Linux: Make dir dentry aliases act like symlinks

Currently, we try to invalidate other dentries that exist for a
particular dir inode when we look up a dentry. This is so we try to
avoid duplicate dentries for a directory, which Linux does not like
(you cannot have hardlinks to a dir).

If we cannot invalidate the other aliases (because they are being
used), right now we just return the alias. This can make it very easy
to panic the client, due to the sanity checks Linux performs when dong
things like 'rmdir'. If we do something like this:

mkdir dir1
fs mkm dir1/mtpt vol
mkdir dir1/mtpt/dir2
fs mkm dir1/mtpt/dir2/mtpt2 vol
cd dir1/mtpt
rmdir dir2/mtpt2

For the 'rmdir', we will lookup 'mtpt2'. Since 'mtpt' and 'mtpt2'
are mountpoints for the same volume, their dentries point to the same
directory inode. So when we lookup 'mtpt2', we will try to invalidate
the other dentry, but we cannot do that since it is the cwd. So we
return the alias dentry (for 'mtpt'). The Linux VFS layer then does a
sanity check for the rmdir operation, checking that the child dentry's
parent inode is the same as the inode we're performing the rmdir for.
Since the dentry we returned was for 'mtpt', whose parent is 'dir1',
and the actual dir we're performing the rmdir for is 'dir2', this
sanity check fails and we BUG.

To avoid this, make the dentry alias act like a symlink when we
encounter an uninvalidateable dentry alias. That is, we allow multiple
dentry aliases for a directory, however, when the dentry aliases are
actually used, we redirect to a common dentry (via d_automount where
possible, and follow_link elsewhere).

This means that such mountpoints will behave similarly to symlinks, in
that we 'point' to a specific mountpoint dentry. This means that if we
have multiple different ways to get to the same volume, and all are
accessed at the same time, all but one of those mountpoints will
behave like symlinks, pointing to the same mountpoint. So, the '..'
entries for each path will all point to the parent dir of one
mountpoint, meaning that the '..' entry will be "wrong", but for most
cases it will still be correct.

In order to try to make the 'target', pointed-to directory consistent,
we add a new field to struct vcache: target_link. This points to the
dentry we should redirect to, whenever that vcache is referenced. To
avoid (possibly not-feasibly-solvable) problems with refcounting, this
pointer is not actually a reference to the target dentry, but just
serves as a pointer to compare to.

FIXES 130273

Reviewed-on: http://gerrit.openafs.org/7741
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Tested-by: Andrew Deason <adeason@sinenomine.net>
Reviewed-by: Derrick Brashear <shadow@dementix.org>
(cherry picked from commit de381aa0d3)

Change-Id: I42959ab0c09e5b752660cd87293c63ca7978d242
Reviewed-on: http://gerrit.openafs.org/7953
Tested-by: BuildBot <buildbot@rampaginggeek.com>
Reviewed-by: Derrick Brashear <shadow@dementix.org>

acinclude.m4

@@ -740,6 +740,7 @@ case $AFS_SYSNAME in *_linux* | *_umlinux*)
                  LINUX_KBUILD_USES_EXTRA_CFLAGS
 		 LINUX_KERNEL_COMPILE_WORKS
 		 LINUX_EXPORTS_FIND_TASK_BY_PID
+		 AC_CHECK_LINUX_STRUCT([dentry_operations], [d_automount], [dcache.h])
 		 LINUX_EXPORTS_PROC_ROOT_FS
                  LINUX_HAVE_CURRENT_KERNEL_TIME
                  LINUX_HAVE_WRITE_BEGIN_AOP

src/afs/LINUX/osi_vnodeops.c

@@ -734,6 +734,61 @@ struct file_operations afs_file_fops = {
   .llseek = 	default_llseek,
 };
 
+static struct dentry *
+canonical_dentry(struct inode *ip)
+{
+    struct vcache *vcp = VTOAFS(ip);
+    struct dentry *first = NULL, *ret = NULL, *cur;
+
+    /* general strategy:
+     * if vcp->target_link is set, and can be found in ip->i_dentry, use that.
+     * otherwise, use the first dentry in ip->i_dentry.
+     * if ip->i_dentry is empty, use the 'dentry' argument we were given.
+     */
+    /* note that vcp->target_link specifies which dentry to use, but we have
+     * no reference held on that dentry. so, we cannot use or dereference
+     * vcp->target_link itself, since it may have been freed. instead, we only
+     * use it to compare to pointers in the ip->i_dentry list. */
+
+    d_prune_aliases(ip);
+
+# ifdef HAVE_DCACHE_LOCK
+    spin_lock(&dcache_lock);
+# else
+    spin_lock(&ip->i_lock);
+# endif
+
+    list_for_each_entry_reverse(cur, &ip->i_dentry, d_alias) {
+
+	if (!vcp->target_link || cur == vcp->target_link) {
+	    ret = cur;
+	    break;
+	}
+
+	if (!first) {
+	    first = cur;
+	}
+    }
+    if (!ret && first) {
+	ret = first;
+    }
+
+    vcp->target_link = ret;
+
+# ifdef HAVE_DCACHE_LOCK
+    if (ret) {
+	dget_locked(ret);
+    }
+    spin_unlock(&dcache_lock);
+# else
+    if (ret) {
+	dget(ret);
+    }
+    spin_unlock(&ip->i_lock);
+# endif
+
+    return ret;
+}
 
 /**********************************************************************
  * AFS Linux dentry operations
@@ -1021,10 +1076,40 @@ afs_dentry_delete(struct dentry *dp)
     return 0;
 }
 
+#ifdef STRUCT_DENTRY_OPERATIONS_HAS_D_AUTOMOUNT
+static struct vfsmount *
+afs_dentry_automount(struct path *path)
+{
+    struct dentry *target;
+
+    target = canonical_dentry(path->dentry->d_inode);
+
+    if (target == path->dentry) {
+	dput(target);
+	target = NULL;
+    }
+
+    if (target) {
+	dput(path->dentry);
+	path->dentry = target;
+
+    } else {
+	spin_lock(&path->dentry->d_lock);
+	path->dentry->d_flags &= ~DCACHE_NEED_AUTOMOUNT;
+	spin_unlock(&path->dentry->d_lock);
+    }
+
+    return NULL;
+}
+#endif /* STRUCT_DENTRY_OPERATIONS_HAS_D_AUTOMOUNT */
+
 struct dentry_operations afs_dentry_operations = {
   .d_revalidate =	afs_linux_dentry_revalidate,
   .d_delete =		afs_dentry_delete,
   .d_iput =		afs_dentry_iput,
+#ifdef STRUCT_DENTRY_OPERATIONS_HAS_D_AUTOMOUNT
+  .d_automount =        afs_dentry_automount,
+#endif /* STRUCT_DENTRY_OPERATIONS_HAS_D_AUTOMOUNT */
 };
 
 /**********************************************************************
@@ -1136,21 +1221,27 @@ afs_linux_lookup(struct inode *dip, struct dentry *dp)
 
 #if defined(AFS_LINUX24_ENV)
     if (ip && S_ISDIR(ip->i_mode)) {
+	int retry = 1;
 	struct dentry *alias;
 
-	alias = d_find_alias(ip);
-	if (alias) {
-	    if (d_invalidate(alias) == 0) {
-		dput(alias);
-	    } else {
-		iput(ip);
-#if defined(AFS_LINUX26_ENV)
-		unlock_kernel();
-#endif
-		crfree(credp);
-		return alias;
+	while (retry) {
+	    retry = 0;
+
+	    /* Try to invalidate an existing alias in favor of our new one */
+	    alias = d_find_alias(ip);
+	    if (alias) {
+		if (d_invalidate(alias) == 0) {
+		    /* there may be more aliases; try again until we run out */
+		    retry = 1;
+		}
 	    }
+
+	    dput(alias);
 	}
+
+#ifdef STRUCT_DENTRY_OPERATIONS_HAS_D_AUTOMOUNT
+	ip->i_flags |= S_AUTOMOUNT;
+#endif
     }
 #endif
     d_add(dp, ip);
@@ -1906,6 +1997,33 @@ afs_linux_write_begin(struct file *file, struct address_space *mapping,
 }
 #endif
 
+#ifndef STRUCT_DENTRY_OPERATIONS_HAS_D_AUTOMOUNT
+static void *
+afs_linux_dir_follow_link(struct dentry *dentry, struct nameidata *nd)
+{
+    struct dentry **dpp;
+    struct dentry *target;
+
+    target = canonical_dentry(dentry->d_inode);
+
+# ifdef STRUCT_NAMEIDATA_HAS_PATH
+    dpp = &nd->path.dentry;
+# else
+    dpp = &nd->dentry;
+# endif
+
+    dput(*dpp);
+
+    if (target) {
+	*dpp = target;
+    } else {
+	*dpp = dget(dentry);
+    }
+
+    return NULL;
+}
+#endif /* !STRUCT_DENTRY_OPERATIONS_HAS_D_AUTOMOUNT */
+
 static struct inode_operations afs_file_iops = {
 #if defined(AFS_LINUX26_ENV)
   .permission =		afs_linux_permission,
@@ -1962,6 +2080,9 @@ static struct inode_operations afs_dir_iops = {
   .revalidate =		afs_linux_revalidate,
 #endif
   .permission =		afs_linux_permission,
+#ifndef STRUCT_DENTRY_OPERATIONS_HAS_D_AUTOMOUNT
+  .follow_link =        afs_linux_dir_follow_link,
+#endif
 };
 
 /* We really need a separate symlink set of ops, since do_follow_link()

src/afs/afs.h

@@ -751,6 +751,13 @@ struct vcache {
     struct bhv_desc vc_bhv_desc;	/* vnode's behavior data. */
 #endif
 #endif				/* AFS_SGI_ENV */
+#ifdef AFS_LINUX24_ENV
+    struct dentry *target_link; /* dentry we prefer, when we are redirecting
+                                 * all requests due to duplicate dentry aliases.
+                                 * See LINUX/osi_vnodeops.c. Note that this is
+                                 * NOT an actual reference to a dentry, so this
+                                 * pointer MUST NOT be dereferenced on its own. */
+#endif
     afs_int32 vc_error;		/* stash write error for this vnode. */
     int xlatordv;		/* Used by nfs xlator */
     struct AFS_UCRED *uncred;