6 files changed, 279 insertions, 208 deletions

diff --git a/kernel/fs/jbd2/checkpoint.c b/kernel/fs/jbd2/checkpoint.c
index 78c1545a3..6e18a06aa 100644
--- a/kernel/fs/jbd2/checkpoint.c
+++ b/kernel/fs/jbd2/checkpoint.c
@@ -429,7 +429,6 @@ static int journal_clean_one_cp_list(struct journal_head *jh, bool destroy)
 	struct journal_head *last_jh;
 	struct journal_head *next_jh = jh;
 	int ret;
-	int freed = 0;
 
 	if (!jh)
 		return 0;
@@ -443,10 +442,9 @@ static int journal_clean_one_cp_list(struct journal_head *jh, bool destroy)
 		else
 			ret = __jbd2_journal_remove_checkpoint(jh) + 1;
 		if (!ret)
-			return freed;
+			return 0;
 		if (ret == 2)
 			return 1;
-		freed = 1;
 		/*
 		 * This function only frees up some memory
 		 * if possible so we dont have an obligation
@@ -454,10 +452,10 @@ static int journal_clean_one_cp_list(struct journal_head *jh, bool destroy)
 		 * requested:
 		 */
 		if (need_resched())
-			return freed;
+			return 0;
 	} while (jh != last_jh);
 
-	return freed;
+	return 0;
 }
 
 /*
diff --git a/kernel/fs/jbd2/commit.c b/kernel/fs/jbd2/commit.c
index 362e5f614..36345fefa 100644
--- a/kernel/fs/jbd2/commit.c
+++ b/kernel/fs/jbd2/commit.c
@@ -142,8 +142,7 @@ static int journal_submit_commit_record(journal_t *journal,
 	tmp->h_commit_sec = cpu_to_be64(now.tv_sec);
 	tmp->h_commit_nsec = cpu_to_be32(now.tv_nsec);
 
-	if (JBD2_HAS_COMPAT_FEATURE(journal,
-				    JBD2_FEATURE_COMPAT_CHECKSUM)) {
+	if (jbd2_has_feature_checksum(journal)) {
 		tmp->h_chksum_type 	= JBD2_CRC32_CHKSUM;
 		tmp->h_chksum_size 	= JBD2_CRC32_CHKSUM_SIZE;
 		tmp->h_chksum[0] 	= cpu_to_be32(crc32_sum);
@@ -157,8 +156,7 @@ static int journal_submit_commit_record(journal_t *journal,
 	bh->b_end_io = journal_end_buffer_io_sync;
 
 	if (journal->j_flags & JBD2_BARRIER &&
-	    !JBD2_HAS_INCOMPAT_FEATURE(journal,
-				       JBD2_FEATURE_INCOMPAT_ASYNC_COMMIT))
+	    !jbd2_has_feature_async_commit(journal))
 		ret = submit_bh(WRITE_SYNC | WRITE_FLUSH_FUA, bh);
 	else
 		ret = submit_bh(WRITE_SYNC, bh);
@@ -317,7 +315,7 @@ static void write_tag_block(journal_t *j, journal_block_tag_t *tag,
 				   unsigned long long block)
 {
 	tag->t_blocknr = cpu_to_be32(block & (u32)~0);
-	if (JBD2_HAS_INCOMPAT_FEATURE(j, JBD2_FEATURE_INCOMPAT_64BIT))
+	if (jbd2_has_feature_64bit(j))
 		tag->t_blocknr_high = cpu_to_be32((block >> 31) >> 1);
 }
 
@@ -356,7 +354,7 @@ static void jbd2_block_tag_csum_set(journal_t *j, journal_block_tag_t *tag,
 			     bh->b_size);
 	kunmap_atomic(addr);
 
-	if (JBD2_HAS_INCOMPAT_FEATURE(j, JBD2_FEATURE_INCOMPAT_CSUM_V3))
+	if (jbd2_has_feature_csum3(j))
 		tag3->t_checksum = cpu_to_be32(csum32);
 	else
 		tag->t_checksum = cpu_to_be16(csum32);
@@ -730,8 +728,7 @@ start_journal_io:
 				/*
 				 * Compute checksum.
 				 */
-				if (JBD2_HAS_COMPAT_FEATURE(journal,
-					JBD2_FEATURE_COMPAT_CHECKSUM)) {
+				if (jbd2_has_feature_checksum(journal)) {
 					crc32_sum =
 					    jbd2_checksum_data(crc32_sum, bh);
 				}
@@ -797,8 +794,7 @@ start_journal_io:
 		blkdev_issue_flush(journal->j_fs_dev, GFP_NOFS, NULL);
 
 	/* Done it all: now write the commit record asynchronously. */
-	if (JBD2_HAS_INCOMPAT_FEATURE(journal,
-				      JBD2_FEATURE_INCOMPAT_ASYNC_COMMIT)) {
+	if (jbd2_has_feature_async_commit(journal)) {
 		err = journal_submit_commit_record(journal, commit_transaction,
 						 &cbh, crc32_sum);
 		if (err)
@@ -889,8 +885,7 @@ start_journal_io:
 	commit_transaction->t_state = T_COMMIT_JFLUSH;
 	write_unlock(&journal->j_state_lock);
 
-	if (!JBD2_HAS_INCOMPAT_FEATURE(journal,
-				       JBD2_FEATURE_INCOMPAT_ASYNC_COMMIT)) {
+	if (!jbd2_has_feature_async_commit(journal)) {
 		err = journal_submit_commit_record(journal, commit_transaction,
 						&cbh, crc32_sum);
 		if (err)
@@ -898,8 +893,7 @@ start_journal_io:
 	}
 	if (cbh)
 		err = journal_wait_on_commit_record(journal, cbh);
-	if (JBD2_HAS_INCOMPAT_FEATURE(journal,
-				      JBD2_FEATURE_INCOMPAT_ASYNC_COMMIT) &&
+	if (jbd2_has_feature_async_commit(journal) &&
 	    journal->j_flags & JBD2_BARRIER) {
 		blkdev_issue_flush(journal->j_dev, GFP_NOFS, NULL);
 	}
diff --git a/kernel/fs/jbd2/journal.c b/kernel/fs/jbd2/journal.c
index 7003c0925..81e622681 100644
--- a/kernel/fs/jbd2/journal.c
+++ b/kernel/fs/jbd2/journal.c
@@ -124,7 +124,7 @@ EXPORT_SYMBOL(__jbd2_debug);
 /* Checksumming functions */
 static int jbd2_verify_csum_type(journal_t *j, journal_superblock_t *sb)
 {
-	if (!jbd2_journal_has_csum_v2or3(j))
+	if (!jbd2_journal_has_csum_v2or3_feature(j))
 		return 1;
 
 	return sb->s_checksum_type == JBD2_CRC32C_CHKSUM;
@@ -371,16 +371,7 @@ int jbd2_journal_write_metadata_buffer(transaction_t *transaction,
 	 */
 	J_ASSERT_BH(bh_in, buffer_jbddirty(bh_in));
 
-retry_alloc:
-	new_bh = alloc_buffer_head(GFP_NOFS);
-	if (!new_bh) {
-		/*
-		 * Failure is not an option, but __GFP_NOFAIL is going
-		 * away; so we retry ourselves here.
-		 */
-		congestion_wait(BLK_RW_ASYNC, HZ/50);
-		goto retry_alloc;
-	}
+	new_bh = alloc_buffer_head(GFP_NOFS|__GFP_NOFAIL);
 
 	/* keep subsequent assertions sane */
 	atomic_set(&new_bh->b_count, 1);
@@ -1144,7 +1135,6 @@ journal_t * jbd2_journal_init_dev(struct block_device *bdev,
 {
 	journal_t *journal = journal_init_common();
 	struct buffer_head *bh;
-	char *p;
 	int n;
 
 	if (!journal)
@@ -1157,9 +1147,7 @@ journal_t * jbd2_journal_init_dev(struct block_device *bdev,
 	journal->j_blk_offset = start;
 	journal->j_maxlen = len;
 	bdevname(journal->j_dev, journal->j_devname);
-	p = journal->j_devname;
-	while ((p = strchr(p, '/')))
-		*p = '!';
+	strreplace(journal->j_devname, '/', '!');
 	jbd2_stats_proc_init(journal);
 	n = journal->j_blocksize / sizeof(journal_block_tag_t);
 	journal->j_wbufsize = n;
@@ -1211,10 +1199,7 @@ journal_t * jbd2_journal_init_inode (struct inode *inode)
 	journal->j_dev = journal->j_fs_dev = inode->i_sb->s_bdev;
 	journal->j_inode = inode;
 	bdevname(journal->j_dev, journal->j_devname);
-	p = journal->j_devname;
-	while ((p = strchr(p, '/')))
-		*p = '!';
-	p = journal->j_devname + strlen(journal->j_devname);
+	p = strreplace(journal->j_devname, '/', '!');
 	sprintf(p, "-%lu", journal->j_inode->i_ino);
 	jbd_debug(1,
 		  "journal %p: inode %s/%ld, size %Ld, bits %d, blksize %ld\n",
@@ -1471,7 +1456,7 @@ void jbd2_journal_update_sb_errno(journal_t *journal)
 	sb->s_errno    = cpu_to_be32(journal->j_errno);
 	read_unlock(&journal->j_state_lock);
 
-	jbd2_write_superblock(journal, WRITE_SYNC);
+	jbd2_write_superblock(journal, WRITE_FUA);
 }
 EXPORT_SYMBOL(jbd2_journal_update_sb_errno);
 
@@ -1538,16 +1523,16 @@ static int journal_get_superblock(journal_t *journal)
 		goto out;
 	}
 
-	if (JBD2_HAS_INCOMPAT_FEATURE(journal, JBD2_FEATURE_INCOMPAT_CSUM_V2) &&
-	    JBD2_HAS_INCOMPAT_FEATURE(journal, JBD2_FEATURE_INCOMPAT_CSUM_V3)) {
+	if (jbd2_has_feature_csum2(journal) &&
+	    jbd2_has_feature_csum3(journal)) {
 		/* Can't have checksum v2 and v3 at the same time! */
 		printk(KERN_ERR "JBD2: Can't enable checksumming v2 and v3 "
 		       "at the same time!\n");
 		goto out;
 	}
 
-	if (jbd2_journal_has_csum_v2or3(journal) &&
-	    JBD2_HAS_COMPAT_FEATURE(journal, JBD2_FEATURE_COMPAT_CHECKSUM)) {
+	if (jbd2_journal_has_csum_v2or3_feature(journal) &&
+	    jbd2_has_feature_checksum(journal)) {
 		/* Can't have checksum v1 and v2 on at the same time! */
 		printk(KERN_ERR "JBD2: Can't enable checksumming v1 and v2/3 "
 		       "at the same time!\n");
@@ -1560,7 +1545,7 @@ static int journal_get_superblock(journal_t *journal)
 	}
 
 	/* Load the checksum driver */
-	if (jbd2_journal_has_csum_v2or3(journal)) {
+	if (jbd2_journal_has_csum_v2or3_feature(journal)) {
 		journal->j_chksum_driver = crypto_alloc_shash("crc32c", 0, 0);
 		if (IS_ERR(journal->j_chksum_driver)) {
 			printk(KERN_ERR "JBD2: Cannot load crc32c driver.\n");
@@ -1573,6 +1558,7 @@ static int journal_get_superblock(journal_t *journal)
 	/* Check superblock checksum */
 	if (!jbd2_superblock_csum_verify(journal, sb)) {
 		printk(KERN_ERR "JBD2: journal checksum error\n");
+		err = -EFSBADCRC;
 		goto out;
 	}
 
@@ -1664,7 +1650,7 @@ int jbd2_journal_load(journal_t *journal)
 		printk(KERN_ERR "JBD2: journal transaction %u on %s "
 		       "is corrupt.\n", journal->j_failed_commit,
 		       journal->j_devname);
-		return -EIO;
+		return -EFSCORRUPTED;
 	}
 
 	/* OK, we've finished with the dynamic journal bits:
@@ -2086,8 +2072,12 @@ static void __journal_abort_soft (journal_t *journal, int errno)
 
 	__jbd2_journal_abort_hard(journal);
 
-	if (errno)
+	if (errno) {
 		jbd2_journal_update_sb_errno(journal);
+		write_lock(&journal->j_state_lock);
+		journal->j_flags |= JBD2_REC_ERR;
+		write_unlock(&journal->j_state_lock);
+	}
 }
 
 /**
@@ -2212,15 +2202,15 @@ size_t journal_tag_bytes(journal_t *journal)
 {
 	size_t sz;
 
-	if (JBD2_HAS_INCOMPAT_FEATURE(journal, JBD2_FEATURE_INCOMPAT_CSUM_V3))
+	if (jbd2_has_feature_csum3(journal))
 		return sizeof(journal_block_tag3_t);
 
 	sz = sizeof(journal_block_tag_t);
 
-	if (JBD2_HAS_INCOMPAT_FEATURE(journal, JBD2_FEATURE_INCOMPAT_CSUM_V2))
+	if (jbd2_has_feature_csum2(journal))
 		sz += sizeof(__u16);
 
-	if (JBD2_HAS_INCOMPAT_FEATURE(journal, JBD2_FEATURE_INCOMPAT_64BIT))
+	if (jbd2_has_feature_64bit(journal))
 		return sz;
 	else
 		return sz - sizeof(__u32);
@@ -2363,7 +2353,7 @@ static int jbd2_journal_init_journal_head_cache(void)
 	jbd2_journal_head_cache = kmem_cache_create("jbd2_journal_head",
 				sizeof(struct journal_head),
 				0,		/* offset */
-				SLAB_TEMPORARY,	/* flags */
+				SLAB_TEMPORARY | SLAB_DESTROY_BY_RCU,
 				NULL);		/* ctor */
 	retval = 0;
 	if (!jbd2_journal_head_cache) {
@@ -2395,10 +2385,8 @@ static struct journal_head *journal_alloc_journal_head(void)
 	if (!ret) {
 		jbd_debug(1, "out of memory for journal_head\n");
 		pr_notice_ratelimited("ENOMEM in %s, retrying.\n", __func__);
-		while (!ret) {
-			yield();
-			ret = kmem_cache_zalloc(jbd2_journal_head_cache, GFP_NOFS);
-		}
+		ret = kmem_cache_zalloc(jbd2_journal_head_cache,
+				GFP_NOFS | __GFP_NOFAIL);
 	}
 	return ret;
 }
diff --git a/kernel/fs/jbd2/recovery.c b/kernel/fs/jbd2/recovery.c
index a9079d035..7f277e49f 100644
--- a/kernel/fs/jbd2/recovery.c
+++ b/kernel/fs/jbd2/recovery.c
@@ -140,7 +140,7 @@ static int jread(struct buffer_head **bhp, journal_t *journal,
 
 	if (offset >= journal->j_maxlen) {
 		printk(KERN_ERR "JBD2: corrupted journal superblock\n");
-		return -EIO;
+		return -EFSCORRUPTED;
 	}
 
 	err = jbd2_journal_bmap(journal, offset, &blocknr);
@@ -342,7 +342,7 @@ static inline unsigned long long read_tag_block(journal_t *journal,
 						journal_block_tag_t *tag)
 {
 	unsigned long long block = be32_to_cpu(tag->t_blocknr);
-	if (JBD2_HAS_INCOMPAT_FEATURE(journal, JBD2_FEATURE_INCOMPAT_64BIT))
+	if (jbd2_has_feature_64bit(journal))
 		block |= (u64)be32_to_cpu(tag->t_blocknr_high) << 32;
 	return block;
 }
@@ -411,7 +411,7 @@ static int jbd2_block_tag_csum_verify(journal_t *j, journal_block_tag_t *tag,
 	csum32 = jbd2_chksum(j, j->j_csum_seed, (__u8 *)&seq, sizeof(seq));
 	csum32 = jbd2_chksum(j, csum32, buf, j->j_blocksize);
 
-	if (JBD2_HAS_INCOMPAT_FEATURE(j, JBD2_FEATURE_INCOMPAT_CSUM_V3))
+	if (jbd2_has_feature_csum3(j))
 		return tag3->t_checksum == cpu_to_be32(csum32);
 	else
 		return tag->t_checksum == cpu_to_be16(csum32);
@@ -527,7 +527,7 @@ static int do_one_pass(journal_t *journal,
 				printk(KERN_ERR "JBD2: Invalid checksum "
 				       "recovering block %lu in log\n",
 				       next_log_block);
-				err = -EIO;
+				err = -EFSBADCRC;
 				brelse(bh);
 				goto failed;
 			}
@@ -538,8 +538,7 @@ static int do_one_pass(journal_t *journal,
 			 * just skip over the blocks it describes. */
 			if (pass != PASS_REPLAY) {
 				if (pass == PASS_SCAN &&
-				    JBD2_HAS_COMPAT_FEATURE(journal,
-					    JBD2_FEATURE_COMPAT_CHECKSUM) &&
+				    jbd2_has_feature_checksum(journal) &&
 				    !info->end_transaction) {
 					if (calc_chksums(journal, bh,
 							&next_log_block,
@@ -602,7 +601,7 @@ static int do_one_pass(journal_t *journal,
 						journal, tag, obh->b_data,
 						be32_to_cpu(tmp->h_sequence))) {
 						brelse(obh);
-						success = -EIO;
+						success = -EFSBADCRC;
 						printk(KERN_ERR "JBD2: Invalid "
 						       "checksum recovering "
 						       "block %llu in log\n",
@@ -694,8 +693,7 @@ static int do_one_pass(journal_t *journal,
 			 * much to do other than move on to the next sequence
 			 * number. */
 			if (pass == PASS_SCAN &&
-			    JBD2_HAS_COMPAT_FEATURE(journal,
-				    JBD2_FEATURE_COMPAT_CHECKSUM)) {
+			    jbd2_has_feature_checksum(journal)) {
 				int chksum_err, chksum_seen;
 				struct commit_header *cbh =
 					(struct commit_header *)bh->b_data;
@@ -735,8 +733,7 @@ static int do_one_pass(journal_t *journal,
 				if (chksum_err) {
 					info->end_transaction = next_commit_ID;
 
-					if (!JBD2_HAS_INCOMPAT_FEATURE(journal,
-					   JBD2_FEATURE_INCOMPAT_ASYNC_COMMIT)){
+					if (!jbd2_has_feature_async_commit(journal)) {
 						journal->j_failed_commit =
 							next_commit_ID;
 						brelse(bh);
@@ -750,8 +747,7 @@ static int do_one_pass(journal_t *journal,
 							   bh->b_data)) {
 				info->end_transaction = next_commit_ID;
 
-				if (!JBD2_HAS_INCOMPAT_FEATURE(journal,
-				     JBD2_FEATURE_INCOMPAT_ASYNC_COMMIT)) {
+				if (!jbd2_has_feature_async_commit(journal)) {
 					journal->j_failed_commit =
 						next_commit_ID;
 					brelse(bh);
@@ -851,7 +847,7 @@ static int scan_revoke_records(journal_t *journal, struct buffer_head *bh,
 	rcount = be32_to_cpu(header->r_count);
 
 	if (!jbd2_revoke_block_csum_verify(journal, header))
-		return -EINVAL;
+		return -EFSBADCRC;
 
 	if (jbd2_journal_has_csum_v2or3(journal))
 		csum_size = sizeof(struct jbd2_journal_revoke_tail);
@@ -859,7 +855,7 @@ static int scan_revoke_records(journal_t *journal, struct buffer_head *bh,
 		return -EINVAL;
 	max = rcount;
 
-	if (JBD2_HAS_INCOMPAT_FEATURE(journal, JBD2_FEATURE_INCOMPAT_64BIT))
+	if (jbd2_has_feature_64bit(journal))
 		record_len = 8;
 
 	while (offset + record_len <= max) {
diff --git a/kernel/fs/jbd2/revoke.c b/kernel/fs/jbd2/revoke.c
index 14214da80..705ae5778 100644
--- a/kernel/fs/jbd2/revoke.c
+++ b/kernel/fs/jbd2/revoke.c
@@ -141,11 +141,13 @@ static int insert_revoke_hash(journal_t *journal, unsigned long long blocknr,
 {
 	struct list_head *hash_list;
 	struct jbd2_revoke_record_s *record;
+	gfp_t gfp_mask = GFP_NOFS;
 
-repeat:
-	record = kmem_cache_alloc(jbd2_revoke_record_cache, GFP_NOFS);
+	if (journal_oom_retry)
+		gfp_mask |= __GFP_NOFAIL;
+	record = kmem_cache_alloc(jbd2_revoke_record_cache, gfp_mask);
 	if (!record)
-		goto oom;
+		return -ENOMEM;
 
 	record->sequence = seq;
 	record->blocknr = blocknr;
@@ -154,13 +156,6 @@ repeat:
 	list_add(&record->hash, hash_list);
 	spin_unlock(&journal->j_revoke_lock);
 	return 0;
-
-oom:
-	if (!journal_oom_retry)
-		return -ENOMEM;
-	jbd_debug(1, "ENOMEM in %s, retrying\n", __func__);
-	yield();
-	goto repeat;
 }
 
 /* Find a revoke record in the journal's hash table. */
@@ -594,7 +589,7 @@ static void write_one_revoke_record(journal_t *journal,
 	if (jbd2_journal_has_csum_v2or3(journal))
 		csum_size = sizeof(struct jbd2_journal_revoke_tail);
 
-	if (JBD2_HAS_INCOMPAT_FEATURE(journal, JBD2_FEATURE_INCOMPAT_64BIT))
+	if (jbd2_has_feature_64bit(journal))
 		sz = 8;
 	else
 		sz = 4;
@@ -624,7 +619,7 @@ static void write_one_revoke_record(journal_t *journal,
 		*descriptorp = descriptor;
 	}
 
-	if (JBD2_HAS_INCOMPAT_FEATURE(journal, JBD2_FEATURE_INCOMPAT_64BIT))
+	if (jbd2_has_feature_64bit(journal))
 		* ((__be64 *)(&descriptor->b_data[offset])) =
 			cpu_to_be64(record->blocknr);
 	else
diff --git a/kernel/fs/jbd2/transaction.c b/kernel/fs/jbd2/transaction.c
index ff2f2e6ad..ca181e81c 100644
--- a/kernel/fs/jbd2/transaction.c
+++ b/kernel/fs/jbd2/transaction.c
@@ -204,6 +204,20 @@ static int add_transaction_credits(journal_t *journal, int blocks,
 		 * attach this handle to a new transaction.
 		 */
 		atomic_sub(total, &t->t_outstanding_credits);
+
+		/*
+		 * Is the number of reserved credits in the current transaction too
+		 * big to fit this handle? Wait until reserved credits are freed.
+		 */
+		if (atomic_read(&journal->j_reserved_credits) + total >
+		    journal->j_max_transaction_buffers) {
+			read_unlock(&journal->j_state_lock);
+			wait_event(journal->j_wait_reserved,
+				   atomic_read(&journal->j_reserved_credits) + total <=
+				   journal->j_max_transaction_buffers);
+			return 1;
+		}
+
 		wait_transaction_locked(journal);
 		return 1;
 	}
@@ -262,38 +276,36 @@ static int start_this_handle(journal_t *journal, handle_t *handle,
 	int		rsv_blocks = 0;
 	unsigned long ts = jiffies;
 
+	if (handle->h_rsv_handle)
+		rsv_blocks = handle->h_rsv_handle->h_buffer_credits;
+
 	/*
-	 * 1/2 of transaction can be reserved so we can practically handle
-	 * only 1/2 of maximum transaction size per operation
+	 * Limit the number of reserved credits to 1/2 of maximum transaction
+	 * size and limit the number of total credits to not exceed maximum
+	 * transaction size per operation.
 	 */
-	if (WARN_ON(blocks > journal->j_max_transaction_buffers / 2)) {
-		printk(KERN_ERR "JBD2: %s wants too many credits (%d > %d)\n",
-		       current->comm, blocks,
-		       journal->j_max_transaction_buffers / 2);
+	if ((rsv_blocks > journal->j_max_transaction_buffers / 2) ||
+	    (rsv_blocks + blocks > journal->j_max_transaction_buffers)) {
+		printk(KERN_ERR "JBD2: %s wants too many credits "
+		       "credits:%d rsv_credits:%d max:%d\n",
+		       current->comm, blocks, rsv_blocks,
+		       journal->j_max_transaction_buffers);
+		WARN_ON(1);
 		return -ENOSPC;
 	}
 
-	if (handle->h_rsv_handle)
-		rsv_blocks = handle->h_rsv_handle->h_buffer_credits;
-
 alloc_transaction:
 	if (!journal->j_running_transaction) {
+		/*
+		 * If __GFP_FS is not present, then we may be being called from
+		 * inside the fs writeback layer, so we MUST NOT fail.
+		 */
+		if ((gfp_mask & __GFP_FS) == 0)
+			gfp_mask |= __GFP_NOFAIL;
 		new_transaction = kmem_cache_zalloc(transaction_cache,
 						    gfp_mask);
-		if (!new_transaction) {
-			/*
-			 * If __GFP_FS is not present, then we may be
-			 * being called from inside the fs writeback
-			 * layer, so we MUST NOT fail.  Since
-			 * __GFP_NOFAIL is going away, we will arrange
-			 * to retry the allocation ourselves.
-			 */
-			if ((gfp_mask & __GFP_FS) == 0) {
-				congestion_wait(BLK_RW_ASYNC, HZ/50);
-				goto alloc_transaction;
-			}
+		if (!new_transaction)
 			return -ENOMEM;
-		}
 	}
 
 	jbd_debug(3, "New handle %p going live.\n", handle);
@@ -761,6 +773,30 @@ static void warn_dirty_buffer(struct buffer_head *bh)
 	       bdevname(bh->b_bdev, b), (unsigned long long)bh->b_blocknr);
 }
 
+/* Call t_frozen trigger and copy buffer data into jh->b_frozen_data. */
+static void jbd2_freeze_jh_data(struct journal_head *jh)
+{
+	struct page *page;
+	int offset;
+	char *source;
+	struct buffer_head *bh = jh2bh(jh);
+
+	J_EXPECT_JH(jh, buffer_uptodate(bh), "Possible IO failure.\n");
+	page = bh->b_page;
+	offset = offset_in_page(bh->b_data);
+	source = kmap_atomic(page);
+	/* Fire data frozen trigger just before we copy the data */
+	jbd2_buffer_frozen_trigger(jh, source + offset, jh->b_triggers);
+	memcpy(jh->b_frozen_data, source + offset, bh->b_size);
+	kunmap_atomic(source);
+
+	/*
+	 * Now that the frozen data is saved off, we need to store any matching
+	 * triggers.
+	 */
+	jh->b_frozen_triggers = jh->b_triggers;
+}
+
 /*
  * If the buffer is already part of the current transaction, then there
  * is nothing we need to do.  If it is already part of a prior
@@ -780,7 +816,6 @@ do_get_write_access(handle_t *handle, struct journal_head *jh,
 	journal_t *journal;
 	int error;
 	char *frozen_buffer = NULL;
-	int need_copy = 0;
 	unsigned long start_lock, time_lock;
 
 	if (is_handle_aborted(handle))
@@ -867,119 +902,96 @@ repeat:
        jh->b_modified = 0;
 
 	/*
+	 * If the buffer is not journaled right now, we need to make sure it
+	 * doesn't get written to disk before the caller actually commits the
+	 * new data
+	 */
+	if (!jh->b_transaction) {
+		JBUFFER_TRACE(jh, "no transaction");
+		J_ASSERT_JH(jh, !jh->b_next_transaction);
+		JBUFFER_TRACE(jh, "file as BJ_Reserved");
+		/*
+		 * Make sure all stores to jh (b_modified, b_frozen_data) are
+		 * visible before attaching it to the running transaction.
+		 * Paired with barrier in jbd2_write_access_granted()
+		 */
+		smp_wmb();
+		spin_lock(&journal->j_list_lock);
+		__jbd2_journal_file_buffer(jh, transaction, BJ_Reserved);
+		spin_unlock(&journal->j_list_lock);
+		goto done;
+	}
+	/*
 	 * If there is already a copy-out version of this buffer, then we don't
 	 * need to make another one
 	 */
 	if (jh->b_frozen_data) {
 		JBUFFER_TRACE(jh, "has frozen data");
 		J_ASSERT_JH(jh, jh->b_next_transaction == NULL);
-		jh->b_next_transaction = transaction;
-		goto done;
+		goto attach_next;
 	}
 
-	/* Is there data here we need to preserve? */
+	JBUFFER_TRACE(jh, "owned by older transaction");
+	J_ASSERT_JH(jh, jh->b_next_transaction == NULL);
+	J_ASSERT_JH(jh, jh->b_transaction == journal->j_committing_transaction);
 
-	if (jh->b_transaction && jh->b_transaction != transaction) {
-		JBUFFER_TRACE(jh, "owned by older transaction");
-		J_ASSERT_JH(jh, jh->b_next_transaction == NULL);
-		J_ASSERT_JH(jh, jh->b_transaction ==
-					journal->j_committing_transaction);
+	/*
+	 * There is one case we have to be very careful about.  If the
+	 * committing transaction is currently writing this buffer out to disk
+	 * and has NOT made a copy-out, then we cannot modify the buffer
+	 * contents at all right now.  The essence of copy-out is that it is
+	 * the extra copy, not the primary copy, which gets journaled.  If the
+	 * primary copy is already going to disk then we cannot do copy-out
+	 * here.
+	 */
+	if (buffer_shadow(bh)) {
+		JBUFFER_TRACE(jh, "on shadow: sleep");
+		jbd_unlock_bh_state(bh);
+		wait_on_bit_io(&bh->b_state, BH_Shadow, TASK_UNINTERRUPTIBLE);
+		goto repeat;
+	}
 
-		/* There is one case we have to be very careful about.
-		 * If the committing transaction is currently writing
-		 * this buffer out to disk and has NOT made a copy-out,
-		 * then we cannot modify the buffer contents at all
-		 * right now.  The essence of copy-out is that it is the
-		 * extra copy, not the primary copy, which gets
-		 * journaled.  If the primary copy is already going to
-		 * disk then we cannot do copy-out here. */
-
-		if (buffer_shadow(bh)) {
-			JBUFFER_TRACE(jh, "on shadow: sleep");
+	/*
+	 * Only do the copy if the currently-owning transaction still needs it.
+	 * If buffer isn't on BJ_Metadata list, the committing transaction is
+	 * past that stage (here we use the fact that BH_Shadow is set under
+	 * bh_state lock together with refiling to BJ_Shadow list and at this
+	 * point we know the buffer doesn't have BH_Shadow set).
+	 *
+	 * Subtle point, though: if this is a get_undo_access, then we will be
+	 * relying on the frozen_data to contain the new value of the
+	 * committed_data record after the transaction, so we HAVE to force the
+	 * frozen_data copy in that case.
+	 */
+	if (jh->b_jlist == BJ_Metadata || force_copy) {
+		JBUFFER_TRACE(jh, "generate frozen data");
+		if (!frozen_buffer) {
+			JBUFFER_TRACE(jh, "allocate memory for buffer");
 			jbd_unlock_bh_state(bh);
-			wait_on_bit_io(&bh->b_state, BH_Shadow,
-				       TASK_UNINTERRUPTIBLE);
-			goto repeat;
-		}
-
-		/*
-		 * Only do the copy if the currently-owning transaction still
-		 * needs it. If buffer isn't on BJ_Metadata list, the
-		 * committing transaction is past that stage (here we use the
-		 * fact that BH_Shadow is set under bh_state lock together with
-		 * refiling to BJ_Shadow list and at this point we know the
-		 * buffer doesn't have BH_Shadow set).
-		 *
-		 * Subtle point, though: if this is a get_undo_access,
-		 * then we will be relying on the frozen_data to contain
-		 * the new value of the committed_data record after the
-		 * transaction, so we HAVE to force the frozen_data copy
-		 * in that case.
-		 */
-		if (jh->b_jlist == BJ_Metadata || force_copy) {
-			JBUFFER_TRACE(jh, "generate frozen data");
+			frozen_buffer = jbd2_alloc(jh2bh(jh)->b_size, GFP_NOFS);
 			if (!frozen_buffer) {
-				JBUFFER_TRACE(jh, "allocate memory for buffer");
-				jbd_unlock_bh_state(bh);
-				frozen_buffer =
-					jbd2_alloc(jh2bh(jh)->b_size,
-							 GFP_NOFS);
-				if (!frozen_buffer) {
-					printk(KERN_ERR
-					       "%s: OOM for frozen_buffer\n",
-					       __func__);
-					JBUFFER_TRACE(jh, "oom!");
-					error = -ENOMEM;
-					jbd_lock_bh_state(bh);
-					goto done;
-				}
-				goto repeat;
+				printk(KERN_ERR "%s: OOM for frozen_buffer\n",
+				       __func__);
+				JBUFFER_TRACE(jh, "oom!");
+				error = -ENOMEM;
+				goto out;
 			}
-			jh->b_frozen_data = frozen_buffer;
-			frozen_buffer = NULL;
-			need_copy = 1;
+			goto repeat;
 		}
-		jh->b_next_transaction = transaction;
+		jh->b_frozen_data = frozen_buffer;
+		frozen_buffer = NULL;
+		jbd2_freeze_jh_data(jh);
 	}
-
-
+attach_next:
 	/*
-	 * Finally, if the buffer is not journaled right now, we need to make
-	 * sure it doesn't get written to disk before the caller actually
-	 * commits the new data
+	 * Make sure all stores to jh (b_modified, b_frozen_data) are visible
+	 * before attaching it to the running transaction. Paired with barrier
+	 * in jbd2_write_access_granted()
 	 */
-	if (!jh->b_transaction) {
-		JBUFFER_TRACE(jh, "no transaction");
-		J_ASSERT_JH(jh, !jh->b_next_transaction);
-		JBUFFER_TRACE(jh, "file as BJ_Reserved");
-		spin_lock(&journal->j_list_lock);
-		__jbd2_journal_file_buffer(jh, transaction, BJ_Reserved);
-		spin_unlock(&journal->j_list_lock);
-	}
+	smp_wmb();
+	jh->b_next_transaction = transaction;
 
 done:
-	if (need_copy) {
-		struct page *page;
-		int offset;
-		char *source;
-
-		J_EXPECT_JH(jh, buffer_uptodate(jh2bh(jh)),
-			    "Possible IO failure.\n");
-		page = jh2bh(jh)->b_page;
-		offset = offset_in_page(jh2bh(jh)->b_data);
-		source = kmap_atomic(page);
-		/* Fire data frozen trigger just before we copy the data */
-		jbd2_buffer_frozen_trigger(jh, source + offset,
-					   jh->b_triggers);
-		memcpy(jh->b_frozen_data, source+offset, jh2bh(jh)->b_size);
-		kunmap_atomic(source);
-
-		/*
-		 * Now that the frozen data is saved off, we need to store
-		 * any matching triggers.
-		 */
-		jh->b_frozen_triggers = jh->b_triggers;
-	}
 	jbd_unlock_bh_state(bh);
 
 	/*
@@ -996,6 +1008,59 @@ out:
 	return error;
 }
 
+/* Fast check whether buffer is already attached to the required transaction */
+static bool jbd2_write_access_granted(handle_t *handle, struct buffer_head *bh,
+							bool undo)
+{
+	struct journal_head *jh;
+	bool ret = false;
+
+	/* Dirty buffers require special handling... */
+	if (buffer_dirty(bh))
+		return false;
+
+	/*
+	 * RCU protects us from dereferencing freed pages. So the checks we do
+	 * are guaranteed not to oops. However the jh slab object can get freed
+	 * & reallocated while we work with it. So we have to be careful. When
+	 * we see jh attached to the running transaction, we know it must stay
+	 * so until the transaction is committed. Thus jh won't be freed and
+	 * will be attached to the same bh while we run.  However it can
+	 * happen jh gets freed, reallocated, and attached to the transaction
+	 * just after we get pointer to it from bh. So we have to be careful
+	 * and recheck jh still belongs to our bh before we return success.
+	 */
+	rcu_read_lock();
+	if (!buffer_jbd(bh))
+		goto out;
+	/* This should be bh2jh() but that doesn't work with inline functions */
+	jh = READ_ONCE(bh->b_private);
+	if (!jh)
+		goto out;
+	/* For undo access buffer must have data copied */
+	if (undo && !jh->b_committed_data)
+		goto out;
+	if (jh->b_transaction != handle->h_transaction &&
+	    jh->b_next_transaction != handle->h_transaction)
+		goto out;
+	/*
+	 * There are two reasons for the barrier here:
+	 * 1) Make sure to fetch b_bh after we did previous checks so that we
+	 * detect when jh went through free, realloc, attach to transaction
+	 * while we were checking. Paired with implicit barrier in that path.
+	 * 2) So that access to bh done after jbd2_write_access_granted()
+	 * doesn't get reordered and see inconsistent state of concurrent
+	 * do_get_write_access().
+	 */
+	smp_mb();
+	if (unlikely(jh->b_bh != bh))
+		goto out;
+	ret = true;
+out:
+	rcu_read_unlock();
+	return ret;
+}
+
 /**
  * int jbd2_journal_get_write_access() - notify intent to modify a buffer for metadata (not data) update.
  * @handle: transaction to add buffer modifications to
@@ -1009,9 +1074,13 @@ out:
 
 int jbd2_journal_get_write_access(handle_t *handle, struct buffer_head *bh)
 {
-	struct journal_head *jh = jbd2_journal_add_journal_head(bh);
+	struct journal_head *jh;
 	int rc;
 
+	if (jbd2_write_access_granted(handle, bh, false))
+		return 0;
+
+	jh = jbd2_journal_add_journal_head(bh);
 	/* We do not want to get caught playing with fields which the
 	 * log thread also manipulates.  Make sure that the buffer
 	 * completes any outstanding IO before proceeding. */
@@ -1141,11 +1210,14 @@ out:
 int jbd2_journal_get_undo_access(handle_t *handle, struct buffer_head *bh)
 {
 	int err;
-	struct journal_head *jh = jbd2_journal_add_journal_head(bh);
+	struct journal_head *jh;
 	char *committed_data = NULL;
 
 	JBUFFER_TRACE(jh, "entry");
+	if (jbd2_write_access_granted(handle, bh, true))
+		return 0;
 
+	jh = jbd2_journal_add_journal_head(bh);
 	/*
 	 * Do this first --- it can drop the journal lock, so we want to
 	 * make sure that obtaining the committed_data is done
@@ -1230,8 +1302,6 @@ void jbd2_buffer_abort_trigger(struct journal_head *jh,
 	triggers->t_abort(triggers, jh2bh(jh));
 }
 
-
-
 /**
  * int jbd2_journal_dirty_metadata() -  mark a buffer as containing dirty metadata
  * @handle: transaction to add buffer to.
@@ -1264,12 +1334,41 @@ int jbd2_journal_dirty_metadata(handle_t *handle, struct buffer_head *bh)
 
 	if (is_handle_aborted(handle))
 		return -EROFS;
-	journal = transaction->t_journal;
-	jh = jbd2_journal_grab_journal_head(bh);
-	if (!jh) {
+	if (!buffer_jbd(bh)) {
 		ret = -EUCLEAN;
 		goto out;
 	}
+	/*
+	 * We don't grab jh reference here since the buffer must be part
+	 * of the running transaction.
+	 */
+	jh = bh2jh(bh);
+	/*
+	 * This and the following assertions are unreliable since we may see jh
+	 * in inconsistent state unless we grab bh_state lock. But this is
+	 * crucial to catch bugs so let's do a reliable check until the
+	 * lockless handling is fully proven.
+	 */
+	if (jh->b_transaction != transaction &&
+	    jh->b_next_transaction != transaction) {
+		jbd_lock_bh_state(bh);
+		J_ASSERT_JH(jh, jh->b_transaction == transaction ||
+				jh->b_next_transaction == transaction);
+		jbd_unlock_bh_state(bh);
+	}
+	if (jh->b_modified == 1) {
+		/* If it's in our transaction it must be in BJ_Metadata list. */
+		if (jh->b_transaction == transaction &&
+		    jh->b_jlist != BJ_Metadata) {
+			jbd_lock_bh_state(bh);
+			J_ASSERT_JH(jh, jh->b_transaction != transaction ||
+					jh->b_jlist == BJ_Metadata);
+			jbd_unlock_bh_state(bh);
+		}
+		goto out;
+	}
+
+	journal = transaction->t_journal;
 	jbd_debug(5, "journal_head %p\n", jh);
 	JBUFFER_TRACE(jh, "entry");
 
@@ -1360,7 +1459,6 @@ int jbd2_journal_dirty_metadata(handle_t *handle, struct buffer_head *bh)
 	spin_unlock(&journal->j_list_lock);
 out_unlock_bh:
 	jbd_unlock_bh_state(bh);
-	jbd2_journal_put_journal_head(jh);
 out:
 	JBUFFER_TRACE(jh, "exit");
 	return ret;
@@ -1843,8 +1941,8 @@ out:
  * @journal: journal for operation
  * @page: to try and free
  * @gfp_mask: we use the mask to detect how hard should we try to release
- * buffers. If __GFP_WAIT and __GFP_FS is set, we wait for commit code to
- * release the buffers.
+ * buffers. If __GFP_DIRECT_RECLAIM and __GFP_FS is set, we wait for commit
+ * code to release the buffers.
  *
  *
  * For all the buffers on this page,
@@ -2058,6 +2156,7 @@ static int journal_unmap_buffer(journal_t *journal, struct buffer_head *bh,
 
 		if (!buffer_dirty(bh)) {
 			/* bdflush has written it.  We can drop it now */
+			__jbd2_journal_remove_checkpoint(jh);
 			goto zap_buffer;
 		}
 
@@ -2087,6 +2186,7 @@ static int journal_unmap_buffer(journal_t *journal, struct buffer_head *bh,
 				/* The orphan record's transaction has
 				 * committed.  We can cleanse this buffer */
 				clear_buffer_jbddirty(bh);
+				__jbd2_journal_remove_checkpoint(jh);
 				goto zap_buffer;
 			}
 		}