblocks + nblocks);
if (status < 0) {
mlog_errno(status);
goto bail;
@@ -454,11 +453,44 @@ int ocfs2_extend_trans(handle_t *handle, int nblocks)
status = 0;
bail:
+ return status;
+}
+
+/*
+ * If we have fewer than thresh credits, extend by OCFS2_MAX_TRANS_DATA.
+ * If that fails, restart the transaction & regain write access for the
+ * buffer head which is used for metadata modifications.
+ * Taken from Ext4: extend_or_restart_transaction()
+ */
+int ocfs2_allocate_extend_trans(handle_t *handle, int thresh)
+{
+ int status, old_nblks;
+
+ BUG_ON(!handle);
+
+ old_nblks = handle->h_buffer_credits;
+ trace_ocfs2_allocate_extend_trans(old_nblks, thresh);
+
+ if (old_nblks < thresh)
+ return 0;
+
+ status = jbd2_journal_extend(handle, OCFS2_MAX_TRANS_DATA);
+ if (status < 0) {
+ mlog_errno(status);
+ goto bail;
+ }
+
+ if (status > 0) {
+ status = jbd2_journal_restart(handle, OCFS2_MAX_TRANS_DATA);
+ if (status < 0)
+ mlog_errno(status);
+ }
- mlog_exit(status);
+bail:
return status;
}
+
struct ocfs2_triggers {
struct jbd2_buffer_trigger_type ot_triggers;
int ot_offset;
@@ -469,7 +501,7 @@ static inline struct ocfs2_triggers *to_ocfs2_trigger(struct jbd2_buffer_trigger
return container_of(triggers, struct ocfs2_triggers, ot_triggers);
}
-static void ocfs2_commit_trigger(struct jbd2_buffer_trigger_type *triggers,
+static void ocfs2_frozen_trigger(struct jbd2_buffer_trigger_type *triggers,
struct buffer_head *bh,
void *data, size_t size)
{
@@ -488,7 +520,7 @@ static void ocfs2_commit_trigger(struct jbd2_buffer_trigger_type *triggers,
* Quota blocks have their own trigger because the struct ocfs2_block_check
* offset depends on the blocksize.
*/
-static void ocfs2_dq_commit_trigger(struct jbd2_buffer_trigger_type *triggers,
+static void ocfs2_dq_frozen_trigger(struct jbd2_buffer_trigger_type *triggers,
struct buffer_head *bh,
void *data, size_t size)
{
@@ -508,7 +540,7 @@ static void ocfs2_dq_commit_trigger(struct jbd2_buffer_trigger_type *triggers,
* Directory blocks also have their own trigger because the
* struct ocfs2_block_check offset depends on the blocksize.
*/
-static void ocfs2_db_commit_trigger(struct jbd2_buffer_trigger_type *triggers,
+static void ocfs2_db_frozen_trigger(struct jbd2_buffer_trigger_type *triggers,
struct buffer_head *bh,
void *data, size_t size)
{
@@ -541,7 +573,7 @@ static void ocfs2_abort_trigger(struct jbd2_buffer_trigger_type *triggers,
static struct ocfs2_triggers di_triggers = {
.ot_triggers = {
- .t_commit = ocfs2_commit_trigger,
+ .t_frozen = ocfs2_frozen_trigger,
.t_abort = ocfs2_abort_trigger,
},
.ot_offset = offsetof(struct ocfs2_dinode, i_check),
@@ -549,7 +581,7 @@ static struct ocfs2_triggers di_triggers = {
static struct ocfs2_triggers eb_triggers = {
.ot_triggers = {
- .t_commit = ocfs2_commit_trigger,
+ .t_frozen = ocfs2_frozen_trigger,
.t_abort = ocfs2_abort_trigger,
},
.ot_offset = offsetof(struct ocfs2_extent_block, h_check),
@@ -557,7 +589,7 @@ static struct ocfs2_triggers eb_triggers = {
static struct ocfs2_triggers rb_triggers = {
.ot_triggers = {
- .t_commit = ocfs2_commit_trigger,
+ .t_frozen = ocfs2_frozen_trigger,
.t_abort = ocfs2_abort_trigger,
},
.ot_offset = offsetof(struct ocfs2_refcount_block, rf_check),
@@ -565,7 +597,7 @@ static struct ocfs2_triggers rb_triggers = {
static struct ocfs2_triggers gd_triggers = {
.ot_triggers = {
- .t_commit = ocfs2_commit_trigger,
+ .t_frozen = ocfs2_frozen_trigger,
.t_abort = ocfs2_abort_trigger,
},
.ot_offset = offsetof(struct ocfs2_group_desc, bg_check),
@@ -573,14 +605,14 @@ static struct ocfs2_triggers gd_triggers = {
static struct ocfs2_triggers db_triggers = {
.ot_triggers = {
- .t_commit = ocfs2_db_commit_trigger,
+ .t_frozen = ocfs2_db_frozen_trigger,
.t_abort = ocfs2_abort_trigger,
},
};
static struct ocfs2_triggers xb_triggers = {
.ot_triggers = {
- .t_commit = ocfs2_commit_trigger,
+ .t_frozen = ocfs2_frozen_trigger,
.t_abort = ocfs2_abort_trigger,
},
.ot_offset = offsetof(struct ocfs2_xattr_block, xb_check),
@@ -588,14 +620,14 @@ static struct ocfs2_triggers xb_triggers = {
static struct ocfs2_triggers dq_triggers = {
.ot_triggers = {
- .t_commit = ocfs2_dq_commit_trigger,
+ .t_frozen = ocfs2_dq_frozen_trigger,
.t_abort = ocfs2_abort_trigger,
},
};
static struct ocfs2_triggers dr_triggers = {
.ot_triggers = {
- .t_commit = ocfs2_commit_trigger,
+ .t_frozen = ocfs2_frozen_trigger,
.t_abort = ocfs2_abort_trigger,
},
.ot_offset = offsetof(struct ocfs2_dx_root_block, dr_check),
@@ -603,7 +635,7 @@ static struct ocfs2_triggers dr_triggers = {
static struct ocfs2_triggers dl_triggers = {
.ot_triggers = {
- .t_commit = ocfs2_commit_trigger,
+ .t_frozen = ocfs2_frozen_trigger,
.t_abort = ocfs2_abort_trigger,
},
.ot_offset = offsetof(struct ocfs2_dx_leaf, dl_check),
@@ -623,12 +655,9 @@ static int __ocfs2_journal_access(handle_t *handle,
BUG_ON(!handle);
BUG_ON(!bh);
- mlog_entry("bh->b_blocknr=%llu, type=%d (\"%s\"), bh->b_size = %zu\n",
- (unsigned long long)bh->b_blocknr, type,
- (type == OCFS2_JOURNAL_ACCESS_CREATE) ?
- "OCFS2_JOURNAL_ACCESS_CREATE" :
- "OCFS2_JOURNAL_ACCESS_WRITE",
- bh->b_size);
+ trace_ocfs2_journal_access(
+ (unsigned long long)ocfs2_metadata_cache_owner(ci),
+ (unsigned long long)bh->b_blocknr, type, bh->b_size);
/* we can safely remove this assertion after testing. */
if (!buffer_uptodate(bh)) {
@@ -669,7 +698,6 @@ static int __ocfs2_journal_access(handle_t *handle,
mlog(ML_ERROR, "Error %d getting %d access to buffer!\n",
status, type);
- mlog_exit(status);
return status;
}
@@ -734,22 +762,14 @@ int ocfs2_journal_access(handle_t *handle, struct ocfs2_caching_info *ci,
return __ocfs2_journal_access(handle, ci, bh, NULL, type);
}
-int ocfs2_journal_dirty(handle_t *handle,
- struct buffer_head *bh)
+void ocfs2_journal_dirty(handle_t *handle, struct buffer_head *bh)
{
int status;
- mlog_entry("(bh->b_blocknr=%llu)\n",
- (unsigned long long)bh->b_blocknr);
+ trace_ocfs2_journal_dirty((unsigned long long)bh->b_blocknr);
status = jbd2_journal_dirty_metadata(handle, bh);
- if (status < 0)
- mlog(ML_ERROR, "Could not dirty metadata buffer. "
- "(bh->b_blocknr=%llu)\n",
- (unsigned long long)bh->b_blocknr);
-
- mlog_exit(status);
- return status;
+ BUG_ON(status);
}
#define OCFS2_DEFAULT_COMMIT_INTERVAL (HZ * JBD2_DEFAULT_MAX_COMMIT_AGE)
@@ -762,13 +782,13 @@ void ocfs2_set_journal_params(struct ocfs2_super *osb)
if (osb->osb_commit_interval)
commit_interval = osb->osb_commit_interval;
- spin_lock(&journal->j_state_lock);
+ write_lock(&journal->j_state_lock);
journal->j_commit_interval = commit_interval;
if (osb->s_mount_opt & OCFS2_MOUNT_BARRIER)
journal->j_flags |= JBD2_BARRIER;
else
journal->j_flags &= ~JBD2_BARRIER;
- spin_unlock(&journal->j_state_lock);
+ write_unlock(&journal->j_state_lock);
}
int ocfs2_journal_init(struct ocfs2_journal *journal, int *dirty)
@@ -781,8 +801,6 @@ int ocfs2_journal_init(struct ocfs2_journal *journal, int *dirty)
struct ocfs2_super *osb;
int inode_lock = 0;
- mlog_entry_void();
-
BUG_ON(!journal);
osb = journal->j_osb;
@@ -819,17 +837,16 @@ int ocfs2_journal_init(struct ocfs2_journal *journal, int *dirty)
inode_lock = 1;
di = (struct ocfs2_dinode *)bh->b_data;
- if (inode->i_size < OCFS2_MIN_JOURNAL_SIZE) {
+ if (i_size_read(inode) < OCFS2_MIN_JOURNAL_SIZE) {
mlog(ML_ERROR, "Journal file size (%lld) is too small!\n",
- inode->i_size);
+ i_size_read(inode));
status = -EINVAL;
goto done;
}
- mlog(0, "inode->i_size = %lld\n", inode->i_size);
- mlog(0, "inode->i_blocks = %llu\n",
- (unsigned long long)inode->i_blocks);
- mlog(0, "inode->ip_clusters = %u\n", OCFS2_I(inode)->ip_clusters);
+ trace_ocfs2_journal_init(i_size_read(inode),
+ (unsigned long long)inode->i_blocks,
+ OCFS2_I(inode)->ip_clusters);
/* call the kernels journal init function now */
j_journal = jbd2_journal_init_inode(inode);
@@ -839,8 +856,7 @@ int ocfs2_journal_init(struct ocfs2_journal *journal, int *dirty)
goto done;
}
- mlog(0, "Returned from jbd2_journal_init_inode\n");
- mlog(0, "j_journal->j_maxlen = %u\n", j_journal->j_maxlen);
+ trace_ocfs2_journal_init_maxlen(j_journal->j_maxlen);
*dirty = (le32_to_cpu(di->id1.journal1.ij_flags) &
OCFS2_JOURNAL_DIRTY_FL);
@@ -865,7 +881,6 @@ done:
}
}
- mlog_exit(status);
return status;
}
@@ -888,8 +903,6 @@ static int ocfs2_journal_toggle_dirty(struct ocfs2_super *osb,
struct buffer_head *bh = journal->j_bh;
struct ocfs2_dinode *fe;
- mlog_entry_void();
-
fe = (struct ocfs2_dinode *)bh->b_data;
/* The journal bh on the osb always comes from ocfs2_journal_init()
@@ -912,7 +925,6 @@ static int ocfs2_journal_toggle_dirty(struct ocfs2_super *osb,
if (status < 0)
mlog_errno(status);
- mlog_exit(status);
return status;
}
@@ -927,8 +939,6 @@ void ocfs2_journal_shutdown(struct ocfs2_super *osb)
struct inode *inode = NULL;
int num_running_trans = 0;
- mlog_entry_void();
-
BUG_ON(!osb);
journal = osb->journal;
@@ -945,10 +955,7 @@ void ocfs2_journal_shutdown(struct ocfs2_super *osb)
BUG();
num_running_trans = atomic_read(&(osb->journal->j_num_trans));
- if (num_running_trans > 0)
- mlog(0, "Shutting down journal: must wait on %d "
- "running transactions!\n",
- num_running_trans);
+ trace_ocfs2_journal_shutdown(num_running_trans);
/* Do a commit_cache here. It will flush our journal, *and*
* release any locks that are still held.
@@ -961,7 +968,7 @@ void ocfs2_journal_shutdown(struct ocfs2_super *osb)
* completely destroy the journal. */
if (osb->commit_task) {
/* Wait for the commit thread */
- mlog(0, "Waiting for ocfs2commit to exit....\n");
+ trace_ocfs2_journal_shutdown_wait(osb->commit_task);
kthread_stop(osb->commit_task);
osb->commit_task = NULL;
}
@@ -1004,7 +1011,6 @@ void ocfs2_journal_shutdown(struct ocfs2_super *osb)
done:
if (inode)
iput(inode);
- mlog_exit_void();
}
static void ocfs2_clear_journal_error(struct super_block *sb,
@@ -1030,8 +1036,6 @@ int ocfs2_journal_load(struct ocfs2_journal *journal, int local, int replayed)
int status = 0;
struct ocfs2_super *osb;
- mlog_entry_void();
-
BUG_ON(!journal);
osb = journal->j_osb;
@@ -1065,7 +1069,6 @@ int ocfs2_journal_load(struct ocfs2_journal *journal, int local, int replayed)
osb->commit_task = NULL;
done:
- mlog_exit(status);
return status;
}
@@ -1076,8 +1079,6 @@ int ocfs2_journal_wipe(struct ocfs2_journal *journal, int full)
{
int status;
- mlog_entry_void();
-
BUG_ON(!journal);
status = jbd2_journal_wipe(journal->j_journal, full);
@@ -1091,7 +1092,6 @@ int ocfs2_journal_wipe(struct ocfs2_journal *journal, int full)
mlog_errno(status);
bail:
- mlog_exit(status);
return status;
}
@@ -1130,11 +1130,9 @@ static int ocfs2_force_read_journal(struct inode *inode)
#define CONCURRENT_JOURNAL_FILL 32ULL
struct buffer_head *bhs[CONCURRENT_JOURNAL_FILL];
- mlog_entry_void();
-
memset(bhs, 0, sizeof(struct buffer_head *) * CONCURRENT_JOURNAL_FILL);
- num_blocks = ocfs2_blocks_for_bytes(inode->i_sb, inode->i_size);
+ num_blocks = ocfs2_blocks_for_bytes(inode->i_sb, i_size_read(inode));
v_blkno = 0;
while (v_blkno < num_blocks) {
status = ocfs2_extent_map_get_blocks(inode, v_blkno,
@@ -1167,7 +1165,6 @@ static int ocfs2_force_read_journal(struct inode *inode)
bail:
for(i = 0; i < CONCURRENT_JOURNAL_FILL; i++)
brelse(bhs[i]);
- mlog_exit(status);
return status;
}
@@ -1191,7 +1188,7 @@ struct ocfs2_la_recovery_item {
*/
void ocfs2_complete_recovery(struct work_struct *work)
{
- int ret;
+ int ret = 0;
struct ocfs2_journal *journal =
container_of(work, struct ocfs2_journal, j_recovery_work);
struct ocfs2_super *osb = journal->j_osb;
@@ -1200,9 +1197,8 @@ void ocfs2_complete_recovery(struct work_struct *work)
struct ocfs2_quota_recovery *qrec;
LIST_HEAD(tmp_la_list);
- mlog_entry_void();
-
- mlog(0, "completing recovery from keventd\n");
+ trace_ocfs2_complete_recovery(
+ (unsigned long long)OCFS2_I(journal->j_inode)->ip_blkno);
spin_lock(&journal->j_lock);
list_splice_init(&journal->j_la_cleanups, &tmp_la_list);
@@ -1211,15 +1207,18 @@ void ocfs2_complete_recovery(struct work_struct *work)
list_for_each_entry_safe(item, n, &tmp_la_list, lri_list) {
list_del_init(&item->lri_list);
- mlog(0, "Complete recovery for slot %d\n", item->lri_slot);
-
ocfs2_wait_on_quotas(osb);
la_dinode = item->lri_la_dinode;
- if (la_dinode) {
- mlog(0, "Clean up local alloc %llu\n",
- (unsigned long long)le64_to_cpu(la_dinode->i_blkno));
+ tl_dinode = item->lri_tl_dinode;
+ qrec = item->lri_qrec;
+
+ trace_ocfs2_complete_recovery_slot(item->lri_slot,
+ la_dinode ? le64_to_cpu(la_dinode->i_blkno) : 0,
+ tl_dinode ? le64_to_cpu(tl_dinode->i_blkno) : 0,
+ qrec);
+ if (la_dinode) {
ret = ocfs2_complete_local_alloc_recovery(osb,
la_dinode);
if (ret < 0)
@@ -1228,11 +1227,7 @@ void ocfs2_complete_recovery(struct work_struct *work)
kfree(la_dinode);
}
- tl_dinode = item->lri_tl_dinode;
if (tl_dinode) {
- mlog(0, "Clean up truncate log %llu\n",
- (unsigned long long)le64_to_cpu(tl_dinode->i_blkno));
-
ret = ocfs2_complete_truncate_log_recovery(osb,
tl_dinode);
if (ret < 0)
@@ -1245,9 +1240,7 @@ void ocfs2_complete_recovery(struct work_struct *work)
if (ret < 0)
mlog_errno(ret);
- qrec = item->lri_qrec;
if (qrec) {
- mlog(0, "Recovering quota files");
ret = ocfs2_finish_quota_recovery(osb, qrec,
item->lri_slot);
if (ret < 0)
@@ -1258,8 +1251,7 @@ void ocfs2_complete_recovery(struct work_struct *work)
kfree(item);
}
- mlog(0, "Recovery completion\n");
- mlog_exit_void();
+ trace_ocfs2_complete_recovery_end(ret);
}
/* NOTE: This function always eats your references to la_dinode and
@@ -1278,11 +1270,8 @@ static void ocfs2_queue_recovery_completion(struct ocfs2_journal *journal,
/* Though we wish to avoid it, we are in fact safe in
* skipping local alloc cleanup as fsck.ocfs2 is more
* than capable of reclaiming unused space. */
- if (la_dinode)
- kfree(la_dinode);
-
- if (tl_dinode)
- kfree(tl_dinode);
+ kfree(la_dinode);
+ kfree(tl_dinode);
if (qrec)
ocfs2_free_quota_recovery(qrec);
@@ -1309,6 +1298,9 @@ void ocfs2_complete_mount_recovery(struct ocfs2_super *osb)
{
struct ocfs2_journal *journal = osb->journal;
+ if (ocfs2_is_hard_readonly(osb))
+ return;
+
/* No need to queue up our truncate_log as regular cleanup will catch
* that */
ocfs2_queue_recovery_completion(journal, osb->slot_num,
@@ -1345,8 +1337,6 @@ static int __ocfs2_recovery_thread(void *arg)
int rm_quota_used = 0, i;
struct ocfs2_quota_recovery *qrec;
- mlog_entry_void();
-
status = ocfs2_wait_on_mount(osb);
if (status < 0) {
goto bail;
@@ -1378,15 +1368,12 @@ restart:
* clear it until ocfs2_recover_node() has succeeded. */
node_num = rm->rm_entries[0];
spin_unlock(&osb->osb_lock);
- mlog(0, "checking node %d\n", node_num);
slot_num = ocfs2_node_num_to_slot(osb, node_num);
+ trace_ocfs2_recovery_thread_node(node_num, slot_num);
if (slot_num == -ENOENT) {
status = 0;
- mlog(0, "no slot for this node, so no recovery"
- "required.\n");
goto skip_recovery;
}
- mlog(0, "node %d was using slot %d\n", node_num, slot_num);
/* It is a bit subtle with quota recovery. We cannot do it
* immediately because we have to obtain cluster locks from
@@ -1413,7 +1400,7 @@ skip_recovery:
spin_lock(&osb->osb_lock);
}
spin_unlock(&osb->osb_lock);
- mlog(0, "All nodes recovered\n");
+ trace_ocfs2_recovery_thread_end(status);
/* Refresh all journal recovery generations from disk */
status = ocfs2_check_journals_nolocks(osb);
@@ -1422,7 +1409,7 @@ skip_recovery:
mlog_errno(status);
/* Now it is right time to recover quotas... We have to do this under
- * superblock lock so that noone can start using the slot (and crash)
+ * superblock lock so that no one can start using the slot (and crash)
* before we recover it */
for (i = 0; i < rm_quota_used; i++) {
qrec = ocfs2_begin_quota_recovery(osb, rm_quota[i]);
@@ -1454,10 +1441,8 @@ bail:
mutex_unlock(&osb->recovery_lock);
- if (rm_quota)
- kfree(rm_quota);
+ kfree(rm_quota);
- mlog_exit(status);
/* no one is callint kthread_stop() for us so the kthread() api
* requires that we call do_exit(). And it isn't exported, but
* complete_and_exit() seems to be a minimal wrapper around it. */
@@ -1467,19 +1452,15 @@ bail:
void ocfs2_recovery_thread(struct ocfs2_super *osb, int node_num)
{
- mlog_entry("(node_num=%d, osb->node_num = %d)\n",
- node_num, osb->node_num);
-
mutex_lock(&osb->recovery_lock);
- if (osb->disable_recovery)
- goto out;
- /* People waiting on recovery will wait on
- * the recovery map to empty. */
- if (ocfs2_recovery_map_set(osb, node_num))
- mlog(0, "node %d already in recovery map.\n", node_num);
+ trace_ocfs2_recovery_thread(node_num, osb->node_num,
+ osb->disable_recovery, osb->recovery_thread_task,
+ osb->disable_recovery ?
+ -1 : ocfs2_recovery_map_set(osb, node_num));
- mlog(0, "starting recovery thread...\n");
+ if (osb->disable_recovery)
+ goto out;
if (osb->recovery_thread_task)
goto out;
@@ -1494,8 +1475,6 @@ void ocfs2_recovery_thread(struct ocfs2_super *osb, int node_num)
out:
mutex_unlock(&osb->recovery_lock);
wake_up(&osb->recovery_event);
-
- mlog_exit_void();
}
static int ocfs2_read_journal_inode(struct ocfs2_super *osb,
@@ -1569,7 +1548,7 @@ static int ocfs2_replay_journal(struct ocfs2_super *osb,
* If not, it needs recovery.
*/
if (osb->slot_recovery_generations[slot_num] != slot_reco_gen) {
- mlog(0, "Slot %u already recovered (old/new=%u/%u)\n", slot_num,
+ trace_ocfs2_replay_journal_recovered(slot_num,
osb->slot_recovery_generations[slot_num], slot_reco_gen);
osb->slot_recovery_generations[slot_num] = slot_reco_gen;
status = -EBUSY;
@@ -1580,7 +1559,7 @@ static int ocfs2_replay_journal(struct ocfs2_super *osb,
status = ocfs2_inode_lock_full(inode, &bh, 1, OCFS2_META_LOCK_RECOVERY);
if (status < 0) {
- mlog(0, "status returned from ocfs2_inode_lock=%d\n", status);
+ trace_ocfs2_replay_journal_lock_err(status);
if (status != -ERESTARTSYS)
mlog(ML_ERROR, "Could not lock journal!\n");
goto done;
@@ -1593,7 +1572,7 @@ static int ocfs2_replay_journal(struct ocfs2_super *osb,
slot_reco_gen = ocfs2_get_recovery_generation(fe);
if (!(flags & OCFS2_JOURNAL_DIRTY_FL)) {
- mlog(0, "No recovery required for node %d\n", node_num);
+ trace_ocfs2_replay_journal_skip(node_num);
/* Refresh recovery generation for the slot */
osb->slot_recovery_generations[slot_num] = slot_reco_gen;
goto done;
@@ -1602,9 +1581,9 @@ 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);
@@ -1614,7 +1593,6 @@ static int ocfs2_replay_journal(struct ocfs2_super *osb,
goto done;
}
- mlog(0, "calling journal_init_inode\n");
journal = jbd2_journal_init_inode(inode);
if (journal == NULL) {
mlog(ML_ERROR, "Linux journal layer error\n");
@@ -1634,7 +1612,6 @@ static int ocfs2_replay_journal(struct ocfs2_super *osb,
ocfs2_clear_journal_error(osb->sb, journal, slot_num);
/* wipe the journal */
- mlog(0, "flushing the journal.\n");
jbd2_journal_lock_updates(journal);
status = jbd2_journal_flush(journal);
jbd2_journal_unlock_updates(journal);
@@ -1661,6 +1638,9 @@ 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)
@@ -1671,7 +1651,6 @@ done:
brelse(bh);
- mlog_exit(status);
return status;
}
@@ -1694,8 +1673,7 @@ static int ocfs2_recover_node(struct ocfs2_super *osb,
struct ocfs2_dinode *la_copy = NULL;
struct ocfs2_dinode *tl_copy = NULL;
- mlog_entry("(node_num=%d, slot_num=%d, osb->node_num = %d)\n",
- node_num, slot_num, osb->node_num);
+ trace_ocfs2_recover_node(node_num, slot_num, osb->node_num);
/* Should not ever be called to recover ourselves -- in that
* case we should've called ocfs2_journal_load instead. */
@@ -1704,9 +1682,7 @@ static int ocfs2_recover_node(struct ocfs2_super *osb,
status = ocfs2_replay_journal(osb, node_num, slot_num);
if (status < 0) {
if (status == -EBUSY) {
- mlog(0, "Skipping recovery for slot %u (node %u) "
- "as another node has recovered it\n", slot_num,
- node_num);
+ trace_ocfs2_recover_node_skip(slot_num, node_num);
status = 0;
goto done;
}
@@ -1741,7 +1717,6 @@ static int ocfs2_recover_node(struct ocfs2_super *osb,
status = 0;
done:
- mlog_exit(status);
return status;
}
@@ -1814,8 +1789,8 @@ int ocfs2_mark_dead_nodes(struct ocfs2_super *osb)
spin_lock(&osb->osb_lock);
osb->slot_recovery_generations[i] = gen;
- mlog(0, "Slot %u recovery generation is %u\n", i,
- osb->slot_recovery_generations[i]);
+ trace_ocfs2_mark_dead_nodes(i,
+ osb->slot_recovery_generations[i]);
if (i == osb->slot_num) {
spin_unlock(&osb->osb_lock);
@@ -1851,7 +1826,6 @@ int ocfs2_mark_dead_nodes(struct ocfs2_super *osb)
status = 0;
bail:
- mlog_exit(status);
return status;
}
@@ -1874,6 +1848,20 @@ 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
@@ -1893,6 +1881,9 @@ void ocfs2_queue_orphan_scan(struct ocfs2_super *osb)
if (atomic_read(&os->os_state) == ORPHAN_SCAN_INACTIVE)
goto out;
+ trace_ocfs2_queue_orphan_scan_begin(os->os_count, os->os_seqno,
+ atomic_read(&os->os_state));
+
status = ocfs2_orphan_scan_lock(osb, &seqno);
if (status < 0) {
if (status != -EAGAIN)
@@ -1922,6 +1913,8 @@ void ocfs2_queue_orphan_scan(struct ocfs2_super *osb)
unlock:
ocfs2_orphan_scan_unlock(osb, seqno);
out:
+ trace_ocfs2_queue_orphan_scan_end(os->os_count, os->os_seqno,
+ atomic_read(&os->os_state));
return;
}
@@ -1938,7 +1931,7 @@ void ocfs2_orphan_scan_work(struct work_struct *work)
mutex_lock(&os->os_lock);
ocfs2_queue_orphan_scan(osb);
if (atomic_read(&os->os_state) == ORPHAN_SCAN_ACTIVE)
- schedule_delayed_work(&os->os_orphan_scan_work,
+ queue_delayed_work(ocfs2_wq, &os->os_orphan_scan_work,
ocfs2_orphan_scan_timeout());
mutex_unlock(&os->os_lock);
}
@@ -1978,12 +1971,13 @@ void ocfs2_orphan_scan_start(struct ocfs2_super *osb)
atomic_set(&os->os_state, ORPHAN_SCAN_INACTIVE);
else {
atomic_set(&os->os_state, ORPHAN_SCAN_ACTIVE);
- schedule_delayed_work(&os->os_orphan_scan_work,
- ocfs2_orphan_scan_timeout());
+ queue_delayed_work(ocfs2_wq, &os->os_orphan_scan_work,
+ ocfs2_orphan_scan_timeout());
}
}
struct ocfs2_orphan_filldir_priv {
+ struct dir_context ctx;
struct inode *head;
struct ocfs2_super *osb;
};
@@ -2005,8 +1999,7 @@ static int ocfs2_orphan_filldir(void *priv, const char *name, int name_len,
if (IS_ERR(iter))
return 0;
- mlog(0, "queue orphan %llu\n",
- (unsigned long long)OCFS2_I(iter)->ip_blkno);
+ trace_ocfs2_orphan_filldir((unsigned long long)OCFS2_I(iter)->ip_blkno);
/* No locking is required for the next_orphan queue as there
* is only ever a single process doing orphan recovery. */
OCFS2_I(iter)->ip_next_orphan = p->head;
@@ -2021,11 +2014,11 @@ static int ocfs2_queue_orphans(struct ocfs2_super *osb,
{
int status;
struct inode *orphan_dir_inode = NULL;
- struct ocfs2_orphan_filldir_priv priv;
- loff_t pos = 0;
-
- priv.osb = osb;
- priv.head = *head;
+ struct ocfs2_orphan_filldir_priv priv = {
+ .ctx.actor = ocfs2_orphan_filldir,
+ .osb = osb,
+ .head = *head
+ };
orphan_dir_inode = ocfs2_get_system_file_inode(osb,
ORPHAN_DIR_SYSTEM_INODE,
@@ -2034,7 +2027,7 @@ static int ocfs2_queue_orphans(struct ocfs2_super *osb,
status = -ENOENT;
mlog_errno(status);
return status;
- }
+ }
mutex_lock(&orphan_dir_inode->i_mutex);
status = ocfs2_inode_lock(orphan_dir_inode, NULL, 0);
@@ -2043,8 +2036,7 @@ static int ocfs2_queue_orphans(struct ocfs2_super *osb,
goto out;
}
- status = ocfs2_dir_foreach(orphan_dir_inode, &pos, &priv,
- ocfs2_orphan_filldir);
+ status = ocfs2_dir_foreach(orphan_dir_inode, &priv.ctx);
if (status) {
mlog_errno(status);
goto out_cluster;
@@ -2122,7 +2114,7 @@ static int ocfs2_recover_orphans(struct ocfs2_super *osb,
struct inode *iter;
struct ocfs2_inode_info *oi;
- mlog(0, "Recover inodes from orphan dir in slot %d\n", slot);
+ trace_ocfs2_recover_orphans(slot);
ocfs2_mark_recovering_orphan_dir(osb, slot);
ret = ocfs2_queue_orphans(osb, slot, &inode);
@@ -2135,17 +2127,12 @@ static int ocfs2_recover_orphans(struct ocfs2_super *osb,
while (inode) {
oi = OCFS2_I(inode);
- mlog(0, "iput orphan %llu\n", (unsigned long long)oi->ip_blkno);
+ trace_ocfs2_recover_orphans_iput(
+ (unsigned long long)oi->ip_blkno);
iter = oi->ip_next_orphan;
spin_lock(&oi->ip_lock);
- /* The remote delete code may have set these on the
- * assumption that the other node would wipe them
- * successfully. If they are still in the node's
- * orphan dir, we need to reset that state. */
- oi->ip_flags &= ~(OCFS2_INODE_DELETED|OCFS2_INODE_SKIP_DELETE);
-
/* Set the proper information to get us going into
* ocfs2_delete_inode. */
oi->ip_flags |= OCFS2_INODE_MAYBE_ORPHANED;
@@ -2173,6 +2160,7 @@ static int __ocfs2_wait_on_mount(struct ocfs2_super *osb, int quota)
* MOUNTED flag, but this is set right before
* dismount_volume() so we can trust it. */
if (atomic_read(&osb->vol_state) == VOLUME_DISABLED) {
+ trace_ocfs2_wait_on_mount(VOLUME_DISABLED);
mlog(0, "mount error, exiting!\n");
return -EBUSY;
}
@@ -2198,8 +2186,20 @@ static int ocfs2_commit_thread(void *arg)
|| kthread_should_stop());
status = ocfs2_commit_cache(osb);
- if (status < 0)
- mlog_errno(status);
+ if (status < 0) {
+ static unsigned long abort_warn_time;
+
+ /* Warn about this once per minute */
+ if (printk_timed_ratelimit(&abort_warn_time, 60*HZ))
+ mlog(ML_ERROR, "status = %d, journal is "
+ "already aborted.\n", status);
+ /*
+ * After ocfs2_commit_cache() fails, j_num_trans has a
+ * non-zero value. Sleep here to avoid a busy-wait
+ * loop.
+ */
+ msleep_interruptible(1000);
+ }
if (kthread_should_stop() && atomic_read(&journal->j_num_trans)){
mlog(ML_KTHREAD,
generated by cgit v1.2.3-70-g09d2 (git 2.46.0) at 2025-11-10 14:17:39 +0000