<html>
<head>
<meta content="text/html; charset=ISO-8859-1"
http-equiv="Content-Type">
</head>
<body text="#000000" bgcolor="#FFFFFF">
<div class="moz-cite-prefix">On 07/31/12 08:13, Dave Chinner wrote:<br>
</div>
<blockquote cite="mid:20120731001322.GN2877@dastard" type="cite">
<pre wrap="">On Thu, Jul 26, 2012 at 04:56:22PM +0800, Jeff Liu wrote:
</pre>
<blockquote type="cite">
<pre wrap="">Search possbile hole offset for given range from page cache for unwritten and delayed-allocation extents.
Signed-off-by: Jie Liu <a class="moz-txt-link-rfc2396E" href="mailto:jeff.liu@oracle.com"><jeff.liu@oracle.com></a>
Reviewed-by: Mark Tinguely <a class="moz-txt-link-rfc2396E" href="mailto:tinguely@sgi.com"><tinguely@sgi.com></a>
Reviewed-by: Christoph Hellwig <a class="moz-txt-link-rfc2396E" href="mailto:hch@lst.de"><hch@lst.de></a>
Reviewed-by: Dave Chinner <a class="moz-txt-link-rfc2396E" href="mailto:dchinner@redhat.com"><dchinner@redhat.com></a>
---
fs/xfs/xfs_file.c | 90 +++++++++++++++++++++++++++++++++++++++++++++++------
1 files changed, 80 insertions(+), 10 deletions(-)
diff --git a/fs/xfs/xfs_file.c b/fs/xfs/xfs_file.c
index 15acd4d..1b83549 100644
--- a/fs/xfs/xfs_file.c
+++ b/fs/xfs/xfs_file.c
@@ -1296,9 +1296,9 @@ xfs_seek_hole(
        struct xfs_inode        *ip = XFS_I(inode);
        struct xfs_mount        *mp = ip->i_mount;
        loff_t                        uninitialized_var(offset);
-        loff_t                        holeoff;
        xfs_fsize_t                isize;
        xfs_fileoff_t                fsbno;
+        xfs_filblks_t                end;
        uint                        lock;
        int                        error;
@@ -1314,20 +1314,90 @@ xfs_seek_hole(
        }
        fsbno = XFS_B_TO_FSBT(mp, start);
-        error = xfs_bmap_first_unused(NULL, ip, 1, &fsbno, XFS_DATA_FORK);
-        if (error)
-                goto out_unlock;
+        end = XFS_B_TO_FSB(mp, isize);
+
+        for (;;) {
+                struct xfs_bmbt_irec        map[2];
+                int                        nmap = 2;
+
+                error = xfs_bmapi_read(ip, fsbno, end - fsbno, map, &nmap,
+                                 XFS_BMAPI_ENTIRE);
+                if (error)
+                        goto out_unlock;
+
+                /* No extents at given offset, must be beyond EOF */
+                if (nmap == 0) {
+                        error = ENXIO;
+                        goto out_unlock;
+                }
-        holeoff = XFS_FSB_TO_B(mp, fsbno);
-        if (holeoff <= start)
                offset = start;
-        else {
+                /* Landed in a hole, return */
+                if (map[0].br_startblock == HOLESTARTBLOCK)
+                        break;
+                else {
+                        /*
+                         * Landed in an unwritten extent, try to lookup page
+                         * cache to find holes.
+                         */
</pre>
</blockquote>
<pre wrap="">
Laned in a data extent is more correct here.
</pre>
<blockquote type="cite">
<pre wrap="">+                        if (map[0].br_state == XFS_EXT_UNWRITTEN ||
+                         isnullstartblock(map[0].br_startblock)) {
+                                /* Probing page cache start from offset */
+                                if (xfs_find_get_desired_pgoff(inode, &map[0],
+                                                        HOLE_OFF, &offset))
+                                        break;
+                        }
</pre>
</blockquote>
<pre wrap="">
Same logic changes as previous patch:
                if (map[0].br_startblock == HOLESTARTBLOCK)
                        break;
                else if (map[0].br_state == XFS_EXT_UNWRITTEN) {
                        ASSERT(!isnullstartblock(map[0].br_startblock));
                        /* Probing page cache start from offset */
                        if (xfs_find_get_desired_pgoff(inode, &map[0],
                                                HOLE_OFF, &offset))
                                break;
                } else {
                        .....
</pre>
<blockquote type="cite">
<pre wrap="">                 /*
-                 * xfs_bmap_first_unused() could return a value bigger than
-                 * isize if there are no more holes past the supplied offset.
+                 * Both mappings contains data, proceed to the next round of
+                 * search if the current reading offset not beyond or hit EOF.
                 */
-                offset = min_t(loff_t, holeoff, isize);
+                fsbno = map[1].br_startoff + map[1].br_blockcount;
+                start = XFS_FSB_TO_B(mp, fsbno);
+                if (start >= isize) {
+                        offset = isize;
+                        break;
+                }
</pre>
</blockquote>
<pre wrap="">
And that check can move to the start of the loop, too.</pre>
</blockquote>
Thanks for a comprehensive review, I'll fix them according to your
comments.<br>
<br>
Thanks,<br>
-Jeff<br>
<span class="st"><em></em></span>
<blockquote cite="mid:20120731001322.GN2877@dastard" type="cite">
<pre wrap="">
Cheers,
Dave.
</pre>
</blockquote>
<br>
</body>
</html>