Merge branch 'for-next' of git://git.kernel.org/pub/scm/linux/kernel/git/hch/vfs-queue

* 'for-next' of git://git.kernel.org/pub/scm/linux/kernel/git/hch/vfs-queue: (21 commits) leases: fix write-open/read-lease race nfs: drop unnecessary locking in llseek ext4: replace cut'n'pasted llseek code with generic_file_llseek_size vfs: add generic_file_llseek_size vfs: do (nearly) lockless generic_file_llseek direct-io: merge direct_io_walker into __blockdev_direct_IO direct-io: inline the complete submission path direct-io: separate map_bh from dio direct-io: use a slab cache for struct dio direct-io: rearrange fields in dio/dio_submit to avoid holes direct-io: fix a wrong comment direct-io: separate fields only used in the submission path from struct dio vfs: fix spinning prevention in prune_icache_sb vfs: add a comment to inode_permission() vfs: pass all mask flags check_acl and posix_acl_permission vfs: add hex format for MAY_* flag values vfs: indicate that the permission functions take all the MAY_* flags compat: sync compat_stats with statfs. vfs: add "device" tag to /proc/self/mountstats cleanup: vfs: small comment fix for block_invalidatepage ... Fix up trivial conflict in fs/gfs2/file.c (llseek changes)

Merge branch 'for-next' of git://git.kernel.org/pub/scm/linux/kernel/git/hch/vfs-queue
* 'for-next' of git://git.kernel.org/pub/scm/linux/kernel/git/hch/vfs-queue: (21 commits) leases: fix write-open/read-lease race nfs: drop unnecessary locking in llseek ext4: replace cut'n'pasted llseek code with generic_file_llseek_size vfs: add generic_file_llseek_size vfs: do (nearly) lockless generic_file_llseek direct-io: merge direct_io_walker into __blockdev_direct_IO direct-io: inline the complete submission path direct-io: separate map_bh from dio direct-io: use a slab cache for struct dio direct-io: rearrange fields in dio/dio_submit to avoid holes direct-io: fix a wrong comment direct-io: separate fields only used in the submission path from struct dio vfs: fix spinning prevention in prune_icache_sb vfs: add a comment to inode_permission() vfs: pass all mask flags check_acl and posix_acl_permission vfs: add hex format for MAY_* flag values vfs: indicate that the permission functions take all the MAY_* flags compat: sync compat_stats with statfs. vfs: add "device" tag to /proc/self/mountstats cleanup: vfs: small comment fix for block_invalidatepage ... Fix up trivial conflict in fs/gfs2/file.c (llseek changes)
Linus Torvalds
2 parents f793f29611 f3c7691e8d
Showing 22 changed files Side-by-side Diff
arch/mips/include/asm/compat.h
arch/parisc/include/asm/compat.h
arch/powerpc/include/asm/compat.h
arch/s390/include/asm/compat.h
arch/sparc/include/asm/compat.h
arch/x86/include/asm/compat.h
fs/btrfs/file.c
fs/buffer.c
fs/cifs/cifsfs.c
fs/compat.c
fs/direct-io.c
fs/ext4/file.c
fs/gfs2/file.c
fs/inode.c
fs/namei.c
fs/namespace.c
fs/nfs/file.c
fs/open.c
fs/posix_acl.c
fs/read_write.c
@@ -111,7 +111,8 @@
 	int		f_bavail;
 	compat_fsid_t	f_fsid;
 	int		f_namelen;
-	int		f_spare[6];
+	int		f_flags;
+	int		f_spare[5];
 };
  
 #define COMPAT_RLIM_INFINITY	0x7fffffffUL
@@ -105,7 +105,8 @@
 	__kernel_fsid_t	f_fsid;
 	s32		f_namelen;
 	s32		f_frsize;
-	s32		f_spare[5];
+	s32		f_flags;
+	s32		f_spare[4];
 };
  
 struct compat_sigcontext {
@@ -100,7 +100,8 @@
 	compat_fsid_t	f_fsid;
 	int		f_namelen;	/* SunOS ignores this field. */
 	int		f_frsize;
-	int		f_spare[5];
+	int		f_flags;
+	int		f_spare[4];
 };
  
 #define COMPAT_RLIM_OLD_INFINITY	0x7fffffff
@@ -131,7 +131,8 @@
 	compat_fsid_t	f_fsid;
 	s32		f_namelen;
 	s32		f_frsize;
-	s32		f_spare[6];
+	s32		f_flags;
+	s32		f_spare[5];
 };
  
 #define COMPAT_RLIM_OLD_INFINITY	0x7fffffff
@@ -134,7 +134,8 @@
 	compat_fsid_t	f_fsid;
 	int		f_namelen;	/* SunOS ignores this field. */
 	int		f_frsize;
-	int		f_spare[5];
+	int		f_flags;
+	int		f_spare[4];
 };
  
 #define COMPAT_RLIM_INFINITY 0x7fffffff
@@ -108,7 +108,8 @@
 	compat_fsid_t	f_fsid;
 	int		f_namelen;	/* SunOS ignores this field. */
 	int		f_frsize;
-	int		f_spare[5];
+	int		f_flags;
+	int		f_spare[4];
 };
  
 #define COMPAT_RLIM_OLD_INFINITY	0x7fffffff
@@ -1821,7 +1821,7 @@
 	switch (origin) {
 	case SEEK_END:
 	case SEEK_CUR:
-		offset = generic_file_llseek_unlocked(file, offset, origin);
+		offset = generic_file_llseek(file, offset, origin);
 		goto out;
 	case SEEK_DATA:
 	case SEEK_HOLE:
@@ -1470,13 +1470,13 @@
 }
  
 /**
- * block_invalidatepage - invalidate part of all of a buffer-backed page
+ * block_invalidatepage - invalidate part or all of a buffer-backed page
  *
  * @page: the page which is affected
  * @offset: the index of the truncation point
  *
  * block_invalidatepage() is called when all or part of the page has become
- * invalidatedby a truncate operation.
+ * invalidated by a truncate operation.
  *
  * block_invalidatepage() does not have to release all buffers, but it must
  * ensure that no dirty buffer is left outside @offset and that no I/O
@@ -730,7 +730,7 @@
 		if (rc < 0)
 			return (loff_t)rc;
 	}
-	return generic_file_llseek_unlocked(file, offset, origin);
+	return generic_file_llseek(file, offset, origin);
 }
  
 static int cifs_setlease(struct file *file, long arg, struct file_lock **lease)
@@ -246,11 +246,8 @@
 	    __put_user(kbuf->f_fsid.val[0], &ubuf->f_fsid.val[0]) ||
 	    __put_user(kbuf->f_fsid.val[1], &ubuf->f_fsid.val[1]) ||
 	    __put_user(kbuf->f_frsize, &ubuf->f_frsize) ||
-	    __put_user(0, &ubuf->f_spare[0]) || 
-	    __put_user(0, &ubuf->f_spare[1]) || 
-	    __put_user(0, &ubuf->f_spare[2]) || 
-	    __put_user(0, &ubuf->f_spare[3]) || 
-	    __put_user(0, &ubuf->f_spare[4]))
+	    __put_user(kbuf->f_flags, &ubuf->f_flags) ||
+	    __clear_user(ubuf->f_spare, sizeof(ubuf->f_spare)))
 		return -EFAULT;
 	return 0;
 }
@@ -39,7 +39,7 @@
  
 /*
  * How many user pages to map in one call to get_user_pages().  This determines
- * the size of a structure on the stack.
+ * the size of a structure in the slab cache
  */
 #define DIO_PAGES	64
  
  
@@ -55,13 +55,10 @@
  * blocksize.
  */
  
-struct dio {
-	/* BIO submission state */
+/* dio_state only used in the submission path */
+
+struct dio_submit {
 	struct bio *bio;		/* bio under assembly */
-	struct inode *inode;
-	int rw;
-	loff_t i_size;			/* i_size when submitted */
-	int flags;			/* doesn't change */
 	unsigned blkbits;		/* doesn't change */
 	unsigned blkfactor;		/* When we're using an alignment which
 					   is finer than the filesystem's soft
  
  
  
  
@@ -76,18 +73,17 @@
 	sector_t block_in_file;		/* Current offset into the underlying
 					   file in dio_block units. */
 	unsigned blocks_available;	/* At block_in_file.  changes */
+	int reap_counter;		/* rate limit reaping */
 	sector_t final_block_in_request;/* doesn't change */
 	unsigned first_block_in_page;	/* doesn't change, Used only once */
 	int boundary;			/* prev block is at a boundary */
-	int reap_counter;		/* rate limit reaping */
 	get_block_t *get_block;		/* block mapping function */
-	dio_iodone_t *end_io;		/* IO completion function */
 	dio_submit_t *submit_io;	/* IO submition function */
+
 	loff_t logical_offset_in_bio;	/* current first logical block in bio */
 	sector_t final_block_in_bio;	/* current final block in bio + 1 */
 	sector_t next_block_for_io;	/* next block to be put under IO,
 					   in dio_blocks units */
-	struct buffer_head map_bh;	/* last get_block() result */
  
 	/*
 	 * Deferred addition of a page to the dio.  These variables are
@@ -100,18 +96,6 @@
 	sector_t cur_page_block;	/* Where it starts */
 	loff_t cur_page_fs_offset;	/* Offset in file */
  
-	/* BIO completion state */
-	spinlock_t bio_lock;		/* protects BIO fields below */
-	unsigned long refcount;		/* direct_io_worker() and bios */
-	struct bio *bio_list;		/* singly linked via bi_private */
-	struct task_struct *waiter;	/* waiting task (NULL if none) */
-
-	/* AIO related stuff */
-	struct kiocb *iocb;		/* kiocb */
-	int is_async;			/* is IO async ? */
-	int io_error;			/* IO error in completion path */
-	ssize_t result;                 /* IO result */
-
 	/*
 	 * Page fetching state. These variables belong to dio_refill_pages().
 	 */
  
  
  
  
@@ -125,16 +109,41 @@
 	 */
 	unsigned head;			/* next page to process */
 	unsigned tail;			/* last valid page + 1 */
+};
+
+/* dio_state communicated between submission path and end_io */
+struct dio {
+	int flags;			/* doesn't change */
+	int rw;
+	struct inode *inode;
+	loff_t i_size;			/* i_size when submitted */
+	dio_iodone_t *end_io;		/* IO completion function */
+
+	void *private;			/* copy from map_bh.b_private */
+
+	/* BIO completion state */
+	spinlock_t bio_lock;		/* protects BIO fields below */
 	int page_errors;		/* errno from get_user_pages() */
+	int is_async;			/* is IO async ? */
+	int io_error;			/* IO error in completion path */
+	unsigned long refcount;		/* direct_io_worker() and bios */
+	struct bio *bio_list;		/* singly linked via bi_private */
+	struct task_struct *waiter;	/* waiting task (NULL if none) */
  
+	/* AIO related stuff */
+	struct kiocb *iocb;		/* kiocb */
+	ssize_t result;                 /* IO result */
+
 	/*
 	 * pages[] (and any fields placed after it) are not zeroed out at
 	 * allocation time.  Don't add new fields after pages[] unless you
 	 * wish that they not be zeroed.
 	 */
 	struct page *pages[DIO_PAGES];	/* page buffer */
-};
+} ____cacheline_aligned_in_smp;
  
+static struct kmem_cache *dio_cache __read_mostly;
+
 static void __inode_dio_wait(struct inode *inode)
 {
 	wait_queue_head_t *wq = bit_waitqueue(&inode->i_state, __I_DIO_WAKEUP);
  
  
  
  
  
@@ -182,27 +191,27 @@
 /*
  * How many pages are in the queue?
  */
-static inline unsigned dio_pages_present(struct dio *dio)
+static inline unsigned dio_pages_present(struct dio_submit *sdio)
 {
-	return dio->tail - dio->head;
+	return sdio->tail - sdio->head;
 }
  
 /*
  * Go grab and pin some userspace pages.   Typically we'll get 64 at a time.
  */
-static int dio_refill_pages(struct dio *dio)
+static inline int dio_refill_pages(struct dio *dio, struct dio_submit *sdio)
 {
 	int ret;
 	int nr_pages;
  
-	nr_pages = min(dio->total_pages - dio->curr_page, DIO_PAGES);
+	nr_pages = min(sdio->total_pages - sdio->curr_page, DIO_PAGES);
 	ret = get_user_pages_fast(
-		dio->curr_user_address,		/* Where from? */
+		sdio->curr_user_address,		/* Where from? */
 		nr_pages,			/* How many pages? */
 		dio->rw == READ,		/* Write to memory? */
 		&dio->pages[0]);		/* Put results here */
  
-	if (ret < 0 && dio->blocks_available && (dio->rw & WRITE)) {
+	if (ret < 0 && sdio->blocks_available && (dio->rw & WRITE)) {
 		struct page *page = ZERO_PAGE(0);
 		/*
 		 * A memory fault, but the filesystem has some outstanding
  
@@ -213,17 +222,17 @@
 			dio->page_errors = ret;
 		page_cache_get(page);
 		dio->pages[0] = page;
-		dio->head = 0;
-		dio->tail = 1;
+		sdio->head = 0;
+		sdio->tail = 1;
 		ret = 0;
 		goto out;
 	}
  
 	if (ret >= 0) {
-		dio->curr_user_address += ret * PAGE_SIZE;
-		dio->curr_page += ret;
-		dio->head = 0;
-		dio->tail = ret;
+		sdio->curr_user_address += ret * PAGE_SIZE;
+		sdio->curr_page += ret;
+		sdio->head = 0;
+		sdio->tail = ret;
 		ret = 0;
 	}
 out:
  
  
  
  
@@ -236,17 +245,18 @@
  * decent number of pages, less frequently.  To provide nicer use of the
  * L1 cache.
  */
-static struct page *dio_get_page(struct dio *dio)
+static inline struct page *dio_get_page(struct dio *dio,
+		struct dio_submit *sdio)
 {
-	if (dio_pages_present(dio) == 0) {
+	if (dio_pages_present(sdio) == 0) {
 		int ret;
  
-		ret = dio_refill_pages(dio);
+		ret = dio_refill_pages(dio, sdio);
 		if (ret)
 			return ERR_PTR(ret);
-		BUG_ON(dio_pages_present(dio) == 0);
+		BUG_ON(dio_pages_present(sdio) == 0);
 	}
-	return dio->pages[dio->head++];
+	return dio->pages[sdio->head++];
 }
  
 /**
@@ -292,7 +302,7 @@
  
 	if (dio->end_io && dio->result) {
 		dio->end_io(dio->iocb, offset, transferred,
-			    dio->map_bh.b_private, ret, is_async);
+			    dio->private, ret, is_async);
 	} else {
 		if (is_async)
 			aio_complete(dio->iocb, ret, 0);
@@ -323,7 +333,7 @@
  
 	if (remaining == 0) {
 		dio_complete(dio, dio->iocb->ki_pos, 0, true);
-		kfree(dio);
+		kmem_cache_free(dio_cache, dio);
 	}
 }
  
@@ -367,9 +377,10 @@
 }
 EXPORT_SYMBOL_GPL(dio_end_io);
  
-static void
-dio_bio_alloc(struct dio *dio, struct block_device *bdev,
-		sector_t first_sector, int nr_vecs)
+static inline void
+dio_bio_alloc(struct dio *dio, struct dio_submit *sdio,
+	      struct block_device *bdev,
+	      sector_t first_sector, int nr_vecs)
 {
 	struct bio *bio;
  
@@ -386,8 +397,8 @@
 	else
 		bio->bi_end_io = dio_bio_end_io;
  
-	dio->bio = bio;
-	dio->logical_offset_in_bio = dio->cur_page_fs_offset;
+	sdio->bio = bio;
+	sdio->logical_offset_in_bio = sdio->cur_page_fs_offset;
 }
  
 /*
  
@@ -397,9 +408,9 @@
  *
  * bios hold a dio reference between submit_bio and ->end_io.
  */
-static void dio_bio_submit(struct dio *dio)
+static inline void dio_bio_submit(struct dio *dio, struct dio_submit *sdio)
 {
-	struct bio *bio = dio->bio;
+	struct bio *bio = sdio->bio;
 	unsigned long flags;
  
 	bio->bi_private = dio;
  
  
  
@@ -411,24 +422,24 @@
 	if (dio->is_async && dio->rw == READ)
 		bio_set_pages_dirty(bio);
  
-	if (dio->submit_io)
-		dio->submit_io(dio->rw, bio, dio->inode,
-			       dio->logical_offset_in_bio);
+	if (sdio->submit_io)
+		sdio->submit_io(dio->rw, bio, dio->inode,
+			       sdio->logical_offset_in_bio);
 	else
 		submit_bio(dio->rw, bio);
  
-	dio->bio = NULL;
-	dio->boundary = 0;
-	dio->logical_offset_in_bio = 0;
+	sdio->bio = NULL;
+	sdio->boundary = 0;
+	sdio->logical_offset_in_bio = 0;
 }
  
 /*
  * Release any resources in case of a failure
  */
-static void dio_cleanup(struct dio *dio)
+static inline void dio_cleanup(struct dio *dio, struct dio_submit *sdio)
 {
-	while (dio_pages_present(dio))
-		page_cache_release(dio_get_page(dio));
+	while (dio_pages_present(sdio))
+		page_cache_release(dio_get_page(dio, sdio));
 }
  
 /*
  
@@ -518,11 +529,11 @@
  *
  * This also helps to limit the peak amount of pinned userspace memory.
  */
-static int dio_bio_reap(struct dio *dio)
+static inline int dio_bio_reap(struct dio *dio, struct dio_submit *sdio)
 {
 	int ret = 0;
  
-	if (dio->reap_counter++ >= 64) {
+	if (sdio->reap_counter++ >= 64) {
 		while (dio->bio_list) {
 			unsigned long flags;
 			struct bio *bio;
  
@@ -536,14 +547,14 @@
 			if (ret == 0)
 				ret = ret2;
 		}
-		dio->reap_counter = 0;
+		sdio->reap_counter = 0;
 	}
 	return ret;
 }
  
 /*
  * Call into the fs to map some more disk blocks.  We record the current number
- * of available blocks at dio->blocks_available.  These are in units of the
+ * of available blocks at sdio->blocks_available.  These are in units of the
  * fs blocksize, (1 << inode->i_blkbits).
  *
  * The fs is allowed to map lots of blocks at once.  If it wants to do that,
  
@@ -564,10 +575,10 @@
  * buffer_mapped().  However the direct-io code will only process holes one
  * block at a time - it will repeatedly call get_block() as it walks the hole.
  */
-static int get_more_blocks(struct dio *dio)
+static int get_more_blocks(struct dio *dio, struct dio_submit *sdio,
+			   struct buffer_head *map_bh)
 {
 	int ret;
-	struct buffer_head *map_bh = &dio->map_bh;
 	sector_t fs_startblk;	/* Into file, in filesystem-sized blocks */
 	unsigned long fs_count;	/* Number of filesystem-sized blocks */
 	unsigned long dio_count;/* Number of dio_block-sized blocks */
@@ -580,11 +591,11 @@
 	 */
 	ret = dio->page_errors;
 	if (ret == 0) {
-		BUG_ON(dio->block_in_file >= dio->final_block_in_request);
-		fs_startblk = dio->block_in_file >> dio->blkfactor;
-		dio_count = dio->final_block_in_request - dio->block_in_file;
-		fs_count = dio_count >> dio->blkfactor;
-		blkmask = (1 << dio->blkfactor) - 1;
+		BUG_ON(sdio->block_in_file >= sdio->final_block_in_request);
+		fs_startblk = sdio->block_in_file >> sdio->blkfactor;
+		dio_count = sdio->final_block_in_request - sdio->block_in_file;
+		fs_count = dio_count >> sdio->blkfactor;
+		blkmask = (1 << sdio->blkfactor) - 1;
 		if (dio_count & blkmask)	
 			fs_count++;
  
  
  
@@ -604,13 +615,16 @@
 		 */
 		create = dio->rw & WRITE;
 		if (dio->flags & DIO_SKIP_HOLES) {
-			if (dio->block_in_file < (i_size_read(dio->inode) >>
-							dio->blkbits))
+			if (sdio->block_in_file < (i_size_read(dio->inode) >>
+							sdio->blkbits))
 				create = 0;
 		}
  
-		ret = (*dio->get_block)(dio->inode, fs_startblk,
+		ret = (*sdio->get_block)(dio->inode, fs_startblk,
 						map_bh, create);
+
+		/* Store for completion */
+		dio->private = map_bh->b_private;
 	}
 	return ret;
 }
  
  
  
@@ -618,20 +632,21 @@
 /*
  * There is no bio.  Make one now.
  */
-static int dio_new_bio(struct dio *dio, sector_t start_sector)
+static inline int dio_new_bio(struct dio *dio, struct dio_submit *sdio,
+		sector_t start_sector, struct buffer_head *map_bh)
 {
 	sector_t sector;
 	int ret, nr_pages;
  
-	ret = dio_bio_reap(dio);
+	ret = dio_bio_reap(dio, sdio);
 	if (ret)
 		goto out;
-	sector = start_sector << (dio->blkbits - 9);
-	nr_pages = min(dio->pages_in_io, bio_get_nr_vecs(dio->map_bh.b_bdev));
+	sector = start_sector << (sdio->blkbits - 9);
+	nr_pages = min(sdio->pages_in_io, bio_get_nr_vecs(map_bh->b_bdev));
 	nr_pages = min(nr_pages, BIO_MAX_PAGES);
 	BUG_ON(nr_pages <= 0);
-	dio_bio_alloc(dio, dio->map_bh.b_bdev, sector, nr_pages);
-	dio->boundary = 0;
+	dio_bio_alloc(dio, sdio, map_bh->b_bdev, sector, nr_pages);
+	sdio->boundary = 0;
 out:
 	return ret;
 }
  
  
@@ -643,21 +658,21 @@
  *
  * Return zero on success.  Non-zero means the caller needs to start a new BIO.
  */
-static int dio_bio_add_page(struct dio *dio)
+static inline int dio_bio_add_page(struct dio_submit *sdio)
 {
 	int ret;
  
-	ret = bio_add_page(dio->bio, dio->cur_page,
-			dio->cur_page_len, dio->cur_page_offset);
-	if (ret == dio->cur_page_len) {
+	ret = bio_add_page(sdio->bio, sdio->cur_page,
+			sdio->cur_page_len, sdio->cur_page_offset);
+	if (ret == sdio->cur_page_len) {
 		/*
 		 * Decrement count only, if we are done with this page
 		 */
-		if ((dio->cur_page_len + dio->cur_page_offset) == PAGE_SIZE)
-			dio->pages_in_io--;
-		page_cache_get(dio->cur_page);
-		dio->final_block_in_bio = dio->cur_page_block +
-			(dio->cur_page_len >> dio->blkbits);
+		if ((sdio->cur_page_len + sdio->cur_page_offset) == PAGE_SIZE)
+			sdio->pages_in_io--;
+		page_cache_get(sdio->cur_page);
+		sdio->final_block_in_bio = sdio->cur_page_block +
+			(sdio->cur_page_len >> sdio->blkbits);
 		ret = 0;
 	} else {
 		ret = 1;
  
@@ -675,14 +690,15 @@
  * The caller of this function is responsible for removing cur_page from the
  * dio, and for dropping the refcount which came from that presence.
  */
-static int dio_send_cur_page(struct dio *dio)
+static inline int dio_send_cur_page(struct dio *dio, struct dio_submit *sdio,
+		struct buffer_head *map_bh)
 {
 	int ret = 0;
  
-	if (dio->bio) {
-		loff_t cur_offset = dio->cur_page_fs_offset;
-		loff_t bio_next_offset = dio->logical_offset_in_bio +
-			dio->bio->bi_size;
+	if (sdio->bio) {
+		loff_t cur_offset = sdio->cur_page_fs_offset;
+		loff_t bio_next_offset = sdio->logical_offset_in_bio +
+			sdio->bio->bi_size;
  
 		/*
 		 * See whether this new request is contiguous with the old.
  
  
  
  
  
@@ -698,28 +714,28 @@
 		 * be the next logical offset in the bio, submit the bio we
 		 * have.
 		 */
-		if (dio->final_block_in_bio != dio->cur_page_block ||
+		if (sdio->final_block_in_bio != sdio->cur_page_block ||
 		    cur_offset != bio_next_offset)
-			dio_bio_submit(dio);
+			dio_bio_submit(dio, sdio);
 		/*
 		 * Submit now if the underlying fs is about to perform a
 		 * metadata read
 		 */
-		else if (dio->boundary)
-			dio_bio_submit(dio);
+		else if (sdio->boundary)
+			dio_bio_submit(dio, sdio);
 	}
  
-	if (dio->bio == NULL) {
-		ret = dio_new_bio(dio, dio->cur_page_block);
+	if (sdio->bio == NULL) {
+		ret = dio_new_bio(dio, sdio, sdio->cur_page_block, map_bh);
 		if (ret)
 			goto out;
 	}
  
-	if (dio_bio_add_page(dio) != 0) {
-		dio_bio_submit(dio);
-		ret = dio_new_bio(dio, dio->cur_page_block);
+	if (dio_bio_add_page(sdio) != 0) {
+		dio_bio_submit(dio, sdio);
+		ret = dio_new_bio(dio, sdio, sdio->cur_page_block, map_bh);
 		if (ret == 0) {
-			ret = dio_bio_add_page(dio);
+			ret = dio_bio_add_page(sdio);
 			BUG_ON(ret != 0);
 		}
 	}
@@ -744,9 +760,10 @@
  * If that doesn't work out then we put the old page into the bio and add this
  * page to the dio instead.
  */
-static int
-submit_page_section(struct dio *dio, struct page *page,
-		unsigned offset, unsigned len, sector_t blocknr)
+static inline int
+submit_page_section(struct dio *dio, struct dio_submit *sdio, struct page *page,
+		    unsigned offset, unsigned len, sector_t blocknr,
+		    struct buffer_head *map_bh)
 {
 	int ret = 0;
  
  
  
@@ -760,20 +777,20 @@
 	/*
 	 * Can we just grow the current page's presence in the dio?
 	 */
-	if (	(dio->cur_page == page) &&
-		(dio->cur_page_offset + dio->cur_page_len == offset) &&
-		(dio->cur_page_block +
-			(dio->cur_page_len >> dio->blkbits) == blocknr)) {
-		dio->cur_page_len += len;
+	if (sdio->cur_page == page &&
+	    sdio->cur_page_offset + sdio->cur_page_len == offset &&
+	    sdio->cur_page_block +
+	    (sdio->cur_page_len >> sdio->blkbits) == blocknr) {
+		sdio->cur_page_len += len;
  
 		/*
-		 * If dio->boundary then we want to schedule the IO now to
+		 * If sdio->boundary then we want to schedule the IO now to
 		 * avoid metadata seeks.
 		 */
-		if (dio->boundary) {
-			ret = dio_send_cur_page(dio);
-			page_cache_release(dio->cur_page);
-			dio->cur_page = NULL;
+		if (sdio->boundary) {
+			ret = dio_send_cur_page(dio, sdio, map_bh);
+			page_cache_release(sdio->cur_page);
+			sdio->cur_page = NULL;
 		}
 		goto out;
 	}
  
@@ -781,20 +798,20 @@
 	/*
 	 * If there's a deferred page already there then send it.
 	 */
-	if (dio->cur_page) {
-		ret = dio_send_cur_page(dio);
-		page_cache_release(dio->cur_page);
-		dio->cur_page = NULL;
+	if (sdio->cur_page) {
+		ret = dio_send_cur_page(dio, sdio, map_bh);
+		page_cache_release(sdio->cur_page);
+		sdio->cur_page = NULL;
 		if (ret)
 			goto out;
 	}
  
 	page_cache_get(page);		/* It is in dio */
-	dio->cur_page = page;
-	dio->cur_page_offset = offset;
-	dio->cur_page_len = len;
-	dio->cur_page_block = blocknr;
-	dio->cur_page_fs_offset = dio->block_in_file << dio->blkbits;
+	sdio->cur_page = page;
+	sdio->cur_page_offset = offset;
+	sdio->cur_page_len = len;
+	sdio->cur_page_block = blocknr;
+	sdio->cur_page_fs_offset = sdio->block_in_file << sdio->blkbits;
 out:
 	return ret;
 }
  
  
@@ -804,16 +821,16 @@
  * file blocks.  Only called for S_ISREG files - blockdevs do not set
  * buffer_new
  */
-static void clean_blockdev_aliases(struct dio *dio)
+static void clean_blockdev_aliases(struct dio *dio, struct buffer_head *map_bh)
 {
 	unsigned i;
 	unsigned nblocks;
  
-	nblocks = dio->map_bh.b_size >> dio->inode->i_blkbits;
+	nblocks = map_bh->b_size >> dio->inode->i_blkbits;
  
 	for (i = 0; i < nblocks; i++) {
-		unmap_underlying_metadata(dio->map_bh.b_bdev,
-					dio->map_bh.b_blocknr + i);
+		unmap_underlying_metadata(map_bh->b_bdev,
+					  map_bh->b_blocknr + i);
 	}
 }
  
  
  
@@ -826,19 +843,20 @@
  * `end' is zero if we're doing the start of the IO, 1 at the end of the
  * IO.
  */
-static void dio_zero_block(struct dio *dio, int end)
+static inline void dio_zero_block(struct dio *dio, struct dio_submit *sdio,
+		int end, struct buffer_head *map_bh)
 {
 	unsigned dio_blocks_per_fs_block;
 	unsigned this_chunk_blocks;	/* In dio_blocks */
 	unsigned this_chunk_bytes;
 	struct page *page;
  
-	dio->start_zero_done = 1;
-	if (!dio->blkfactor || !buffer_new(&dio->map_bh))
+	sdio->start_zero_done = 1;
+	if (!sdio->blkfactor || !buffer_new(map_bh))
 		return;
  
-	dio_blocks_per_fs_block = 1 << dio->blkfactor;
-	this_chunk_blocks = dio->block_in_file & (dio_blocks_per_fs_block - 1);
+	dio_blocks_per_fs_block = 1 << sdio->blkfactor;
+	this_chunk_blocks = sdio->block_in_file & (dio_blocks_per_fs_block - 1);
  
 	if (!this_chunk_blocks)
 		return;
  
  
@@ -850,14 +868,14 @@
 	if (end) 
 		this_chunk_blocks = dio_blocks_per_fs_block - this_chunk_blocks;
  
-	this_chunk_bytes = this_chunk_blocks << dio->blkbits;
+	this_chunk_bytes = this_chunk_blocks << sdio->blkbits;
  
 	page = ZERO_PAGE(0);
-	if (submit_page_section(dio, page, 0, this_chunk_bytes, 
-				dio->next_block_for_io))
+	if (submit_page_section(dio, sdio, page, 0, this_chunk_bytes,
+				sdio->next_block_for_io, map_bh))
 		return;
  
-	dio->next_block_for_io += this_chunk_blocks;
+	sdio->next_block_for_io += this_chunk_blocks;
 }
  
 /*
  
  
  
  
@@ -876,20 +894,20 @@
  * it should set b_size to PAGE_SIZE or more inside get_block().  This gives
  * fine alignment but still allows this function to work in PAGE_SIZE units.
  */
-static int do_direct_IO(struct dio *dio)
+static int do_direct_IO(struct dio *dio, struct dio_submit *sdio,
+			struct buffer_head *map_bh)
 {
-	const unsigned blkbits = dio->blkbits;
+	const unsigned blkbits = sdio->blkbits;
 	const unsigned blocks_per_page = PAGE_SIZE >> blkbits;
 	struct page *page;
 	unsigned block_in_page;
-	struct buffer_head *map_bh = &dio->map_bh;
 	int ret = 0;
  
 	/* The I/O can start at any block offset within the first page */
-	block_in_page = dio->first_block_in_page;
+	block_in_page = sdio->first_block_in_page;
  
-	while (dio->block_in_file < dio->final_block_in_request) {
-		page = dio_get_page(dio);
+	while (sdio->block_in_file < sdio->final_block_in_request) {
+		page = dio_get_page(dio, sdio);
 		if (IS_ERR(page)) {
 			ret = PTR_ERR(page);
 			goto out;
  
@@ -901,14 +919,14 @@
 			unsigned this_chunk_blocks;	/* # of blocks */
 			unsigned u;
  
-			if (dio->blocks_available == 0) {
+			if (sdio->blocks_available == 0) {
 				/*
 				 * Need to go and map some more disk
 				 */
 				unsigned long blkmask;
 				unsigned long dio_remainder;
  
-				ret = get_more_blocks(dio);
+				ret = get_more_blocks(dio, sdio, map_bh);
 				if (ret) {
 					page_cache_release(page);
 					goto out;
  
  
  
@@ -916,18 +934,18 @@
 				if (!buffer_mapped(map_bh))
 					goto do_holes;
  
-				dio->blocks_available =
-						map_bh->b_size >> dio->blkbits;
-				dio->next_block_for_io =
-					map_bh->b_blocknr << dio->blkfactor;
+				sdio->blocks_available =
+						map_bh->b_size >> sdio->blkbits;
+				sdio->next_block_for_io =
+					map_bh->b_blocknr << sdio->blkfactor;
 				if (buffer_new(map_bh))
-					clean_blockdev_aliases(dio);
+					clean_blockdev_aliases(dio, map_bh);
  
-				if (!dio->blkfactor)
+				if (!sdio->blkfactor)
 					goto do_holes;
  
-				blkmask = (1 << dio->blkfactor) - 1;
-				dio_remainder = (dio->block_in_file & blkmask);
+				blkmask = (1 << sdio->blkfactor) - 1;
+				dio_remainder = (sdio->block_in_file & blkmask);
  
 				/*
 				 * If we are at the start of IO and that IO
@@ -941,8 +959,8 @@
 				 * on-disk
 				 */
 				if (!buffer_new(map_bh))
-					dio->next_block_for_io += dio_remainder;
-				dio->blocks_available -= dio_remainder;
+					sdio->next_block_for_io += dio_remainder;
+				sdio->blocks_available -= dio_remainder;
 			}
 do_holes:
 			/* Handle holes */
@@ -961,7 +979,7 @@
 				 */
 				i_size_aligned = ALIGN(i_size_read(dio->inode),
 							1 << blkbits);
-				if (dio->block_in_file >=
+				if (sdio->block_in_file >=
 						i_size_aligned >> blkbits) {
 					/* We hit eof */
 					page_cache_release(page);
@@ -969,7 +987,7 @@
 				}
 				zero_user(page, block_in_page << blkbits,
 						1 << blkbits);
-				dio->block_in_file++;
+				sdio->block_in_file++;
 				block_in_page++;
 				goto next_block;
 			}
  
  
  
  
  
  
  
@@ -979,38 +997,41 @@
 			 * is finer than the underlying fs, go check to see if
 			 * we must zero out the start of this block.
 			 */
-			if (unlikely(dio->blkfactor && !dio->start_zero_done))
-				dio_zero_block(dio, 0);
+			if (unlikely(sdio->blkfactor && !sdio->start_zero_done))
+				dio_zero_block(dio, sdio, 0, map_bh);
  
 			/*
 			 * Work out, in this_chunk_blocks, how much disk we
 			 * can add to this page
 			 */
-			this_chunk_blocks = dio->blocks_available;
+			this_chunk_blocks = sdio->blocks_available;
 			u = (PAGE_SIZE - offset_in_page) >> blkbits;
 			if (this_chunk_blocks > u)
 				this_chunk_blocks = u;
-			u = dio->final_block_in_request - dio->block_in_file;
+			u = sdio->final_block_in_request - sdio->block_in_file;
 			if (this_chunk_blocks > u)
 				this_chunk_blocks = u;
 			this_chunk_bytes = this_chunk_blocks << blkbits;
 			BUG_ON(this_chunk_bytes == 0);
  
-			dio->boundary = buffer_boundary(map_bh);
-			ret = submit_page_section(dio, page, offset_in_page,
-				this_chunk_bytes, dio->next_block_for_io);
+			sdio->boundary = buffer_boundary(map_bh);
+			ret = submit_page_section(dio, sdio, page,
+						  offset_in_page,
+						  this_chunk_bytes,
+						  sdio->next_block_for_io,
+						  map_bh);
 			if (ret) {
 				page_cache_release(page);
 				goto out;
 			}
-			dio->next_block_for_io += this_chunk_blocks;
+			sdio->next_block_for_io += this_chunk_blocks;
  
-			dio->block_in_file += this_chunk_blocks;
+			sdio->block_in_file += this_chunk_blocks;
 			block_in_page += this_chunk_blocks;
-			dio->blocks_available -= this_chunk_blocks;
+			sdio->blocks_available -= this_chunk_blocks;
 next_block:
-			BUG_ON(dio->block_in_file > dio->final_block_in_request);
-			if (dio->block_in_file == dio->final_block_in_request)
+			BUG_ON(sdio->block_in_file > sdio->final_block_in_request);
+			if (sdio->block_in_file == sdio->final_block_in_request)
 				break;
 		}
  
  
  
  
  
@@ -1022,137 +1043,12 @@
 	return ret;
 }
  
-static ssize_t
-direct_io_worker(int rw, struct kiocb *iocb, struct inode *inode, 
-	const struct iovec *iov, loff_t offset, unsigned long nr_segs, 
-	unsigned blkbits, get_block_t get_block, dio_iodone_t end_io,
-	dio_submit_t submit_io, struct dio *dio)
+static inline int drop_refcount(struct dio *dio)
 {
-	unsigned long user_addr; 
+	int ret2;
 	unsigned long flags;
-	int seg;
-	ssize_t ret = 0;
-	ssize_t ret2;
-	size_t bytes;
  
-	dio->inode = inode;
-	dio->rw = rw;
-	dio->blkbits = blkbits;
-	dio->blkfactor = inode->i_blkbits - blkbits;
-	dio->block_in_file = offset >> blkbits;
-
-	dio->get_block = get_block;
-	dio->end_io = end_io;
-	dio->submit_io = submit_io;
-	dio->final_block_in_bio = -1;
-	dio->next_block_for_io = -1;
-
-	dio->iocb = iocb;
-	dio->i_size = i_size_read(inode);
-
-	spin_lock_init(&dio->bio_lock);
-	dio->refcount = 1;
-
 	/*
-	 * In case of non-aligned buffers, we may need 2 more
-	 * pages since we need to zero out first and last block.
-	 */
-	if (unlikely(dio->blkfactor))
-		dio->pages_in_io = 2;
-
-	for (seg = 0; seg < nr_segs; seg++) {
-		user_addr = (unsigned long)iov[seg].iov_base;
-		dio->pages_in_io +=
-			((user_addr+iov[seg].iov_len +PAGE_SIZE-1)/PAGE_SIZE
-				- user_addr/PAGE_SIZE);
-	}
-
-	for (seg = 0; seg < nr_segs; seg++) {
-		user_addr = (unsigned long)iov[seg].iov_base;
-		dio->size += bytes = iov[seg].iov_len;
-
-		/* Index into the first page of the first block */
-		dio->first_block_in_page = (user_addr & ~PAGE_MASK) >> blkbits;
-		dio->final_block_in_request = dio->block_in_file +
-						(bytes >> blkbits);
-		/* Page fetching state */
-		dio->head = 0;
-		dio->tail = 0;
-		dio->curr_page = 0;
-
-		dio->total_pages = 0;
-		if (user_addr & (PAGE_SIZE-1)) {
-			dio->total_pages++;
-			bytes -= PAGE_SIZE - (user_addr & (PAGE_SIZE - 1));
-		}
-		dio->total_pages += (bytes + PAGE_SIZE - 1) / PAGE_SIZE;
-		dio->curr_user_address = user_addr;
-	
-		ret = do_direct_IO(dio);
-
-		dio->result += iov[seg].iov_len -
-			((dio->final_block_in_request - dio->block_in_file) <<
-					blkbits);
-
-		if (ret) {
-			dio_cleanup(dio);
-			break;
-		}
-	} /* end iovec loop */
-
-	if (ret == -ENOTBLK) {
-		/*
-		 * The remaining part of the request will be
-		 * be handled by buffered I/O when we return
-		 */
-		ret = 0;
-	}
-	/*
-	 * There may be some unwritten disk at the end of a part-written
-	 * fs-block-sized block.  Go zero that now.
-	 */
-	dio_zero_block(dio, 1);
-
-	if (dio->cur_page) {
-		ret2 = dio_send_cur_page(dio);
-		if (ret == 0)
-			ret = ret2;
-		page_cache_release(dio->cur_page);
-		dio->cur_page = NULL;
-	}
-	if (dio->bio)
-		dio_bio_submit(dio);
-
-	/*
-	 * It is possible that, we return short IO due to end of file.
-	 * In that case, we need to release all the pages we got hold on.
-	 */
-	dio_cleanup(dio);
-
-	/*
-	 * All block lookups have been performed. For READ requests
-	 * we can let i_mutex go now that its achieved its purpose
-	 * of protecting us from looking up uninitialized blocks.
-	 */
-	if (rw == READ && (dio->flags & DIO_LOCKING))
-		mutex_unlock(&dio->inode->i_mutex);
-
-	/*
-	 * The only time we want to leave bios in flight is when a successful
-	 * partial aio read or full aio write have been setup.  In that case
-	 * bio completion will call aio_complete.  The only time it's safe to
-	 * call aio_complete is when we return -EIOCBQUEUED, so we key on that.
-	 * This had *better* be the only place that raises -EIOCBQUEUED.
-	 */
-	BUG_ON(ret == -EIOCBQUEUED);
-	if (dio->is_async && ret == 0 && dio->result &&
-	    ((rw & READ) || (dio->result == dio->size)))
-		ret = -EIOCBQUEUED;
-
-	if (ret != -EIOCBQUEUED)
-		dio_await_completion(dio);
-
-	/*
 	 * Sync will always be dropping the final ref and completing the
 	 * operation.  AIO can if it was a broken operation described above or
 	 * in fact if all the bios race to complete before we get here.  In
@@ -1166,14 +1062,7 @@
 	spin_lock_irqsave(&dio->bio_lock, flags);
 	ret2 = --dio->refcount;
 	spin_unlock_irqrestore(&dio->bio_lock, flags);
-
-	if (ret2 == 0) {
-		ret = dio_complete(dio, offset, ret, false);
-		kfree(dio);
-	} else
-		BUG_ON(ret != -EIOCBQUEUED);
-
-	return ret;
+	return ret2;
 }
  
 /*
@@ -1195,6 +1084,11 @@
  * expected that filesystem provide exclusion between new direct I/O
  * and truncates.  For DIO_LOCKING filesystems this is done by i_mutex,
  * but other filesystems need to take care of this on their own.
+ *
+ * NOTE: if you pass "sdio" to anything by pointer make sure that function
+ * is always inlined. Otherwise gcc is unable to split the structure into
+ * individual fields and will generate much worse code. This is important
+ * for the whole file.
  */
 ssize_t
 __blockdev_direct_IO(int rw, struct kiocb *iocb, struct inode *inode,
@@ -1211,6 +1105,10 @@
 	ssize_t retval = -EINVAL;
 	loff_t end = offset;
 	struct dio *dio;
+	struct dio_submit sdio = { 0, };
+	unsigned long user_addr;
+	size_t bytes;
+	struct buffer_head map_bh = { 0, };
  
 	if (rw & WRITE)
 		rw = WRITE_ODIRECT;
@@ -1244,7 +1142,7 @@
 	if (rw == READ && end == offset)
 		return 0;
  
-	dio = kmalloc(sizeof(*dio), GFP_KERNEL);
+	dio = kmem_cache_alloc(dio_cache, GFP_KERNEL);
 	retval = -ENOMEM;
 	if (!dio)
 		goto out;
@@ -1268,7 +1166,7 @@
 							      end - 1);
 			if (retval) {
 				mutex_unlock(&inode->i_mutex);
-				kfree(dio);
+				kmem_cache_free(dio_cache, dio);
 				goto out;
 			}
 		}
  
  
@@ -1288,12 +1186,142 @@
 	dio->is_async = !is_sync_kiocb(iocb) && !((rw & WRITE) &&
 		(end > i_size_read(inode)));
  
-	retval = direct_io_worker(rw, iocb, inode, iov, offset,
-				nr_segs, blkbits, get_block, end_io,
-				submit_io, dio);
+	retval = 0;
  
+	dio->inode = inode;
+	dio->rw = rw;
+	sdio.blkbits = blkbits;
+	sdio.blkfactor = inode->i_blkbits - blkbits;
+	sdio.block_in_file = offset >> blkbits;
+
+	sdio.get_block = get_block;
+	dio->end_io = end_io;
+	sdio.submit_io = submit_io;
+	sdio.final_block_in_bio = -1;
+	sdio.next_block_for_io = -1;
+
+	dio->iocb = iocb;
+	dio->i_size = i_size_read(inode);
+
+	spin_lock_init(&dio->bio_lock);
+	dio->refcount = 1;
+
+	/*
+	 * In case of non-aligned buffers, we may need 2 more
+	 * pages since we need to zero out first and last block.
+	 */
+	if (unlikely(sdio.blkfactor))
+		sdio.pages_in_io = 2;
+
+	for (seg = 0; seg < nr_segs; seg++) {
+		user_addr = (unsigned long)iov[seg].iov_base;
+		sdio.pages_in_io +=
+			((user_addr + iov[seg].iov_len + PAGE_SIZE-1) /
+				PAGE_SIZE - user_addr / PAGE_SIZE);
+	}
+
+	for (seg = 0; seg < nr_segs; seg++) {
+		user_addr = (unsigned long)iov[seg].iov_base;
+		sdio.size += bytes = iov[seg].iov_len;
+
+		/* Index into the first page of the first block */
+		sdio.first_block_in_page = (user_addr & ~PAGE_MASK) >> blkbits;
+		sdio.final_block_in_request = sdio.block_in_file +
+						(bytes >> blkbits);
+		/* Page fetching state */
+		sdio.head = 0;
+		sdio.tail = 0;
+		sdio.curr_page = 0;
+
+		sdio.total_pages = 0;
+		if (user_addr & (PAGE_SIZE-1)) {
+			sdio.total_pages++;
+			bytes -= PAGE_SIZE - (user_addr & (PAGE_SIZE - 1));
+		}
+		sdio.total_pages += (bytes + PAGE_SIZE - 1) / PAGE_SIZE;
+		sdio.curr_user_address = user_addr;
+
+		retval = do_direct_IO(dio, &sdio, &map_bh);
+
+		dio->result += iov[seg].iov_len -
+			((sdio.final_block_in_request - sdio.block_in_file) <<
+					blkbits);
+
+		if (retval) {
+			dio_cleanup(dio, &sdio);
+			break;
+		}
+	} /* end iovec loop */
+
+	if (retval == -ENOTBLK) {
+		/*
+		 * The remaining part of the request will be
+		 * be handled by buffered I/O when we return
+		 */
+		retval = 0;
+	}
+	/*
+	 * There may be some unwritten disk at the end of a part-written
+	 * fs-block-sized block.  Go zero that now.
+	 */
+	dio_zero_block(dio, &sdio, 1, &map_bh);
+
+	if (sdio.cur_page) {
+		ssize_t ret2;
+
+		ret2 = dio_send_cur_page(dio, &sdio, &map_bh);
+		if (retval == 0)
+			retval = ret2;
+		page_cache_release(sdio.cur_page);
+		sdio.cur_page = NULL;
+	}
+	if (sdio.bio)
+		dio_bio_submit(dio, &sdio);
+
+	/*
+	 * It is possible that, we return short IO due to end of file.
+	 * In that case, we need to release all the pages we got hold on.
+	 */
+	dio_cleanup(dio, &sdio);
+
+	/*
+	 * All block lookups have been performed. For READ requests
+	 * we can let i_mutex go now that its achieved its purpose
+	 * of protecting us from looking up uninitialized blocks.
+	 */
+	if (rw == READ && (dio->flags & DIO_LOCKING))
+		mutex_unlock(&dio->inode->i_mutex);
+
+	/*
+	 * The only time we want to leave bios in flight is when a successful
+	 * partial aio read or full aio write have been setup.  In that case
+	 * bio completion will call aio_complete.  The only time it's safe to
+	 * call aio_complete is when we return -EIOCBQUEUED, so we key on that.
+	 * This had *better* be the only place that raises -EIOCBQUEUED.
+	 */
+	BUG_ON(retval == -EIOCBQUEUED);
+	if (dio->is_async && retval == 0 && dio->result &&
+	    ((rw & READ) || (dio->result == sdio.size)))
+		retval = -EIOCBQUEUED;
+
+	if (retval != -EIOCBQUEUED)
+		dio_await_completion(dio);
+
+	if (drop_refcount(dio) == 0) {
+		retval = dio_complete(dio, offset, retval, false);
+		kmem_cache_free(dio_cache, dio);
+	} else
+		BUG_ON(retval != -EIOCBQUEUED);
+
 out:
 	return retval;
 }
 EXPORT_SYMBOL(__blockdev_direct_IO);
+
+static __init int dio_init(void)
+{
+	dio_cache = KMEM_CACHE(dio, SLAB_PANIC);
+	return 0;
+}
+module_init(dio_init)
@@ -224,53 +224,8 @@
 		maxbytes = EXT4_SB(inode->i_sb)->s_bitmap_maxbytes;
 	else
 		maxbytes = inode->i_sb->s_maxbytes;
-	mutex_lock(&inode->i_mutex);
-	switch (origin) {
-	case SEEK_END:
-		offset += inode->i_size;
-		break;
-	case SEEK_CUR:
-		if (offset == 0) {
-			mutex_unlock(&inode->i_mutex);
-			return file->f_pos;
-		}
-		offset += file->f_pos;
-		break;
-	case SEEK_DATA:
-		/*
-		 * In the generic case the entire file is data, so as long as
-		 * offset isn't at the end of the file then the offset is data.
-		 */
-		if (offset >= inode->i_size) {
-			mutex_unlock(&inode->i_mutex);
-			return -ENXIO;
-		}
-		break;
-	case SEEK_HOLE:
-		/*
-		 * There is a virtual hole at the end of the file, so as long as
-		 * offset isn't i_size or larger, return i_size.
-		 */
-		if (offset >= inode->i_size) {
-			mutex_unlock(&inode->i_mutex);
-			return -ENXIO;
-		}
-		offset = inode->i_size;
-		break;
-	}
  
-	if (offset < 0 || offset > maxbytes) {
-		mutex_unlock(&inode->i_mutex);
-		return -EINVAL;
-	}
-
-	if (offset != file->f_pos) {
-		file->f_pos = offset;
-		file->f_version = 0;
-	}
-	mutex_unlock(&inode->i_mutex);
-
-	return offset;
+	return generic_file_llseek_size(file, offset, origin, maxbytes);
 }
  
 const struct file_operations ext4_file_operations = {
@@ -66,13 +66,13 @@
 		error = gfs2_glock_nq_init(ip->i_gl, LM_ST_SHARED, LM_FLAG_ANY,
 					   &i_gh);
 		if (!error) {
-			error = generic_file_llseek_unlocked(file, offset, origin);
+			error = generic_file_llseek(file, offset, origin);
 			gfs2_glock_dq_uninit(&i_gh);
 		}
 		break;
 	case SEEK_CUR:
 	case SEEK_SET:
-		error = generic_file_llseek_unlocked(file, offset, origin);
+		error = generic_file_llseek(file, offset, origin);
 		break;
 	default:
 		error = -EINVAL;
@@ -634,7 +634,7 @@
 		 * inode to the back of the list so we don't spin on it.
 		 */
 		if (!spin_trylock(&inode->i_lock)) {
-			list_move(&inode->i_lru, &sb->s_inode_lru);
+			list_move_tail(&inode->i_lru, &sb->s_inode_lru);
 			continue;
 		}
  
@@ -221,14 +221,12 @@
 }
  
 /*
- * This does basic POSIX ACL permission checking
+ * This does the basic permission checking
  */
 static int acl_permission_check(struct inode *inode, int mask)
 {
 	unsigned int mode = inode->i_mode;
  
-	mask &= MAY_READ | MAY_WRITE | MAY_EXEC | MAY_NOT_BLOCK;
-
 	if (current_user_ns() != inode_userns(inode))
 		goto other_perms;
  
@@ -257,7 +255,7 @@
 /**
  * generic_permission -  check for access rights on a Posix-like filesystem
  * @inode:	inode to check access rights for
- * @mask:	right to check for (%MAY_READ, %MAY_WRITE, %MAY_EXEC)
+ * @mask:	right to check for (%MAY_READ, %MAY_WRITE, %MAY_EXEC, ...)
  *
  * Used to check for read/write/execute permissions on a file.
  * We use "fsuid" for this, letting us set arbitrary permissions
@@ -273,7 +271,7 @@
 	int ret;
  
 	/*
-	 * Do the basic POSIX ACL permission checks.
+	 * Do the basic permission checks.
 	 */
 	ret = acl_permission_check(inode, mask);
 	if (ret != -EACCES)
  
@@ -331,12 +329,14 @@
 /**
  * inode_permission  -  check for access rights to a given inode
  * @inode:	inode to check permission on
- * @mask:	right to check for (%MAY_READ, %MAY_WRITE, %MAY_EXEC)
+ * @mask:	right to check for (%MAY_READ, %MAY_WRITE, %MAY_EXEC, ...)
  *
  * Used to check for read/write/execute permissions on an inode.
  * We use "fsuid" for this, letting us set arbitrary permissions
  * for filesystem access without changing the "normal" uids which
  * are used for other things.
+ *
+ * When checking for MAY_APPEND, MAY_WRITE must also be set in @mask.
  */
 int inode_permission(struct inode *inode, int mask)
 {
@@ -2035,10 +2035,7 @@
 	if (flag & O_NOATIME && !inode_owner_or_capable(inode))
 		return -EPERM;
  
-	/*
-	 * Ensure there are no outstanding leases on the file.
-	 */
-	return break_lease(inode, flag);
+	return 0;
 }
  
 static int handle_truncate(struct file *filp)
@@ -1109,6 +1109,7 @@
  
 	/* device */
 	if (mnt->mnt_sb->s_op->show_devname) {
+		seq_puts(m, "device ");
 		err = mnt->mnt_sb->s_op->show_devname(m, mnt);
 	} else {
 		if (mnt->mnt_devname) {
@@ -180,8 +180,6 @@
  
 static loff_t nfs_file_llseek(struct file *filp, loff_t offset, int origin)
 {
-	loff_t loff;
-
 	dprintk("NFS: llseek file(%s/%s, %lld, %d)\n",
 			filp->f_path.dentry->d_parent->d_name.name,
 			filp->f_path.dentry->d_name.name,
  
@@ -197,13 +195,9 @@
 		int retval = nfs_revalidate_file_size(inode, filp);
 		if (retval < 0)
 			return (loff_t)retval;
+	}
  
-		spin_lock(&inode->i_lock);
-		loff = generic_file_llseek_unlocked(filp, offset, origin);
-		spin_unlock(&inode->i_lock);
-	} else
-		loff = generic_file_llseek_unlocked(filp, offset, origin);
-	return loff;
+	return generic_file_llseek(filp, offset, origin);
 }
  
 /*
@@ -685,6 +685,10 @@
 	if (error)
 		goto cleanup_all;
  
+	error = break_lease(inode, f->f_flags);
+	if (error)
+		goto cleanup_all;
+
 	if (!open && f->f_op)
 		open = f->f_op->open;
 	if (open) {
@@ -218,6 +218,8 @@
 	const struct posix_acl_entry *pa, *pe, *mask_obj;
 	int found = 0;
  
+	want &= MAY_READ | MAY_WRITE | MAY_EXEC | MAY_NOT_BLOCK;
+
 	FOREACH_ACL_ENTRY(pa, acl, pe) {
                 switch(pa->e_tag) {
                         case ACL_USER_OBJ:
@@ -35,23 +35,45 @@
 	return file->f_mode & FMODE_UNSIGNED_OFFSET;
 }
  
+static loff_t lseek_execute(struct file *file, struct inode *inode,
+		loff_t offset, loff_t maxsize)
+{
+	if (offset < 0 && !unsigned_offsets(file))
+		return -EINVAL;
+	if (offset > maxsize)
+		return -EINVAL;
+
+	if (offset != file->f_pos) {
+		file->f_pos = offset;
+		file->f_version = 0;
+	}
+	return offset;
+}
+
 /**
- * generic_file_llseek_unlocked - lockless generic llseek implementation
+ * generic_file_llseek_size - generic llseek implementation for regular files
  * @file:	file structure to seek on
  * @offset:	file offset to seek to
  * @origin:	type of seek
+ * @size:	max size of file system
  *
- * Updates the file offset to the value specified by @offset and @origin.
- * Locking must be provided by the caller.
+ * This is a variant of generic_file_llseek that allows passing in a custom
+ * file size.
+ *
+ * Synchronization:
+ * SEEK_SET and SEEK_END are unsynchronized (but atomic on 64bit platforms)
+ * SEEK_CUR is synchronized against other SEEK_CURs, but not read/writes.
+ * read/writes behave like SEEK_SET against seeks.
  */
 loff_t
-generic_file_llseek_unlocked(struct file *file, loff_t offset, int origin)
+generic_file_llseek_size(struct file *file, loff_t offset, int origin,
+		loff_t maxsize)
 {
 	struct inode *inode = file->f_mapping->host;
  
 	switch (origin) {
 	case SEEK_END:
-		offset += inode->i_size;
+		offset += i_size_read(inode);
 		break;
 	case SEEK_CUR:
 		/*
  
@@ -62,14 +84,22 @@
 		 */
 		if (offset == 0)
 			return file->f_pos;
-		offset += file->f_pos;
-		break;
+		/*
+		 * f_lock protects against read/modify/write race with other
+		 * SEEK_CURs. Note that parallel writes and reads behave
+		 * like SEEK_SET.
+		 */
+		spin_lock(&file->f_lock);
+		offset = lseek_execute(file, inode, file->f_pos + offset,
+				       maxsize);
+		spin_unlock(&file->f_lock);
+		return offset;
 	case SEEK_DATA:
 		/*
 		 * In the generic case the entire file is data, so as long as
 		 * offset isn't at the end of the file then the offset is data.
 		 */
-		if (offset >= inode->i_size)
+		if (offset >= i_size_read(inode))
 			return -ENXIO;
 		break;
 	case SEEK_HOLE:
  
  
  
@@ -77,26 +107,15 @@
 		 * There is a virtual hole at the end of the file, so as long as
 		 * offset isn't i_size or larger, return i_size.
 		 */
-		if (offset >= inode->i_size)
+		if (offset >= i_size_read(inode))
 			return -ENXIO;
-		offset = inode->i_size;
+		offset = i_size_read(inode);
 		break;
 	}
  
-	if (offset < 0 && !unsigned_offsets(file))
-		return -EINVAL;
-	if (offset > inode->i_sb->s_maxbytes)
-		return -EINVAL;
-
-	/* Special lock needed here? */
-	if (offset != file->f_pos) {
-		file->f_pos = offset;
-		file->f_version = 0;
-	}
-
-	return offset;
+	return lseek_execute(file, inode, offset, maxsize);
 }
-EXPORT_SYMBOL(generic_file_llseek_unlocked);
+EXPORT_SYMBOL(generic_file_llseek_size);
  
 /**
  * generic_file_llseek - generic llseek implementation for regular files
  
@@ -110,13 +129,10 @@
  */
 loff_t generic_file_llseek(struct file *file, loff_t offset, int origin)
 {
-	loff_t rval;
+	struct inode *inode = file->f_mapping->host;
  
-	mutex_lock(&file->f_dentry->d_inode->i_mutex);
-	rval = generic_file_llseek_unlocked(file, offset, origin);
-	mutex_unlock(&file->f_dentry->d_inode->i_mutex);
-
-	return rval;
+	return generic_file_llseek_size(file, offset, origin,
+					inode->i_sb->s_maxbytes);
 }
 EXPORT_SYMBOL(generic_file_llseek);
  
@@ -58,14 +58,15 @@
  
 #define NR_FILE  8192	/* this can well be larger on a larger system */
  
-#define MAY_EXEC 1
-#define MAY_WRITE 2
-#define MAY_READ 4
-#define MAY_APPEND 8
-#define MAY_ACCESS 16
-#define MAY_OPEN 32
-#define MAY_CHDIR 64
-#define MAY_NOT_BLOCK 128	/* called from RCU mode, don't block */
+#define MAY_EXEC		0x00000001
+#define MAY_WRITE		0x00000002
+#define MAY_READ		0x00000004
+#define MAY_APPEND		0x00000008
+#define MAY_ACCESS		0x00000010
+#define MAY_OPEN		0x00000020
+#define MAY_CHDIR		0x00000040
+/* called from RCU mode, don't block */
+#define MAY_NOT_BLOCK		0x00000080
  
 /*
  * flags in file.f_mode.  Note that FMODE_READ and FMODE_WRITE must correspond
@@ -963,7 +964,12 @@
 #define f_dentry	f_path.dentry
 #define f_vfsmnt	f_path.mnt
 	const struct file_operations	*f_op;
-	spinlock_t		f_lock;  /* f_ep_links, f_flags, no IRQ */
+
+	/*
+	 * Protects f_ep_links, f_flags, f_pos vs i_size in lseek SEEK_CUR.
+	 * Must not be taken from IRQ context.
+	 */
+	spinlock_t		f_lock;
 #ifdef CONFIG_SMP
 	int			f_sb_list_cpu;
 #endif
@@ -2401,8 +2407,8 @@
 extern loff_t noop_llseek(struct file *file, loff_t offset, int origin);
 extern loff_t no_llseek(struct file *file, loff_t offset, int origin);
 extern loff_t generic_file_llseek(struct file *file, loff_t offset, int origin);
-extern loff_t generic_file_llseek_unlocked(struct file *file, loff_t offset,
-			int origin);
+extern loff_t generic_file_llseek_size(struct file *file, loff_t offset,
+		int origin, loff_t maxsize);
 extern int generic_file_open(struct inode * inode, struct file * filp);
 extern int nonseekable_open(struct inode * inode, struct file * filp);
  
@@ -2115,6 +2115,7 @@
 	} else {
 		const struct iovec *iov = i->iov;
 		size_t base = i->iov_offset;
+		unsigned long nr_segs = i->nr_segs;
  
 		/*
 		 * The !iov->iov_len check ensures we skip over unlikely
  
@@ -2130,11 +2131,13 @@
 			base += copy;
 			if (iov->iov_len == base) {
 				iov++;
+				nr_segs--;
 				base = 0;
 			}
 		}
 		i->iov = iov;
 		i->iov_offset = base;
+		i->nr_segs = nr_segs;
 	}
 }
 EXPORT_SYMBOL(iov_iter_advance);
...	...	@@ -111,7 +111,8 @@
111	111	int f_bavail;
112	112	compat_fsid_t f_fsid;
113	113	int f_namelen;
114		- int f_spare[6];
	114	+ int f_flags;
	115	+ int f_spare[5];
115	116	};
116	117
117	118	#define COMPAT_RLIM_INFINITY 0x7fffffffUL
...	...	@@ -105,7 +105,8 @@
105	105	__kernel_fsid_t f_fsid;
106	106	s32 f_namelen;
107	107	s32 f_frsize;
108		- s32 f_spare[5];
	108	+ s32 f_flags;
	109	+ s32 f_spare[4];
109	110	};
110	111
111	112	struct compat_sigcontext {
...	...	@@ -100,7 +100,8 @@
100	100	compat_fsid_t f_fsid;
101	101	int f_namelen; /* SunOS ignores this field. */
102	102	int f_frsize;
103		- int f_spare[5];
	103	+ int f_flags;
	104	+ int f_spare[4];
104	105	};
105	106
106	107	#define COMPAT_RLIM_OLD_INFINITY 0x7fffffff
...	...	@@ -131,7 +131,8 @@
131	131	compat_fsid_t f_fsid;
132	132	s32 f_namelen;
133	133	s32 f_frsize;
134		- s32 f_spare[6];
	134	+ s32 f_flags;
	135	+ s32 f_spare[5];
135	136	};
136	137
137	138	#define COMPAT_RLIM_OLD_INFINITY 0x7fffffff
...	...	@@ -134,7 +134,8 @@
134	134	compat_fsid_t f_fsid;
135	135	int f_namelen; /* SunOS ignores this field. */
136	136	int f_frsize;
137		- int f_spare[5];
	137	+ int f_flags;
	138	+ int f_spare[4];
138	139	};
139	140
140	141	#define COMPAT_RLIM_INFINITY 0x7fffffff
...	...	@@ -108,7 +108,8 @@
108	108	compat_fsid_t f_fsid;
109	109	int f_namelen; /* SunOS ignores this field. */
110	110	int f_frsize;
111		- int f_spare[5];
	111	+ int f_flags;
	112	+ int f_spare[4];
112	113	};
113	114
114	115	#define COMPAT_RLIM_OLD_INFINITY 0x7fffffff
...	...	@@ -1821,7 +1821,7 @@
1821	1821	switch (origin) {
1822	1822	case SEEK_END:
1823	1823	case SEEK_CUR:
1824		- offset = generic_file_llseek_unlocked(file, offset, origin);
	1824	+ offset = generic_file_llseek(file, offset, origin);
1825	1825	goto out;
1826	1826	case SEEK_DATA:
1827	1827	case SEEK_HOLE:
...	...	@@ -1470,13 +1470,13 @@
1470	1470	}
1471	1471
1472	1472	/**
1473		- * block_invalidatepage - invalidate part of all of a buffer-backed page
	1473	+ * block_invalidatepage - invalidate part or all of a buffer-backed page
1474	1474	*
1475	1475	* @page: the page which is affected
1476	1476	* @offset: the index of the truncation point
1477	1477	*
1478	1478	* block_invalidatepage() is called when all or part of the page has become
1479		- * invalidatedby a truncate operation.
	1479	+ * invalidated by a truncate operation.
1480	1480	*
1481	1481	* block_invalidatepage() does not have to release all buffers, but it must
1482	1482	* ensure that no dirty buffer is left outside @offset and that no I/O
...	...	@@ -730,7 +730,7 @@
730	730	if (rc < 0)
731	731	return (loff_t)rc;
732	732	}
733		- return generic_file_llseek_unlocked(file, offset, origin);
	733	+ return generic_file_llseek(file, offset, origin);
734	734	}
735	735
736	736	static int cifs_setlease(struct file file, long arg, struct file_lock *lease)
...	...	@@ -246,11 +246,8 @@
246	246	__put_user(kbuf->f_fsid.val[0], &ubuf->f_fsid.val[0]) \|\|
247	247	__put_user(kbuf->f_fsid.val[1], &ubuf->f_fsid.val[1]) \|\|
248	248	__put_user(kbuf->f_frsize, &ubuf->f_frsize) \|\|
249		- __put_user(0, &ubuf->f_spare[0]) \|\|
250		- __put_user(0, &ubuf->f_spare[1]) \|\|
251		- __put_user(0, &ubuf->f_spare[2]) \|\|
252		- __put_user(0, &ubuf->f_spare[3]) \|\|
253		- __put_user(0, &ubuf->f_spare[4]))
	249	+ __put_user(kbuf->f_flags, &ubuf->f_flags) \|\|
	250	+ __clear_user(ubuf->f_spare, sizeof(ubuf->f_spare)))
254	251	return -EFAULT;
255	252	return 0;
256	253	}