NFS: Fix an unstable write data integrity race
Commit 2c61be0a9478258f77b66208a0c4b1f5f8161c3c (NFS: Ensure that the WRITE
and COMMIT RPC calls are always uninterruptible) exposed a race on file
close. In order to ensure correct close-to-open behaviour, we want to wait
for all outstanding background commit operations to complete.
This patch adds an inode flag that indicates if a commit operation is under
way, and provides a mechanism to allow ->write_inode() to wait for its
completion if this is a data integrity flush.
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
diff --git a/fs/nfs/write.c b/fs/nfs/write.c
index de38d63..ccde2ae 100644
--- a/fs/nfs/write.c
+++ b/fs/nfs/write.c
@@ -1201,6 +1201,25 @@
#if defined(CONFIG_NFS_V3) || defined(CONFIG_NFS_V4)
+static int nfs_commit_set_lock(struct nfs_inode *nfsi, int may_wait)
+{
+ if (!test_and_set_bit(NFS_INO_COMMIT, &nfsi->flags))
+ return 1;
+ if (may_wait && !out_of_line_wait_on_bit_lock(&nfsi->flags,
+ NFS_INO_COMMIT, nfs_wait_bit_killable,
+ TASK_KILLABLE))
+ return 1;
+ return 0;
+}
+
+static void nfs_commit_clear_lock(struct nfs_inode *nfsi)
+{
+ clear_bit(NFS_INO_COMMIT, &nfsi->flags);
+ smp_mb__after_clear_bit();
+ wake_up_bit(&nfsi->flags, NFS_INO_COMMIT);
+}
+
+
static void nfs_commitdata_release(void *data)
{
struct nfs_write_data *wdata = data;
@@ -1262,8 +1281,6 @@
task = rpc_run_task(&task_setup_data);
if (IS_ERR(task))
return PTR_ERR(task);
- if (how & FLUSH_SYNC)
- rpc_wait_for_completion_task(task);
rpc_put_task(task);
return 0;
}
@@ -1294,6 +1311,7 @@
BDI_RECLAIMABLE);
nfs_clear_page_tag_locked(req);
}
+ nfs_commit_clear_lock(NFS_I(inode));
return -ENOMEM;
}
@@ -1349,6 +1367,7 @@
next:
nfs_clear_page_tag_locked(req);
}
+ nfs_commit_clear_lock(NFS_I(data->inode));
nfs_commitdata_release(calldata);
}
@@ -1363,8 +1382,11 @@
static int nfs_commit_inode(struct inode *inode, int how)
{
LIST_HEAD(head);
- int res;
+ int may_wait = how & FLUSH_SYNC;
+ int res = 0;
+ if (!nfs_commit_set_lock(NFS_I(inode), may_wait))
+ goto out;
spin_lock(&inode->i_lock);
res = nfs_scan_commit(inode, &head, 0, 0);
spin_unlock(&inode->i_lock);
@@ -1372,7 +1394,13 @@
int error = nfs_commit_list(inode, &head, how);
if (error < 0)
return error;
- }
+ if (may_wait)
+ wait_on_bit(&NFS_I(inode)->flags, NFS_INO_COMMIT,
+ nfs_wait_bit_killable,
+ TASK_KILLABLE);
+ } else
+ nfs_commit_clear_lock(NFS_I(inode));
+out:
return res;
}