From: Andrew Deason <adeason@sinenomine.net>
Date: Fri, 6 Apr 2012 19:56:07 +0000 (-0500)
Subject: LINUX: Do not lookup immediately recursive mtpts
X-Git-Tag: upstream/1.8.0_pre1^2~2125
X-Git-Url: https://git.michaelhowe.org/gitweb/?a=commitdiff_plain;h=d785abd5ec9302bdc1b3c33368246e573e0cd65d;p=packages%2Fo%2Fopenafs.git

LINUX: Do not lookup immediately recursive mtpts

On Linux, having a mountpoint in a volume root that points to the same
volume can cause serious problems. By 'immediately recursive', I mean
a situation like the following:

fs mkm mtpt vol
fs mkm mtpt/mtpt vol

If there are multiple dentry aliases for the directory (which is
possible if the directory is a mountpoint), an 'rmdir' on the
recursive mountpoint can cause the client to deadlock. Since the
'rmdir' code path in Linux locks the parent directory inode to perform
the rmdir, and locks the child directory inode after performing a
couple of sanity checks. For an immediately recursive mountpoint,
these two inodes are the same, and so we will deadlock.

Change-Id: Icb9bf8a3dd77a2ef6b88856b0d41556541bb1d00
Reviewed-on: http://gerrit.openafs.org/7742
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Derrick Brashear <shadow@dementix.org>
---

diff --git a/src/afs/LINUX/osi_vnodeops.c b/src/afs/LINUX/osi_vnodeops.c
index f8dc925f2..7d5e166b0 100644
--- a/src/afs/LINUX/osi_vnodeops.c
+++ b/src/afs/LINUX/osi_vnodeops.c
@@ -1337,6 +1337,18 @@ afs_linux_lookup(struct inode *dip, struct dentry *dp)
     
     if (vcp) {
 	struct vattr vattr;
+	struct vcache *parent_vc = VTOAFS(dip);
+
+	if (parent_vc == vcp) {
+	    /* This is possible if the parent dir is a mountpoint to a volume,
+	     * and the dir entry we looked up is a mountpoint to the same
+	     * volume. Linux cannot cope with this, so return an error instead
+	     * of risking a deadlock or panic. */
+	    afs_PutVCache(vcp);
+	    code = EDEADLK;
+	    AFS_GUNLOCK();
+	    goto done;
+	}
 
 	ip = AFSTOV(vcp);
 	afs_getattr(vcp, &vattr, credp);
@@ -1376,6 +1388,7 @@ afs_linux_lookup(struct inode *dip, struct dentry *dp)
     }
     newdp = d_splice_alias(ip, dp);
 
+ done:
     crfree(credp);
 
     /* It's ok for the file to not be found. That's noted by the caller by
diff --git a/src/afs/LINUX24/osi_vnodeops.c b/src/afs/LINUX24/osi_vnodeops.c
index 679293fcd..46b82b73d 100644
--- a/src/afs/LINUX24/osi_vnodeops.c
+++ b/src/afs/LINUX24/osi_vnodeops.c
@@ -1273,6 +1273,18 @@ afs_linux_lookup(struct inode *dip, struct dentry *dp)
     
     if (vcp) {
 	struct vattr vattr;
+	struct vcache *parent_vc = VTOAFS(dip);
+
+	if (parent_vc == vcp) {
+	    /* This is possible if the parent dir is a mountpoint to a volume,
+	     * and the dir entry we looked up is a mountpoint to the same
+	     * volume. Linux cannot cope with this, so return an error instead
+	     * of risking a deadlock or panic. */
+	    afs_PutVCache(vcp);
+	    code = EDEADLK;
+	    AFS_GUNLOCK();
+	    goto done;
+	}
 
 	ip = AFSTOV(vcp);
 	afs_getattr(vcp, &vattr, credp);
@@ -1314,6 +1326,7 @@ afs_linux_lookup(struct inode *dip, struct dentry *dp)
 #endif
     d_add(dp, ip);
 
+ done:
     crfree(credp);
 
     /* It's ok for the file to not be found. That's noted by the caller by