Commit 1f214e9c authored Oct 31, 2025 by Trond Myklebust Committed by Anna Schumaker Nov 10, 2025

NFSv4: Fix an incorrect parameter when calling nfs4_call_sync()



The Smatch static checker noted that in _nfs4_proc_lookupp(), the flag
RPC_TASK_TIMEOUT is being passed as an argument to nfs4_init_sequence(),
which is clearly incorrect.
Since LOOKUPP is an idempotent operation, nfs4_init_sequence() should
not ask the server to cache the result. The RPC_TASK_TIMEOUT flag needs
to be passed down to the RPC layer.

Reported-by: Dan Carpenter <dan.carpenter@linaro.org>
Reported-by: Harshit Mogalapalli <harshit.m.mogalapalli@oracle.com>
Fixes: 76998ebb ("NFSv4: Observe the NFS_MOUNT_SOFTREVAL flag in _nfs4_proc_lookupp")
Signed-off-by: Trond Myklebust <trond.myklebust@hammerspace.com>
Signed-off-by: Anna Schumaker <anna.schumaker@oracle.com>

parent 7a7a3456

fs/nfs/nfs4proc.c

+6 −3

Original line number	Diff line number	Diff line
		@@ -4715,15 +4715,18 @@ static int _nfs4_proc_lookupp(struct inode *inode,
		};
		unsigned short task_flags = 0;

		if (NFS_SERVER(inode)->flags & NFS_MOUNT_SOFTREVAL)
		if (server->flags & NFS_MOUNT_SOFTREVAL)
		task_flags \|= RPC_TASK_TIMEOUT;
		if (server->caps & NFS_CAP_MOVEABLE)
		task_flags \|= RPC_TASK_MOVEABLE;

		args.bitmask = nfs4_bitmask(server, fattr->label);

		nfs_fattr_init(fattr);
		nfs4_init_sequence(&args.seq_args, &res.seq_res, 0, 0);

		dprintk("NFS call lookupp ino=0x%lx\n", inode->i_ino);
		status = nfs4_call_sync(clnt, server, &msg, &args.seq_args,
		status = nfs4_do_call_sync(clnt, server, &msg, &args.seq_args,
		&res.seq_res, task_flags);
		dprintk("NFS reply lookupp: %d\n", status);
		return status;