Merge branch 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel…

…/git/tytso/ext4

* 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tytso/ext4:
  jbd2: call __jbd2_log_start_commit with j_state_lock write locked
  ext4: serialize unaligned asynchronous DIO
  ext4: make grpinfo slab cache names static
  ext4: Fix data corruption with multi-block writepages support
  ext4: fix up ext4 error handling
  ext4: unregister features interface on module unload
  ext4: fix panic on module unload when stopping lazyinit thread

fs/ext4/ext4.h

@@ -848,6 +848,7 @@ struct ext4_inode_info {
 	atomic_t i_ioend_count;	/* Number of outstanding io_end structs */
 	/* current io_end structure for async DIO write*/
 	ext4_io_end_t *cur_aio_dio;
+	atomic_t i_aiodio_unwritten; /* Nr. of inflight conversions pending */
 
 	spinlock_t i_block_reservation_lock;
 
@@ -2119,6 +2120,15 @@ static inline void set_bitmap_uptodate(struct buffer_head *bh)
 
 #define in_range(b, first, len)	((b) >= (first) && (b) <= (first) + (len) - 1)
 
+/* For ioend & aio unwritten conversion wait queues */
+#define EXT4_WQ_HASH_SZ		37
+#define ext4_ioend_wq(v)   (&ext4__ioend_wq[((unsigned long)(v)) %\
+					    EXT4_WQ_HASH_SZ])
+#define ext4_aio_mutex(v)  (&ext4__aio_mutex[((unsigned long)(v)) %\
+					     EXT4_WQ_HASH_SZ])
+extern wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ];
+extern struct mutex ext4__aio_mutex[EXT4_WQ_HASH_SZ];
+
 #endif	/* __KERNEL__ */
 
 #endif	/* _EXT4_H */

fs/ext4/extents.c

@@ -3174,9 +3174,10 @@ ext4_ext_handle_uninitialized_extents(handle_t *handle, struct inode *inode,
 		 * that this IO needs to convertion to written when IO is
 		 * completed
 		 */
-		if (io)
+		if (io && !(io->flag & EXT4_IO_END_UNWRITTEN)) {
 			io->flag = EXT4_IO_END_UNWRITTEN;
-		else
+			atomic_inc(&EXT4_I(inode)->i_aiodio_unwritten);
+		} else
 			ext4_set_inode_state(inode, EXT4_STATE_DIO_UNWRITTEN);
 		if (ext4_should_dioread_nolock(inode))
 			map->m_flags |= EXT4_MAP_UNINIT;
@@ -3463,9 +3464,10 @@ int ext4_ext_map_blocks(handle_t *handle, struct inode *inode,
 		 * that we need to perform convertion when IO is done.
 		 */
 		if ((flags & EXT4_GET_BLOCKS_PRE_IO)) {
-			if (io)
+			if (io && !(io->flag & EXT4_IO_END_UNWRITTEN)) {
 				io->flag = EXT4_IO_END_UNWRITTEN;
-			else
+				atomic_inc(&EXT4_I(inode)->i_aiodio_unwritten);
+			} else
 				ext4_set_inode_state(inode,
 						     EXT4_STATE_DIO_UNWRITTEN);
 		}

fs/ext4/file.c

@@ -55,11 +55,47 @@ static int ext4_release_file(struct inode *inode, struct file *filp)
 	return 0;
 }
 
+static void ext4_aiodio_wait(struct inode *inode)
+{
+	wait_queue_head_t *wq = ext4_ioend_wq(inode);
+
+	wait_event(*wq, (atomic_read(&EXT4_I(inode)->i_aiodio_unwritten) == 0));
+}
+
+/*
+ * This tests whether the IO in question is block-aligned or not.
+ * Ext4 utilizes unwritten extents when hole-filling during direct IO, and they
+ * are converted to written only after the IO is complete.  Until they are
+ * mapped, these blocks appear as holes, so dio_zero_block() will assume that
+ * it needs to zero out portions of the start and/or end block.  If 2 AIO
+ * threads are at work on the same unwritten block, they must be synchronized
+ * or one thread will zero the other's data, causing corruption.
+ */
+static int
+ext4_unaligned_aio(struct inode *inode, const struct iovec *iov,
+		   unsigned long nr_segs, loff_t pos)
+{
+	struct super_block *sb = inode->i_sb;
+	int blockmask = sb->s_blocksize - 1;
+	size_t count = iov_length(iov, nr_segs);
+	loff_t final_size = pos + count;
+
+	if (pos >= inode->i_size)
+		return 0;
+
+	if ((pos & blockmask) || (final_size & blockmask))
+		return 1;
+
+	return 0;
+}
+
 static ssize_t
 ext4_file_write(struct kiocb *iocb, const struct iovec *iov,
 		unsigned long nr_segs, loff_t pos)
 {
 	struct inode *inode = iocb->ki_filp->f_path.dentry->d_inode;
+	int unaligned_aio = 0;
+	int ret;
 
 	/*
 	 * If we have encountered a bitmap-format file, the size limit
@@ -78,9 +114,31 @@ ext4_file_write(struct kiocb *iocb, const struct iovec *iov,
 			nr_segs = iov_shorten((struct iovec *)iov, nr_segs,
 					      sbi->s_bitmap_maxbytes - pos);
 		}
+	} else if (unlikely((iocb->ki_filp->f_flags & O_DIRECT) &&
+		   !is_sync_kiocb(iocb))) {
+		unaligned_aio = ext4_unaligned_aio(inode, iov, nr_segs, pos);
 	}
 
-	return generic_file_aio_write(iocb, iov, nr_segs, pos);
+	/* Unaligned direct AIO must be serialized; see comment above */
+	if (unaligned_aio) {
+		static unsigned long unaligned_warn_time;
+
+		/* Warn about this once per day */
+		if (printk_timed_ratelimit(&unaligned_warn_time, 60*60*24*HZ))
+			ext4_msg(inode->i_sb, KERN_WARNING,
+				 "Unaligned AIO/DIO on inode %ld by %s; "
+				 "performance will be poor.",
+				 inode->i_ino, current->comm);
+		mutex_lock(ext4_aio_mutex(inode));
+		ext4_aiodio_wait(inode);
+	}
+
+	ret = generic_file_aio_write(iocb, iov, nr_segs, pos);
+
+	if (unaligned_aio)
+		mutex_unlock(ext4_aio_mutex(inode));
+
+	return ret;
 }
 
 static const struct vm_operations_struct ext4_file_vm_ops = {

fs/ext4/mballoc.c

@@ -342,10 +342,15 @@ static struct kmem_cache *ext4_free_ext_cachep;
 /* We create slab caches for groupinfo data structures based on the
  * superblock block size.  There will be one per mounted filesystem for
  * each unique s_blocksize_bits */
-#define NR_GRPINFO_CACHES	\
-	(EXT4_MAX_BLOCK_LOG_SIZE - EXT4_MIN_BLOCK_LOG_SIZE + 1)
+#define NR_GRPINFO_CACHES 8
 static struct kmem_cache *ext4_groupinfo_caches[NR_GRPINFO_CACHES];
 
+static const char *ext4_groupinfo_slab_names[NR_GRPINFO_CACHES] = {
+	"ext4_groupinfo_1k", "ext4_groupinfo_2k", "ext4_groupinfo_4k",
+	"ext4_groupinfo_8k", "ext4_groupinfo_16k", "ext4_groupinfo_32k",
+	"ext4_groupinfo_64k", "ext4_groupinfo_128k"
+};
+
 static void ext4_mb_generate_from_pa(struct super_block *sb, void *bitmap,
 					ext4_group_t group);
 static void ext4_mb_generate_from_freelist(struct super_block *sb, void *bitmap,
@@ -2414,16 +2419,62 @@ static int ext4_mb_init_backend(struct super_block *sb)
 	return -ENOMEM;
 }
 
+static void ext4_groupinfo_destroy_slabs(void)
+{
+	int i;
+
+	for (i = 0; i < NR_GRPINFO_CACHES; i++) {
+		if (ext4_groupinfo_caches[i])
+			kmem_cache_destroy(ext4_groupinfo_caches[i]);
+		ext4_groupinfo_caches[i] = NULL;
+	}
+}
+
+static int ext4_groupinfo_create_slab(size_t size)
+{
+	static DEFINE_MUTEX(ext4_grpinfo_slab_create_mutex);
+	int slab_size;
+	int blocksize_bits = order_base_2(size);
+	int cache_index = blocksize_bits - EXT4_MIN_BLOCK_LOG_SIZE;
+	struct kmem_cache *cachep;
+
+	if (cache_index >= NR_GRPINFO_CACHES)
+		return -EINVAL;
+
+	if (unlikely(cache_index < 0))
+		cache_index = 0;
+
+	mutex_lock(&ext4_grpinfo_slab_create_mutex);
+	if (ext4_groupinfo_caches[cache_index]) {
+		mutex_unlock(&ext4_grpinfo_slab_create_mutex);
+		return 0;	/* Already created */
+	}
+
+	slab_size = offsetof(struct ext4_group_info,
+				bb_counters[blocksize_bits + 2]);
+
+	cachep = kmem_cache_create(ext4_groupinfo_slab_names[cache_index],
+					slab_size, 0, SLAB_RECLAIM_ACCOUNT,
+					NULL);
+
+	mutex_unlock(&ext4_grpinfo_slab_create_mutex);
+	if (!cachep) {
+		printk(KERN_EMERG "EXT4: no memory for groupinfo slab cache\n");
+		return -ENOMEM;
+	}
+
+	ext4_groupinfo_caches[cache_index] = cachep;
+
+	return 0;
+}
+
 int ext4_mb_init(struct super_block *sb, int needs_recovery)
 {
 	struct ext4_sb_info *sbi = EXT4_SB(sb);
 	unsigned i, j;
 	unsigned offset;
 	unsigned max;
 	int ret;
-	int cache_index;
-	struct kmem_cache *cachep;
-	char *namep = NULL;
 
 	i = (sb->s_blocksize_bits + 2) * sizeof(*sbi->s_mb_offsets);
 
@@ -2440,30 +2491,9 @@ int ext4_mb_init(struct super_block *sb, int needs_recovery)
 		goto out;
 	}
 
-	cache_index = sb->s_blocksize_bits - EXT4_MIN_BLOCK_LOG_SIZE;
-	cachep = ext4_groupinfo_caches[cache_index];
-	if (!cachep) {
-		char name[32];
-		int len = offsetof(struct ext4_group_info,
-					bb_counters[sb->s_blocksize_bits + 2]);
-
-		sprintf(name, "ext4_groupinfo_%d", sb->s_blocksize_bits);
-		namep = kstrdup(name, GFP_KERNEL);
-		if (!namep) {
-			ret = -ENOMEM;
-			goto out;
-		}
-
-		/* Need to free the kmem_cache_name() when we
-		 * destroy the slab */
-		cachep = kmem_cache_create(namep, len, 0,
-					     SLAB_RECLAIM_ACCOUNT, NULL);
-		if (!cachep) {
-			ret = -ENOMEM;
-			goto out;
-		}
-		ext4_groupinfo_caches[cache_index] = cachep;
-	}
+	ret = ext4_groupinfo_create_slab(sb->s_blocksize);
+	if (ret < 0)
+		goto out;
 
 	/* order 0 is regular bitmap */
 	sbi->s_mb_maxs[0] = sb->s_blocksize << 3;
@@ -2520,7 +2550,6 @@ int ext4_mb_init(struct super_block *sb, int needs_recovery)
 	if (ret) {
 		kfree(sbi->s_mb_offsets);
 		kfree(sbi->s_mb_maxs);
-		kfree(namep);
 	}
 	return ret;
 }
@@ -2734,7 +2763,6 @@ int __init ext4_init_mballoc(void)
 
 void ext4_exit_mballoc(void)
 {
-	int i;
 	/*
 	 * Wait for completion of call_rcu()'s on ext4_pspace_cachep
 	 * before destroying the slab cache.
@@ -2743,15 +2771,7 @@ void ext4_exit_mballoc(void)
 	kmem_cache_destroy(ext4_pspace_cachep);
 	kmem_cache_destroy(ext4_ac_cachep);
 	kmem_cache_destroy(ext4_free_ext_cachep);
-
-	for (i = 0; i < NR_GRPINFO_CACHES; i++) {
-		struct kmem_cache *cachep = ext4_groupinfo_caches[i];
-		if (cachep) {
-			char *name = (char *)kmem_cache_name(cachep);
-			kmem_cache_destroy(cachep);
-			kfree(name);
-		}
-	}
+	ext4_groupinfo_destroy_slabs();
 	ext4_remove_debugfs_entry();
 }