X-Git-Url: https://git.whamcloud.com/?p=fs%2Flustre-release.git;a=blobdiff_plain;f=lustre%2Fllite%2Fnamei.c;h=97ea1f173b0f5ffb53b3326a2df47f73a69dcfec;hp=0d282115ebc4f8a748ffebe12644521e59074364;hb=2bc5bcb7efa247fcd8cc65d013ffc9f6c33dd788;hpb=5e314327f35535075f77c0b149e96db4c520fdd9

diff --git a/lustre/llite/namei.c b/lustre/llite/namei.c
index 0d28211..97ea1f1 100644
--- a/lustre/llite/namei.c
+++ b/lustre/llite/namei.c
@@ -27,7 +27,7 @@
  * Copyright (c) 2002, 2010, Oracle and/or its affiliates. All rights reserved.
  * Use is subject to license terms.
  *
- * Copyright (c) 2011, 2012, Whamcloud, Inc.
+ * Copyright (c) 2011, 2014, Intel Corporation.
  */
 /*
  * This file is part of Lustre, http://www.lustre.org/
@@ -46,131 +46,114 @@
 
 #include <obd_support.h>
 #include <lustre_fid.h>
-#include <lustre_lite.h>
 #include <lustre_dlm.h>
 #include <lustre_ver.h>
 #include "llite_internal.h"
 
-/*
- * Check if we have something mounted at the named dchild.
- * In such a case there would always be dentry present.
- */
-static int ll_d_mountpoint(struct dentry *dparent, struct dentry *dchild,
-                           struct qstr *name)
-{
-        int mounted = 0;
-
-        if (unlikely(dchild)) {
-                mounted = d_mountpoint(dchild);
-        } else if (dparent) {
-                dchild = d_lookup(dparent, name);
-                if (dchild) {
-                        mounted = d_mountpoint(dchild);
-                        dput(dchild);
-                }
-        }
-        return mounted;
-}
-
-int ll_unlock(__u32 mode, struct lustre_handle *lockh)
-{
-        ENTRY;
-
-        ldlm_lock_decref(lockh, mode);
-
-        RETURN(0);
-}
-
+static int ll_create_it(struct inode *dir, struct dentry *dentry,
+			struct lookup_intent *it);
 
 /* called from iget5_locked->find_inode() under inode_lock spinlock */
 static int ll_test_inode(struct inode *inode, void *opaque)
 {
-        struct ll_inode_info *lli = ll_i2info(inode);
-        struct lustre_md     *md = opaque;
+	struct ll_inode_info	*lli = ll_i2info(inode);
+	struct lustre_md	*md = opaque;
 
-        if (unlikely(!(md->body->valid & OBD_MD_FLID))) {
-                CERROR("MDS body missing FID\n");
-                return 0;
-        }
+	if (unlikely(!(md->body->mbo_valid & OBD_MD_FLID))) {
+		CERROR("MDS body missing FID\n");
+		return 0;
+	}
 
-        if (!lu_fid_eq(&lli->lli_fid, &md->body->fid1))
-                return 0;
+	if (!lu_fid_eq(&lli->lli_fid, &md->body->mbo_fid1))
+		return 0;
 
-        return 1;
+	return 1;
 }
 
 static int ll_set_inode(struct inode *inode, void *opaque)
 {
-        struct ll_inode_info *lli = ll_i2info(inode);
-        struct mdt_body *body = ((struct lustre_md *)opaque)->body;
+	struct ll_inode_info *lli = ll_i2info(inode);
+	struct mdt_body *body = ((struct lustre_md *)opaque)->body;
 
-        if (unlikely(!(body->valid & OBD_MD_FLID))) {
-                CERROR("MDS body missing FID\n");
-                return -EINVAL;
-        }
+	if (unlikely(!(body->mbo_valid & OBD_MD_FLID))) {
+		CERROR("MDS body missing FID\n");
+		return -EINVAL;
+	}
 
-        lli->lli_fid = body->fid1;
-        if (unlikely(!(body->valid & OBD_MD_FLTYPE))) {
-                CERROR("Can not initialize inode "DFID" without object type: "
-                       "valid = "LPX64"\n", PFID(&lli->lli_fid), body->valid);
-                return -EINVAL;
-        }
+	lli->lli_fid = body->mbo_fid1;
+	if (unlikely(!(body->mbo_valid & OBD_MD_FLTYPE))) {
+		CERROR("Can not initialize inode "DFID" without object type: "
+		       "valid = "LPX64"\n",
+		       PFID(&lli->lli_fid), body->mbo_valid);
+		return -EINVAL;
+	}
 
-        inode->i_mode = (inode->i_mode & ~S_IFMT) | (body->mode & S_IFMT);
-        if (unlikely(inode->i_mode == 0)) {
-                CERROR("Invalid inode "DFID" type\n", PFID(&lli->lli_fid));
-                return -EINVAL;
-        }
+	inode->i_mode = (inode->i_mode & ~S_IFMT) | (body->mbo_mode & S_IFMT);
+	if (unlikely(inode->i_mode == 0)) {
+		CERROR("Invalid inode "DFID" type\n", PFID(&lli->lli_fid));
+		return -EINVAL;
+	}
 
-        ll_lli_init(lli);
+	ll_lli_init(lli);
 
-        return 0;
+	return 0;
 }
 
 
-/*
- * Get an inode by inode number (already instantiated by the intent lookup).
- * Returns inode or NULL
+/**
+ * Get an inode by inode number(@hash), which is already instantiated by
+ * the intent lookup).
  */
 struct inode *ll_iget(struct super_block *sb, ino_t hash,
                       struct lustre_md *md)
 {
-        struct inode         *inode;
-        ENTRY;
+	struct inode	*inode;
+	int		rc = 0;
+
+	ENTRY;
 
         LASSERT(hash != 0);
         inode = iget5_locked(sb, hash, ll_test_inode, ll_set_inode, md);
+	if (inode == NULL)
+		RETURN(ERR_PTR(-ENOMEM));
+
+	if (inode->i_state & I_NEW) {
+		rc = ll_read_inode2(inode, md);
+		if (rc == 0 && S_ISREG(inode->i_mode) &&
+		    ll_i2info(inode)->lli_clob == NULL) {
+			CDEBUG(D_INODE, "%s: apply lsm %p to inode "DFID"\n",
+				ll_get_fsname(sb, NULL, 0), md->lsm,
+				PFID(ll_inode2fid(inode)));
+			rc = cl_file_inode_init(inode, md);
+		}
+		if (rc != 0) {
+			make_bad_inode(inode);
+			unlock_new_inode(inode);
+			iput(inode);
+			inode = ERR_PTR(rc);
+		} else {
+			unlock_new_inode(inode);
+		}
+	} else if (!(inode->i_state & (I_FREEING | I_CLEAR))) {
+		rc = ll_update_inode(inode, md);
+		CDEBUG(D_VFSTRACE, "got inode: "DFID"(%p): rc = %d\n",
+		       PFID(&md->body->mbo_fid1), inode, rc);
+		if (rc != 0) {
+			iput(inode);
+			inode = ERR_PTR(rc);
+		}
+	}
 
-        if (inode) {
-                if (inode->i_state & I_NEW) {
-                        int rc = 0;
-
-                        ll_read_inode2(inode, md);
-                        if (S_ISREG(inode->i_mode) &&
-                            ll_i2info(inode)->lli_clob == NULL)
-                                rc = cl_file_inode_init(inode, md);
-                        if (rc != 0) {
-                                md->lsm = NULL;
-                                make_bad_inode(inode);
-                                unlock_new_inode(inode);
-                                iput(inode);
-                                inode = ERR_PTR(rc);
-                        } else
-                                unlock_new_inode(inode);
-                } else if (!(inode->i_state & (I_FREEING | I_CLEAR)))
-                        ll_update_inode(inode, md);
-                CDEBUG(D_VFSTRACE, "got inode: %p for "DFID"\n",
-                       inode, PFID(&md->body->fid1));
-        }
         RETURN(inode);
 }
 
 static void ll_invalidate_negative_children(struct inode *dir)
 {
 	struct dentry *dentry, *tmp_subdir;
+	DECLARE_LL_D_HLIST_NODE_PTR(p);
 
 	ll_lock_dcache(dir);
-	list_for_each_entry(dentry, &dir->i_dentry, d_alias) {
+	ll_d_hlist_for_each_entry(dentry, p, &dir->i_dentry, d_alias) {
 		spin_lock(&dentry->d_lock);
 		if (!list_empty(&dentry->d_subdirs)) {
 			struct dentry *child;
@@ -179,7 +162,7 @@ static void ll_invalidate_negative_children(struct inode *dir)
 						 &dentry->d_subdirs,
 						 d_u.d_child) {
 				if (child->d_inode == NULL)
-					d_lustre_invalidate(child);
+					d_lustre_invalidate(child, 1);
 			}
 		}
 		spin_unlock(&dentry->d_lock);
@@ -187,111 +170,164 @@ static void ll_invalidate_negative_children(struct inode *dir)
 	ll_unlock_dcache(dir);
 }
 
+int ll_test_inode_by_fid(struct inode *inode, void *opaque)
+{
+	return lu_fid_eq(&ll_i2info(inode)->lli_fid, opaque);
+}
+
 int ll_md_blocking_ast(struct ldlm_lock *lock, struct ldlm_lock_desc *desc,
-                       void *data, int flag)
+		       void *data, int flag)
 {
-        int rc;
-        struct lustre_handle lockh;
-        ENTRY;
+	struct lustre_handle lockh;
+	int rc;
+	ENTRY;
 
-        switch (flag) {
-        case LDLM_CB_BLOCKING:
-                ldlm_lock2handle(lock, &lockh);
-                rc = ldlm_cli_cancel(&lockh);
-                if (rc < 0) {
-                        CDEBUG(D_INODE, "ldlm_cli_cancel: %d\n", rc);
-                        RETURN(rc);
-                }
-                break;
-        case LDLM_CB_CANCELING: {
-                struct inode *inode = ll_inode_from_lock(lock);
-                struct ll_inode_info *lli;
-                __u64 bits = lock->l_policy_data.l_inodebits.bits;
-                struct lu_fid *fid;
-                ldlm_mode_t mode = lock->l_req_mode;
-
-                /* Invalidate all dentries associated with this inode */
-                if (inode == NULL)
-                        break;
-
-                LASSERT(lock->l_flags & LDLM_FL_CANCELING);
-                /* For OPEN locks we differentiate between lock modes
+	switch (flag) {
+	case LDLM_CB_BLOCKING:
+		ldlm_lock2handle(lock, &lockh);
+		rc = ldlm_cli_cancel(&lockh, LCF_ASYNC);
+		if (rc < 0) {
+			CDEBUG(D_INODE, "ldlm_cli_cancel: rc = %d\n", rc);
+			RETURN(rc);
+		}
+		break;
+	case LDLM_CB_CANCELING: {
+		struct inode *inode = ll_inode_from_resource_lock(lock);
+		__u64 bits = lock->l_policy_data.l_inodebits.bits;
+
+		/* Inode is set to lock->l_resource->lr_lvb_inode
+		 * for mdc - bug 24555 */
+		LASSERT(lock->l_ast_data == NULL);
+
+		if (inode == NULL)
+			break;
+
+		/* Invalidate all dentries associated with this inode */
+		LASSERT(ldlm_is_canceling(lock));
+
+		if (!fid_res_name_eq(ll_inode2fid(inode),
+				     &lock->l_resource->lr_name)) {
+			LDLM_ERROR(lock, "data mismatch with object "DFID"(%p)",
+				   PFID(ll_inode2fid(inode)), inode);
+			LBUG();
+		}
+
+		if (bits & MDS_INODELOCK_XATTR) {
+			ll_xattr_cache_destroy(inode);
+			bits &= ~MDS_INODELOCK_XATTR;
+		}
+
+		/* For OPEN locks we differentiate between lock modes
 		 * LCK_CR, LCK_CW, LCK_PR - bug 22891 */
-		if (bits & (MDS_INODELOCK_LOOKUP | MDS_INODELOCK_UPDATE |
-			    MDS_INODELOCK_LAYOUT))
-                        ll_have_md_lock(inode, &bits, LCK_MINMODE);
-
-                if (bits & MDS_INODELOCK_OPEN)
-                        ll_have_md_lock(inode, &bits, mode);
-
-                fid = ll_inode2fid(inode);
-                if (lock->l_resource->lr_name.name[0] != fid_seq(fid) ||
-                    lock->l_resource->lr_name.name[1] != fid_oid(fid) ||
-                    lock->l_resource->lr_name.name[2] != fid_ver(fid)) {
-                        LDLM_ERROR(lock, "data mismatch with object "
-                                   DFID" (%p)", PFID(fid), inode);
-                }
+		if (bits & MDS_INODELOCK_OPEN)
+			ll_have_md_lock(inode, &bits, lock->l_req_mode);
+
+		if (bits & MDS_INODELOCK_OPEN) {
+			fmode_t fmode;
+
+			switch (lock->l_req_mode) {
+			case LCK_CW:
+				fmode = FMODE_WRITE;
+				break;
+			case LCK_PR:
+				fmode = FMODE_EXEC;
+				break;
+			case LCK_CR:
+				fmode = FMODE_READ;
+				break;
+			default:
+				LDLM_ERROR(lock, "bad lock mode for OPEN lock");
+				LBUG();
+			}
 
-                if (bits & MDS_INODELOCK_OPEN) {
-                        int flags = 0;
-                        switch (lock->l_req_mode) {
-                        case LCK_CW:
-                                flags = FMODE_WRITE;
-                                break;
-                        case LCK_PR:
-                                flags = FMODE_EXEC;
-                                break;
-                        case LCK_CR:
-                                flags = FMODE_READ;
-                                break;
-                        default:
-                                CERROR("Unexpected lock mode for OPEN lock "
-                                       "%d, inode %ld\n", lock->l_req_mode,
-                                       inode->i_ino);
-                        }
-                        ll_md_real_close(inode, flags);
-                }
+			ll_md_real_close(inode, fmode);
+
+			bits &= ~MDS_INODELOCK_OPEN;
+		}
+
+		if (bits & (MDS_INODELOCK_LOOKUP | MDS_INODELOCK_UPDATE |
+			    MDS_INODELOCK_LAYOUT | MDS_INODELOCK_PERM))
+			ll_have_md_lock(inode, &bits, LCK_MINMODE);
 
-		lli = ll_i2info(inode);
 		if (bits & MDS_INODELOCK_LAYOUT) {
-			struct cl_object_conf conf = { .coc_inode = inode,
-						       .coc_invalidate = true };
+			struct cl_object_conf conf = {
+				.coc_opc = OBJECT_CONF_INVALIDATE,
+				.coc_inode = inode,
+			};
+
 			rc = ll_layout_conf(inode, &conf);
-			if (rc)
-				CDEBUG(D_INODE, "invaliding layout %d.\n", rc);
+			if (rc < 0)
+				CDEBUG(D_INODE, "cannot invalidate layout of "
+				       DFID": rc = %d\n",
+				       PFID(ll_inode2fid(inode)), rc);
 		}
 
-                if (bits & MDS_INODELOCK_UPDATE)
-                        lli->lli_flags &= ~LLIF_MDS_SIZE_LOCK;
+		if (bits & MDS_INODELOCK_UPDATE) {
+			struct ll_inode_info *lli = ll_i2info(inode);
+
+			spin_lock(&lli->lli_lock);
+			lli->lli_flags &= ~LLIF_MDS_SIZE_LOCK;
+			spin_unlock(&lli->lli_lock);
+		}
 
-                if (S_ISDIR(inode->i_mode) &&
-                     (bits & MDS_INODELOCK_UPDATE)) {
-                        CDEBUG(D_INODE, "invalidating inode %lu\n",
-                               inode->i_ino);
-                        truncate_inode_pages(inode->i_mapping, 0);
-			ll_invalidate_negative_children(inode);
+		if ((bits & MDS_INODELOCK_UPDATE) && S_ISDIR(inode->i_mode)) {
+			struct ll_inode_info *lli = ll_i2info(inode);
+
+			CDEBUG(D_INODE, "invalidating inode "DFID" lli = %p, "
+			       "pfid  = "DFID"\n", PFID(ll_inode2fid(inode)),
+			       lli, PFID(&lli->lli_pfid));
+			truncate_inode_pages(inode->i_mapping, 0);
+
+			if (unlikely(!fid_is_zero(&lli->lli_pfid))) {
+				struct inode *master_inode = NULL;
+				unsigned long hash;
+
+				/* This is slave inode, since all of the child
+				 * dentry is connected on the master inode, so
+				 * we have to invalidate the negative children
+				 * on master inode */
+				CDEBUG(D_INODE, "Invalidate s"DFID" m"DFID"\n",
+				       PFID(ll_inode2fid(inode)),
+				       PFID(&lli->lli_pfid));
+
+				hash = cl_fid_build_ino(&lli->lli_pfid,
+					ll_need_32bit_api(ll_i2sbi(inode)));
+
+				master_inode = ilookup5(inode->i_sb, hash,
+							ll_test_inode_by_fid,
+							(void *)&lli->lli_pfid);
+				if (master_inode != NULL &&
+					!IS_ERR(master_inode)) {
+					ll_invalidate_negative_children(
+								master_inode);
+					iput(master_inode);
+				}
+			} else {
+				ll_invalidate_negative_children(inode);
+			}
 		}
 
-		if (inode->i_sb->s_root &&
-		    inode != inode->i_sb->s_root->d_inode &&
-		    (bits & MDS_INODELOCK_LOOKUP))
+		if ((bits & (MDS_INODELOCK_LOOKUP | MDS_INODELOCK_PERM)) &&
+		    inode->i_sb->s_root != NULL &&
+		    inode != inode->i_sb->s_root->d_inode)
 			ll_invalidate_aliases(inode);
-                iput(inode);
-                break;
-        }
-        default:
-                LBUG();
-        }
 
-        RETURN(0);
+		iput(inode);
+		break;
+	}
+	default:
+		LBUG();
+	}
+
+	RETURN(0);
 }
 
 __u32 ll_i2suppgid(struct inode *i)
 {
-        if (cfs_curproc_is_in_groups(i->i_gid))
-                return (__u32)i->i_gid;
-        else
-                return (__u32)(-1);
+	if (in_group_p(i->i_gid))
+		return (__u32)from_kgid(&init_user_ns, i->i_gid);
+	else
+		return (__u32) __kgid_val(INVALID_GID);
 }
 
 /* Pack the required supplementary groups into the supplied groups array.
@@ -300,35 +336,15 @@ __u32 ll_i2suppgid(struct inode *i)
  * array in case it might be useful.  Not needed if doing an MDS-side upcall. */
 void ll_i2gids(__u32 *suppgids, struct inode *i1, struct inode *i2)
 {
-#if 0
-        int i;
-#endif
+	LASSERT(i1 != NULL);
+	LASSERT(suppgids != NULL);
 
-        LASSERT(i1 != NULL);
-        LASSERT(suppgids != NULL);
+	suppgids[0] = ll_i2suppgid(i1);
 
-        suppgids[0] = ll_i2suppgid(i1);
-
-        if (i2)
-                suppgids[1] = ll_i2suppgid(i2);
-                else
-                        suppgids[1] = -1;
-
-#if 0
-        for (i = 0; i < current_ngroups; i++) {
-                if (suppgids[0] == -1) {
-                        if (current_groups[i] != suppgids[1])
-                                suppgids[0] = current_groups[i];
-                        continue;
-                }
-                if (suppgids[1] == -1) {
-                        if (current_groups[i] != suppgids[0])
-                                suppgids[1] = current_groups[i];
-                        continue;
-                }
-                break;
-        }
-#endif
+	if (i2)
+		suppgids[1] = ll_i2suppgid(i2);
+	else
+		suppgids[1] = -1;
 }
 
 /*
@@ -342,14 +358,15 @@ void ll_i2gids(__u32 *suppgids, struct inode *i1, struct inode *i2)
 static struct dentry *ll_find_alias(struct inode *inode, struct dentry *dentry)
 {
 	struct dentry *alias, *discon_alias, *invalid_alias;
+	DECLARE_LL_D_HLIST_NODE_PTR(p);
 
-	if (list_empty(&inode->i_dentry))
+	if (ll_d_hlist_empty(&inode->i_dentry))
 		return NULL;
 
 	discon_alias = invalid_alias = NULL;
 
 	ll_lock_dcache(inode);
-	list_for_each_entry(alias, &inode->i_dentry, d_alias) {
+	ll_d_hlist_for_each_entry(alias, p, &inode->i_dentry, d_alias) {
 		LASSERT(alias != dentry);
 
 		spin_lock(&alias->d_lock);
@@ -379,48 +396,54 @@ static struct dentry *ll_find_alias(struct inode *inode, struct dentry *dentry)
 }
 
 /*
- * Similar to d_splice_alias(), but lustre treats DCACHE_LUSTRE_INVALID alias
+ * Similar to d_splice_alias(), but lustre treats invalid alias
  * similar to DCACHE_DISCONNECTED, and tries to use it anyway.
  */
 struct dentry *ll_splice_alias(struct inode *inode, struct dentry *de)
 {
 	struct dentry *new;
+	int rc;
 
 	if (inode) {
 		new = ll_find_alias(inode, de);
 		if (new) {
-			ll_dops_init(new, 1, 1);
+			rc = ll_d_init(new);
+			if (rc < 0) {
+				dput(new);
+				return ERR_PTR(rc);
+			}
 			d_move(new, de);
 			iput(inode);
 			CDEBUG(D_DENTRY,
 			       "Reuse dentry %p inode %p refc %d flags %#x\n",
-			      new, new->d_inode, d_refcount(new), new->d_flags);
+			      new, new->d_inode, ll_d_count(new), new->d_flags);
 			return new;
 		}
 	}
-	__d_lustre_invalidate(de);
-	ll_dops_init(de, 1, 1);
+	rc = ll_d_init(de);
+	if (rc < 0)
+		return ERR_PTR(rc);
 	d_add(de, inode);
 	CDEBUG(D_DENTRY, "Add dentry %p inode %p refc %d flags %#x\n",
-	       de, de->d_inode, d_refcount(de), de->d_flags);
+	       de, de->d_inode, ll_d_count(de), de->d_flags);
         return de;
 }
 
-int ll_lookup_it_finish(struct ptlrpc_request *request,
-                        struct lookup_intent *it, void *data)
+static int ll_lookup_it_finish(struct ptlrpc_request *request,
+			       struct lookup_intent *it,
+			       struct inode *parent, struct dentry **de)
 {
-        struct it_cb_data *icbd = data;
-        struct dentry **de = icbd->icbd_childp;
-        struct inode *parent = icbd->icbd_parent;
-        struct inode *inode = NULL;
-	__u64 bits = 0;
-	int rc;
+	struct inode		 *inode = NULL;
+	__u64			  bits = 0;
+	int			  rc;
 	ENTRY;
 
 	/* NB 1 request reference will be taken away by ll_intent_lock()
 	 * when I return */
+	CDEBUG(D_DENTRY, "it %p it_disposition %x\n", it,
+	       it->d.lustre.it_disposition);
 	if (!it_disposition(it, DISP_LOOKUP_NEG)) {
-                rc = ll_prep_inode(&inode, request, (*de)->d_sb);
+                rc = ll_prep_inode(&inode, request, (*de)->d_sb, it);
                 if (rc)
                         RETURN(rc);
 
@@ -436,20 +459,51 @@ int ll_lookup_it_finish(struct ptlrpc_request *request,
                    Also see bug 7198. */
 	}
 
-	*de = ll_splice_alias(inode, *de);
+	/* Only hash *de if it is unhashed (new dentry).
+	 * Atoimc_open may passin hashed dentries for open.
+	 */
+	if (d_unhashed(*de)) {
+		struct dentry *alias;
+
+		alias = ll_splice_alias(inode, *de);
+		if (IS_ERR(alias))
+			RETURN(PTR_ERR(alias));
+		*de = alias;
+	} else if (!it_disposition(it, DISP_LOOKUP_NEG)  &&
+		   !it_disposition(it, DISP_OPEN_CREATE)) {
+		/* With DISP_OPEN_CREATE dentry will
+		   instantiated in ll_create_it. */
+		LASSERT((*de)->d_inode == NULL);
+		d_instantiate(*de, inode);
+	}
 
 	if (!it_disposition(it, DISP_LOOKUP_NEG)) {
 		/* we have lookup look - unhide dentry */
 		if (bits & MDS_INODELOCK_LOOKUP)
 			d_lustre_revalidate(*de);
-	} else {
+	} else if (!it_disposition(it, DISP_OPEN_CREATE)) {
+		/* If file created on server, don't depend on parent UPDATE
+		 * lock to unhide it. It is left hidden and next lookup can
+		 * find it in ll_splice_alias.
+		 */
 		/* Check that parent has UPDATE lock. */
 		struct lookup_intent parent_it = {
 					.it_op = IT_GETATTR,
 					.d.lustre.it_lock_handle = 0 };
+		struct lu_fid	fid = ll_i2info(parent)->lli_fid;
+
+		/* If it is striped directory, get the real stripe parent */
+		if (unlikely(ll_i2info(parent)->lli_lsm_md != NULL)) {
+			rc = md_get_fid_from_lsm(ll_i2mdexp(parent),
+						 ll_i2info(parent)->lli_lsm_md,
+						 (*de)->d_name.name,
+						 (*de)->d_name.len, &fid);
+			if (rc != 0)
+				RETURN(rc);
+		}
 
-		if (md_revalidate_lock(ll_i2mdexp(parent), &parent_it,
-				       &ll_i2info(parent)->lli_fid, NULL)) {
+		if (md_revalidate_lock(ll_i2mdexp(parent), &parent_it, &fid,
+				       NULL)) {
 			d_lustre_revalidate(*de);
 			ll_intent_release(&parent_it);
 		}
@@ -459,13 +513,12 @@ int ll_lookup_it_finish(struct ptlrpc_request *request,
 }
 
 static struct dentry *ll_lookup_it(struct inode *parent, struct dentry *dentry,
-                                   struct lookup_intent *it, int lookup_flags)
+				   struct lookup_intent *it)
 {
-        struct lookup_intent lookup_it = { .it_op = IT_LOOKUP };
-        struct dentry *save = dentry, *retval;
-        struct ptlrpc_request *req = NULL;
-        struct md_op_data *op_data;
-        struct it_cb_data icbd;
+	struct lookup_intent lookup_it = { .it_op = IT_LOOKUP };
+	struct dentry *save = dentry, *retval;
+	struct ptlrpc_request *req = NULL;
+	struct md_op_data *op_data = NULL;
         __u32 opc;
         int rc;
         ENTRY;
@@ -473,58 +526,68 @@ static struct dentry *ll_lookup_it(struct inode *parent, struct dentry *dentry,
         if (dentry->d_name.len > ll_i2sbi(parent)->ll_namelen)
                 RETURN(ERR_PTR(-ENAMETOOLONG));
 
-        CDEBUG(D_VFSTRACE, "VFS Op:name=%.*s,dir=%lu/%u(%p),intent=%s\n",
-               dentry->d_name.len, dentry->d_name.name, parent->i_ino,
-               parent->i_generation, parent, LL_IT2STR(it));
+	CDEBUG(D_VFSTRACE, "VFS Op:name=%.*s, dir="DFID"(%p), intent=%s\n",
+	       dentry->d_name.len, dentry->d_name.name,
+	       PFID(ll_inode2fid(parent)), parent, LL_IT2STR(it));
 
         if (d_mountpoint(dentry))
                 CERROR("Tell Peter, lookup on mtpt, it %s\n", LL_IT2STR(it));
 
-        ll_frob_intent(&it, &lookup_it);
+	if (it == NULL || it->it_op == IT_GETXATTR)
+		it = &lookup_it;
 
-        /* As do_lookup is called before follow_mount, root dentry may be left
-         * not valid, revalidate it here. */
-        if (parent->i_sb->s_root && (parent->i_sb->s_root->d_inode == parent) &&
-            (it->it_op & (IT_OPEN | IT_CREAT))) {
-                rc = ll_inode_revalidate_it(parent->i_sb->s_root, it,
-                                            MDS_INODELOCK_LOOKUP);
-                if (rc)
-                        RETURN(ERR_PTR(rc));
-        }
-
-        if (it->it_op == IT_GETATTR) {
-                rc = ll_statahead_enter(parent, &dentry, 0);
-                if (rc == 1) {
-                        if (dentry == save)
-                                GOTO(out, retval = NULL);
-                        GOTO(out, retval = dentry);
-                }
-        }
-
-        icbd.icbd_childp = &dentry;
-        icbd.icbd_parent = parent;
-
-        if (it->it_op & IT_CREAT ||
-            (it->it_op & IT_OPEN && it->it_create_mode & O_CREAT))
-                opc = LUSTRE_OPC_CREATE;
-        else
-                opc = LUSTRE_OPC_ANY;
-
-        op_data = ll_prep_md_op_data(NULL, parent, NULL, dentry->d_name.name,
-                                     dentry->d_name.len, lookup_flags, opc,
-                                     NULL);
-        if (IS_ERR(op_data))
-                RETURN((void *)op_data);
+	if (it->it_op == IT_GETATTR && dentry_may_statahead(parent, dentry)) {
+		rc = ll_statahead(parent, &dentry, 0);
+		if (rc == 1)
+			RETURN(dentry == save ? NULL : dentry);
+	}
 
-        it->it_create_mode &= ~cfs_curproc_umask();
+	if (it->it_op & IT_CREAT)
+		opc = LUSTRE_OPC_CREATE;
+	else
+		opc = LUSTRE_OPC_ANY;
+
+	op_data = ll_prep_md_op_data(NULL, parent, NULL, dentry->d_name.name,
+				     dentry->d_name.len, 0, opc, NULL);
+	if (IS_ERR(op_data))
+		RETURN((void *)op_data);
+
+	/* enforce umask if acl disabled or MDS doesn't support umask */
+	if (!IS_POSIXACL(parent) || !exp_connect_umask(ll_i2mdexp(parent)))
+		it->it_create_mode &= ~current_umask();
+
+	rc = md_intent_lock(ll_i2mdexp(parent), op_data, it, &req,
+			    &ll_md_blocking_ast, 0);
+	/* If the MDS allows the client to chgrp (CFS_SETGRP_PERM), but the
+	 * client does not know which suppgid should be sent to the MDS, or
+	 * some other(s) changed the target file's GID after this RPC sent
+	 * to the MDS with the suppgid as the original GID, then we should
+	 * try again with right suppgid. */
+	if (rc == -EACCES && it->it_op & IT_OPEN &&
+	    it_disposition(it, DISP_OPEN_DENY)) {
+		struct mdt_body *body;
+
+		LASSERT(req != NULL);
+
+		body = req_capsule_server_get(&req->rq_pill, &RMF_MDT_BODY);
+		if (op_data->op_suppgids[0] == body->mbo_gid ||
+		    op_data->op_suppgids[1] == body->mbo_gid ||
+		    !in_group_p(make_kgid(&init_user_ns, body->mbo_gid)))
+			GOTO(out, retval = ERR_PTR(-EACCES));
+
+		fid_zero(&op_data->op_fid2);
+		op_data->op_suppgids[1] = body->mbo_gid;
+		ptlrpc_req_finished(req);
+		req = NULL;
+		ll_intent_release(it);
+		rc = md_intent_lock(ll_i2mdexp(parent), op_data, it, &req,
+				    &ll_md_blocking_ast, 0);
+	}
 
-        rc = md_intent_lock(ll_i2mdexp(parent), op_data, NULL, 0, it,
-                            lookup_flags, &req, ll_md_blocking_ast, 0);
-        ll_finish_md_op_data(op_data);
-        if (rc < 0)
-                GOTO(out, retval = ERR_PTR(rc));
+	if (rc < 0)
+		GOTO(out, retval = ERR_PTR(rc));
 
-        rc = ll_lookup_it_finish(req, it, &icbd);
+	rc = ll_lookup_it_finish(req, it, parent, &dentry);
         if (rc != 0) {
                 ll_intent_release(it);
                 GOTO(out, retval = ERR_PTR(rc));
@@ -537,38 +600,149 @@ static struct dentry *ll_lookup_it(struct inode *parent, struct dentry *dentry,
         }
         ll_lookup_finish_locks(it, dentry);
 
-        if (dentry == save)
-                GOTO(out, retval = NULL);
-        else
-                GOTO(out, retval = dentry);
- out:
-        if (req)
-                ptlrpc_req_finished(req);
-        if (it->it_op == IT_GETATTR && (retval == NULL || retval == dentry))
-                ll_statahead_mark(parent, dentry);
-        return retval;
+	GOTO(out, retval = (dentry == save) ? NULL : dentry);
+
+out:
+	if (op_data != NULL && !IS_ERR(op_data))
+		ll_finish_md_op_data(op_data);
+
+	ptlrpc_req_finished(req);
+	return retval;
 }
 
-struct lookup_intent *ll_convert_intent(struct open_intent *oit,
-                                        int lookup_flags)
+#ifdef HAVE_IOP_ATOMIC_OPEN
+static struct dentry *ll_lookup_nd(struct inode *parent, struct dentry *dentry,
+				   unsigned int flags)
 {
-        struct lookup_intent *it;
+	struct lookup_intent *itp, it = { .it_op = IT_GETATTR };
+	struct dentry *de;
 
-        OBD_ALLOC(it, sizeof(*it));
-        if (!it)
-                return ERR_PTR(-ENOMEM);
-
-        if (lookup_flags & LOOKUP_OPEN) {
-                it->it_op = IT_OPEN;
-                if (lookup_flags & LOOKUP_CREATE)
-                        it->it_op |= IT_CREAT;
-                it->it_create_mode = (oit->create_mode & S_IALLUGO) | S_IFREG;
-                it->it_flags = oit->flags;
-        } else {
-                it->it_op = IT_GETATTR;
-        }
+	CDEBUG(D_VFSTRACE, "VFS Op:name=%.*s, dir="DFID"(%p), flags=%u\n",
+	       dentry->d_name.len, dentry->d_name.name,
+	       PFID(ll_inode2fid(parent)), parent, flags);
+
+	/* Optimize away (CREATE && !OPEN). Let .create handle the race. */
+	if ((flags & LOOKUP_CREATE) && !(flags & LOOKUP_OPEN))
+		return NULL;
+
+	if (flags & (LOOKUP_PARENT|LOOKUP_OPEN|LOOKUP_CREATE))
+		itp = NULL;
+	else
+		itp = &it;
+	de = ll_lookup_it(parent, dentry, itp);
+
+	if (itp != NULL)
+		ll_intent_release(itp);
+
+	return de;
+}
+
+/*
+ * For cached negative dentry and new dentry, handle lookup/create/open
+ * together.
+ */
+static int ll_atomic_open(struct inode *dir, struct dentry *dentry,
+			  struct file *file, unsigned open_flags,
+			  umode_t mode, int *opened)
+{
+	struct lookup_intent *it;
+	struct dentry *de;
+	long long lookup_flags = LOOKUP_OPEN;
+	int rc = 0;
+	ENTRY;
+
+	CDEBUG(D_VFSTRACE, "VFS Op:name=%.*s, dir="DFID"(%p), file %p,"
+			   "open_flags %x, mode %x opened %d\n",
+	       dentry->d_name.len, dentry->d_name.name,
+	       PFID(ll_inode2fid(dir)), dir, file, open_flags, mode, *opened);
+
+	OBD_ALLOC(it, sizeof(*it));
+	if (!it)
+		RETURN(-ENOMEM);
+
+	it->it_op = IT_OPEN;
+	if (open_flags & O_CREAT) {
+		it->it_op |= IT_CREAT;
+		lookup_flags |= LOOKUP_CREATE;
+	}
+	it->it_create_mode = (mode & S_IALLUGO) | S_IFREG;
+	it->it_flags = (open_flags & ~O_ACCMODE) | OPEN_FMODE(open_flags);
+	it->it_flags &= ~MDS_OPEN_FL_INTERNAL;
+
+	/* Dentry added to dcache tree in ll_lookup_it */
+	de = ll_lookup_it(dir, dentry, it);
+	if (IS_ERR(de))
+		rc = PTR_ERR(de);
+	else if (de != NULL)
+		dentry = de;
+
+	if (!rc) {
+		if (it_disposition(it, DISP_OPEN_CREATE)) {
+			/* Dentry instantiated in ll_create_it. */
+			rc = ll_create_it(dir, dentry, it);
+			if (rc) {
+				/* We dget in ll_splice_alias. */
+				if (de != NULL)
+					dput(de);
+				goto out_release;
+			}
+
+			*opened |= FILE_CREATED;
+		}
+		if (dentry->d_inode && it_disposition(it, DISP_OPEN_OPEN)) {
+			/* Open dentry. */
+			if (S_ISFIFO(dentry->d_inode->i_mode)) {
+				/* We cannot call open here as it would
+				 * deadlock.
+				 */
+				if (it_disposition(it, DISP_ENQ_OPEN_REF))
+					ptlrpc_req_finished(
+						       (struct ptlrpc_request *)
+							  it->d.lustre.it_data);
+				rc = finish_no_open(file, de);
+			} else {
+				file->private_data = it;
+				rc = finish_open(file, dentry, NULL, opened);
+				/* We dget in ll_splice_alias. finish_open takes
+				 * care of dget for fd open.
+				 */
+				if (de != NULL)
+					dput(de);
+			}
+		} else {
+			rc = finish_no_open(file, de);
+		}
+	}
+
+out_release:
+	ll_intent_release(it);
+	OBD_FREE(it, sizeof(*it));
+
+	RETURN(rc);
+}
+
+#else /* !HAVE_IOP_ATOMIC_OPEN */
+static struct lookup_intent *
+ll_convert_intent(struct open_intent *oit, int lookup_flags)
+{
+	struct lookup_intent *it;
+
+	OBD_ALLOC_PTR(it);
+	if (!it)
+		return ERR_PTR(-ENOMEM);
+
+	if (lookup_flags & LOOKUP_OPEN) {
+		it->it_op = IT_OPEN;
+		if (lookup_flags & LOOKUP_CREATE)
+			it->it_op |= IT_CREAT;
+		it->it_create_mode = (oit->create_mode & S_IALLUGO) | S_IFREG;
+		it->it_flags = ll_namei_to_lookup_intent_flag(oit->flags);
+		it->it_flags &= ~MDS_OPEN_FL_INTERNAL;
+	} else {
+		it->it_op = IT_GETATTR;
+	}
 
-        return it;
+	return it;
 }
 
 static struct dentry *ll_lookup_nd(struct inode *parent, struct dentry *dentry,
@@ -584,20 +758,16 @@ static struct dentry *ll_lookup_nd(struct inode *parent, struct dentry *dentry,
                         it = ll_d2d(dentry)->lld_it;
                         ll_d2d(dentry)->lld_it = NULL;
                 } else {
-                        if ((nd->flags & LOOKUP_CREATE ) && !(nd->flags & LOOKUP_OPEN)) {
-                                /* We are sure this is new dentry, so we need to create
-                                   our private data and set the dentry ops */ 
-                                ll_dops_init(dentry, 1, 1);
-				__d_lustre_invalidate(dentry);
-				d_add(dentry, NULL);
+			if ((nd->flags & LOOKUP_CREATE) &&
+			    !(nd->flags & LOOKUP_OPEN))
                                 RETURN(NULL);
-                        }
+
                         it = ll_convert_intent(&nd->intent.open, nd->flags);
                         if (IS_ERR(it))
                                 RETURN((struct dentry *)it);
                 }
 
-                de = ll_lookup_it(parent, dentry, it, nd->flags);
+		de = ll_lookup_it(parent, dentry, it);
                 if (de)
                         dentry = de;
                 if ((nd->flags & LOOKUP_OPEN) && !IS_ERR(dentry)) { /* Open */
@@ -638,17 +808,15 @@ static struct dentry *ll_lookup_nd(struct inode *parent, struct dentry *dentry,
                         OBD_FREE(it, sizeof(*it));
                 }
         } else {
-                de = ll_lookup_it(parent, dentry, NULL, 0);
-        }
+		de = ll_lookup_it(parent, dentry, NULL);
+	}
 
-        RETURN(de);
+	RETURN(de);
 }
+#endif /* HAVE_IOP_ATOMIC_OPEN */
 
 /* We depend on "mode" being set with the proper file type/umask by now */
-static struct inode *ll_create_node(struct inode *dir, const char *name,
-                                    int namelen, const void *data, int datalen,
-                                    int mode, __u64 extra,
-                                    struct lookup_intent *it)
+static struct inode *ll_create_node(struct inode *dir, struct lookup_intent *it)
 {
         struct inode *inode = NULL;
         struct ptlrpc_request *request = NULL;
@@ -661,17 +829,17 @@ static struct inode *ll_create_node(struct inode *dir, const char *name,
         LASSERT(it_disposition(it, DISP_ENQ_CREATE_REF));
         request = it->d.lustre.it_data;
         it_clear_disposition(it, DISP_ENQ_CREATE_REF);
-        rc = ll_prep_inode(&inode, request, dir->i_sb);
+        rc = ll_prep_inode(&inode, request, dir->i_sb, it);
         if (rc)
                 GOTO(out, inode = ERR_PTR(rc));
 
-        LASSERT(list_empty(&inode->i_dentry));
+	LASSERT(ll_d_hlist_empty(&inode->i_dentry));
 
         /* We asked for a lock on the directory, but were granted a
          * lock on the inode.  Since we finally have an inode pointer,
          * stuff it in the lock. */
-        CDEBUG(D_DLMTRACE, "setting l_ast_data to inode %p (%lu/%u)\n",
-               inode, inode->i_ino, inode->i_generation);
+	CDEBUG(D_DLMTRACE, "setting l_ast_data to inode "DFID"(%p)\n",
+	       PFID(ll_inode2fid(inode)), inode);
         ll_set_lock_data(sbi->ll_md_exp, inode, it, NULL);
         EXIT;
  out:
@@ -693,52 +861,52 @@ static struct inode *ll_create_node(struct inode *dir, const char *name,
  * If the create succeeds, we fill in the inode information
  * with d_instantiate().
  */
-static int ll_create_it(struct inode *dir, struct dentry *dentry, int mode,
-                        struct lookup_intent *it)
+static int ll_create_it(struct inode *dir, struct dentry *dentry,
+			struct lookup_intent *it)
 {
-        struct inode *inode;
-        int rc = 0;
-        ENTRY;
+	struct inode *inode;
+	int rc = 0;
+	ENTRY;
 
-        CDEBUG(D_VFSTRACE, "VFS Op:name=%.*s,dir=%lu/%u(%p),intent=%s\n",
-               dentry->d_name.len, dentry->d_name.name, dir->i_ino,
-               dir->i_generation, dir, LL_IT2STR(it));
+	CDEBUG(D_VFSTRACE, "VFS Op:name=%.*s, dir="DFID"(%p), intent=%s\n",
+	       dentry->d_name.len, dentry->d_name.name,
+	       PFID(ll_inode2fid(dir)), dir, LL_IT2STR(it));
 
-        rc = it_open_error(DISP_OPEN_CREATE, it);
-        if (rc)
-                RETURN(rc);
+	rc = it_open_error(DISP_OPEN_CREATE, it);
+	if (rc)
+		RETURN(rc);
 
-        inode = ll_create_node(dir, dentry->d_name.name, dentry->d_name.len,
-                               NULL, 0, mode, 0, it);
-        if (IS_ERR(inode))
-                RETURN(PTR_ERR(inode));
+	inode = ll_create_node(dir, it);
+	if (IS_ERR(inode))
+		RETURN(PTR_ERR(inode));
 
-        d_instantiate(dentry, inode);
-        RETURN(0);
+	d_instantiate(dentry, inode);
+	RETURN(0);
 }
 
-static void ll_update_times(struct ptlrpc_request *request,
-                            struct inode *inode)
+void ll_update_times(struct ptlrpc_request *request, struct inode *inode)
 {
-        struct mdt_body *body = req_capsule_server_get(&request->rq_pill,
-                                                       &RMF_MDT_BODY);
-
-        LASSERT(body);
-        if (body->valid & OBD_MD_FLMTIME &&
-            body->mtime > LTIME_S(inode->i_mtime)) {
-                CDEBUG(D_INODE, "setting ino %lu mtime from %lu to "LPU64"\n",
-                       inode->i_ino, LTIME_S(inode->i_mtime), body->mtime);
-                LTIME_S(inode->i_mtime) = body->mtime;
-        }
-        if (body->valid & OBD_MD_FLCTIME &&
-            body->ctime > LTIME_S(inode->i_ctime))
-                LTIME_S(inode->i_ctime) = body->ctime;
+	struct mdt_body *body = req_capsule_server_get(&request->rq_pill,
+						       &RMF_MDT_BODY);
+
+	LASSERT(body);
+	if (body->mbo_valid & OBD_MD_FLMTIME &&
+	    body->mbo_mtime > LTIME_S(inode->i_mtime)) {
+		CDEBUG(D_INODE, "setting fid "DFID" mtime from %lu to "LPU64
+		       "\n", PFID(ll_inode2fid(inode)),
+		       LTIME_S(inode->i_mtime), body->mbo_mtime);
+		LTIME_S(inode->i_mtime) = body->mbo_mtime;
+	}
+
+	if (body->mbo_valid & OBD_MD_FLCTIME &&
+	    body->mbo_ctime > LTIME_S(inode->i_ctime))
+		LTIME_S(inode->i_ctime) = body->mbo_ctime;
 }
 
-static int ll_new_node(struct inode *dir, struct qstr *name,
-                       const char *tgt, int mode, int rdev,
-                       struct dentry *dchild, __u32 opc)
+static int ll_new_node(struct inode *dir, struct dentry *dchild,
+		       const char *tgt, umode_t mode, int rdev, __u32 opc)
 {
+	struct qstr *name = &dchild->d_name;
         struct ptlrpc_request *request = NULL;
         struct md_op_data *op_data;
         struct inode *inode = NULL;
@@ -755,22 +923,22 @@ static int ll_new_node(struct inode *dir, struct qstr *name,
         if (IS_ERR(op_data))
                 GOTO(err_exit, err = PTR_ERR(op_data));
 
-        err = md_create(sbi->ll_md_exp, op_data, tgt, tgt_len, mode,
-                        cfs_curproc_fsuid(), cfs_curproc_fsgid(),
-                        cfs_curproc_cap_pack(), rdev, &request);
-        ll_finish_md_op_data(op_data);
+	err = md_create(sbi->ll_md_exp, op_data, tgt, tgt_len, mode,
+			from_kuid(&init_user_ns, current_fsuid()),
+			from_kgid(&init_user_ns, current_fsgid()),
+			cfs_curproc_cap_pack(), rdev, &request);
+	ll_finish_md_op_data(op_data);
         if (err)
                 GOTO(err_exit, err);
 
         ll_update_times(request, dir);
 
-        if (dchild) {
-                err = ll_prep_inode(&inode, request, dchild->d_sb);
-                if (err)
-                     GOTO(err_exit, err);
+	err = ll_prep_inode(&inode, request, dchild->d_sb, NULL);
+	if (err)
+		GOTO(err_exit, err);
+
+	d_instantiate(dchild, inode);
 
-                d_instantiate(dchild, inode);
-        }
         EXIT;
 err_exit:
         ptlrpc_req_finished(request);
@@ -778,17 +946,19 @@ err_exit:
         return err;
 }
 
-static int ll_mknod_generic(struct inode *dir, struct qstr *name, int mode,
-                            unsigned rdev, struct dentry *dchild)
+static int ll_mknod(struct inode *dir, struct dentry *dchild, ll_umode_t mode,
+		    dev_t rdev)
 {
-        int err;
+	struct qstr *name = &dchild->d_name;
+	int err;
         ENTRY;
 
-        CDEBUG(D_VFSTRACE, "VFS Op:name=%.*s,dir=%lu/%u(%p) mode %o dev %x\n",
-               name->len, name->name, dir->i_ino, dir->i_generation, dir,
+	CDEBUG(D_VFSTRACE, "VFS Op:name=%.*s, dir="DFID"(%p) mode %o dev %x\n",
+	       name->len, name->name, PFID(ll_inode2fid(dir)), dir,
                mode, rdev);
 
-        mode &= ~cfs_curproc_umask();
+	if (!IS_POSIXACL(dir) || !exp_connect_umask(ll_i2mdexp(dir)))
+		mode &= ~current_umask();
 
         switch (mode & S_IFMT) {
         case 0:
@@ -798,8 +968,8 @@ static int ll_mknod_generic(struct inode *dir, struct qstr *name, int mode,
         case S_IFBLK:
         case S_IFIFO:
         case S_IFSOCK:
-                err = ll_new_node(dir, name, NULL, mode, rdev, dchild,
-                                  LUSTRE_OPC_MKNOD);
+		err = ll_new_node(dir, dchild, NULL, mode, old_encode_dev(rdev),
+				  LUSTRE_OPC_MKNOD);
                 break;
         case S_IFDIR:
                 err = -EPERM;
@@ -814,16 +984,44 @@ static int ll_mknod_generic(struct inode *dir, struct qstr *name, int mode,
         RETURN(err);
 }
 
+#ifdef HAVE_IOP_ATOMIC_OPEN
+/*
+ * Plain create. Intent create is handled in atomic_open.
+ */
+static int ll_create_nd(struct inode *dir, struct dentry *dentry,
+			umode_t mode, bool want_excl)
+{
+	int rc;
+
+	CDEBUG(D_VFSTRACE, "VFS Op:name=%.*s, dir="DFID"(%p), "
+			   "flags=%u, excl=%d\n", dentry->d_name.len,
+	       dentry->d_name.name, PFID(ll_inode2fid(dir)),
+	       dir, mode, want_excl);
+
+	rc = ll_mknod(dir, dentry, mode, 0);
+
+	ll_stats_ops_tally(ll_i2sbi(dir), LPROC_LL_CREATE, 1);
+
+	CDEBUG(D_VFSTRACE, "VFS Op:name=%.*s, unhashed %d\n",
+	       dentry->d_name.len, dentry->d_name.name, d_unhashed(dentry));
+
+	return rc;
+}
+#else /* !HAVE_IOP_ATOMIC_OPEN */
 static int ll_create_nd(struct inode *dir, struct dentry *dentry,
-                        int mode, struct nameidata *nd)
+			ll_umode_t mode, struct nameidata *nd)
 {
-        struct lookup_intent *it = ll_d2d(dentry)->lld_it;
+	struct ll_dentry_data *lld = ll_d2d(dentry);
+	struct lookup_intent *it = NULL;
         int rc;
 
+	if (lld != NULL)
+		it = lld->lld_it;
+
         if (!it)
-                return ll_mknod_generic(dir, &dentry->d_name, mode, 0, dentry);
+		return ll_mknod(dir, dentry, mode, 0);
 
-        ll_d2d(dentry)->lld_it = NULL;
+	lld->lld_it = NULL;
 
         /* Was there an error? Propagate it! */
         if (it->d.lustre.it_status) {
@@ -831,7 +1029,7 @@ static int ll_create_nd(struct inode *dir, struct dentry *dentry,
                 goto out;
         }
 
-        rc = ll_create_it(dir, dentry, mode, it);
+	rc = ll_create_it(dir, dentry, it);
         if (nd && (nd->flags & LOOKUP_OPEN) && dentry->d_inode) { /* Open */
 		struct file *filp;
 
@@ -850,19 +1048,21 @@ out:
 
         return rc;
 }
+#endif /* HAVE_IOP_ATOMIC_OPEN */
 
-static int ll_symlink_generic(struct inode *dir, struct qstr *name,
-                              const char *tgt, struct dentry *dchild)
+static int ll_symlink(struct inode *dir, struct dentry *dchild,
+		      const char *oldpath)
 {
-        int err;
-        ENTRY;
+	struct qstr *name = &dchild->d_name;
+	int err;
+	ENTRY;
 
-        CDEBUG(D_VFSTRACE, "VFS Op:name=%.*s,dir=%lu/%u(%p),target=%.*s\n",
-               name->len, name->name, dir->i_ino, dir->i_generation,
-               dir, 3000, tgt);
+	CDEBUG(D_VFSTRACE, "VFS Op:name=%.*s, dir="DFID"(%p), target=%.*s\n",
+	       name->len, name->name, PFID(ll_inode2fid(dir)),
+	       dir, 3000, oldpath);
 
-        err = ll_new_node(dir, name, (char *)tgt, S_IFLNK | S_IRWXUGO,
-                          0, dchild, LUSTRE_OPC_SYMLINK);
+	err = ll_new_node(dir, dchild, oldpath, S_IFLNK | S_IRWXUGO, 0,
+			  LUSTRE_OPC_SYMLINK);
 
         if (!err)
                 ll_stats_ops_tally(ll_i2sbi(dir), LPROC_LL_SYMLINK, 1);
@@ -870,19 +1070,20 @@ static int ll_symlink_generic(struct inode *dir, struct qstr *name,
         RETURN(err);
 }
 
-static int ll_link_generic(struct inode *src,  struct inode *dir,
-                           struct qstr *name, struct dentry *dchild)
+static int ll_link(struct dentry *old_dentry, struct inode *dir,
+		   struct dentry *new_dentry)
 {
-        struct ll_sb_info *sbi = ll_i2sbi(dir);
-        struct ptlrpc_request *request = NULL;
-        struct md_op_data *op_data;
-        int err;
+	struct inode *src = old_dentry->d_inode;
+	struct qstr *name = &new_dentry->d_name;
+	struct ll_sb_info *sbi = ll_i2sbi(dir);
+	struct ptlrpc_request *request = NULL;
+	struct md_op_data *op_data;
+	int err;
 
-        ENTRY;
-        CDEBUG(D_VFSTRACE,
-               "VFS Op: inode=%lu/%u(%p), dir=%lu/%u(%p), target=%.*s\n",
-               src->i_ino, src->i_generation, src, dir->i_ino,
-               dir->i_generation, dir, name->len, name->name);
+	ENTRY;
+	CDEBUG(D_VFSTRACE, "VFS Op: inode="DFID"(%p), dir="DFID"(%p), "
+	       "target=%.*s\n", PFID(ll_inode2fid(src)), src,
+	       PFID(ll_inode2fid(dir)), dir, name->len, name->name);
 
         op_data = ll_prep_md_op_data(NULL, src, dir, name->name, name->len,
                                      0, LUSTRE_OPC_ANY, NULL);
@@ -902,53 +1103,39 @@ out:
         RETURN(err);
 }
 
-static int ll_mkdir_generic(struct inode *dir, struct qstr *name,
-                            int mode, struct dentry *dchild)
-
+static int ll_mkdir(struct inode *dir, struct dentry *dchild, ll_umode_t mode)
 {
+	struct qstr *name = &dchild->d_name;
         int err;
         ENTRY;
 
-        CDEBUG(D_VFSTRACE, "VFS Op:name=%.*s,dir=%lu/%u(%p)\n",
-               name->len, name->name, dir->i_ino, dir->i_generation, dir);
+	CDEBUG(D_VFSTRACE, "VFS Op:name=%.*s, dir="DFID"(%p)\n",
+	       name->len, name->name, PFID(ll_inode2fid(dir)), dir);
 
-        mode = (mode & (S_IRWXUGO|S_ISVTX) & ~cfs_curproc_umask()) | S_IFDIR;
-        err = ll_new_node(dir, name, NULL, mode, 0, dchild, LUSTRE_OPC_MKDIR);
+	if (!IS_POSIXACL(dir) || !exp_connect_umask(ll_i2mdexp(dir)))
+		mode &= ~current_umask();
 
-        if (!err)
-                ll_stats_ops_tally(ll_i2sbi(dir), LPROC_LL_MKDIR, 1);
+	mode = (mode & (S_IRWXUGO|S_ISVTX)) | S_IFDIR;
 
-        RETURN(err);
-}
+	err = ll_new_node(dir, dchild, NULL, mode, 0, LUSTRE_OPC_MKDIR);
+	if (err == 0)
+		ll_stats_ops_tally(ll_i2sbi(dir), LPROC_LL_MKDIR, 1);
 
-/* Try to find the child dentry by its name.
-   If found, put the result fid into @fid. */
-static void ll_get_child_fid(struct inode * dir, struct qstr *name,
-                             struct lu_fid *fid)
-{
-        struct dentry *parent, *child;
-
-        parent = list_entry(dir->i_dentry.next, struct dentry, d_alias);
-        child = d_lookup(parent, name);
-        if (child) {
-                if (child->d_inode)
-                        *fid = *ll_inode2fid(child->d_inode);
-                dput(child);
-        }
+	RETURN(err);
 }
 
-static int ll_rmdir_generic(struct inode *dir, struct dentry *dparent,
-                            struct dentry *dchild, struct qstr *name)
+static int ll_rmdir(struct inode *dir, struct dentry *dchild)
 {
+	struct qstr *name = &dchild->d_name;
         struct ptlrpc_request *request = NULL;
         struct md_op_data *op_data;
         int rc;
         ENTRY;
 
-        CDEBUG(D_VFSTRACE, "VFS Op:name=%.*s,dir=%lu/%u(%p)\n",
-               name->len, name->name, dir->i_ino, dir->i_generation, dir);
+	CDEBUG(D_VFSTRACE, "VFS Op:name=%.*s, dir="DFID"(%p)\n",
+	       name->len, name->name, PFID(ll_inode2fid(dir)), dir);
 
-        if (unlikely(ll_d_mountpoint(dparent, dchild, name)))
+	if (unlikely(d_mountpoint(dchild)))
                 RETURN(-EBUSY);
 
         op_data = ll_prep_md_op_data(NULL, dir, NULL, name->name, name->len,
@@ -956,7 +1143,10 @@ static int ll_rmdir_generic(struct inode *dir, struct dentry *dparent,
         if (IS_ERR(op_data))
                 RETURN(PTR_ERR(op_data));
 
-        ll_get_child_fid(dir, name, &op_data->op_fid3);
+	if (dchild->d_inode != NULL)
+		op_data->op_fid3 = *ll_inode2fid(dchild->d_inode);
+
+	op_data->op_fid2 = op_data->op_fid3;
         rc = md_unlink(ll_i2sbi(dir)->ll_md_exp, op_data, &request);
         ll_finish_md_op_data(op_data);
         if (rc == 0) {
@@ -968,150 +1158,108 @@ static int ll_rmdir_generic(struct inode *dir, struct dentry *dparent,
         RETURN(rc);
 }
 
-int ll_objects_destroy(struct ptlrpc_request *request, struct inode *dir)
+/**
+ * Remove dir entry
+ **/
+int ll_rmdir_entry(struct inode *dir, char *name, int namelen)
 {
-        struct mdt_body *body;
-        struct lov_mds_md *eadata;
-        struct lov_stripe_md *lsm = NULL;
-        struct obd_trans_info oti = { 0 };
-        struct obdo *oa;
-        struct obd_capa *oc = NULL;
-        int rc;
-        ENTRY;
-
-        /* req is swabbed so this is safe */
-        body = req_capsule_server_get(&request->rq_pill, &RMF_MDT_BODY);
-        if (!(body->valid & OBD_MD_FLEASIZE))
-                RETURN(0);
-
-        if (body->eadatasize == 0) {
-                CERROR("OBD_MD_FLEASIZE set but eadatasize zero\n");
-                GOTO(out, rc = -EPROTO);
-        }
-
-        /* The MDS sent back the EA because we unlinked the last reference
-         * to this file. Use this EA to unlink the objects on the OST.
-         * It's opaque so we don't swab here; we leave it to obd_unpackmd() to
-         * check it is complete and sensible. */
-        eadata = req_capsule_server_sized_get(&request->rq_pill, &RMF_MDT_MD,
-                                              body->eadatasize);
-        LASSERT(eadata != NULL);
-
-        rc = obd_unpackmd(ll_i2dtexp(dir), &lsm, eadata, body->eadatasize);
-        if (rc < 0) {
-                CERROR("obd_unpackmd: %d\n", rc);
-                GOTO(out, rc);
-        }
-        LASSERT(rc >= sizeof(*lsm));
-
-        OBDO_ALLOC(oa);
-        if (oa == NULL)
-                GOTO(out_free_memmd, rc = -ENOMEM);
-
-        oa->o_id = lsm->lsm_object_id;
-        oa->o_seq = lsm->lsm_object_seq;
-        oa->o_mode = body->mode & S_IFMT;
-        oa->o_valid = OBD_MD_FLID | OBD_MD_FLTYPE | OBD_MD_FLGROUP;
-
-        if (body->valid & OBD_MD_FLCOOKIE) {
-                oa->o_valid |= OBD_MD_FLCOOKIE;
-                oti.oti_logcookies =
-                        req_capsule_server_sized_get(&request->rq_pill,
-                                                     &RMF_LOGCOOKIES,
-                                                   sizeof(struct llog_cookie) *
-                                                     lsm->lsm_stripe_count);
-                if (oti.oti_logcookies == NULL) {
-                        oa->o_valid &= ~OBD_MD_FLCOOKIE;
-                        body->valid &= ~OBD_MD_FLCOOKIE;
-                }
-        }
+	struct ptlrpc_request *request = NULL;
+	struct md_op_data *op_data;
+	int rc;
+	ENTRY;
 
-        if (body->valid & OBD_MD_FLOSSCAPA) {
-                rc = md_unpack_capa(ll_i2mdexp(dir), request, &RMF_CAPA2, &oc);
-                if (rc)
-                        GOTO(out_free_memmd, rc);
-        }
+	CDEBUG(D_VFSTRACE, "VFS Op:name=%.*s, dir="DFID"(%p)\n",
+	       namelen, name, PFID(ll_inode2fid(dir)), dir);
+
+	op_data = ll_prep_md_op_data(NULL, dir, NULL, name, strlen(name),
+				     S_IFDIR, LUSTRE_OPC_ANY, NULL);
+	if (IS_ERR(op_data))
+		RETURN(PTR_ERR(op_data));
+	op_data->op_cli_flags |= CLI_RM_ENTRY;
+	rc = md_unlink(ll_i2sbi(dir)->ll_md_exp, op_data, &request);
+	ll_finish_md_op_data(op_data);
+	if (rc == 0) {
+		ll_update_times(request, dir);
+		ll_stats_ops_tally(ll_i2sbi(dir), LPROC_LL_RMDIR, 1);
+	}
 
-        rc = obd_destroy(NULL, ll_i2dtexp(dir), oa, lsm, &oti,
-                         ll_i2mdexp(dir), oc);
-        capa_put(oc);
-        OBDO_FREE(oa);
-        if (rc)
-                CERROR("obd destroy objid "LPX64" error %d\n",
-                       lsm->lsm_object_id, rc);
- out_free_memmd:
-        obd_free_memmd(ll_i2dtexp(dir), &lsm);
- out:
-        return rc;
+	ptlrpc_req_finished(request);
+	RETURN(rc);
 }
 
-/* ll_unlink_generic() doesn't update the inode with the new link count.
- * Instead, ll_ddelete() and ll_d_iput() will update it based upon if there
- * is any lock existing. They will recycle dentries and inodes based upon locks
- * too. b=20433 */
-static int ll_unlink_generic(struct inode *dir, struct dentry *dparent,
-                             struct dentry *dchild, struct qstr *name)
+/* ll_unlink() doesn't update the inode with the new link count.
+ * Instead, ll_ddelete() and ll_d_iput() will update it based upon if
+ * there is any lock existing. They will recycle dentries and inodes
+ * based upon locks too. b=20433 */
+static int ll_unlink(struct inode *dir, struct dentry *dchild)
 {
+	struct qstr *name = &dchild->d_name;
         struct ptlrpc_request *request = NULL;
         struct md_op_data *op_data;
         int rc;
         ENTRY;
-        CDEBUG(D_VFSTRACE, "VFS Op:name=%.*s,dir=%lu/%u(%p)\n",
-               name->len, name->name, dir->i_ino, dir->i_generation, dir);
+	CDEBUG(D_VFSTRACE, "VFS Op:name=%.*s, dir="DFID"(%p)\n",
+	       name->len, name->name, PFID(ll_inode2fid(dir)), dir);
 
         /*
          * XXX: unlink bind mountpoint maybe call to here,
          * just check it as vfs_unlink does.
          */
-        if (unlikely(ll_d_mountpoint(dparent, dchild, name)))
-                RETURN(-EBUSY);
+	if (unlikely(d_mountpoint(dchild)))
+		RETURN(-EBUSY);
 
-        op_data = ll_prep_md_op_data(NULL, dir, NULL, name->name,
-                                     name->len, 0, LUSTRE_OPC_ANY, NULL);
-        if (IS_ERR(op_data))
-                RETURN(PTR_ERR(op_data));
+	op_data = ll_prep_md_op_data(NULL, dir, NULL, name->name, name->len, 0,
+				     LUSTRE_OPC_ANY, NULL);
+	if (IS_ERR(op_data))
+		RETURN(PTR_ERR(op_data));
 
-        ll_get_child_fid(dir, name, &op_data->op_fid3);
-        rc = md_unlink(ll_i2sbi(dir)->ll_md_exp, op_data, &request);
-        ll_finish_md_op_data(op_data);
-        if (rc)
-                GOTO(out, rc);
+	if (dchild->d_inode != NULL)
+		op_data->op_fid3 = *ll_inode2fid(dchild->d_inode);
+
+	op_data->op_fid2 = op_data->op_fid3;
+	rc = md_unlink(ll_i2sbi(dir)->ll_md_exp, op_data, &request);
+	ll_finish_md_op_data(op_data);
+	if (rc)
+		GOTO(out, rc);
 
         ll_update_times(request, dir);
         ll_stats_ops_tally(ll_i2sbi(dir), LPROC_LL_UNLINK, 1);
 
-        rc = ll_objects_destroy(request, dir);
  out:
         ptlrpc_req_finished(request);
         RETURN(rc);
 }
 
-static int ll_rename_generic(struct inode *src, struct dentry *src_dparent,
-                             struct dentry *src_dchild, struct qstr *src_name,
-                             struct inode *tgt, struct dentry *tgt_dparent,
-                             struct dentry *tgt_dchild, struct qstr *tgt_name)
+static int ll_rename(struct inode *src, struct dentry *src_dchild,
+		     struct inode *tgt, struct dentry *tgt_dchild)
 {
+	struct qstr *src_name = &src_dchild->d_name;
+	struct qstr *tgt_name = &tgt_dchild->d_name;
         struct ptlrpc_request *request = NULL;
         struct ll_sb_info *sbi = ll_i2sbi(src);
         struct md_op_data *op_data;
         int err;
         ENTRY;
-        CDEBUG(D_VFSTRACE,"VFS Op:oldname=%.*s,src_dir=%lu/%u(%p),newname=%.*s,"
-               "tgt_dir=%lu/%u(%p)\n", src_name->len, src_name->name,
-               src->i_ino, src->i_generation, src, tgt_name->len,
-               tgt_name->name, tgt->i_ino, tgt->i_generation, tgt);
+	CDEBUG(D_VFSTRACE, "VFS Op:oldname=%.*s, src_dir="DFID
+	       "(%p), newname=%.*s, tgt_dir="DFID"(%p)\n",
+	       src_name->len, src_name->name,
+	       PFID(ll_inode2fid(src)), src, tgt_name->len,
+	       tgt_name->name, PFID(ll_inode2fid(tgt)), tgt);
 
-        if (unlikely(ll_d_mountpoint(src_dparent, src_dchild, src_name) ||
-            ll_d_mountpoint(tgt_dparent, tgt_dchild, tgt_name)))
-                RETURN(-EBUSY);
+	if (unlikely(d_mountpoint(src_dchild) || d_mountpoint(tgt_dchild)))
+		RETURN(-EBUSY);
 
-        op_data = ll_prep_md_op_data(NULL, src, tgt, NULL, 0, 0,
-                                     LUSTRE_OPC_ANY, NULL);
-        if (IS_ERR(op_data))
-                RETURN(PTR_ERR(op_data));
+	op_data = ll_prep_md_op_data(NULL, src, tgt, NULL, 0, 0,
+				     LUSTRE_OPC_ANY, NULL);
+	if (IS_ERR(op_data))
+		RETURN(PTR_ERR(op_data));
+
+	if (src_dchild->d_inode != NULL)
+		op_data->op_fid3 = *ll_inode2fid(src_dchild->d_inode);
+
+	if (tgt_dchild->d_inode != NULL)
+		op_data->op_fid4 = *ll_inode2fid(tgt_dchild->d_inode);
 
-        ll_get_child_fid(src, src_name, &op_data->op_fid3);
-        ll_get_child_fid(tgt, tgt_name, &op_data->op_fid4);
         err = md_rename(sbi->ll_md_exp, op_data,
                         src_name->name, src_name->len,
                         tgt_name->name, tgt_name->len, &request);
@@ -1120,87 +1268,51 @@ static int ll_rename_generic(struct inode *src, struct dentry *src_dparent,
                 ll_update_times(request, src);
                 ll_update_times(request, tgt);
                 ll_stats_ops_tally(sbi, LPROC_LL_RENAME, 1);
-                err = ll_objects_destroy(request, src);
         }
 
         ptlrpc_req_finished(request);
 
-        RETURN(err);
-}
+	if (err == 0)
+		d_move(src_dchild, tgt_dchild);
 
-static int ll_mknod(struct inode *dir, struct dentry *dchild, int mode,
-		    dev_t rdev)
-{
-	return ll_mknod_generic(dir, &dchild->d_name, mode,
-				old_encode_dev(rdev), dchild);
+	RETURN(err);
 }
 
-static int ll_unlink(struct inode * dir, struct dentry *dentry)
-{
-        return ll_unlink_generic(dir, NULL, dentry, &dentry->d_name);
-}
-static int ll_mkdir(struct inode *dir, struct dentry *dentry, int mode)
-{
-        return ll_mkdir_generic(dir, &dentry->d_name, mode, dentry);
-}
-static int ll_rmdir(struct inode *dir, struct dentry *dentry)
-{
-        return ll_rmdir_generic(dir, NULL, dentry, &dentry->d_name);
-}
-static int ll_symlink(struct inode *dir, struct dentry *dentry,
-                      const char *oldname)
-{
-        return ll_symlink_generic(dir, &dentry->d_name, oldname, dentry);
-}
-static int ll_link(struct dentry *old_dentry, struct inode *dir,
-                   struct dentry *new_dentry)
-{
-        return ll_link_generic(old_dentry->d_inode, dir, &new_dentry->d_name,
-                               new_dentry);
-}
-static int ll_rename(struct inode *old_dir, struct dentry *old_dentry,
-                     struct inode *new_dir, struct dentry *new_dentry)
-{
-        int err;
-        err = ll_rename_generic(old_dir, NULL,
-                                 old_dentry, &old_dentry->d_name,
-                                 new_dir, NULL, new_dentry,
-                                 &new_dentry->d_name);
-        if (!err) {
-#ifndef HAVE_FS_RENAME_DOES_D_MOVE
-                if (!S_ISDIR(old_dentry->d_inode->i_mode))
+const struct inode_operations ll_dir_inode_operations = {
+	.mknod              = ll_mknod,
+#ifdef HAVE_IOP_ATOMIC_OPEN
+	.atomic_open	    = ll_atomic_open,
+#endif
+	.lookup             = ll_lookup_nd,
+	.create             = ll_create_nd,
+	/* We need all these non-raw things for NFSD, to not patch it. */
+	.unlink             = ll_unlink,
+	.mkdir              = ll_mkdir,
+	.rmdir              = ll_rmdir,
+	.symlink            = ll_symlink,
+	.link               = ll_link,
+	.rename             = ll_rename,
+	.setattr            = ll_setattr,
+	.getattr            = ll_getattr,
+	.permission         = ll_inode_permission,
+	.setxattr           = ll_setxattr,
+	.getxattr           = ll_getxattr,
+	.listxattr          = ll_listxattr,
+	.removexattr        = ll_removexattr,
+#ifdef HAVE_IOP_GET_ACL
+	.get_acl	    = ll_get_acl,
 #endif
-                        d_move(old_dentry, new_dentry);
-        }
-        return err;
-}
-
-struct inode_operations ll_dir_inode_operations = {
-        .mknod              = ll_mknod,
-        .lookup             = ll_lookup_nd,
-        .create             = ll_create_nd,
-        /* We need all these non-raw things for NFSD, to not patch it. */
-        .unlink             = ll_unlink,
-        .mkdir              = ll_mkdir,
-        .rmdir              = ll_rmdir,
-        .symlink            = ll_symlink,
-        .link               = ll_link,
-        .rename             = ll_rename,
-        .setattr            = ll_setattr,
-        .getattr            = ll_getattr,
-        .permission         = ll_inode_permission,
-        .setxattr           = ll_setxattr,
-        .getxattr           = ll_getxattr,
-        .listxattr          = ll_listxattr,
-        .removexattr        = ll_removexattr,
 };
 
-struct inode_operations ll_special_inode_operations = {
-        .setattr        = ll_setattr,
-        .getattr        = ll_getattr,
-        .permission     = ll_inode_permission,
-        .setxattr       = ll_setxattr,
-        .getxattr       = ll_getxattr,
-        .listxattr      = ll_listxattr,
-        .removexattr    = ll_removexattr,
+const struct inode_operations ll_special_inode_operations = {
+	.setattr        = ll_setattr,
+	.getattr        = ll_getattr,
+	.permission     = ll_inode_permission,
+	.setxattr       = ll_setxattr,
+	.getxattr       = ll_getxattr,
+	.listxattr      = ll_listxattr,
+	.removexattr    = ll_removexattr,
+#ifdef HAVE_IOP_GET_ACL
+	.get_acl	    = ll_get_acl,
+#endif
 };