Add the rt linux 4.1.3-rt3 as base
[kvmfornfv.git] / kernel / drivers / staging / lustre / lustre / llite / llite_nfs.c
diff --git a/kernel/drivers/staging/lustre/lustre/llite/llite_nfs.c b/kernel/drivers/staging/lustre/lustre/llite/llite_nfs.c
new file mode 100644 (file)
index 0000000..db43b81
--- /dev/null
@@ -0,0 +1,335 @@
+/*
+ * GPL HEADER START
+ *
+ * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 only,
+ * as published by the Free Software Foundation.
+ *
+ * This program is distributed in the hope that it will be useful, but
+ * WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * General Public License version 2 for more details (a copy is included
+ * in the LICENSE file that accompanied this code).
+ *
+ * You should have received a copy of the GNU General Public License
+ * version 2 along with this program; If not, see
+ * http://www.sun.com/software/products/lustre/docs/GPLv2.pdf
+ *
+ * Please contact Sun Microsystems, Inc., 4150 Network Circle, Santa Clara,
+ * CA 95054 USA or visit www.sun.com if you need additional information or
+ * have any questions.
+ *
+ * GPL HEADER END
+ */
+/*
+ * Copyright (c) 2003, 2010, Oracle and/or its affiliates. All rights reserved.
+ * Use is subject to license terms.
+ *
+ * Copyright (c) 2011, 2012, Intel Corporation.
+ */
+/*
+ * This file is part of Lustre, http://www.lustre.org/
+ * Lustre is a trademark of Sun Microsystems, Inc.
+ *
+ * lustre/lustre/llite/llite_nfs.c
+ *
+ * NFS export of Lustre Light File System
+ *
+ * Author: Yury Umanets <umka@clusterfs.com>
+ * Author: Huang Hua <huanghua@clusterfs.com>
+ */
+
+#define DEBUG_SUBSYSTEM S_LLITE
+#include "../include/lustre_lite.h"
+#include "llite_internal.h"
+#include <linux/exportfs.h>
+
+__u32 get_uuid2int(const char *name, int len)
+{
+       __u32 key0 = 0x12a3fe2d, key1 = 0x37abe8f9;
+       while (len--) {
+               __u32 key = key1 + (key0 ^ (*name++ * 7152373));
+
+               if (key & 0x80000000)
+                       key -= 0x7fffffff;
+               key1 = key0;
+               key0 = key;
+       }
+       return (key0 << 1);
+}
+
+void get_uuid2fsid(const char *name, int len, __kernel_fsid_t *fsid)
+{
+       __u64 key = 0, key0 = 0x12a3fe2d, key1 = 0x37abe8f9;
+
+       while (len--) {
+               key = key1 + (key0 ^ (*name++ * 7152373));
+               if (key & 0x8000000000000000ULL)
+                       key -= 0x7fffffffffffffffULL;
+               key1 = key0;
+               key0 = key;
+       }
+
+       fsid->val[0] = key;
+       fsid->val[1] = key >> 32;
+}
+
+static int ll_nfs_test_inode(struct inode *inode, void *opaque)
+{
+       return lu_fid_eq(&ll_i2info(inode)->lli_fid,
+                        (struct lu_fid *)opaque);
+}
+
+struct inode *search_inode_for_lustre(struct super_block *sb,
+                                     const struct lu_fid *fid)
+{
+       struct ll_sb_info     *sbi = ll_s2sbi(sb);
+       struct ptlrpc_request *req = NULL;
+       struct inode      *inode = NULL;
+       int                eadatalen = 0;
+       unsigned long         hash = cl_fid_build_ino(fid,
+                                                     ll_need_32bit_api(sbi));
+       struct  md_op_data    *op_data;
+       int                rc;
+
+       CDEBUG(D_INFO, "searching inode for:(%lu,"DFID")\n", hash, PFID(fid));
+
+       inode = ilookup5(sb, hash, ll_nfs_test_inode, (void *)fid);
+       if (inode)
+               return inode;
+
+       rc = ll_get_default_mdsize(sbi, &eadatalen);
+       if (rc)
+               return ERR_PTR(rc);
+
+       /* Because inode is NULL, ll_prep_md_op_data can not
+        * be used here. So we allocate op_data ourselves */
+       op_data = kzalloc(sizeof(*op_data), GFP_NOFS);
+       if (!op_data)
+               return ERR_PTR(-ENOMEM);
+
+       op_data->op_fid1 = *fid;
+       op_data->op_mode = eadatalen;
+       op_data->op_valid = OBD_MD_FLEASIZE;
+
+       /* mds_fid2dentry ignores f_type */
+       rc = md_getattr(sbi->ll_md_exp, op_data, &req);
+       OBD_FREE_PTR(op_data);
+       if (rc) {
+               CERROR("can't get object attrs, fid "DFID", rc %d\n",
+                      PFID(fid), rc);
+               return ERR_PTR(rc);
+       }
+       rc = ll_prep_inode(&inode, req, sb, NULL);
+       ptlrpc_req_finished(req);
+       if (rc)
+               return ERR_PTR(rc);
+
+       return inode;
+}
+
+struct lustre_nfs_fid {
+       struct lu_fid   lnf_child;
+       struct lu_fid   lnf_parent;
+};
+
+static struct dentry *
+ll_iget_for_nfs(struct super_block *sb, struct lu_fid *fid, struct lu_fid *parent)
+{
+       struct inode  *inode;
+       struct dentry *result;
+
+       CDEBUG(D_INFO, "Get dentry for fid: "DFID"\n", PFID(fid));
+       if (!fid_is_sane(fid))
+               return ERR_PTR(-ESTALE);
+
+       inode = search_inode_for_lustre(sb, fid);
+       if (IS_ERR(inode))
+               return ERR_CAST(inode);
+
+       if (is_bad_inode(inode)) {
+               /* we didn't find the right inode.. */
+               iput(inode);
+               return ERR_PTR(-ESTALE);
+       }
+
+       /**
+        * It is an anonymous dentry without OST objects created yet.
+        * We have to find the parent to tell MDS how to init lov objects.
+        */
+       if (S_ISREG(inode->i_mode) && !ll_i2info(inode)->lli_has_smd &&
+           parent != NULL) {
+               struct ll_inode_info *lli = ll_i2info(inode);
+
+               spin_lock(&lli->lli_lock);
+               lli->lli_pfid = *parent;
+               spin_unlock(&lli->lli_lock);
+       }
+
+       result = d_obtain_alias(inode);
+       if (IS_ERR(result)) {
+               iput(inode);
+               return result;
+       }
+
+       return result;
+}
+
+#define LUSTRE_NFS_FID   0x97
+
+/**
+ * \a connectable - is nfsd will connect himself or this should be done
+ *               at lustre
+ *
+ * The return value is file handle type:
+ * 1 -- contains child file handle;
+ * 2 -- contains child file handle and parent file handle;
+ * 255 -- error.
+ */
+static int ll_encode_fh(struct inode *inode, __u32 *fh, int *plen,
+                       struct inode *parent)
+{
+       struct lustre_nfs_fid *nfs_fid = (void *)fh;
+
+       CDEBUG(D_INFO, "encoding for (%lu,"DFID") maxlen=%d minlen=%d\n",
+             inode->i_ino, PFID(ll_inode2fid(inode)), *plen,
+             (int)sizeof(struct lustre_nfs_fid));
+
+       if (*plen < sizeof(struct lustre_nfs_fid) / 4)
+               return 255;
+
+       nfs_fid->lnf_child = *ll_inode2fid(inode);
+       nfs_fid->lnf_parent = *ll_inode2fid(parent);
+       *plen = sizeof(struct lustre_nfs_fid) / 4;
+
+       return LUSTRE_NFS_FID;
+}
+
+static int ll_nfs_get_name_filldir(struct dir_context *ctx, const char *name,
+                                  int namelen, loff_t hash, u64 ino,
+                                  unsigned type)
+{
+       /* It is hack to access lde_fid for comparison with lgd_fid.
+        * So the input 'name' must be part of the 'lu_dirent'. */
+       struct lu_dirent *lde = container_of0(name, struct lu_dirent, lde_name);
+       struct ll_getname_data *lgd =
+               container_of(ctx, struct ll_getname_data, ctx);
+       struct lu_fid fid;
+
+       fid_le_to_cpu(&fid, &lde->lde_fid);
+       if (lu_fid_eq(&fid, &lgd->lgd_fid)) {
+               memcpy(lgd->lgd_name, name, namelen);
+               lgd->lgd_name[namelen] = 0;
+               lgd->lgd_found = 1;
+       }
+       return lgd->lgd_found;
+}
+
+static int ll_get_name(struct dentry *dentry, char *name,
+                      struct dentry *child)
+{
+       struct inode *dir = d_inode(dentry);
+       int rc;
+       struct ll_getname_data lgd = {
+               .lgd_name = name,
+               .lgd_fid = ll_i2info(d_inode(child))->lli_fid,
+               .ctx.actor = ll_nfs_get_name_filldir,
+       };
+
+       if (!dir || !S_ISDIR(dir->i_mode)) {
+               rc = -ENOTDIR;
+               goto out;
+       }
+
+       if (!dir->i_fop) {
+               rc = -EINVAL;
+               goto out;
+       }
+
+       mutex_lock(&dir->i_mutex);
+       rc = ll_dir_read(dir, &lgd.ctx);
+       mutex_unlock(&dir->i_mutex);
+       if (!rc && !lgd.lgd_found)
+               rc = -ENOENT;
+out:
+       return rc;
+}
+
+static struct dentry *ll_fh_to_dentry(struct super_block *sb, struct fid *fid,
+                                     int fh_len, int fh_type)
+{
+       struct lustre_nfs_fid *nfs_fid = (struct lustre_nfs_fid *)fid;
+
+       if (fh_type != LUSTRE_NFS_FID)
+               return ERR_PTR(-EPROTO);
+
+       return ll_iget_for_nfs(sb, &nfs_fid->lnf_child, &nfs_fid->lnf_parent);
+}
+
+static struct dentry *ll_fh_to_parent(struct super_block *sb, struct fid *fid,
+                                     int fh_len, int fh_type)
+{
+       struct lustre_nfs_fid *nfs_fid = (struct lustre_nfs_fid *)fid;
+
+       if (fh_type != LUSTRE_NFS_FID)
+               return ERR_PTR(-EPROTO);
+
+       return ll_iget_for_nfs(sb, &nfs_fid->lnf_parent, NULL);
+}
+
+static struct dentry *ll_get_parent(struct dentry *dchild)
+{
+       struct ptlrpc_request *req = NULL;
+       struct inode      *dir = d_inode(dchild);
+       struct ll_sb_info     *sbi;
+       struct dentry    *result = NULL;
+       struct mdt_body       *body;
+       static char        dotdot[] = "..";
+       struct md_op_data     *op_data;
+       int                rc;
+       int                   lmmsize;
+
+       LASSERT(dir && S_ISDIR(dir->i_mode));
+
+       sbi = ll_s2sbi(dir->i_sb);
+
+       CDEBUG(D_INFO, "getting parent for (%lu,"DFID")\n",
+                       dir->i_ino, PFID(ll_inode2fid(dir)));
+
+       rc = ll_get_default_mdsize(sbi, &lmmsize);
+       if (rc != 0)
+               return ERR_PTR(rc);
+
+       op_data = ll_prep_md_op_data(NULL, dir, NULL, dotdot,
+                                    strlen(dotdot), lmmsize,
+                                    LUSTRE_OPC_ANY, NULL);
+       if (IS_ERR(op_data))
+               return (void *)op_data;
+
+       rc = md_getattr_name(sbi->ll_md_exp, op_data, &req);
+       ll_finish_md_op_data(op_data);
+       if (rc) {
+               CERROR("failure %d inode %lu get parent\n", rc, dir->i_ino);
+               return ERR_PTR(rc);
+       }
+       body = req_capsule_server_get(&req->rq_pill, &RMF_MDT_BODY);
+       LASSERT(body->valid & OBD_MD_FLID);
+
+       CDEBUG(D_INFO, "parent for "DFID" is "DFID"\n",
+               PFID(ll_inode2fid(dir)), PFID(&body->fid1));
+
+       result = ll_iget_for_nfs(dir->i_sb, &body->fid1, NULL);
+
+       ptlrpc_req_finished(req);
+       return result;
+}
+
+struct export_operations lustre_export_operations = {
+       .get_parent = ll_get_parent,
+       .encode_fh  = ll_encode_fh,
+       .get_name   = ll_get_name,
+       .fh_to_dentry = ll_fh_to_dentry,
+       .fh_to_parent = ll_fh_to_parent,
+};