xfs: report iomap_offset and iomap_bsize in block base
Christoph Hellwig [Wed, 28 Apr 2010 12:28:54 +0000 (12:28 +0000)]
Report the iomap_offset and iomap_bsize fields of struct xfs_iomap
in terms of fsblocks instead of in terms of disk blocks.  Shift the
byte conversions into the callers temporarily, but they will
disappear or get cleaned up later.

Signed-off-by: Christoph Hellwig <hch@lst.de>
Signed-off-by: Alex Elder <aelder@sgi.com>

fs/xfs/linux-2.6/xfs_aops.c
fs/xfs/xfs_iomap.c

index 577d0b4..303a779 100644 (file)
@@ -320,11 +320,16 @@ xfs_map_blocks(
 
 STATIC int
 xfs_iomap_valid(
+       struct inode            *inode,
        xfs_iomap_t             *iomapp,
        loff_t                  offset)
 {
-       return offset >= iomapp->iomap_offset &&
-               offset < iomapp->iomap_offset + iomapp->iomap_bsize;
+       struct xfs_mount        *mp = XFS_I(inode)->i_mount;
+       xfs_off_t               iomap_offset = XFS_FSB_TO_B(mp, iomapp->iomap_offset);
+       xfs_off_t               iomap_bsize = XFS_FSB_TO_B(mp, iomapp->iomap_bsize);
+
+       return offset >= iomap_offset &&
+               offset < iomap_offset + iomap_bsize;
 }
 
 /*
@@ -561,11 +566,13 @@ xfs_map_buffer(
        xfs_off_t               offset)
 {
        sector_t                bn;
+       struct xfs_mount        *m = XFS_I(inode)->i_mount;
+       xfs_off_t               iomap_offset = XFS_FSB_TO_B(m, mp->iomap_offset);
 
        ASSERT(mp->iomap_bn != IOMAP_DADDR_NULL);
 
        bn = (mp->iomap_bn >> (inode->i_blkbits - BBSHIFT)) +
-             ((offset - mp->iomap_offset) >> inode->i_blkbits);
+             ((offset - iomap_offset) >> inode->i_blkbits);
 
        ASSERT(bn || XFS_IS_REALTIME_INODE(XFS_I(inode)));
 
@@ -806,7 +813,7 @@ xfs_convert_page(
                        else
                                type = IOMAP_DELAY;
 
-                       if (!xfs_iomap_valid(mp, offset)) {
+                       if (!xfs_iomap_valid(inode, mp, offset)) {
                                done = 1;
                                continue;
                        }
@@ -1116,7 +1123,7 @@ xfs_page_state_convert(
                }
 
                if (iomap_valid)
-                       iomap_valid = xfs_iomap_valid(&iomap, offset);
+                       iomap_valid = xfs_iomap_valid(inode, &iomap, offset);
 
                /*
                 * First case, map an unwritten extent and prepare for
@@ -1171,7 +1178,7 @@ xfs_page_state_convert(
                                                &iomap, flags);
                                if (err)
                                        goto error;
-                               iomap_valid = xfs_iomap_valid(&iomap, offset);
+                               iomap_valid = xfs_iomap_valid(inode, &iomap, offset);
                        }
                        if (iomap_valid) {
                                xfs_map_at_offset(inode, bh, &iomap, offset);
@@ -1201,7 +1208,7 @@ xfs_page_state_convert(
                                                &iomap, flags);
                                if (err)
                                        goto error;
-                               iomap_valid = xfs_iomap_valid(&iomap, offset);
+                               iomap_valid = xfs_iomap_valid(inode, &iomap, offset);
                        }
 
                        /*
@@ -1241,7 +1248,11 @@ xfs_page_state_convert(
                xfs_start_page_writeback(page, 1, count);
 
        if (ioend && iomap_valid) {
-               offset = (iomap.iomap_offset + iomap.iomap_bsize - 1) >>
+               struct xfs_mount        *m = XFS_I(inode)->i_mount;
+               xfs_off_t               iomap_offset = XFS_FSB_TO_B(m, iomap.iomap_offset);
+               xfs_off_t               iomap_bsize = XFS_FSB_TO_B(m, iomap.iomap_bsize);
+
+               offset = (iomap_offset + iomap_bsize - 1) >>
                                        PAGE_CACHE_SHIFT;
                tlast = min_t(pgoff_t, offset, last_index);
                xfs_cluster_write(inode, page->index + 1, &iomap, &ioend,
@@ -1512,11 +1523,14 @@ __xfs_get_blocks(
        }
 
        if (direct || size > (1 << inode->i_blkbits)) {
-               xfs_off_t iomap_delta = offset - iomap.iomap_offset;
+               struct xfs_mount        *mp = XFS_I(inode)->i_mount;
+               xfs_off_t               iomap_offset = XFS_FSB_TO_B(mp, iomap.iomap_offset);
+               xfs_off_t               iomap_delta = offset - iomap_offset;
+               xfs_off_t               iomap_bsize = XFS_FSB_TO_B(mp, iomap.iomap_bsize);
 
-               ASSERT(iomap.iomap_bsize - iomap_delta > 0);
+               ASSERT(iomap_bsize - iomap_delta > 0);
                offset = min_t(xfs_off_t,
-                               iomap.iomap_bsize - iomap_delta, size);
+                               iomap_bsize - iomap_delta, size);
                bh_result->b_size = (ssize_t)min_t(xfs_off_t, LONG_MAX, offset);
        }
 
index c6b409e..49b5ad2 100644 (file)
@@ -64,11 +64,10 @@ xfs_imap_to_bmap(
        int             imaps,                  /* Number of imap entries */
        int             flags)
 {
-       xfs_mount_t     *mp = ip->i_mount;
        xfs_fsblock_t   start_block;
 
-       iomapp->iomap_offset = XFS_FSB_TO_B(mp, imap->br_startoff);
-       iomapp->iomap_bsize = XFS_FSB_TO_B(mp, imap->br_blockcount);
+       iomapp->iomap_offset = imap->br_startoff;
+       iomapp->iomap_bsize = imap->br_blockcount;
        iomapp->iomap_flags = flags;
 
        start_block = imap->br_startblock;