aboutsummaryrefslogtreecommitdiffstats
path: root/fs/jbd2
diff options
context:
space:
mode:
authorJan Kara <jack@suse.cz>2011-06-13 15:38:22 -0400
committerTheodore Ts'o <tytso@mit.edu>2011-06-13 15:38:22 -0400
commitde1b794130b130e77ffa975bb58cb843744f9ae5 (patch)
tree8c4b37582128dc36c2b7385294fba58d017ce3e8 /fs/jbd2
parent1fb74cda1b5e9c6207225fda5ef7504e815ce0e0 (diff)
jbd2: Fix oops in jbd2_journal_remove_journal_head()
jbd2_journal_remove_journal_head() can oops when trying to access journal_head returned by bh2jh(). This is caused for example by the following race: TASK1 TASK2 jbd2_journal_commit_transaction() ... processing t_forget list __jbd2_journal_refile_buffer(jh); if (!jh->b_transaction) { jbd_unlock_bh_state(bh); jbd2_journal_try_to_free_buffers() jbd2_journal_grab_journal_head(bh) jbd_lock_bh_state(bh) __journal_try_to_free_buffer() jbd2_journal_put_journal_head(jh) jbd2_journal_remove_journal_head(bh); jbd2_journal_put_journal_head() in TASK2 sees that b_jcount == 0 and buffer is not part of any transaction and thus frees journal_head before TASK1 gets to doing so. Note that even buffer_head can be released by try_to_free_buffers() after jbd2_journal_put_journal_head() which adds even larger opportunity for oops (but I didn't see this happen in reality). Fix the problem by making transactions hold their own journal_head reference (in b_jcount). That way we don't have to remove journal_head explicitely via jbd2_journal_remove_journal_head() and instead just remove journal_head when b_jcount drops to zero. The result of this is that [__]jbd2_journal_refile_buffer(), [__]jbd2_journal_unfile_buffer(), and __jdb2_journal_remove_checkpoint() can free journal_head which needs modification of a few callers. Also we have to be careful because once journal_head is removed, buffer_head might be freed as well. So we have to get our own buffer_head reference where it matters. Signed-off-by: Jan Kara <jack@suse.cz> Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
Diffstat (limited to 'fs/jbd2')
-rw-r--r--fs/jbd2/checkpoint.c28
-rw-r--r--fs/jbd2/commit.c33
-rw-r--r--fs/jbd2/journal.c91
-rw-r--r--fs/jbd2/transaction.c67
4 files changed, 99 insertions, 120 deletions
diff --git a/fs/jbd2/checkpoint.c b/fs/jbd2/checkpoint.c
index 6a79fd0a1a32..2c62c5aae82f 100644
--- a/fs/jbd2/checkpoint.c
+++ b/fs/jbd2/checkpoint.c
@@ -97,10 +97,14 @@ static int __try_to_free_cp_buf(struct journal_head *jh)
97 97
98 if (jh->b_jlist == BJ_None && !buffer_locked(bh) && 98 if (jh->b_jlist == BJ_None && !buffer_locked(bh) &&
99 !buffer_dirty(bh) && !buffer_write_io_error(bh)) { 99 !buffer_dirty(bh) && !buffer_write_io_error(bh)) {
100 /*
101 * Get our reference so that bh cannot be freed before
102 * we unlock it
103 */
104 get_bh(bh);
100 JBUFFER_TRACE(jh, "remove from checkpoint list"); 105 JBUFFER_TRACE(jh, "remove from checkpoint list");
101 ret = __jbd2_journal_remove_checkpoint(jh) + 1; 106 ret = __jbd2_journal_remove_checkpoint(jh) + 1;
102 jbd_unlock_bh_state(bh); 107 jbd_unlock_bh_state(bh);
103 jbd2_journal_remove_journal_head(bh);
104 BUFFER_TRACE(bh, "release"); 108 BUFFER_TRACE(bh, "release");
105 __brelse(bh); 109 __brelse(bh);
106 } else { 110 } else {
@@ -223,8 +227,8 @@ restart:
223 spin_lock(&journal->j_list_lock); 227 spin_lock(&journal->j_list_lock);
224 goto restart; 228 goto restart;
225 } 229 }
230 get_bh(bh);
226 if (buffer_locked(bh)) { 231 if (buffer_locked(bh)) {
227 atomic_inc(&bh->b_count);
228 spin_unlock(&journal->j_list_lock); 232 spin_unlock(&journal->j_list_lock);
229 jbd_unlock_bh_state(bh); 233 jbd_unlock_bh_state(bh);
230 wait_on_buffer(bh); 234 wait_on_buffer(bh);
@@ -243,7 +247,6 @@ restart:
243 */ 247 */
244 released = __jbd2_journal_remove_checkpoint(jh); 248 released = __jbd2_journal_remove_checkpoint(jh);
245 jbd_unlock_bh_state(bh); 249 jbd_unlock_bh_state(bh);
246 jbd2_journal_remove_journal_head(bh);
247 __brelse(bh); 250 __brelse(bh);
248 } 251 }
249 252
@@ -284,7 +287,7 @@ static int __process_buffer(journal_t *journal, struct journal_head *jh,
284 int ret = 0; 287 int ret = 0;
285 288
286 if (buffer_locked(bh)) { 289 if (buffer_locked(bh)) {
287 atomic_inc(&bh->b_count); 290 get_bh(bh);
288 spin_unlock(&journal->j_list_lock); 291 spin_unlock(&journal->j_list_lock);
289 jbd_unlock_bh_state(bh); 292 jbd_unlock_bh_state(bh);
290 wait_on_buffer(bh); 293 wait_on_buffer(bh);
@@ -316,12 +319,12 @@ static int __process_buffer(journal_t *journal, struct journal_head *jh,
316 ret = 1; 319 ret = 1;
317 if (unlikely(buffer_write_io_error(bh))) 320 if (unlikely(buffer_write_io_error(bh)))
318 ret = -EIO; 321 ret = -EIO;
322 get_bh(bh);
319 J_ASSERT_JH(jh, !buffer_jbddirty(bh)); 323 J_ASSERT_JH(jh, !buffer_jbddirty(bh));
320 BUFFER_TRACE(bh, "remove from checkpoint"); 324 BUFFER_TRACE(bh, "remove from checkpoint");
321 __jbd2_journal_remove_checkpoint(jh); 325 __jbd2_journal_remove_checkpoint(jh);
322 spin_unlock(&journal->j_list_lock); 326 spin_unlock(&journal->j_list_lock);
323 jbd_unlock_bh_state(bh); 327 jbd_unlock_bh_state(bh);
324 jbd2_journal_remove_journal_head(bh);
325 __brelse(bh); 328 __brelse(bh);
326 } else { 329 } else {
327 /* 330 /*
@@ -554,7 +557,8 @@ int jbd2_cleanup_journal_tail(journal_t *journal)
554/* 557/*
555 * journal_clean_one_cp_list 558 * journal_clean_one_cp_list
556 * 559 *
557 * Find all the written-back checkpoint buffers in the given list and release them. 560 * Find all the written-back checkpoint buffers in the given list and
561 * release them.
558 * 562 *
559 * Called with the journal locked. 563 * Called with the journal locked.
560 * Called with j_list_lock held. 564 * Called with j_list_lock held.
@@ -663,8 +667,8 @@ out:
663 * checkpoint lists. 667 * checkpoint lists.
664 * 668 *
665 * The function returns 1 if it frees the transaction, 0 otherwise. 669 * The function returns 1 if it frees the transaction, 0 otherwise.
670 * The function can free jh and bh.
666 * 671 *
667 * This function is called with the journal locked.
668 * This function is called with j_list_lock held. 672 * This function is called with j_list_lock held.
669 * This function is called with jbd_lock_bh_state(jh2bh(jh)) 673 * This function is called with jbd_lock_bh_state(jh2bh(jh))
670 */ 674 */
@@ -684,13 +688,14 @@ int __jbd2_journal_remove_checkpoint(struct journal_head *jh)
684 } 688 }
685 journal = transaction->t_journal; 689 journal = transaction->t_journal;
686 690
691 JBUFFER_TRACE(jh, "removing from transaction");
687 __buffer_unlink(jh); 692 __buffer_unlink(jh);
688 jh->b_cp_transaction = NULL; 693 jh->b_cp_transaction = NULL;
694 jbd2_journal_put_journal_head(jh);
689 695
690 if (transaction->t_checkpoint_list != NULL || 696 if (transaction->t_checkpoint_list != NULL ||
691 transaction->t_checkpoint_io_list != NULL) 697 transaction->t_checkpoint_io_list != NULL)
692 goto out; 698 goto out;
693 JBUFFER_TRACE(jh, "transaction has no more buffers");
694 699
695 /* 700 /*
696 * There is one special case to worry about: if we have just pulled the 701 * There is one special case to worry about: if we have just pulled the
@@ -701,10 +706,8 @@ int __jbd2_journal_remove_checkpoint(struct journal_head *jh)
701 * The locking here around t_state is a bit sleazy. 706 * The locking here around t_state is a bit sleazy.
702 * See the comment at the end of jbd2_journal_commit_transaction(). 707 * See the comment at the end of jbd2_journal_commit_transaction().
703 */ 708 */
704 if (transaction->t_state != T_FINISHED) { 709 if (transaction->t_state != T_FINISHED)
705 JBUFFER_TRACE(jh, "belongs to running/committing transaction");
706 goto out; 710 goto out;
707 }
708 711
709 /* OK, that was the last buffer for the transaction: we can now 712 /* OK, that was the last buffer for the transaction: we can now
710 safely remove this transaction from the log */ 713 safely remove this transaction from the log */
@@ -723,7 +726,6 @@ int __jbd2_journal_remove_checkpoint(struct journal_head *jh)
723 wake_up(&journal->j_wait_logspace); 726 wake_up(&journal->j_wait_logspace);
724 ret = 1; 727 ret = 1;
725out: 728out:
726 JBUFFER_TRACE(jh, "exit");
727 return ret; 729 return ret;
728} 730}
729 731
@@ -742,6 +744,8 @@ void __jbd2_journal_insert_checkpoint(struct journal_head *jh,
742 J_ASSERT_JH(jh, buffer_dirty(jh2bh(jh)) || buffer_jbddirty(jh2bh(jh))); 744 J_ASSERT_JH(jh, buffer_dirty(jh2bh(jh)) || buffer_jbddirty(jh2bh(jh)));
743 J_ASSERT_JH(jh, jh->b_cp_transaction == NULL); 745 J_ASSERT_JH(jh, jh->b_cp_transaction == NULL);
744 746
747 /* Get reference for checkpointing transaction */
748 jbd2_journal_grab_journal_head(jh2bh(jh));
745 jh->b_cp_transaction = transaction; 749 jh->b_cp_transaction = transaction;
746 750
747 if (!transaction->t_checkpoint_list) { 751 if (!transaction->t_checkpoint_list) {
diff --git a/fs/jbd2/commit.c b/fs/jbd2/commit.c
index 7f21cf3aaf92..eef6979821a4 100644
--- a/fs/jbd2/commit.c
+++ b/fs/jbd2/commit.c
@@ -848,10 +848,16 @@ restart_loop:
848 while (commit_transaction->t_forget) { 848 while (commit_transaction->t_forget) {
849 transaction_t *cp_transaction; 849 transaction_t *cp_transaction;
850 struct buffer_head *bh; 850 struct buffer_head *bh;
851 int try_to_free = 0;
851 852
852 jh = commit_transaction->t_forget; 853 jh = commit_transaction->t_forget;
853 spin_unlock(&journal->j_list_lock); 854 spin_unlock(&journal->j_list_lock);
854 bh = jh2bh(jh); 855 bh = jh2bh(jh);
856 /*
857 * Get a reference so that bh cannot be freed before we are
858 * done with it.
859 */
860 get_bh(bh);
855 jbd_lock_bh_state(bh); 861 jbd_lock_bh_state(bh);
856 J_ASSERT_JH(jh, jh->b_transaction == commit_transaction); 862 J_ASSERT_JH(jh, jh->b_transaction == commit_transaction);
857 863
@@ -914,28 +920,27 @@ restart_loop:
914 __jbd2_journal_insert_checkpoint(jh, commit_transaction); 920 __jbd2_journal_insert_checkpoint(jh, commit_transaction);
915 if (is_journal_aborted(journal)) 921 if (is_journal_aborted(journal))
916 clear_buffer_jbddirty(bh); 922 clear_buffer_jbddirty(bh);
917 JBUFFER_TRACE(jh, "refile for checkpoint writeback");
918 __jbd2_journal_refile_buffer(jh);
919 jbd_unlock_bh_state(bh);
920 } else { 923 } else {
921 J_ASSERT_BH(bh, !buffer_dirty(bh)); 924 J_ASSERT_BH(bh, !buffer_dirty(bh));
922 /* The buffer on BJ_Forget list and not jbddirty means 925 /*
926 * The buffer on BJ_Forget list and not jbddirty means
923 * it has been freed by this transaction and hence it 927 * it has been freed by this transaction and hence it
924 * could not have been reallocated until this 928 * could not have been reallocated until this
925 * transaction has committed. *BUT* it could be 929 * transaction has committed. *BUT* it could be
926 * reallocated once we have written all the data to 930 * reallocated once we have written all the data to
927 * disk and before we process the buffer on BJ_Forget 931 * disk and before we process the buffer on BJ_Forget
928 * list. */ 932 * list.
929 JBUFFER_TRACE(jh, "refile or unfile freed buffer"); 933 */
930 __jbd2_journal_refile_buffer(jh); 934 if (!jh->b_next_transaction)
931 if (!jh->b_transaction) { 935 try_to_free = 1;
932 jbd_unlock_bh_state(bh);
933 /* needs a brelse */
934 jbd2_journal_remove_journal_head(bh);
935 release_buffer_page(bh);
936 } else
937 jbd_unlock_bh_state(bh);
938 } 936 }
937 JBUFFER_TRACE(jh, "refile or unfile buffer");
938 __jbd2_journal_refile_buffer(jh);
939 jbd_unlock_bh_state(bh);
940 if (try_to_free)
941 release_buffer_page(bh); /* Drops bh reference */
942 else
943 __brelse(bh);
939 cond_resched_lock(&journal->j_list_lock); 944 cond_resched_lock(&journal->j_list_lock);
940 } 945 }
941 spin_unlock(&journal->j_list_lock); 946 spin_unlock(&journal->j_list_lock);
diff --git a/fs/jbd2/journal.c b/fs/jbd2/journal.c
index 9a7826990304..0dfa5b598e68 100644
--- a/fs/jbd2/journal.c
+++ b/fs/jbd2/journal.c
@@ -2078,10 +2078,9 @@ static void journal_free_journal_head(struct journal_head *jh)
2078 * When a buffer has its BH_JBD bit set it is immune from being released by 2078 * When a buffer has its BH_JBD bit set it is immune from being released by
2079 * core kernel code, mainly via ->b_count. 2079 * core kernel code, mainly via ->b_count.
2080 * 2080 *
2081 * A journal_head may be detached from its buffer_head when the journal_head's 2081 * A journal_head is detached from its buffer_head when the journal_head's
2082 * b_transaction, b_cp_transaction and b_next_transaction pointers are NULL. 2082 * b_jcount reaches zero. Running transaction (b_transaction) and checkpoint
2083 * Various places in JBD call jbd2_journal_remove_journal_head() to indicate that the 2083 * transaction (b_cp_transaction) hold their references to b_jcount.
2084 * journal_head can be dropped if needed.
2085 * 2084 *
2086 * Various places in the kernel want to attach a journal_head to a buffer_head 2085 * Various places in the kernel want to attach a journal_head to a buffer_head
2087 * _before_ attaching the journal_head to a transaction. To protect the 2086 * _before_ attaching the journal_head to a transaction. To protect the
@@ -2094,17 +2093,16 @@ static void journal_free_journal_head(struct journal_head *jh)
2094 * (Attach a journal_head if needed. Increments b_jcount) 2093 * (Attach a journal_head if needed. Increments b_jcount)
2095 * struct journal_head *jh = jbd2_journal_add_journal_head(bh); 2094 * struct journal_head *jh = jbd2_journal_add_journal_head(bh);
2096 * ... 2095 * ...
2096 * (Get another reference for transaction)
2097 * jbd2_journal_grab_journal_head(bh);
2097 * jh->b_transaction = xxx; 2098 * jh->b_transaction = xxx;
2099 * (Put original reference)
2098 * jbd2_journal_put_journal_head(jh); 2100 * jbd2_journal_put_journal_head(jh);
2099 *
2100 * Now, the journal_head's b_jcount is zero, but it is safe from being released
2101 * because it has a non-zero b_transaction.
2102 */ 2101 */
2103 2102
2104/* 2103/*
2105 * Give a buffer_head a journal_head. 2104 * Give a buffer_head a journal_head.
2106 * 2105 *
2107 * Doesn't need the journal lock.
2108 * May sleep. 2106 * May sleep.
2109 */ 2107 */
2110struct journal_head *jbd2_journal_add_journal_head(struct buffer_head *bh) 2108struct journal_head *jbd2_journal_add_journal_head(struct buffer_head *bh)
@@ -2168,61 +2166,29 @@ static void __journal_remove_journal_head(struct buffer_head *bh)
2168 struct journal_head *jh = bh2jh(bh); 2166 struct journal_head *jh = bh2jh(bh);
2169 2167
2170 J_ASSERT_JH(jh, jh->b_jcount >= 0); 2168 J_ASSERT_JH(jh, jh->b_jcount >= 0);
2171 2169 J_ASSERT_JH(jh, jh->b_transaction == NULL);
2172 get_bh(bh); 2170 J_ASSERT_JH(jh, jh->b_next_transaction == NULL);
2173 if (jh->b_jcount == 0) { 2171 J_ASSERT_JH(jh, jh->b_cp_transaction == NULL);
2174 if (jh->b_transaction == NULL && 2172 J_ASSERT_JH(jh, jh->b_jlist == BJ_None);
2175 jh->b_next_transaction == NULL && 2173 J_ASSERT_BH(bh, buffer_jbd(bh));
2176 jh->b_cp_transaction == NULL) { 2174 J_ASSERT_BH(bh, jh2bh(jh) == bh);
2177 J_ASSERT_JH(jh, jh->b_jlist == BJ_None); 2175 BUFFER_TRACE(bh, "remove journal_head");
2178 J_ASSERT_BH(bh, buffer_jbd(bh)); 2176 if (jh->b_frozen_data) {
2179 J_ASSERT_BH(bh, jh2bh(jh) == bh); 2177 printk(KERN_WARNING "%s: freeing b_frozen_data\n", __func__);
2180 BUFFER_TRACE(bh, "remove journal_head"); 2178 jbd2_free(jh->b_frozen_data, bh->b_size);
2181 if (jh->b_frozen_data) {
2182 printk(KERN_WARNING "%s: freeing "
2183 "b_frozen_data\n",
2184 __func__);
2185 jbd2_free(jh->b_frozen_data, bh->b_size);
2186 }
2187 if (jh->b_committed_data) {
2188 printk(KERN_WARNING "%s: freeing "
2189 "b_committed_data\n",
2190 __func__);
2191 jbd2_free(jh->b_committed_data, bh->b_size);
2192 }
2193 bh->b_private = NULL;
2194 jh->b_bh = NULL; /* debug, really */
2195 clear_buffer_jbd(bh);
2196 __brelse(bh);
2197 journal_free_journal_head(jh);
2198 } else {
2199 BUFFER_TRACE(bh, "journal_head was locked");
2200 }
2201 } 2179 }
2180 if (jh->b_committed_data) {
2181 printk(KERN_WARNING "%s: freeing b_committed_data\n", __func__);
2182 jbd2_free(jh->b_committed_data, bh->b_size);
2183 }
2184 bh->b_private = NULL;
2185 jh->b_bh = NULL; /* debug, really */
2186 clear_buffer_jbd(bh);
2187 journal_free_journal_head(jh);
2202} 2188}
2203 2189
2204/* 2190/*
2205 * jbd2_journal_remove_journal_head(): if the buffer isn't attached to a transaction 2191 * Drop a reference on the passed journal_head. If it fell to zero then
2206 * and has a zero b_jcount then remove and release its journal_head. If we did
2207 * see that the buffer is not used by any transaction we also "logically"
2208 * decrement ->b_count.
2209 *
2210 * We in fact take an additional increment on ->b_count as a convenience,
2211 * because the caller usually wants to do additional things with the bh
2212 * after calling here.
2213 * The caller of jbd2_journal_remove_journal_head() *must* run __brelse(bh) at some
2214 * time. Once the caller has run __brelse(), the buffer is eligible for
2215 * reaping by try_to_free_buffers().
2216 */
2217void jbd2_journal_remove_journal_head(struct buffer_head *bh)
2218{
2219 jbd_lock_bh_journal_head(bh);
2220 __journal_remove_journal_head(bh);
2221 jbd_unlock_bh_journal_head(bh);
2222}
2223
2224/*
2225 * Drop a reference on the passed journal_head. If it fell to zero then try to
2226 * release the journal_head from the buffer_head. 2192 * release the journal_head from the buffer_head.
2227 */ 2193 */
2228void jbd2_journal_put_journal_head(struct journal_head *jh) 2194void jbd2_journal_put_journal_head(struct journal_head *jh)
@@ -2232,11 +2198,12 @@ void jbd2_journal_put_journal_head(struct journal_head *jh)
2232 jbd_lock_bh_journal_head(bh); 2198 jbd_lock_bh_journal_head(bh);
2233 J_ASSERT_JH(jh, jh->b_jcount > 0); 2199 J_ASSERT_JH(jh, jh->b_jcount > 0);
2234 --jh->b_jcount; 2200 --jh->b_jcount;
2235 if (!jh->b_jcount && !jh->b_transaction) { 2201 if (!jh->b_jcount) {
2236 __journal_remove_journal_head(bh); 2202 __journal_remove_journal_head(bh);
2203 jbd_unlock_bh_journal_head(bh);
2237 __brelse(bh); 2204 __brelse(bh);
2238 } 2205 } else
2239 jbd_unlock_bh_journal_head(bh); 2206 jbd_unlock_bh_journal_head(bh);
2240} 2207}
2241 2208
2242/* 2209/*
diff --git a/fs/jbd2/transaction.c b/fs/jbd2/transaction.c
index 547b101049e5..2d7109414cdd 100644
--- a/fs/jbd2/transaction.c
+++ b/fs/jbd2/transaction.c
@@ -30,6 +30,7 @@
30#include <linux/module.h> 30#include <linux/module.h>
31 31
32static void __jbd2_journal_temp_unlink_buffer(struct journal_head *jh); 32static void __jbd2_journal_temp_unlink_buffer(struct journal_head *jh);
33static void __jbd2_journal_unfile_buffer(struct journal_head *jh);
33 34
34/* 35/*
35 * jbd2_get_transaction: obtain a new transaction_t object. 36 * jbd2_get_transaction: obtain a new transaction_t object.
@@ -764,7 +765,6 @@ repeat:
764 if (!jh->b_transaction) { 765 if (!jh->b_transaction) {
765 JBUFFER_TRACE(jh, "no transaction"); 766 JBUFFER_TRACE(jh, "no transaction");
766 J_ASSERT_JH(jh, !jh->b_next_transaction); 767 J_ASSERT_JH(jh, !jh->b_next_transaction);
767 jh->b_transaction = transaction;
768 JBUFFER_TRACE(jh, "file as BJ_Reserved"); 768 JBUFFER_TRACE(jh, "file as BJ_Reserved");
769 spin_lock(&journal->j_list_lock); 769 spin_lock(&journal->j_list_lock);
770 __jbd2_journal_file_buffer(jh, transaction, BJ_Reserved); 770 __jbd2_journal_file_buffer(jh, transaction, BJ_Reserved);
@@ -895,8 +895,6 @@ int jbd2_journal_get_create_access(handle_t *handle, struct buffer_head *bh)
895 * committed and so it's safe to clear the dirty bit. 895 * committed and so it's safe to clear the dirty bit.
896 */ 896 */
897 clear_buffer_dirty(jh2bh(jh)); 897 clear_buffer_dirty(jh2bh(jh));
898 jh->b_transaction = transaction;
899
900 /* first access by this transaction */ 898 /* first access by this transaction */
901 jh->b_modified = 0; 899 jh->b_modified = 0;
902 900
@@ -1230,8 +1228,6 @@ int jbd2_journal_forget (handle_t *handle, struct buffer_head *bh)
1230 __jbd2_journal_file_buffer(jh, transaction, BJ_Forget); 1228 __jbd2_journal_file_buffer(jh, transaction, BJ_Forget);
1231 } else { 1229 } else {
1232 __jbd2_journal_unfile_buffer(jh); 1230 __jbd2_journal_unfile_buffer(jh);
1233 jbd2_journal_remove_journal_head(bh);
1234 __brelse(bh);
1235 if (!buffer_jbd(bh)) { 1231 if (!buffer_jbd(bh)) {
1236 spin_unlock(&journal->j_list_lock); 1232 spin_unlock(&journal->j_list_lock);
1237 jbd_unlock_bh_state(bh); 1233 jbd_unlock_bh_state(bh);
@@ -1554,19 +1550,32 @@ void __jbd2_journal_temp_unlink_buffer(struct journal_head *jh)
1554 mark_buffer_dirty(bh); /* Expose it to the VM */ 1550 mark_buffer_dirty(bh); /* Expose it to the VM */
1555} 1551}
1556 1552
1557void __jbd2_journal_unfile_buffer(struct journal_head *jh) 1553/*
1554 * Remove buffer from all transactions.
1555 *
1556 * Called with bh_state lock and j_list_lock
1557 *
1558 * jh and bh may be already freed when this function returns.
1559 */
1560static void __jbd2_journal_unfile_buffer(struct journal_head *jh)
1558{ 1561{
1559 __jbd2_journal_temp_unlink_buffer(jh); 1562 __jbd2_journal_temp_unlink_buffer(jh);
1560 jh->b_transaction = NULL; 1563 jh->b_transaction = NULL;
1564 jbd2_journal_put_journal_head(jh);
1561} 1565}
1562 1566
1563void jbd2_journal_unfile_buffer(journal_t *journal, struct journal_head *jh) 1567void jbd2_journal_unfile_buffer(journal_t *journal, struct journal_head *jh)
1564{ 1568{
1565 jbd_lock_bh_state(jh2bh(jh)); 1569 struct buffer_head *bh = jh2bh(jh);
1570
1571 /* Get reference so that buffer cannot be freed before we unlock it */
1572 get_bh(bh);
1573 jbd_lock_bh_state(bh);
1566 spin_lock(&journal->j_list_lock); 1574 spin_lock(&journal->j_list_lock);
1567 __jbd2_journal_unfile_buffer(jh); 1575 __jbd2_journal_unfile_buffer(jh);
1568 spin_unlock(&journal->j_list_lock); 1576 spin_unlock(&journal->j_list_lock);
1569 jbd_unlock_bh_state(jh2bh(jh)); 1577 jbd_unlock_bh_state(bh);
1578 __brelse(bh);
1570} 1579}
1571 1580
1572/* 1581/*
@@ -1593,8 +1602,6 @@ __journal_try_to_free_buffer(journal_t *journal, struct buffer_head *bh)
1593 if (jh->b_jlist == BJ_None) { 1602 if (jh->b_jlist == BJ_None) {
1594 JBUFFER_TRACE(jh, "remove from checkpoint list"); 1603 JBUFFER_TRACE(jh, "remove from checkpoint list");
1595 __jbd2_journal_remove_checkpoint(jh); 1604 __jbd2_journal_remove_checkpoint(jh);
1596 jbd2_journal_remove_journal_head(bh);
1597 __brelse(bh);
1598 } 1605 }
1599 } 1606 }
1600 spin_unlock(&journal->j_list_lock); 1607 spin_unlock(&journal->j_list_lock);
@@ -1657,7 +1664,6 @@ int jbd2_journal_try_to_free_buffers(journal_t *journal,
1657 /* 1664 /*
1658 * We take our own ref against the journal_head here to avoid 1665 * We take our own ref against the journal_head here to avoid
1659 * having to add tons of locking around each instance of 1666 * having to add tons of locking around each instance of
1660 * jbd2_journal_remove_journal_head() and
1661 * jbd2_journal_put_journal_head(). 1667 * jbd2_journal_put_journal_head().
1662 */ 1668 */
1663 jh = jbd2_journal_grab_journal_head(bh); 1669 jh = jbd2_journal_grab_journal_head(bh);
@@ -1695,10 +1701,9 @@ static int __dispose_buffer(struct journal_head *jh, transaction_t *transaction)
1695 int may_free = 1; 1701 int may_free = 1;
1696 struct buffer_head *bh = jh2bh(jh); 1702 struct buffer_head *bh = jh2bh(jh);
1697 1703
1698 __jbd2_journal_unfile_buffer(jh);
1699
1700 if (jh->b_cp_transaction) { 1704 if (jh->b_cp_transaction) {
1701 JBUFFER_TRACE(jh, "on running+cp transaction"); 1705 JBUFFER_TRACE(jh, "on running+cp transaction");
1706 __jbd2_journal_temp_unlink_buffer(jh);
1702 /* 1707 /*
1703 * We don't want to write the buffer anymore, clear the 1708 * We don't want to write the buffer anymore, clear the
1704 * bit so that we don't confuse checks in 1709 * bit so that we don't confuse checks in
@@ -1709,8 +1714,7 @@ static int __dispose_buffer(struct journal_head *jh, transaction_t *transaction)
1709 may_free = 0; 1714 may_free = 0;
1710 } else { 1715 } else {
1711 JBUFFER_TRACE(jh, "on running transaction"); 1716 JBUFFER_TRACE(jh, "on running transaction");
1712 jbd2_journal_remove_journal_head(bh); 1717 __jbd2_journal_unfile_buffer(jh);
1713 __brelse(bh);
1714 } 1718 }
1715 return may_free; 1719 return may_free;
1716} 1720}
@@ -1988,6 +1992,8 @@ void __jbd2_journal_file_buffer(struct journal_head *jh,
1988 1992
1989 if (jh->b_transaction) 1993 if (jh->b_transaction)
1990 __jbd2_journal_temp_unlink_buffer(jh); 1994 __jbd2_journal_temp_unlink_buffer(jh);
1995 else
1996 jbd2_journal_grab_journal_head(bh);
1991 jh->b_transaction = transaction; 1997 jh->b_transaction = transaction;
1992 1998
1993 switch (jlist) { 1999 switch (jlist) {
@@ -2039,9 +2045,10 @@ void jbd2_journal_file_buffer(struct journal_head *jh,
2039 * already started to be used by a subsequent transaction, refile the 2045 * already started to be used by a subsequent transaction, refile the
2040 * buffer on that transaction's metadata list. 2046 * buffer on that transaction's metadata list.
2041 * 2047 *
2042 * Called under journal->j_list_lock 2048 * Called under j_list_lock
2043 *
2044 * Called under jbd_lock_bh_state(jh2bh(jh)) 2049 * Called under jbd_lock_bh_state(jh2bh(jh))
2050 *
2051 * jh and bh may be already free when this function returns
2045 */ 2052 */
2046void __jbd2_journal_refile_buffer(struct journal_head *jh) 2053void __jbd2_journal_refile_buffer(struct journal_head *jh)
2047{ 2054{
@@ -2065,6 +2072,11 @@ void __jbd2_journal_refile_buffer(struct journal_head *jh)
2065 2072
2066 was_dirty = test_clear_buffer_jbddirty(bh); 2073 was_dirty = test_clear_buffer_jbddirty(bh);
2067 __jbd2_journal_temp_unlink_buffer(jh); 2074 __jbd2_journal_temp_unlink_buffer(jh);
2075 /*
2076 * We set b_transaction here because b_next_transaction will inherit
2077 * our jh reference and thus __jbd2_journal_file_buffer() must not
2078 * take a new one.
2079 */
2068 jh->b_transaction = jh->b_next_transaction; 2080 jh->b_transaction = jh->b_next_transaction;
2069 jh->b_next_transaction = NULL; 2081 jh->b_next_transaction = NULL;
2070 if (buffer_freed(bh)) 2082 if (buffer_freed(bh))
@@ -2081,30 +2093,21 @@ void __jbd2_journal_refile_buffer(struct journal_head *jh)
2081} 2093}
2082 2094
2083/* 2095/*
2084 * For the unlocked version of this call, also make sure that any 2096 * __jbd2_journal_refile_buffer() with necessary locking added. We take our
2085 * hanging journal_head is cleaned up if necessary. 2097 * bh reference so that we can safely unlock bh.
2086 * 2098 *
2087 * __jbd2_journal_refile_buffer is usually called as part of a single locked 2099 * The jh and bh may be freed by this call.
2088 * operation on a buffer_head, in which the caller is probably going to
2089 * be hooking the journal_head onto other lists. In that case it is up
2090 * to the caller to remove the journal_head if necessary. For the
2091 * unlocked jbd2_journal_refile_buffer call, the caller isn't going to be
2092 * doing anything else to the buffer so we need to do the cleanup
2093 * ourselves to avoid a jh leak.
2094 *
2095 * *** The journal_head may be freed by this call! ***
2096 */ 2100 */
2097void jbd2_journal_refile_buffer(journal_t *journal, struct journal_head *jh) 2101void jbd2_journal_refile_buffer(journal_t *journal, struct journal_head *jh)
2098{ 2102{
2099 struct buffer_head *bh = jh2bh(jh); 2103 struct buffer_head *bh = jh2bh(jh);
2100 2104
2105 /* Get reference so that buffer cannot be freed before we unlock it */
2106 get_bh(bh);
2101 jbd_lock_bh_state(bh); 2107 jbd_lock_bh_state(bh);
2102 spin_lock(&journal->j_list_lock); 2108 spin_lock(&journal->j_list_lock);
2103
2104 __jbd2_journal_refile_buffer(jh); 2109 __jbd2_journal_refile_buffer(jh);
2105 jbd_unlock_bh_state(bh); 2110 jbd_unlock_bh_state(bh);
2106 jbd2_journal_remove_journal_head(bh);
2107
2108 spin_unlock(&journal->j_list_lock); 2111 spin_unlock(&journal->j_list_lock);
2109 __brelse(bh); 2112 __brelse(bh);
2110} 2113}