[PATCH] autofs4 - bad lookup fix

[PATCH] autofs4 - bad lookup fix

[raven]

For browsable autofs maps, a mount request that arrives at the same time 
an expire is happening can fail to perform the needed mount.

This happens becuase the directory exists and so the revalidate succeeds 
when we need it to fail so that lookup is called on the same dentry to do 
the mount. Instead lookup is called on the next path component which 
should be whithin the mount, but the parent isn't mounted. 

The solution is to allow the revalidate to continue and perform the mount 
as no directory creation (at mount time) is needed for browsable mount 
entries.


diff -Nurp linux-2.6.12-rc5-mm1.orig/fs/autofs4/root.c linux-2.6.12-rc5-mm1/fs/autofs4/root.c
--- linux-2.6.12-rc5-mm1.orig/fs/autofs4/root.c	2005-05-29 14:46:30.000000000 +0800
+++ linux-2.6.12-rc5-mm1/fs/autofs4/root.c	2005-05-29 14:47:04.000000000 +0800
@@ -306,7 +306,14 @@ static int try_to_fill_dentry(struct den
 		
 		DPRINTK("expire done status=%d", status);
 		
-		return 0;
+		/*
+		 * If the directory still exists the mount request must
+		 * continue otherwise it can't be followed at the right
+		 * time during the walk.
+		 */
+		status = d_invalidate(dentry);
+		if (status != -EBUSY)
+			return 0;
 	}
 
 	DPRINTK("dentry=%p %.*s ino=%p",

Re: [PATCH] autofs4 - bad lookup fix

[Jeff Moyer]

==> Regarding [PATCH] autofs4 - bad lookup fix; raven@themaw.net adds:

raven> For browsable autofs maps, a mount request that arrives at the same
raven> time an expire is happening can fail to perform the needed mount.

raven> This happens becuase the directory exists and so the revalidate
raven> succeeds when we need it to fail so that lookup is called on the
raven> same dentry to do the mount. Instead lookup is called on the next
raven> path component which should be whithin the mount, but the parent
raven> isn't mounted.

raven> The solution is to allow the revalidate to continue and perform the
raven> mount as no directory creation (at mount time) is needed for
raven> browsable mount entries.

I have a reproducer case for this, and the patch provided fixes it in my
environment.

-Jeff