autofs-5.0.5 - dont fail mount on access fail

From: Ian Kent <raven@themaw.net>

If we encounter a recursive autofs mount in the mount location

path, and call access(2) to perform the mount, the mount may

succeed but the access(2) call return a fail. In the case where

there are multiple processes waiting on this mount they will all

get the failure return which may be incorrect for other waiters.

Ignoring the return code from the access(2) call and allowing the

mount to go ahead anyway should give the VFS the chance to check

the access for each individual process and so return an accurate

retult.

---

 CHANGELOG      |    1 +

 daemon/spawn.c |   12 +++++++++---

 2 files changed, 10 insertions(+), 3 deletions(-)

diff --git a/CHANGELOG b/CHANGELOG

index 8429e20..88bcc1b 100644

--- a/CHANGELOG

+++ b/CHANGELOG

@@ -14,6 +14,7 @@

 - fix timeout in connect_nb().

 - fix pidof init script usage.

 - check for path mount location in generic module.

+- dont fail mount on access fail.

 03/09/2009 autofs-5.0.5

 -----------------------

diff --git a/daemon/spawn.c b/daemon/spawn.c

index 7c254d9..285f4d7 100644

--- a/daemon/spawn.c

+++ b/daemon/spawn.c

@@ -189,9 +189,15 @@ static int do_spawn(unsigned logopt, unsigned int wait,

 			}

 			setpgrp();

-			/* Trigger the recursive mount */

-			if (access(argv[loc], F_OK) == -1)

-				_exit(errno);

+			/*

+			 * Trigger the recursive mount.

+			 *

+			 * Ignore the access(2) return code as there may be

+			 * multiple waiters for this mount and we need to

+			 * let the  VFS handle access returns to each

+			 * individual waiter.

+			 */

+			access(argv[loc], F_OK);

 			seteuid(0);

 			setegid(0);