Commit e9e3bcec authored by Eric Sandeen's avatar Eric Sandeen Committed by Theodore Ts'o

ext4: serialize unaligned asynchronous DIO

ext4 has a data corruption case when doing non-block-aligned
asynchronous direct IO into a sparse file, as demonstrated
by xfstest 240.

The root cause is that while ext4 preallocates space in the
hole, mappings of that space still look "new" and 
dio_zero_block() will zero out the unwritten portions.  When
more than one AIO thread is going, they both find this "new"
block and race to zero out their portion; this is uncoordinated
and causes data corruption.

Dave Chinner fixed this for xfs by simply serializing all
unaligned asynchronous direct IO.  I've done the same here.
The difference is that we only wait on conversions, not all IO.
This is a very big hammer, and I'm not very pleased with
stuffing this into ext4_file_write().  But since ext4 is
DIO_LOCKING, we need to serialize it at this high level.

I tried to move this into ext4_ext_direct_IO, but by then
we have the i_mutex already, and we will wait on the
work queue to do conversions - which must also take the
i_mutex.  So that won't work.

This was originally exposed by qemu-kvm installing to
a raw disk image with a normal sector-63 alignment.  I've
tested a backport of this patch with qemu, and it does
avoid the corruption.  It is also quite a lot slower
(14 min for package installs, vs. 8 min for well-aligned)
but I'll take slow correctness over fast corruption any day.

Mingming suggested that we can track outstanding
conversions, and wait on those so that non-sparse
files won't be affected, and I've implemented that here;
unaligned AIO to nonsparse files won't take a perf hit.

[tytso@mit.edu: Keep the mutex as a hashed array instead
 of bloating the ext4 inode]

[tytso@mit.edu: Fix up namespace issues so that global
 variables are protected with an "ext4_" prefix.]
Signed-off-by: default avatarEric Sandeen <sandeen@redhat.com>
Signed-off-by: default avatar"Theodore Ts'o" <tytso@mit.edu>
parent 2892c15d
...@@ -848,6 +848,7 @@ struct ext4_inode_info { ...@@ -848,6 +848,7 @@ struct ext4_inode_info {
atomic_t i_ioend_count; /* Number of outstanding io_end structs */ atomic_t i_ioend_count; /* Number of outstanding io_end structs */
/* current io_end structure for async DIO write*/ /* current io_end structure for async DIO write*/
ext4_io_end_t *cur_aio_dio; ext4_io_end_t *cur_aio_dio;
atomic_t i_aiodio_unwritten; /* Nr. of inflight conversions pending */
spinlock_t i_block_reservation_lock; spinlock_t i_block_reservation_lock;
...@@ -2119,6 +2120,15 @@ static inline void set_bitmap_uptodate(struct buffer_head *bh) ...@@ -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) #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 /* __KERNEL__ */
#endif /* _EXT4_H */ #endif /* _EXT4_H */
...@@ -3174,9 +3174,10 @@ ext4_ext_handle_uninitialized_extents(handle_t *handle, struct inode *inode, ...@@ -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 * that this IO needs to convertion to written when IO is
* completed * completed
*/ */
if (io) if (io && !(io->flag & EXT4_IO_END_UNWRITTEN)) {
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); ext4_set_inode_state(inode, EXT4_STATE_DIO_UNWRITTEN);
if (ext4_should_dioread_nolock(inode)) if (ext4_should_dioread_nolock(inode))
map->m_flags |= EXT4_MAP_UNINIT; map->m_flags |= EXT4_MAP_UNINIT;
...@@ -3463,9 +3464,10 @@ int ext4_ext_map_blocks(handle_t *handle, struct inode *inode, ...@@ -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. * that we need to perform convertion when IO is done.
*/ */
if ((flags & EXT4_GET_BLOCKS_PRE_IO)) { if ((flags & EXT4_GET_BLOCKS_PRE_IO)) {
if (io) if (io && !(io->flag & EXT4_IO_END_UNWRITTEN)) {
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_set_inode_state(inode,
EXT4_STATE_DIO_UNWRITTEN); EXT4_STATE_DIO_UNWRITTEN);
} }
......
...@@ -55,11 +55,47 @@ static int ext4_release_file(struct inode *inode, struct file *filp) ...@@ -55,11 +55,47 @@ static int ext4_release_file(struct inode *inode, struct file *filp)
return 0; 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 static ssize_t
ext4_file_write(struct kiocb *iocb, const struct iovec *iov, ext4_file_write(struct kiocb *iocb, const struct iovec *iov,
unsigned long nr_segs, loff_t pos) unsigned long nr_segs, loff_t pos)
{ {
struct inode *inode = iocb->ki_filp->f_path.dentry->d_inode; 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 * 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, ...@@ -78,9 +114,31 @@ ext4_file_write(struct kiocb *iocb, const struct iovec *iov,
nr_segs = iov_shorten((struct iovec *)iov, nr_segs, nr_segs = iov_shorten((struct iovec *)iov, nr_segs,
sbi->s_bitmap_maxbytes - pos); 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 = { static const struct vm_operations_struct ext4_file_vm_ops = {
......
...@@ -32,14 +32,8 @@ ...@@ -32,14 +32,8 @@
static struct kmem_cache *io_page_cachep, *io_end_cachep; static struct kmem_cache *io_page_cachep, *io_end_cachep;
#define WQ_HASH_SZ 37
#define to_ioend_wq(v) (&ioend_wq[((unsigned long)v) % WQ_HASH_SZ])
static wait_queue_head_t ioend_wq[WQ_HASH_SZ];
int __init ext4_init_pageio(void) int __init ext4_init_pageio(void)
{ {
int i;
io_page_cachep = KMEM_CACHE(ext4_io_page, SLAB_RECLAIM_ACCOUNT); io_page_cachep = KMEM_CACHE(ext4_io_page, SLAB_RECLAIM_ACCOUNT);
if (io_page_cachep == NULL) if (io_page_cachep == NULL)
return -ENOMEM; return -ENOMEM;
...@@ -48,9 +42,6 @@ int __init ext4_init_pageio(void) ...@@ -48,9 +42,6 @@ int __init ext4_init_pageio(void)
kmem_cache_destroy(io_page_cachep); kmem_cache_destroy(io_page_cachep);
return -ENOMEM; return -ENOMEM;
} }
for (i = 0; i < WQ_HASH_SZ; i++)
init_waitqueue_head(&ioend_wq[i]);
return 0; return 0;
} }
...@@ -62,7 +53,7 @@ void ext4_exit_pageio(void) ...@@ -62,7 +53,7 @@ void ext4_exit_pageio(void)
void ext4_ioend_wait(struct inode *inode) void ext4_ioend_wait(struct inode *inode)
{ {
wait_queue_head_t *wq = to_ioend_wq(inode); wait_queue_head_t *wq = ext4_ioend_wq(inode);
wait_event(*wq, (atomic_read(&EXT4_I(inode)->i_ioend_count) == 0)); wait_event(*wq, (atomic_read(&EXT4_I(inode)->i_ioend_count) == 0));
} }
...@@ -87,7 +78,7 @@ void ext4_free_io_end(ext4_io_end_t *io) ...@@ -87,7 +78,7 @@ void ext4_free_io_end(ext4_io_end_t *io)
for (i = 0; i < io->num_io_pages; i++) for (i = 0; i < io->num_io_pages; i++)
put_io_page(io->pages[i]); put_io_page(io->pages[i]);
io->num_io_pages = 0; io->num_io_pages = 0;
wq = to_ioend_wq(io->inode); wq = ext4_ioend_wq(io->inode);
if (atomic_dec_and_test(&EXT4_I(io->inode)->i_ioend_count) && if (atomic_dec_and_test(&EXT4_I(io->inode)->i_ioend_count) &&
waitqueue_active(wq)) waitqueue_active(wq))
wake_up_all(wq); wake_up_all(wq);
...@@ -102,6 +93,7 @@ int ext4_end_io_nolock(ext4_io_end_t *io) ...@@ -102,6 +93,7 @@ int ext4_end_io_nolock(ext4_io_end_t *io)
struct inode *inode = io->inode; struct inode *inode = io->inode;
loff_t offset = io->offset; loff_t offset = io->offset;
ssize_t size = io->size; ssize_t size = io->size;
wait_queue_head_t *wq;
int ret = 0; int ret = 0;
ext4_debug("ext4_end_io_nolock: io 0x%p from inode %lu,list->next 0x%p," ext4_debug("ext4_end_io_nolock: io 0x%p from inode %lu,list->next 0x%p,"
...@@ -126,7 +118,16 @@ int ext4_end_io_nolock(ext4_io_end_t *io) ...@@ -126,7 +118,16 @@ int ext4_end_io_nolock(ext4_io_end_t *io)
if (io->iocb) if (io->iocb)
aio_complete(io->iocb, io->result, 0); aio_complete(io->iocb, io->result, 0);
/* clear the DIO AIO unwritten flag */ /* clear the DIO AIO unwritten flag */
if (io->flag & EXT4_IO_END_UNWRITTEN) {
io->flag &= ~EXT4_IO_END_UNWRITTEN; io->flag &= ~EXT4_IO_END_UNWRITTEN;
/* Wake up anyone waiting on unwritten extent conversion */
wq = ext4_ioend_wq(io->inode);
if (atomic_dec_and_test(&EXT4_I(inode)->i_aiodio_unwritten) &&
waitqueue_active(wq)) {
wake_up_all(wq);
}
}
return ret; return ret;
} }
......
...@@ -833,6 +833,7 @@ static struct inode *ext4_alloc_inode(struct super_block *sb) ...@@ -833,6 +833,7 @@ static struct inode *ext4_alloc_inode(struct super_block *sb)
ei->i_sync_tid = 0; ei->i_sync_tid = 0;
ei->i_datasync_tid = 0; ei->i_datasync_tid = 0;
atomic_set(&ei->i_ioend_count, 0); atomic_set(&ei->i_ioend_count, 0);
atomic_set(&ei->i_aiodio_unwritten, 0);
return &ei->vfs_inode; return &ei->vfs_inode;
} }
...@@ -4800,11 +4801,21 @@ static void ext4_exit_feat_adverts(void) ...@@ -4800,11 +4801,21 @@ static void ext4_exit_feat_adverts(void)
kfree(ext4_feat); kfree(ext4_feat);
} }
/* Shared across all ext4 file systems */
wait_queue_head_t ext4__ioend_wq[EXT4_WQ_HASH_SZ];
struct mutex ext4__aio_mutex[EXT4_WQ_HASH_SZ];
static int __init ext4_init_fs(void) static int __init ext4_init_fs(void)
{ {
int err; int i, err;
ext4_check_flag_values(); ext4_check_flag_values();
for (i = 0; i < EXT4_WQ_HASH_SZ; i++) {
mutex_init(&ext4__aio_mutex[i]);
init_waitqueue_head(&ext4__ioend_wq[i]);
}
err = ext4_init_pageio(); err = ext4_init_pageio();
if (err) if (err)
return err; return err;
......
Markdown is supported
0%
or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment