From ffc161df9f271bdc1237f555be1edc663566f2ee Mon Sep 17 00:00:00 2001
From: Konstantin Belousov <kib@FreeBSD.org>
Date: Fri, 11 Mar 2016 11:51:38 +0000
Subject: [PATCH] Do not perform unneccessary shared recursion on the
 allproc_lock in pfs_visible().  The recursion does not cause deadlock because
 the sx implementation does not prefer exclusive waiters over the shared, but
 this is an implementation detail.

Reported by:	pho, Matthew Bryan <matthew.bryan@isilon.com>
Reviewed by:	jhb
Tested by:	pho
Approved by:	des (pseudofs maintainer)
Sponsored by:	The FreeBSD Foundation
MFC after:	2 weeks
---
 sys/fs/pseudofs/pseudofs_vnops.c | 22 ++++++++++++----------
 1 file changed, 12 insertions(+), 10 deletions(-)

diff --git a/sys/fs/pseudofs/pseudofs_vnops.c b/sys/fs/pseudofs/pseudofs_vnops.c
index f00b4b2ad5e4..7a9ec7b5e7f8 100644
--- a/sys/fs/pseudofs/pseudofs_vnops.c
+++ b/sys/fs/pseudofs/pseudofs_vnops.c
@@ -104,7 +104,8 @@ pfs_visible_proc(struct thread *td, struct pfs_node *pn, struct proc *proc)
 }
 
 static int
-pfs_visible(struct thread *td, struct pfs_node *pn, pid_t pid, struct proc **p)
+pfs_visible(struct thread *td, struct pfs_node *pn, pid_t pid,
+    bool allproc_locked, struct proc **p)
 {
 	struct proc *proc;
 
@@ -115,7 +116,8 @@ pfs_visible(struct thread *td, struct pfs_node *pn, pid_t pid, struct proc **p)
 		*p = NULL;
 	if (pid == NO_PID)
 		PFS_RETURN (1);
-	if ((proc = pfind(pid)) == NULL)
+	proc = allproc_locked ? pfind_locked(pid) : pfind(pid);
+	if (proc == NULL)
 		PFS_RETURN (0);
 	if (pfs_visible_proc(td, pn, proc)) {
 		if (p)
@@ -202,7 +204,7 @@ pfs_getattr(struct vop_getattr_args *va)
 	PFS_TRACE(("%s", pn->pn_name));
 	pfs_assert_not_owned(pn);
 
-	if (!pfs_visible(curthread, pn, pvd->pvd_pid, &proc))
+	if (!pfs_visible(curthread, pn, pvd->pvd_pid, false, &proc))
 		PFS_RETURN (ENOENT);
 
 	vap->va_type = vn->v_type;
@@ -293,7 +295,7 @@ pfs_ioctl(struct vop_ioctl_args *va)
 	 * This is necessary because process' privileges may
 	 * have changed since the open() call.
 	 */
-	if (!pfs_visible(curthread, pn, pvd->pvd_pid, &proc)) {
+	if (!pfs_visible(curthread, pn, pvd->pvd_pid, false, &proc)) {
 		VOP_UNLOCK(vn, 0);
 		PFS_RETURN (EIO);
 	}
@@ -326,7 +328,7 @@ pfs_getextattr(struct vop_getextattr_args *va)
 	 * This is necessary because either process' privileges may
 	 * have changed since the open() call.
 	 */
-	if (!pfs_visible(curthread, pn, pvd->pvd_pid, &proc))
+	if (!pfs_visible(curthread, pn, pvd->pvd_pid, false, &proc))
 		PFS_RETURN (EIO);
 
 	if (pn->pn_getextattr == NULL)
@@ -462,7 +464,7 @@ pfs_lookup(struct vop_cachedlookup_args *va)
 		PFS_RETURN (ENOENT);
 
 	/* check that parent directory is visible... */
-	if (!pfs_visible(curthread, pd, pvd->pvd_pid, NULL))
+	if (!pfs_visible(curthread, pd, pvd->pvd_pid, false, NULL))
 		PFS_RETURN (ENOENT);
 
 	/* self */
@@ -546,7 +548,7 @@ pfs_lookup(struct vop_cachedlookup_args *va)
  got_pnode:
 	pfs_assert_not_owned(pd);
 	pfs_assert_not_owned(pn);
-	visible = pfs_visible(curthread, pn, pid, NULL);
+	visible = pfs_visible(curthread, pn, pid, false, NULL);
 	if (!visible) {
 		error = ENOENT;
 		goto failed;
@@ -635,7 +637,7 @@ pfs_read(struct vop_read_args *va)
 	 * This is necessary because either process' privileges may
 	 * have changed since the open() call.
 	 */
-	if (!pfs_visible(curthread, pn, pvd->pvd_pid, &proc))
+	if (!pfs_visible(curthread, pn, pvd->pvd_pid, false, &proc))
 		PFS_RETURN (EIO);
 	if (proc != NULL) {
 		_PHOLD(proc);
@@ -791,7 +793,7 @@ pfs_readdir(struct vop_readdir_args *va)
 	pfs_lock(pd);
 
         /* check if the directory is visible to the caller */
-        if (!pfs_visible(curthread, pd, pid, &proc)) {
+        if (!pfs_visible(curthread, pd, pid, true, &proc)) {
 		sx_sunlock(&allproc_lock);
 		pfs_unlock(pd);
                 PFS_RETURN (ENOENT);
@@ -995,7 +997,7 @@ pfs_write(struct vop_write_args *va)
 	 * This is necessary because either process' privileges may
 	 * have changed since the open() call.
 	 */
-	if (!pfs_visible(curthread, pn, pvd->pvd_pid, &proc))
+	if (!pfs_visible(curthread, pn, pvd->pvd_pid, false, &proc))
 		PFS_RETURN (EIO);
 	if (proc != NULL) {
 		_PHOLD(proc);