X-Git-Url: https://git.whamcloud.com/?p=fs%2Flustre-release.git;a=blobdiff_plain;f=lustre%2Fllite%2Fllite_nfs.c;h=3dd11838ec3fd24aba8cfc97c4b897798fed3d42;hp=afd0c9b8c01f252df759a12cbf89e506f8c33361;hb=0d8c5ccc4ecfe7c0d10a0a4f92fd291320a97190;hpb=3192e52a89946f12fd36d28a686c169d01d36e64 diff --git a/lustre/llite/llite_nfs.c b/lustre/llite/llite_nfs.c index afd0c9b..3dd1183 100644 --- a/lustre/llite/llite_nfs.c +++ b/lustre/llite/llite_nfs.c @@ -1,76 +1,107 @@ -/* -*- mode: c; c-basic-offset: 8; indent-tabs-mode: nil; -*- - * vim:expandtab:shiftwidth=8:tabstop=8: +/* + * 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). * - * NFS export of Lustre Light File System + * 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 * - * Copyright (c) 2002, 2006 Cluster File Systems, Inc. + * 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. * - * Author: Yury Umanets - * Huang Hua + * GPL HEADER END + */ +/* + * Copyright (c) 2003, 2010, Oracle and/or its affiliates. All rights reserved. + * Use is subject to license terms. * - * This file is part of Lustre, http://www.lustre.org. + * Copyright (c) 2011, 2013, Intel Corporation. + */ +/* + * This file is part of Lustre, http://www.lustre.org/ + * Lustre is a trademark of Sun Microsystems, Inc. * - * Lustre is free software; you can redistribute it and/or - * modify it under the terms of version 2 of the GNU General Public - * License as published by the Free Software Foundation. + * lustre/lustre/llite/llite_nfs.c * - * Lustre 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 for more details. + * NFS export of Lustre Light File System * - * You should have received a copy of the GNU General Public License - * along with Lustre; if not, write to the Free Software - * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. + * Author: Yury Umanets + * Author: Huang Hua */ #define DEBUG_SUBSYSTEM S_LLITE #include #include "llite_internal.h" -#ifdef HAVE_LINUX_EXPORTFS_H #include -#endif -static int ll_nfs_test_inode(struct inode *inode, void *opaque) +__u32 get_uuid2int(const char *name, int len) { - return lu_fid_eq(&ll_i2info(inode)->lli_fid, - (struct lu_fid *)opaque); + __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); } -static struct inode *search_inode_for_lustre(struct super_block *sb, - struct lu_fid *fid, - int mode) +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; - unsigned long valid = 0; int eadatalen = 0; - ino_t ino = ll_fid_build_ino(sbi, fid); + unsigned long hash = cl_fid_build_ino(fid, + ll_need_32bit_api(sbi)); + struct md_op_data *op_data; int rc; ENTRY; - CDEBUG(D_INFO, "searching inode for:(%lu,"DFID")\n", ino, PFID(fid)); + CDEBUG(D_INFO, "searching inode for:(%lu,"DFID")\n", hash, PFID(fid)); - inode = ILOOKUP(sb, ino, ll_nfs_test_inode, fid); - if (inode) - RETURN(inode); + inode = ilookup5(sb, hash, ll_test_inode_by_fid, (void *)fid); + if (inode) + RETURN(inode); - if (S_ISREG(mode)) { - rc = ll_get_max_mdsize(sbi, &eadatalen); - if (rc) - RETURN(ERR_PTR(rc)); - valid |= OBD_MD_FLEASIZE; - } + 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 */ + OBD_ALLOC_PTR(op_data); + if (op_data == NULL) + return ERR_PTR(-ENOMEM); - rc = md_getattr(sbi->ll_md_exp, fid, NULL, valid, eadatalen, &req); + 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); + /* Suppress erroneous/confusing messages when NFS + * is out of sync and requests old data. */ + CDEBUG(D_INFO, "can't get object attrs, fid "DFID", rc %d\n", + PFID(fid), rc); RETURN(ERR_PTR(rc)); } - - rc = ll_prep_inode(&inode, req, REPLY_REC_OFF, sb); + rc = ll_prep_inode(&inode, req, sb, NULL); ptlrpc_req_finished(req); if (rc) RETURN(ERR_PTR(rc)); @@ -78,163 +109,262 @@ static struct inode *search_inode_for_lustre(struct super_block *sb, RETURN(inode); } -extern struct dentry_operations ll_d_ops; +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, - umode_t mode) +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; ENTRY; - CDEBUG(D_INFO, "Get dentry for fid: "DFID"\n", PFID(fid)); - if (!fid_is_sane(fid)) - RETURN(ERR_PTR(-ESTALE)); + if (!fid_is_sane(fid)) + RETURN(ERR_PTR(-ESTALE)); - inode = search_inode_for_lustre(sb, fid, mode); + CDEBUG(D_INFO, "Get dentry for fid: "DFID"\n", PFID(fid)); + + inode = search_inode_for_lustre(sb, fid); if (IS_ERR(inode)) RETURN(ERR_PTR(PTR_ERR(inode))); if (is_bad_inode(inode)) { /* we didn't find the right inode.. */ - CERROR("can't get inode by fid "DFID"\n", - PFID(fid)); iput(inode); RETURN(ERR_PTR(-ESTALE)); } - result = d_alloc_anon(inode); - if (!result) { - iput(inode); - RETURN(ERR_PTR(-ENOMEM)); - } - ll_set_dd(result); - result->d_op = &ll_d_ops; + /** + * 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 && !fid_is_zero(parent)) { + 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); } -/* - * This length is counted as amount of __u32, - * It is composed of a fid and a mode +#ifndef FILEID_INVALID +#define FILEID_INVALID 0xff +#endif +#ifndef FILEID_LUSTRE +#define FILEID_LUSTRE 0x97 +#endif + +/** + * \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. */ -#define ONE_FH_LEN (sizeof(struct lu_fid)/4 + 1) +#ifndef HAVE_ENCODE_FH_PARENT +static int ll_encode_fh(struct dentry *de, __u32 *fh, int *plen, + int connectable) +{ + struct inode *inode = de->d_inode; + struct inode *parent = de->d_parent->d_inode; +#else +static int ll_encode_fh(struct inode *inode, __u32 *fh, int *plen, + struct inode *parent) +{ +#endif + int fileid_len = sizeof(struct lustre_nfs_fid) / 4; + struct lustre_nfs_fid *nfs_fid = (void *)fh; + ENTRY; + + CDEBUG(D_INFO, "%s: encoding for ("DFID") maxlen=%d minlen=%d\n", + ll_get_fsname(inode->i_sb, NULL, 0), + PFID(ll_inode2fid(inode)), *plen, fileid_len); + + if (*plen < fileid_len) { + *plen = fileid_len; + RETURN(FILEID_INVALID); + } + + nfs_fid->lnf_child = *ll_inode2fid(inode); + if (parent != NULL) + nfs_fid->lnf_parent = *ll_inode2fid(parent); + else + fid_zero(&nfs_fid->lnf_parent); + *plen = fileid_len; + + RETURN(FILEID_LUSTRE); +} -static struct dentry *ll_decode_fh(struct super_block *sb, __u32 *fh, int fh_len, - int fh_type, - int (*acceptable)(void *, struct dentry *), - void *context) +static int ll_nfs_get_name_filldir(void *cookie, const char *name, int namelen, + loff_t hash, u64 ino, unsigned type) { - struct lu_fid *parent = NULL; - struct lu_fid *child; - struct dentry *entry; - ENTRY; + /* 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 = cookie; + 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 = dentry->d_inode; + struct ll_getname_data lgd = { + .lgd_name = name, + .lgd_fid = ll_i2info(child->d_inode)->lli_fid, +#ifdef HAVE_DIR_CONTEXT + .ctx.actor = ll_nfs_get_name_filldir, +#endif + .lgd_found = 0, + }; + struct md_op_data *op_data; + int rc; + ENTRY; + + if (!dir || !S_ISDIR(dir->i_mode)) + GOTO(out, rc = -ENOTDIR); + + if (!dir->i_fop) + GOTO(out, rc = -EINVAL); + + op_data = ll_prep_md_op_data(NULL, dir, dir, NULL, 0, 0, + LUSTRE_OPC_ANY, dir); + if (IS_ERR(op_data)) + GOTO(out, rc = PTR_ERR(op_data)); + + op_data->op_hash_offset = 0; + op_data->op_max_pages = ll_i2sbi(dir)->ll_md_brw_pages; + mutex_lock(&dir->i_mutex); +#ifdef HAVE_DIR_CONTEXT + rc = ll_dir_read(dir, op_data, &lgd.ctx); +#else + rc = ll_dir_read(dir, op_data, &lgd, ll_nfs_get_name_filldir); +#endif + mutex_unlock(&dir->i_mutex); + ll_finish_md_op_data(op_data); + if (!rc && !lgd.lgd_found) + rc = -ENOENT; + EXIT; +out: + return rc; +} - CDEBUG(D_INFO, "decoding for "DFID" fh_len=%d fh_type=%d\n", - PFID((struct lu_fid*)fh), fh_len, fh_type); +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 != 1 && fh_type != 2) - RETURN(ERR_PTR(-ESTALE)); - if (fh_len < ONE_FH_LEN * fh_type) - RETURN(ERR_PTR(-ESTALE)); + if (fh_type != FILEID_LUSTRE) + RETURN(ERR_PTR(-EPROTO)); - child = (struct lu_fid*)fh; - if (fh_type == 2) - parent = (struct lu_fid*)(fh + ONE_FH_LEN); - - entry = sb->s_export_op->find_exported_dentry(sb, child, parent, - acceptable, context); - RETURN(entry); + RETURN(ll_iget_for_nfs(sb, &nfs_fid->lnf_child, &nfs_fid->lnf_parent)); } -/* 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 dentry *de, __u32 *fh, int *plen, int connectable) +static struct dentry *ll_fh_to_parent(struct super_block *sb, struct fid *fid, + int fh_len, int fh_type) { - struct inode *inode = de->d_inode; - struct lu_fid *fid = ll_inode2fid(inode); - ENTRY; + struct lustre_nfs_fid *nfs_fid = (struct lustre_nfs_fid *)fid; - CDEBUG(D_INFO, "encoding for (%lu,"DFID") maxlen=%d minlen=%d\n", - inode->i_ino, PFID(fid), *plen, ONE_FH_LEN); - - if (*plen < ONE_FH_LEN) - RETURN(255); - - memcpy((char*)fh, fid, sizeof(*fid)); - *(fh + ONE_FH_LEN - 1) = (__u32)(S_IFMT & inode->i_mode); - - if (de->d_parent && *plen >= ONE_FH_LEN * 2) { - struct inode *parent = de->d_parent->d_inode; - fh += ONE_FH_LEN; - memcpy((char*)fh, &ll_i2info(parent)->lli_fid, sizeof(*fid)); - *(fh + ONE_FH_LEN - 1) = (__u32)(S_IFMT & parent->i_mode); - *plen = ONE_FH_LEN * 2; - RETURN(2); - } else { - *plen = ONE_FH_LEN; - RETURN(1); - } + if (fh_type != FILEID_LUSTRE) + RETURN(ERR_PTR(-EPROTO)); + + RETURN(ll_iget_for_nfs(sb, &nfs_fid->lnf_parent, NULL)); } -static struct dentry *ll_get_dentry(struct super_block *sb, void *data) +int ll_dir_get_parent_fid(struct inode *dir, struct lu_fid *parent_fid) { - struct lu_fid *fid; - struct dentry *entry; - __u32 mode; - ENTRY; - - fid = (struct lu_fid *)data; - mode = *((__u32*)data + ONE_FH_LEN - 1); - - entry = ll_iget_for_nfs(sb, fid, mode); - RETURN(entry); + struct ptlrpc_request *req = NULL; + struct ll_sb_info *sbi; + struct mdt_body *body; + static const char dotdot[] = ".."; + struct md_op_data *op_data; + int rc; + int lmmsize; + ENTRY; + + LASSERT(dir && S_ISDIR(dir->i_mode)); + + sbi = ll_s2sbi(dir->i_sb); + + CDEBUG(D_INFO, "%s: getting parent for ("DFID")\n", + ll_get_fsname(dir->i_sb, NULL, 0), + PFID(ll_inode2fid(dir))); + + rc = ll_get_default_mdsize(sbi, &lmmsize); + if (rc != 0) + RETURN(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(PTR_ERR(op_data)); + + rc = md_getattr_name(sbi->ll_md_exp, op_data, &req); + ll_finish_md_op_data(op_data); + if (rc != 0) { + CERROR("%s: failure inode "DFID" get parent: rc = %d\n", + ll_get_fsname(dir->i_sb, NULL, 0), + PFID(ll_inode2fid(dir)), rc); + RETURN(rc); + } + body = req_capsule_server_get(&req->rq_pill, &RMF_MDT_BODY); + + /* + * LU-3952: MDT may lost the FID of its parent, we should not crash + * the NFS server, ll_iget_for_nfs() will handle the error. + */ + if (body->valid & OBD_MD_FLID) { + CDEBUG(D_INFO, "parent for "DFID" is "DFID"\n", + PFID(ll_inode2fid(dir)), PFID(&body->fid1)); + *parent_fid = body->fid1; + } + + ptlrpc_req_finished(req); + RETURN(0); } static struct dentry *ll_get_parent(struct dentry *dchild) { - struct ptlrpc_request *req = NULL; - struct inode *dir = dchild->d_inode; - struct ll_sb_info *sbi; - struct dentry *result = NULL; - struct mdt_body *body; - static char dotdot[] = ".."; - int rc; - ENTRY; - - 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 = md_getattr_name(sbi->ll_md_exp, ll_inode2fid(dir), NULL, - dotdot, strlen(dotdot) + 1, 0, 0, - ll_i2suppgid(dir), &req); - if (rc) { - CERROR("failure %d inode %lu get parent\n", rc, dir->i_ino); - RETURN(ERR_PTR(rc)); - } - body = lustre_msg_buf(req->rq_repmsg, REPLY_REC_OFF, sizeof(*body)); - - LASSERT(body->valid & OBD_MD_FLID); - - CDEBUG(D_INFO, "parent for "DFID" is "DFID"\n", - PFID(ll_inode2fid(dir)), PFID(&body->fid1)); + struct lu_fid parent_fid = { 0 }; + int rc; + struct dentry *dentry; + ENTRY; - result = ll_iget_for_nfs(dir->i_sb, &body->fid1, S_IFDIR); + rc = ll_dir_get_parent_fid(dchild->d_inode, &parent_fid); + if (rc != 0) + RETURN(ERR_PTR(rc)); - ptlrpc_req_finished(req); - RETURN(result); -} + dentry = ll_iget_for_nfs(dchild->d_inode->i_sb, &parent_fid, NULL); + + RETURN(dentry); +} struct export_operations lustre_export_operations = { .get_parent = ll_get_parent, - .get_dentry = ll_get_dentry, .encode_fh = ll_encode_fh, - .decode_fh = ll_decode_fh, + .get_name = ll_get_name, + .fh_to_dentry = ll_fh_to_dentry, + .fh_to_parent = ll_fh_to_parent, };