- The NFS client was incorrectly masking SIGSTOP (which is

  non-maskable).
- The NFS client needs to guard against spurious wakeups
  while waiting for the response. ltrace causes the process
  under question to wakeup (possibly from ptrace()), which
  causes NFS to wakeup from tsleep without the response being
  delivered.

Submitted by:	Mohan Srinivasan

1 files changed, 6 insertions, 19 deletions

diff --git a/sys/nfsclient/nfs_socket.c b/sys/nfsclient/nfs_socket.c
index 290e8ad..de58cd1 100644
--- a/sys/nfsclient/nfs_socket.c
+++ b/sys/nfsclient/nfs_socket.c
@@ -528,7 +528,7 @@ nfs_reply(struct nfsreq *rep)
 {
 	register struct socket *so;
 	register struct mbuf *m;
-	int error, sotype, slpflag;
+	int error = 0, sotype, slpflag;
 
 	NET_ASSERT_GIANT();
 
@@ -582,23 +582,9 @@ tryagain:
 	if (rep->r_nmp->nm_flag & NFSMNT_INT)
 		slpflag = PCATCH;
 	mtx_lock(&nfs_reply_mtx);
-	if (rep->r_mrep != NULL) {
-		/*
-		 * This is a very rare race, but it does occur. The reply 
-		 * could come in and the wakeup could happen before the 
-		 * process tsleeps(). Blocking here without checking for 
-		 * this results in a missed wakeup(), blocking this request 
-		 * forever. The 2 reasons why this could happen are a context
-		 * switch in the stack after the request is sent out, or heavy
-		 * interrupt activity pinning down the process within the window.
-		 * (after the request is sent).
-		 */
-		mtx_unlock(&nfs_reply_mtx);
-		nfs_mrep_before_tsleep++;
-		return (0);
-	}
-	error = msleep((caddr_t)rep, &nfs_reply_mtx, 
-		       slpflag | (PZERO - 1), "nfsreq", 0);
+	while ((rep->r_mrep == NULL) && (error == 0))
+		error = msleep((caddr_t)rep, &nfs_reply_mtx, 
+			       slpflag | (PZERO - 1), "nfsreq", 0);
 	mtx_unlock(&nfs_reply_mtx);
 	if (error == EINTR || error == ERESTART)
 		/* NFS operations aren't restartable. Map ERESTART to EINTR */
@@ -1359,13 +1345,14 @@ nfs_softterm(struct nfsreq *rep)
 
 /*
  * Any signal that can interrupt an NFS operation in an intr mount
- * should be added to this set.
+ * should be added to this set. SIGSTOP and SIGKILL cannot be masked.
  */
 int nfs_sig_set[] = {
 	SIGINT,
 	SIGTERM,
 	SIGHUP,
 	SIGKILL,
+	SIGSTOP,
 	SIGQUIT
 };