asynchronous unlock on exit

Hi.

In the course of investigating testing failures in the locking phase of the Connectathon testsuite, I discovered a couple of things. One was that one of the tests in the locking tests was racy when it didn't seem to need to be and two, that the NFS client asynchronously releases locks when a process is exiting.

I developed a patch for the testsuite locking test and am working to try to find a way to get it included into the generally available testsuite.

The harder problem is that the NFS client kernel code asynchronously releasing locks when a process is exiting. I presume that this was added so that a process, which was holding locks when a server goes down, can be completely killed. This sounds good, but leads to several problems.

The Single UNIX Specification Version 3 specifies that: "All locks associated with a file for a given process shall be removed when a file descriptor for that file is closed by that process or the process holding that file descriptor terminates.".

This does not specify whether those locks must be released prior to the completion of the exit processing for the process or not. However, general assumptions seem to be that those locks will be released. This leads to more deterministic behavior under normal circumstances.

The test, test12(), in the Connectathon locking tests, tests the assumption that if a process is holding a lock and gets signaled with a signal that causes it to be killed, that it releases its lock. The file, which was locked, can then be successfully locked by another process.

After further analysis, it appears that only the NFSv2 and NFSv3 clients release locks asynchronously when a process is exiting. The NFSv4 client releases any locks synchronously.

I would propose to make all three versions match and to modify the NFSv2/NFSv3 behavior to match the NFSv4 behavior. The attached patch implements this change.

Thanx...

ps

@@ -155,8 +155,6 @@ static void nlmclnt_release_lockargs(str
 int nlmclnt_proc(struct nlm_host *host, int cmd, struct file_lock *fl)
 {
 	struct nlm_rqst		*call;
-	sigset_t		oldset;
-	unsigned long		flags;
 	int			status;
 
 	nlm_get_host(host);

@@ -168,22 +166,6 @@ int nlmclnt_proc(struct nlm_host *host, 
 	/* Set up the argument struct */
 	nlmclnt_setlockargs(call, fl);
 
-	/* Keep the old signal mask */
-	spin_lock_irqsave(&current->sighand->siglock, flags);
-	oldset = current->blocked;
-
-	/* If we're cleaning up locks because the process is exiting,
-	 * perform the RPC call asynchronously. */
-	if ((IS_SETLK(cmd) || IS_SETLKW(cmd))
-	    && fl->fl_type == F_UNLCK
-	    && (current->flags & PF_EXITING)) {
-		sigfillset(&current->blocked);	/* Mask all signals */
-		recalc_sigpending();
-
-		call->a_flags = RPC_TASK_ASYNC;
-	}
-	spin_unlock_irqrestore(&current->sighand->siglock, flags);
-
 	if (IS_SETLK(cmd) || IS_SETLKW(cmd)) {
 		if (fl->fl_type != F_UNLCK) {
 			call->a_args.block = IS_SETLKW(cmd) ? 1 : 0;

@@ -198,11 +180,6 @@ int nlmclnt_proc(struct nlm_host *host, 
 	fl->fl_ops->fl_release_private(fl);
 	fl->fl_ops = NULL;
 
-	spin_lock_irqsave(&current->sighand->siglock, flags);
-	current->blocked = oldset;
-	recalc_sigpending();
-	spin_unlock_irqrestore(&current->sighand->siglock, flags);
-
 	dprintk("lockd: clnt proc returns %d\n", status);
 	return status;
 }

Hi Peter,

It looks to me as if that the approach you've chosen would basically return us to the old problem that locks don't get cleaned up on the server if/when someone sends the client process a fatal signal. I think a better approach would be to take the one used in NFSv4 of keeping the RPC call asynchronous, and then having the client wait for it to complete.

Typically, the way you would do this is to convert nlmclnt_unlock to use rpc_run_task(), so that you can recover a reference to the rpc_task pointer and then use rpc_wait_for_completion_task() to wait for the RPC call to complete.

Cheers
Trond