NFS: Fix a deadlock with lazy umount
We can't allow rpc callback functions like task->tk_ops->rpc_call_prepare()
and task->tk_ops->rpc_call_done() to call mntput() in any way, since
that will cause a deadlock when the call to rpc_shutdown_client() attempts
to wait on 'task' to complete.
We can avoid the above deadlock by moving calls to mntput to
task->tk_ops->rpc_release() callback, since at that time the task will be
marked as completed, and so rpc_shutdown_client won't attempt to wait on
it.
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
diff --git a/fs/nfs/direct.c b/fs/nfs/direct.c
index 16844f9..e017040 100644
--- a/fs/nfs/direct.c
+++ b/fs/nfs/direct.c
@@ -323,7 +323,7 @@
data->inode = inode;
data->cred = msg.rpc_cred;
data->args.fh = NFS_FH(inode);
- data->args.context = ctx;
+ data->args.context = get_nfs_open_context(ctx);
data->args.offset = pos;
data->args.pgbase = pgbase;
data->args.pages = data->pagevec;
@@ -546,6 +546,7 @@
data->args.fh = NFS_FH(data->inode);
data->args.offset = 0;
data->args.count = 0;
+ data->args.context = get_nfs_open_context(dreq->ctx);
data->res.count = 0;
data->res.fattr = &data->fattr;
data->res.verf = &data->verf;
@@ -728,7 +729,7 @@
data->inode = inode;
data->cred = msg.rpc_cred;
data->args.fh = NFS_FH(inode);
- data->args.context = ctx;
+ data->args.context = get_nfs_open_context(ctx);
data->args.offset = pos;
data->args.pgbase = pgbase;
data->args.pages = data->pagevec;
diff --git a/fs/nfs/inode.c b/fs/nfs/inode.c
index 966a885..a499fb5 100644
--- a/fs/nfs/inode.c
+++ b/fs/nfs/inode.c
@@ -521,8 +521,12 @@
static void __put_nfs_open_context(struct nfs_open_context *ctx, int wait)
{
- struct inode *inode = ctx->path.dentry->d_inode;
+ struct inode *inode;
+ if (ctx == NULL)
+ return;
+
+ inode = ctx->path.dentry->d_inode;
if (!atomic_dec_and_lock(&ctx->count, &inode->i_lock))
return;
list_del(&ctx->list);
diff --git a/fs/nfs/read.c b/fs/nfs/read.c
index 3d7d963..fab0d37 100644
--- a/fs/nfs/read.c
+++ b/fs/nfs/read.c
@@ -73,7 +73,10 @@
void nfs_readdata_release(void *data)
{
- nfs_readdata_free(data);
+ struct nfs_read_data *rdata = data;
+
+ put_nfs_open_context(rdata->args.context);
+ nfs_readdata_free(rdata);
}
static
@@ -186,7 +189,7 @@
data->args.pgbase = req->wb_pgbase + offset;
data->args.pages = data->pagevec;
data->args.count = count;
- data->args.context = req->wb_context;
+ data->args.context = get_nfs_open_context(req->wb_context);
data->res.fattr = &data->fattr;
data->res.count = count;
diff --git a/fs/nfs/write.c b/fs/nfs/write.c
index 80c61fdb..69b4158 100644
--- a/fs/nfs/write.c
+++ b/fs/nfs/write.c
@@ -105,8 +105,11 @@
call_rcu_bh(&wdata->task.u.tk_rcu, nfs_writedata_rcu_free);
}
-void nfs_writedata_release(void *wdata)
+void nfs_writedata_release(void *data)
{
+ struct nfs_write_data *wdata = data;
+
+ put_nfs_open_context(wdata->args.context);
nfs_writedata_free(wdata);
}
@@ -816,7 +819,7 @@
data->args.pgbase = req->wb_pgbase + offset;
data->args.pages = data->pagevec;
data->args.count = count;
- data->args.context = req->wb_context;
+ data->args.context = get_nfs_open_context(req->wb_context);
data->args.stable = NFS_UNSTABLE;
if (how & FLUSH_STABLE) {
data->args.stable = NFS_DATA_SYNC;
@@ -1153,8 +1156,11 @@
#if defined(CONFIG_NFS_V3) || defined(CONFIG_NFS_V4)
-void nfs_commit_release(void *wdata)
+void nfs_commit_release(void *data)
{
+ struct nfs_write_data *wdata = data;
+
+ put_nfs_open_context(wdata->args.context);
nfs_commit_free(wdata);
}
@@ -1197,6 +1203,7 @@
/* Note: we always request a commit of the entire inode */
data->args.offset = 0;
data->args.count = 0;
+ data->args.context = get_nfs_open_context(first->wb_context);
data->res.count = 0;
data->res.fattr = &data->fattr;
data->res.verf = &data->verf;
