1 files changed, 55 insertions, 7 deletions

diff --git a/fs/direct-io.c b/fs/direct-io.c
index e82adc2debb7..da111aacb46e 100644
--- a/fs/direct-io.c
+++ b/fs/direct-io.c
@@ -82,6 +82,8 @@ struct dio {
         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 */
@@ -96,6 +98,7 @@ struct dio {
         unsigned cur_page_offset;       /* Offset into it, in bytes */
         unsigned cur_page_len;          /* Nr of bytes at cur_page_offset */
         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 */
@@ -300,6 +303,26 @@ static void dio_bio_end_io(struct bio *bio, int error)
         spin_unlock_irqrestore(&dio->bio_lock, flags);
 }
 
+/**
+ * dio_end_io - handle the end io action for the given bio
+ * @bio: The direct io bio thats being completed
+ * @error: Error if there was one
+ *
+ * This is meant to be called by any filesystem that uses their own dio_submit_t
+ * so that the DIO specific endio actions are dealt with after the filesystem
+ * has done it's completion work.
+ */
+void dio_end_io(struct bio *bio, int error)
+{
+        struct dio *dio = bio->bi_private;
+
+        if (dio->is_async)
+                dio_bio_end_aio(bio, error);
+        else
+                dio_bio_end_io(bio, error);
+}
+EXPORT_SYMBOL_GPL(dio_end_io);
+
 static int
 dio_bio_alloc(struct dio *dio, struct block_device *bdev,
                 sector_t first_sector, int nr_vecs)
@@ -316,6 +339,7 @@ dio_bio_alloc(struct dio *dio, struct block_device *bdev,
                 bio->bi_end_io = dio_bio_end_io;
 
         dio->bio = bio;
+        dio->logical_offset_in_bio = dio->cur_page_fs_offset;
         return 0;
 }
 
@@ -340,10 +364,15 @@ static void dio_bio_submit(struct dio *dio)
         if (dio->is_async && dio->rw == READ)
                 bio_set_pages_dirty(bio);
 
-        submit_bio(dio->rw, bio);
+        if (dio->submit_io)
+                dio->submit_io(dio->rw, bio, dio->inode,
+                               dio->logical_offset_in_bio);
+        else
+                submit_bio(dio->rw, bio);
 
         dio->bio = NULL;
         dio->boundary = 0;
+        dio->logical_offset_in_bio = 0;
 }
 
 /*
@@ -603,10 +632,26 @@ static int dio_send_cur_page(struct dio *dio)
         int ret = 0;
 
         if (dio->bio) {
+                loff_t cur_offset = dio->block_in_file << dio->blkbits;
+                loff_t bio_next_offset = dio->logical_offset_in_bio +
+                        dio->bio->bi_size;
+
                 /*
-                 * See whether this new request is contiguous with the old
+                 * See whether this new request is contiguous with the old.
+                 *
+                 * Btrfs cannot handl having logically non-contiguous requests
+                 * submitted.  For exmple if you have
+                 *
+                 * Logical:  [0-4095][HOLE][8192-12287]
+                 * Phyiscal: [0-4095]      [4096-8181]
+                 *
+                 * We cannot submit those pages together as one BIO.  So if our
+                 * current logical offset in the file does not equal what would
+                 * be the next logical offset in the bio, submit the bio we
+                 * have.
                  */
-                if (dio->final_block_in_bio != dio->cur_page_block)
+                if (dio->final_block_in_bio != dio->cur_page_block ||
+                    cur_offset != bio_next_offset)
                         dio_bio_submit(dio);
                 /*
                  * Submit now if the underlying fs is about to perform a
@@ -701,6 +746,7 @@ submit_page_section(struct dio *dio, struct 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;
 out:
         return ret;
 }
@@ -935,7 +981,7 @@ 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,
-        struct dio *dio)
+        dio_submit_t submit_io, struct dio *dio)
 {
         unsigned long user_addr; 
         unsigned long flags;
@@ -952,6 +998,7 @@ direct_io_worker(int rw, struct kiocb *iocb, struct inode *inode,
 
         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;
 
@@ -1008,7 +1055,7 @@ direct_io_worker(int rw, struct kiocb *iocb, struct inode *inode,
                 }
         } /* end iovec loop */
 
-        if (ret == -ENOTBLK && (rw & WRITE)) {
+        if (ret == -ENOTBLK) {
                 /*
                  * The remaining part of the request will be
                  * be handled by buffered I/O when we return
@@ -1110,7 +1157,7 @@ ssize_t
 __blockdev_direct_IO(int rw, struct kiocb *iocb, struct inode *inode,
         struct block_device *bdev, const struct iovec *iov, loff_t offset, 
         unsigned long nr_segs, get_block_t get_block, dio_iodone_t end_io,
-        int flags)
+        dio_submit_t submit_io, int flags)
 {
         int seg;
         size_t size;
@@ -1197,7 +1244,8 @@ __blockdev_direct_IO(int rw, struct kiocb *iocb, struct inode *inode,
                 (end > i_size_read(inode)));
 
         retval = direct_io_worker(rw, iocb, inode, iov, offset,
-                                nr_segs, blkbits, get_block, end_io, dio);
+                                nr_segs, blkbits, get_block, end_io,
+                                submit_io, dio);
 
         /*
          * In case of error extending write may have instantiated a few