xfs
[Top] [All Lists]

[PATCH 16/71] xfs: adjust refcount when unmapping file blocks

To: david@xxxxxxxxxxxxx, darrick.wong@xxxxxxxxxx
Subject: [PATCH 16/71] xfs: adjust refcount when unmapping file blocks
From: "Darrick J. Wong" <darrick.wong@xxxxxxxxxx>
Date: Thu, 25 Aug 2016 16:48:14 -0700
Cc: linux-xfs@xxxxxxxxxxxxxxx, xfs@xxxxxxxxxxx
Delivered-to: xfs@xxxxxxxxxxx
In-reply-to: <147216879156.4420.2446767701729565218.stgit@xxxxxxxxxxxxxxxx>
References: <147216879156.4420.2446767701729565218.stgit@xxxxxxxxxxxxxxxx>
User-agent: StGit/0.17.1-dirty
When we're unmapping blocks from a reflinked file, decrease the
refcount of the affected blocks and free the extents that are no
longer in use.

v2: Use deferred ops system to avoid deadlocks and running out of
transaction reservation.

Signed-off-by: Darrick J. Wong <darrick.wong@xxxxxxxxxx>
---
 libxfs/xfs_bmap.c |   14 +++++++++++---
 1 file changed, 11 insertions(+), 3 deletions(-)


diff --git a/libxfs/xfs_bmap.c b/libxfs/xfs_bmap.c
index a50a8af..4045340 100644
--- a/libxfs/xfs_bmap.c
+++ b/libxfs/xfs_bmap.c
@@ -40,6 +40,7 @@
 #include "xfs_quota_defs.h"
 #include "xfs_rmap.h"
 #include "xfs_ag_resv.h"
+#include "xfs_refcount.h"
 
 
 kmem_zone_t            *xfs_bmap_free_item_zone;
@@ -5054,9 +5055,16 @@ xfs_bmap_del_extent(
        /*
         * If we need to, add to list of extents to delete.
         */
-       if (do_fx)
-               xfs_bmap_add_free(mp, dfops, del->br_startblock,
-                               del->br_blockcount, NULL);
+       if (do_fx) {
+               if (xfs_is_reflink_inode(ip) && whichfork == XFS_DATA_FORK) {
+                       error = xfs_refcount_decrease_extent(mp, dfops, del);
+                       if (error)
+                               goto done;
+               } else
+                       xfs_bmap_add_free(mp, dfops, del->br_startblock,
+                                       del->br_blockcount, NULL);
+       }
+
        /*
         * Adjust inode # blocks in the file.
         */

<Prev in Thread] Current Thread [Next in Thread>