aboutsummaryrefslogtreecommitdiffstats
path: root/fs
diff options
context:
space:
mode:
authorTheodore Ts'o <tytso@mit.edu>2008-11-06 22:37:59 -0500
committerTheodore Ts'o <tytso@mit.edu>2008-11-06 22:37:59 -0500
commite219cca082f52e7dfea41f3be264b7b5eb204227 (patch)
tree6d67c967064eed255b602626348e78119f84fb8b /fs
parent45beca08dd8b6d6a65c5ffd730af2eac7a2c7a03 (diff)
jbd: don't give up looking for space so easily in __log_wait_for_space
Commit be07c4ed introducd a regression because it assumed that if there were no transactions ready to be checkpointed, that no progress could be made on making space available in the journal, and so the journal should be aborted. This assumption is false; it could be the case that simply calling cleanup_journal_tail() will recover the necessary space, or, for small journals, the currently committing transaction could be responsible for chewing up the required space in the log, so we need to wait for the currently committing transaction to finish before trying to force a checkpoint operation. This patch fixes the bug reported by Meelis Roos at: http://bugzilla.kernel.org/show_bug.cgi?id=11937 Signed-off-by: "Theodore Ts'o" <tytso@mit.edu> Cc: Duane Griffin <duaneg@dghda.com> Cc: Toshiyuki Okajima <toshi.okajima@jp.fujitsu.com>
Diffstat (limited to 'fs')
-rw-r--r--fs/jbd/checkpoint.c31
1 files changed, 24 insertions, 7 deletions
diff --git a/fs/jbd/checkpoint.c b/fs/jbd/checkpoint.c
index 1bd8d4acc6f2..61f32f3868cd 100644
--- a/fs/jbd/checkpoint.c
+++ b/fs/jbd/checkpoint.c
@@ -115,7 +115,7 @@ static int __try_to_free_cp_buf(struct journal_head *jh)
115 */ 115 */
116void __log_wait_for_space(journal_t *journal) 116void __log_wait_for_space(journal_t *journal)
117{ 117{
118 int nblocks; 118 int nblocks, space_left;
119 assert_spin_locked(&journal->j_state_lock); 119 assert_spin_locked(&journal->j_state_lock);
120 120
121 nblocks = jbd_space_needed(journal); 121 nblocks = jbd_space_needed(journal);
@@ -128,25 +128,42 @@ void __log_wait_for_space(journal_t *journal)
128 /* 128 /*
129 * Test again, another process may have checkpointed while we 129 * Test again, another process may have checkpointed while we
130 * were waiting for the checkpoint lock. If there are no 130 * were waiting for the checkpoint lock. If there are no
131 * outstanding transactions there is nothing to checkpoint and 131 * transactions ready to be checkpointed, try to recover
132 * we can't make progress. Abort the journal in this case. 132 * journal space by calling cleanup_journal_tail(), and if
133 * that doesn't work, by waiting for the currently committing
134 * transaction to complete. If there is absolutely no way
135 * to make progress, this is either a BUG or corrupted
136 * filesystem, so abort the journal and leave a stack
137 * trace for forensic evidence.
133 */ 138 */
134 spin_lock(&journal->j_state_lock); 139 spin_lock(&journal->j_state_lock);
135 spin_lock(&journal->j_list_lock); 140 spin_lock(&journal->j_list_lock);
136 nblocks = jbd_space_needed(journal); 141 nblocks = jbd_space_needed(journal);
137 if (__log_space_left(journal) < nblocks) { 142 space_left = __log_space_left(journal);
143 if (space_left < nblocks) {
138 int chkpt = journal->j_checkpoint_transactions != NULL; 144 int chkpt = journal->j_checkpoint_transactions != NULL;
145 tid_t tid = 0;
139 146
147 if (journal->j_committing_transaction)
148 tid = journal->j_committing_transaction->t_tid;
140 spin_unlock(&journal->j_list_lock); 149 spin_unlock(&journal->j_list_lock);
141 spin_unlock(&journal->j_state_lock); 150 spin_unlock(&journal->j_state_lock);
142 if (chkpt) { 151 if (chkpt) {
143 log_do_checkpoint(journal); 152 log_do_checkpoint(journal);
153 } else if (cleanup_journal_tail(journal) == 0) {
154 /* We were able to recover space; yay! */
155 ;
156 } else if (tid) {
157 log_wait_commit(journal, tid);
144 } else { 158 } else {
145 printk(KERN_ERR "%s: no transactions\n", 159 printk(KERN_ERR "%s: needed %d blocks and "
146 __func__); 160 "only had %d space available\n",
161 __func__, nblocks, space_left);
162 printk(KERN_ERR "%s: no way to get more "
163 "journal space\n", __func__);
164 WARN_ON(1);
147 journal_abort(journal, 0); 165 journal_abort(journal, 0);
148 } 166 }
149
150 spin_lock(&journal->j_state_lock); 167 spin_lock(&journal->j_state_lock);
151 } else { 168 } else {
152 spin_unlock(&journal->j_list_lock); 169 spin_unlock(&journal->j_list_lock);