ext4: simplify mpage_add_bh_to_extent()
authorJan Kara <jack@suse.cz>
Mon, 28 Jan 2013 18:06:48 +0000 (13:06 -0500)
committerTheodore Ts'o <tytso@mit.edu>
Mon, 28 Jan 2013 18:06:48 +0000 (13:06 -0500)
The argument b_size of mpage_add_bh_to_extent() was bogus since it was
always == blocksize (which we can easily derive from inode->i_blkbits).
Also second branch of condition:
if (nrblocks >= EXT4_MAX_TRANS_DATA) {
} else if ((nrblocks + (b_size >> mpd->inode->i_blkbits)) >
EXT4_MAX_TRANS_DATA) {
}
was never taken because (b_size >> mpd->inode->i_blkbits) == 1.

Signed-off-by: Jan Kara <jack@suse.cz>
Signed-off-by: "Theodore Ts'o" <tytso@mit.edu>
fs/ext4/inode.c

index 8a89cbbf0f1a822a8a4e01076c65a4237cd70788..6824cb1bd1bb4f14a53271f255793badbf1bbade 100644 (file)
@@ -1647,16 +1647,16 @@ submit_io:
  *
  * @mpd->lbh - extent of blocks
  * @logical - logical number of the block in the file
- * @bh - bh of the block (used to access block's state)
+ * @b_state - b_state of the buffer head added
  *
  * the function is used to collect contig. blocks in same state
  */
-static void mpage_add_bh_to_extent(struct mpage_da_data *mpd,
-                                  sector_t logical, size_t b_size,
+static void mpage_add_bh_to_extent(struct mpage_da_data *mpd, sector_t logical,
                                   unsigned long b_state)
 {
        sector_t next;
-       int nrblocks = mpd->b_size >> mpd->inode->i_blkbits;
+       int blkbits = mpd->inode->i_blkbits;
+       int nrblocks = mpd->b_size >> blkbits;
 
        /*
         * XXX Don't go larger than mballoc is willing to allocate
@@ -1664,11 +1664,11 @@ static void mpage_add_bh_to_extent(struct mpage_da_data *mpd,
         * mpage_da_submit_io() into this function and then call
         * ext4_map_blocks() multiple times in a loop
         */
-       if (nrblocks >= 8*1024*1024/mpd->inode->i_sb->s_blocksize)
+       if (nrblocks >= (8*1024*1024 >> blkbits))
                goto flush_it;
 
-       /* check if thereserved journal credits might overflow */
-       if (!(ext4_test_inode_flag(mpd->inode, EXT4_INODE_EXTENTS))) {
+       /* check if the reserved journal credits might overflow */
+       if (!ext4_test_inode_flag(mpd->inode, EXT4_INODE_EXTENTS)) {
                if (nrblocks >= EXT4_MAX_TRANS_DATA) {
                        /*
                         * With non-extent format we are limited by the journal
@@ -1677,16 +1677,6 @@ static void mpage_add_bh_to_extent(struct mpage_da_data *mpd,
                         * nrblocks.  So limit nrblocks.
                         */
                        goto flush_it;
-               } else if ((nrblocks + (b_size >> mpd->inode->i_blkbits)) >
-                               EXT4_MAX_TRANS_DATA) {
-                       /*
-                        * Adding the new buffer_head would make it cross the
-                        * allowed limit for which we have journal credit
-                        * reserved. So limit the new bh->b_size
-                        */
-                       b_size = (EXT4_MAX_TRANS_DATA - nrblocks) <<
-                                               mpd->inode->i_blkbits;
-                       /* we will do mpage_da_submit_io in the next loop */
                }
        }
        /*
@@ -1694,7 +1684,7 @@ static void mpage_add_bh_to_extent(struct mpage_da_data *mpd,
         */
        if (mpd->b_size == 0) {
                mpd->b_blocknr = logical;
-               mpd->b_size = b_size;
+               mpd->b_size = 1 << blkbits;
                mpd->b_state = b_state & BH_FLAGS;
                return;
        }
@@ -1704,7 +1694,7 @@ static void mpage_add_bh_to_extent(struct mpage_da_data *mpd,
         * Can we merge the block to our big extent?
         */
        if (logical == next && (b_state & BH_FLAGS) == mpd->b_state) {
-               mpd->b_size += b_size;
+               mpd->b_size += 1 << blkbits;
                return;
        }
 
@@ -2156,7 +2146,6 @@ static int write_cache_pages_da(handle_t *handle,
                                 */
                                if (ext4_bh_delay_or_unwritten(NULL, bh)) {
                                        mpage_add_bh_to_extent(mpd, logical,
-                                                              bh->b_size,
                                                               bh->b_state);
                                        if (mpd->io_done)
                                                goto ret_extent_tail;