aboutsummaryrefslogtreecommitdiff
path: root/fs/jbd
diff options
context:
space:
mode:
Diffstat (limited to 'fs/jbd')
-rw-r--r--fs/jbd/commit.c72
-rw-r--r--fs/jbd/journal.c59
-rw-r--r--fs/jbd/recovery.c7
-rw-r--r--fs/jbd/revoke.c12
-rw-r--r--fs/jbd/transaction.c98
5 files changed, 176 insertions, 72 deletions
diff --git a/fs/jbd/commit.c b/fs/jbd/commit.c
index 52c15c77602..bb217dcb41a 100644
--- a/fs/jbd/commit.c
+++ b/fs/jbd/commit.c
@@ -86,7 +86,12 @@ nope:
static void release_data_buffer(struct buffer_head *bh)
{
if (buffer_freed(bh)) {
+ WARN_ON_ONCE(buffer_dirty(bh));
clear_buffer_freed(bh);
+ clear_buffer_mapped(bh);
+ clear_buffer_new(bh);
+ clear_buffer_req(bh);
+ bh->b_bdev = NULL;
release_buffer_page(bh);
} else
put_bh(bh);
@@ -157,8 +162,17 @@ static void journal_do_submit_data(struct buffer_head **wbuf, int bufs,
for (i = 0; i < bufs; i++) {
wbuf[i]->b_end_io = end_buffer_write_sync;
- /* We use-up our safety reference in submit_bh() */
- submit_bh(write_op, wbuf[i]);
+ /*
+ * Here we write back pagecache data that may be mmaped. Since
+ * we cannot afford to clean the page and set PageWriteback
+ * here due to lock ordering (page lock ranks above transaction
+ * start), the data can change while IO is in flight. Tell the
+ * block layer it should bounce the bio pages if stable data
+ * during write is required.
+ *
+ * We use up our safety reference in submit_bh().
+ */
+ _submit_bh(write_op, wbuf[i], 1 << BIO_SNAP_STABLE);
}
}
@@ -326,13 +340,13 @@ void journal_commit_transaction(journal_t *journal)
J_ASSERT(journal->j_committing_transaction == NULL);
commit_transaction = journal->j_running_transaction;
- J_ASSERT(commit_transaction->t_state == T_RUNNING);
trace_jbd_start_commit(journal, commit_transaction);
jbd_debug(1, "JBD: starting commit of transaction %d\n",
commit_transaction->t_tid);
spin_lock(&journal->j_state_lock);
+ J_ASSERT(commit_transaction->t_state == T_RUNNING);
commit_transaction->t_state = T_LOCKED;
trace_jbd_commit_locking(journal, commit_transaction);
@@ -662,7 +676,17 @@ start_journal_io:
clear_buffer_dirty(bh);
set_buffer_uptodate(bh);
bh->b_end_io = journal_end_buffer_io_sync;
- submit_bh(write_op, bh);
+ /*
+ * In data=journal mode, here we can end up
+ * writing pagecache data that might be
+ * mmapped. Since we can't afford to clean the
+ * page and set PageWriteback (see the comment
+ * near the other use of _submit_bh()), the
+ * data can change while the write is in
+ * flight. Tell the block layer to bounce the
+ * bio pages if stable pages are required.
+ */
+ _submit_bh(write_op, bh, 1 << BIO_SNAP_STABLE);
}
cond_resched();
@@ -866,17 +890,35 @@ restart_loop:
* there's no point in keeping a checkpoint record for
* it. */
- /* A buffer which has been freed while still being
- * journaled by a previous transaction may end up still
- * being dirty here, but we want to avoid writing back
- * that buffer in the future after the "add to orphan"
- * operation been committed, That's not only a performance
- * gain, it also stops aliasing problems if the buffer is
- * left behind for writeback and gets reallocated for another
- * use in a different page. */
- if (buffer_freed(bh) && !jh->b_next_transaction) {
- clear_buffer_freed(bh);
- clear_buffer_jbddirty(bh);
+ /*
+ * A buffer which has been freed while still being journaled by
+ * a previous transaction.
+ */
+ if (buffer_freed(bh)) {
+ /*
+ * If the running transaction is the one containing
+ * "add to orphan" operation (b_next_transaction !=
+ * NULL), we have to wait for that transaction to
+ * commit before we can really get rid of the buffer.
+ * So just clear b_modified to not confuse transaction
+ * credit accounting and refile the buffer to
+ * BJ_Forget of the running transaction. If the just
+ * committed transaction contains "add to orphan"
+ * operation, we can completely invalidate the buffer
+ * now. We are rather throughout in that since the
+ * buffer may be still accessible when blocksize <
+ * pagesize and it is attached to the last partial
+ * page.
+ */
+ jh->b_modified = 0;
+ if (!jh->b_next_transaction) {
+ clear_buffer_freed(bh);
+ clear_buffer_jbddirty(bh);
+ clear_buffer_mapped(bh);
+ clear_buffer_new(bh);
+ clear_buffer_req(bh);
+ bh->b_bdev = NULL;
+ }
}
if (buffer_jbddirty(bh)) {
diff --git a/fs/jbd/journal.c b/fs/jbd/journal.c
index 425c2f2cf17..06fe11e0abf 100644
--- a/fs/jbd/journal.c
+++ b/fs/jbd/journal.c
@@ -90,6 +90,24 @@ static int journal_convert_superblock_v1(journal_t *, journal_superblock_t *);
static void __journal_abort_soft (journal_t *journal, int errno);
static const char *journal_dev_name(journal_t *journal, char *buffer);
+#ifdef CONFIG_JBD_DEBUG
+void __jbd_debug(int level, const char *file, const char *func,
+ unsigned int line, const char *fmt, ...)
+{
+ struct va_format vaf;
+ va_list args;
+
+ if (level > journal_enable_debug)
+ return;
+ va_start(args, fmt);
+ vaf.fmt = fmt;
+ vaf.va = &args;
+ printk(KERN_DEBUG "%s: (%s, %u): %pV\n", file, func, line, &vaf);
+ va_end(args);
+}
+EXPORT_SYMBOL(__jbd_debug);
+#endif
+
/*
* Helper function used to manage commit timeouts
*/
@@ -310,8 +328,6 @@ int journal_write_metadata_buffer(transaction_t *transaction,
new_bh = alloc_buffer_head(GFP_NOFS|__GFP_NOFAIL);
/* keep subsequent assertions sane */
- new_bh->b_state = 0;
- init_buffer(new_bh, NULL, NULL);
atomic_set(&new_bh->b_count, 1);
new_jh = journal_add_journal_head(new_bh); /* This sleeps */
@@ -446,7 +462,8 @@ int __log_start_commit(journal_t *journal, tid_t target)
* currently running transaction (if it exists). Otherwise,
* the target tid must be an old one.
*/
- if (journal->j_running_transaction &&
+ if (journal->j_commit_request != target &&
+ journal->j_running_transaction &&
journal->j_running_transaction->t_tid == target) {
/*
* We want a new commit: OK, mark the request and wakeup the
@@ -534,8 +551,8 @@ int journal_start_commit(journal_t *journal, tid_t *ptid)
ret = 1;
} else if (journal->j_committing_transaction) {
/*
- * If ext3_write_super() recently started a commit, then we
- * have to wait for completion of that transaction
+ * If commit has been started, then we have to wait for
+ * completion of that transaction.
*/
if (ptid)
*ptid = journal->j_committing_transaction->t_tid;
@@ -556,13 +573,23 @@ int log_wait_commit(journal_t *journal, tid_t tid)
#ifdef CONFIG_JBD_DEBUG
spin_lock(&journal->j_state_lock);
if (!tid_geq(journal->j_commit_request, tid)) {
- printk(KERN_EMERG
+ printk(KERN_ERR
"%s: error: j_commit_request=%d, tid=%d\n",
__func__, journal->j_commit_request, tid);
}
spin_unlock(&journal->j_state_lock);
#endif
spin_lock(&journal->j_state_lock);
+ /*
+ * Not running or committing trans? Must be already committed. This
+ * saves us from waiting for a *long* time when tid overflows.
+ */
+ if (!((journal->j_running_transaction &&
+ journal->j_running_transaction->t_tid == tid) ||
+ (journal->j_committing_transaction &&
+ journal->j_committing_transaction->t_tid == tid)))
+ goto out_unlock;
+
if (!tid_geq(journal->j_commit_waited, tid))
journal->j_commit_waited = tid;
while (tid_gt(tid, journal->j_commit_sequence)) {
@@ -574,12 +601,11 @@ int log_wait_commit(journal_t *journal, tid_t tid)
!tid_gt(tid, journal->j_commit_sequence));
spin_lock(&journal->j_state_lock);
}
+out_unlock:
spin_unlock(&journal->j_state_lock);
- if (unlikely(is_journal_aborted(journal))) {
- printk(KERN_EMERG "journal commit I/O error\n");
+ if (unlikely(is_journal_aborted(journal)))
err = -EIO;
- }
return err;
}
@@ -1113,6 +1139,11 @@ static void mark_journal_empty(journal_t *journal)
BUG_ON(!mutex_is_locked(&journal->j_checkpoint_mutex));
spin_lock(&journal->j_state_lock);
+ /* Is it already empty? */
+ if (sb->s_start == 0) {
+ spin_unlock(&journal->j_state_lock);
+ return;
+ }
jbd_debug(1, "JBD: Marking journal as empty (seq %d)\n",
journal->j_tail_sequence);
@@ -1839,7 +1870,7 @@ static struct journal_head *journal_alloc_journal_head(void)
#ifdef CONFIG_JBD_DEBUG
atomic_inc(&nr_journal_heads);
#endif
- ret = kmem_cache_alloc(journal_head_cache, GFP_NOFS);
+ ret = kmem_cache_zalloc(journal_head_cache, GFP_NOFS);
if (ret == NULL) {
jbd_debug(1, "out of memory for journal_head\n");
printk_ratelimited(KERN_NOTICE "ENOMEM in %s, retrying.\n",
@@ -1847,7 +1878,7 @@ static struct journal_head *journal_alloc_journal_head(void)
while (ret == NULL) {
yield();
- ret = kmem_cache_alloc(journal_head_cache, GFP_NOFS);
+ ret = kmem_cache_zalloc(journal_head_cache, GFP_NOFS);
}
}
return ret;
@@ -1909,10 +1940,8 @@ struct journal_head *journal_add_journal_head(struct buffer_head *bh)
struct journal_head *new_jh = NULL;
repeat:
- if (!buffer_jbd(bh)) {
+ if (!buffer_jbd(bh))
new_jh = journal_alloc_journal_head();
- memset(new_jh, 0, sizeof(*new_jh));
- }
jbd_lock_bh_journal_head(bh);
if (buffer_jbd(bh)) {
@@ -2105,7 +2134,7 @@ static void __exit journal_exit(void)
#ifdef CONFIG_JBD_DEBUG
int n = atomic_read(&nr_journal_heads);
if (n)
- printk(KERN_EMERG "JBD: leaked %d journal_heads!\n", n);
+ printk(KERN_ERR "JBD: leaked %d journal_heads!\n", n);
#endif
jbd_remove_debugfs_entry();
journal_destroy_caches();
diff --git a/fs/jbd/recovery.c b/fs/jbd/recovery.c
index 008bf062fd2..a748fe21465 100644
--- a/fs/jbd/recovery.c
+++ b/fs/jbd/recovery.c
@@ -265,8 +265,11 @@ int journal_recover(journal_t *journal)
if (!err)
err = err2;
/* Flush disk caches to get replayed data on the permanent storage */
- if (journal->j_flags & JFS_BARRIER)
- blkdev_issue_flush(journal->j_fs_dev, GFP_KERNEL, NULL);
+ if (journal->j_flags & JFS_BARRIER) {
+ err2 = blkdev_issue_flush(journal->j_fs_dev, GFP_KERNEL, NULL);
+ if (!err)
+ err = err2;
+ }
return err;
}
diff --git a/fs/jbd/revoke.c b/fs/jbd/revoke.c
index 25c713e7071..8898bbd2b61 100644
--- a/fs/jbd/revoke.c
+++ b/fs/jbd/revoke.c
@@ -231,19 +231,15 @@ record_cache_failure:
static struct jbd_revoke_table_s *journal_init_revoke_table(int hash_size)
{
- int shift = 0;
- int tmp = hash_size;
+ int i;
struct jbd_revoke_table_s *table;
table = kmem_cache_alloc(revoke_table_cache, GFP_KERNEL);
if (!table)
goto out;
- while((tmp >>= 1UL) != 0UL)
- shift++;
-
table->hash_size = hash_size;
- table->hash_shift = shift;
+ table->hash_shift = ilog2(hash_size);
table->hash_table =
kmalloc(hash_size * sizeof(struct list_head), GFP_KERNEL);
if (!table->hash_table) {
@@ -252,8 +248,8 @@ static struct jbd_revoke_table_s *journal_init_revoke_table(int hash_size)
goto out;
}
- for (tmp = 0; tmp < hash_size; tmp++)
- INIT_LIST_HEAD(&table->hash_table[tmp]);
+ for (i = 0; i < hash_size; i++)
+ INIT_LIST_HEAD(&table->hash_table[i]);
out:
return table;
diff --git a/fs/jbd/transaction.c b/fs/jbd/transaction.c
index febc10db5ce..1695ba8334a 100644
--- a/fs/jbd/transaction.c
+++ b/fs/jbd/transaction.c
@@ -26,7 +26,6 @@
#include <linux/mm.h>
#include <linux/highmem.h>
#include <linux/hrtimer.h>
-#include <linux/backing-dev.h>
static void __journal_temp_unlink_buffer(struct journal_head *jh);
@@ -100,10 +99,11 @@ static int start_this_handle(journal_t *journal, handle_t *handle)
alloc_transaction:
if (!journal->j_running_transaction) {
- new_transaction = kzalloc(sizeof(*new_transaction), GFP_NOFS);
+ new_transaction = kzalloc(sizeof(*new_transaction),
+ GFP_NOFS|__GFP_NOFAIL);
if (!new_transaction) {
- congestion_wait(BLK_RW_ASYNC, HZ/50);
- goto alloc_transaction;
+ ret = -ENOMEM;
+ goto out;
}
}
@@ -245,7 +245,6 @@ static handle_t *new_handle(int nblocks)
handle_t *handle = jbd_alloc_handle(GFP_NOFS);
if (!handle)
return NULL;
- memset(handle, 0, sizeof(*handle));
handle->h_buffer_credits = nblocks;
handle->h_ref = 1;
@@ -676,7 +675,7 @@ repeat:
jbd_alloc(jh2bh(jh)->b_size,
GFP_NOFS);
if (!frozen_buffer) {
- printk(KERN_EMERG
+ printk(KERN_ERR
"%s: OOM for frozen_buffer\n",
__func__);
JBUFFER_TRACE(jh, "oom!");
@@ -899,7 +898,7 @@ repeat:
if (!jh->b_committed_data) {
committed_data = jbd_alloc(jh2bh(jh)->b_size, GFP_NOFS);
if (!committed_data) {
- printk(KERN_EMERG "%s: No memory for committed data\n",
+ printk(KERN_ERR "%s: No memory for committed data\n",
__func__);
err = -ENOMEM;
goto out;
@@ -1259,7 +1258,7 @@ int journal_forget (handle_t *handle, struct buffer_head *bh)
goto not_jbd;
}
- /* keep track of wether or not this transaction modified us */
+ /* keep track of whether or not this transaction modified us */
was_modified = jh->b_modified;
/*
@@ -1843,15 +1842,16 @@ static int __dispose_buffer(struct journal_head *jh, transaction_t *transaction)
* We're outside-transaction here. Either or both of j_running_transaction
* and j_committing_transaction may be NULL.
*/
-static int journal_unmap_buffer(journal_t *journal, struct buffer_head *bh)
+static int journal_unmap_buffer(journal_t *journal, struct buffer_head *bh,
+ int partial_page)
{
transaction_t *transaction;
struct journal_head *jh;
int may_free = 1;
- int ret;
BUFFER_TRACE(bh, "entry");
+retry:
/*
* It is safe to proceed here without the j_list_lock because the
* buffers cannot be stolen by try_to_free_buffers as long as we are
@@ -1879,10 +1879,18 @@ static int journal_unmap_buffer(journal_t *journal, struct buffer_head *bh)
* clear the buffer dirty bit at latest at the moment when the
* transaction marking the buffer as freed in the filesystem
* structures is committed because from that moment on the
- * buffer can be reallocated and used by a different page.
+ * block can be reallocated and used by a different page.
* Since the block hasn't been freed yet but the inode has
* already been added to orphan list, it is safe for us to add
* the buffer to BJ_Forget list of the newest transaction.
+ *
+ * Also we have to clear buffer_mapped flag of a truncated buffer
+ * because the buffer_head may be attached to the page straddling
+ * i_size (can happen only when blocksize < pagesize) and thus the
+ * buffer_head can be reused when the file is extended again. So we end
+ * up keeping around invalidated buffers attached to transactions'
+ * BJ_Forget list just to stop checkpointing code from cleaning up
+ * the transaction this buffer was modified in.
*/
transaction = jh->b_transaction;
if (transaction == NULL) {
@@ -1909,13 +1917,9 @@ static int journal_unmap_buffer(journal_t *journal, struct buffer_head *bh)
* committed, the buffer won't be needed any
* longer. */
JBUFFER_TRACE(jh, "checkpointed: add to BJ_Forget");
- ret = __dispose_buffer(jh,
+ may_free = __dispose_buffer(jh,
journal->j_running_transaction);
- journal_put_journal_head(jh);
- spin_unlock(&journal->j_list_lock);
- jbd_unlock_bh_state(bh);
- spin_unlock(&journal->j_state_lock);
- return ret;
+ goto zap_buffer;
} else {
/* There is no currently-running transaction. So the
* orphan record which we wrote for this file must have
@@ -1923,13 +1927,9 @@ static int journal_unmap_buffer(journal_t *journal, struct buffer_head *bh)
* the committing transaction, if it exists. */
if (journal->j_committing_transaction) {
JBUFFER_TRACE(jh, "give to committing trans");
- ret = __dispose_buffer(jh,
+ may_free = __dispose_buffer(jh,
journal->j_committing_transaction);
- journal_put_journal_head(jh);
- spin_unlock(&journal->j_list_lock);
- jbd_unlock_bh_state(bh);
- spin_unlock(&journal->j_state_lock);
- return ret;
+ goto zap_buffer;
} else {
/* The orphan record's transaction has
* committed. We can cleanse this buffer */
@@ -1950,10 +1950,26 @@ static int journal_unmap_buffer(journal_t *journal, struct buffer_head *bh)
}
/*
* The buffer is committing, we simply cannot touch
- * it. So we just set j_next_transaction to the
- * running transaction (if there is one) and mark
- * buffer as freed so that commit code knows it should
- * clear dirty bits when it is done with the buffer.
+ * it. If the page is straddling i_size we have to wait
+ * for commit and try again.
+ */
+ if (partial_page) {
+ tid_t tid = journal->j_committing_transaction->t_tid;
+
+ journal_put_journal_head(jh);
+ spin_unlock(&journal->j_list_lock);
+ jbd_unlock_bh_state(bh);
+ spin_unlock(&journal->j_state_lock);
+ unlock_buffer(bh);
+ log_wait_commit(journal, tid);
+ lock_buffer(bh);
+ goto retry;
+ }
+ /*
+ * OK, buffer won't be reachable after truncate. We just set
+ * j_next_transaction to the running transaction (if there is
+ * one) and mark buffer as freed so that commit code knows it
+ * should clear dirty bits when it is done with the buffer.
*/
set_buffer_freed(bh);
if (journal->j_running_transaction && buffer_jbddirty(bh))
@@ -1976,6 +1992,14 @@ static int journal_unmap_buffer(journal_t *journal, struct buffer_head *bh)
}
zap_buffer:
+ /*
+ * This is tricky. Although the buffer is truncated, it may be reused
+ * if blocksize < pagesize and it is attached to the page straddling
+ * EOF. Since the buffer might have been added to BJ_Forget list of the
+ * running transaction, journal_get_write_access() won't clear
+ * b_modified and credit accounting gets confused. So clear b_modified
+ * here. */
+ jh->b_modified = 0;
journal_put_journal_head(jh);
zap_buffer_no_jh:
spin_unlock(&journal->j_list_lock);
@@ -1995,16 +2019,20 @@ zap_buffer_unlocked:
* void journal_invalidatepage() - invalidate a journal page
* @journal: journal to use for flush
* @page: page to flush
- * @offset: length of page to invalidate.
+ * @offset: offset of the range to invalidate
+ * @length: length of the range to invalidate
*
- * Reap page buffers containing data after offset in page.
+ * Reap page buffers containing data in specified range in page.
*/
void journal_invalidatepage(journal_t *journal,
struct page *page,
- unsigned long offset)
+ unsigned int offset,
+ unsigned int length)
{
struct buffer_head *head, *bh, *next;
+ unsigned int stop = offset + length;
unsigned int curr_off = 0;
+ int partial_page = (offset || length < PAGE_CACHE_SIZE);
int may_free = 1;
if (!PageLocked(page))
@@ -2012,6 +2040,8 @@ void journal_invalidatepage(journal_t *journal,
if (!page_has_buffers(page))
return;
+ BUG_ON(stop > PAGE_CACHE_SIZE || stop < length);
+
/* We will potentially be playing with lists other than just the
* data lists (especially for journaled data mode), so be
* cautious in our locking. */
@@ -2021,10 +2051,14 @@ void journal_invalidatepage(journal_t *journal,
unsigned int next_off = curr_off + bh->b_size;
next = bh->b_this_page;
+ if (next_off > stop)
+ return;
+
if (offset <= curr_off) {
/* This block is wholly outside the truncation point */
lock_buffer(bh);
- may_free &= journal_unmap_buffer(journal, bh);
+ may_free &= journal_unmap_buffer(journal, bh,
+ partial_page);
unlock_buffer(bh);
}
curr_off = next_off;
@@ -2032,7 +2066,7 @@ void journal_invalidatepage(journal_t *journal,
} while (bh != head);
- if (!offset) {
+ if (!partial_page) {
if (may_free && try_to_free_buffers(page))
J_ASSERT(!page_has_buffers(page));
}
generated by cgit v1.2.3-70-g09d2 (git 2.46.0) at 2025-11-11 13:53:15 +0000