X-Git-Url: https://git.whamcloud.com/?p=fs%2Flustre-release.git;a=blobdiff_plain;f=lustre%2Fllite%2Fdir.c;h=4a8fc75d85201fecea31ca300c0616116dd69036;hp=b6be8321604ff8668e4468ca1e2800fe857c6c5d;hb=1515e409cc57af5eaef809eee6d8f8d6725d092b;hpb=70b252b806508fcaa95f5e71319e1e968f6b6cb0 diff --git a/lustre/llite/dir.c b/lustre/llite/dir.c index b6be832..4a8fc75 100644 --- a/lustre/llite/dir.c +++ b/lustre/llite/dir.c @@ -1,39 +1,54 @@ /* -*- mode: c; c-basic-offset: 8; indent-tabs-mode: nil; -*- * vim:expandtab:shiftwidth=8:tabstop=8: * - * Copyright (C) 1992, 1993, 1994, 1995 - * Remy Card (card@masi.ibp.fr) - * Laboratoire MASI - Institut Blaise Pascal - * Universite Pierre et Marie Curie (Paris VI) + * GPL HEADER START * - * from + * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER. * - * linux/fs/minix/dir.c - * linux/fs/ext2/dir.c + * 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. * - * Copyright (C) 1991, 1992 Linus Torvalds + * 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). * - * ext2 directory handling functions + * 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 * - * Big-endian to little-endian byte-swapping/bitmaps by - * David S. Miller (davem@caip.rutgers.edu), 1995 + * 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. * - * All code that works with directory layout had been switched to pagecache - * and moved here. AV + * GPL HEADER END + */ +/* + * Copyright (c) 2002, 2010, Oracle and/or its affiliates. All rights reserved. + * Use is subject to license terms. + */ +/* + * Copyright (c) 2011 Whamcloud, Inc. + */ +/* + * This file is part of Lustre, http://www.lustre.org/ + * Lustre is a trademark of Sun Microsystems, Inc. * - * Adapted for Lustre Light - * Copyright (C) 2002-2003, Cluster File Systems, Inc. + * lustre/llite/dir.c * + * Directory code for lustre client. */ #include -#include #include #include #include #include #include #include // for wait_on_buffer +#include #define DEBUG_SUBSYSTEM S_LLITE @@ -127,63 +142,136 @@ * * page format * - * - * - * + * Page in MDS_READPAGE RPC is packed in LU_PAGE_SIZE, and each page contains + * a header lu_dirpage which describes the start/end hash, and whether this + * page is empty (contains no dir entry) or hash collide with next page. + * After client receives reply, several pages will be integrated into dir page + * in CFS_PAGE_SIZE (if CFS_PAGE_SIZE greater than LU_PAGE_SIZE), and the + * lu_dirpage for this integrated page will be adjusted. * */ /* returns the page unlocked, but with a reference */ -static int ll_dir_readpage(struct file *file, struct page *page) +static int ll_dir_readpage(struct file *file, struct page *page0) { - struct inode *inode = page->mapping->host; + struct inode *inode = page0->mapping->host; + int hash64 = ll_i2sbi(inode)->ll_flags & LL_SBI_64BIT_HASH; + struct obd_export *exp = ll_i2sbi(inode)->ll_md_exp; struct ptlrpc_request *request; struct mdt_body *body; struct obd_capa *oc; __u64 hash; + struct page **page_pool; + struct page *page; +#ifndef HAVE_ADD_TO_PAGE_CACHE_LRU + struct pagevec lru_pvec; +#endif + struct lu_dirpage *dp; + int max_pages = ll_i2sbi(inode)->ll_md_brw_size >> CFS_PAGE_SHIFT; + int nrdpgs = 0; /* number of pages read actually */ + int npages; + int i; int rc; ENTRY; - hash = hash_x_index(page->index); - CDEBUG(D_VFSTRACE, "VFS Op:inode=%lu/%u(%p) off %lu\n", - inode->i_ino, inode->i_generation, inode, (unsigned long)hash); + if (file) { + struct ll_file_data *fd = LUSTRE_FPRIVATE(file); + + hash = fd->fd_dir.lfd_next; + } else { + hash = ll_i2info(inode)->lli_sa_pos; + } + CDEBUG(D_VFSTRACE, "VFS Op:inode=%lu/%u(%p) hash "LPU64"\n", + inode->i_ino, inode->i_generation, inode, hash); + + LASSERT(max_pages > 0 && max_pages <= PTLRPC_MAX_BRW_PAGES); + + OBD_ALLOC(page_pool, sizeof(page) * max_pages); + if (page_pool != NULL) { + page_pool[0] = page0; + } else { + page_pool = &page0; + max_pages = 1; + } + for (npages = 1; npages < max_pages; npages++) { + page = page_cache_alloc_cold(inode->i_mapping); + if (!page) + break; + page_pool[npages] = page; + } oc = ll_mdscapa_get(inode); - rc = md_readpage(ll_i2sbi(inode)->ll_md_exp, ll_inode2fid(inode), - oc, hash, page, &request); + rc = md_readpage(exp, ll_inode2fid(inode), oc, hash, page_pool, npages, + &request); capa_put(oc); - if (!rc) { + if (rc == 0) { body = req_capsule_server_get(&request->rq_pill, &RMF_MDT_BODY); /* Checked by mdc_readpage() */ LASSERT(body != NULL); - if (body->valid & OBD_MD_FLSIZE) { - ll_inode_size_lock(inode, 0); - i_size_write(inode, body->size); - ll_inode_size_unlock(inode, 0); + if (body->valid & OBD_MD_FLSIZE) + cl_isize_write(inode, body->size); + + nrdpgs = (request->rq_bulk->bd_nob_transferred+CFS_PAGE_SIZE-1) + >> CFS_PAGE_SHIFT; + SetPageUptodate(page0); + } + unlock_page(page0); + ptlrpc_req_finished(request); + + CDEBUG(D_VFSTRACE, "read %d/%d pages\n", nrdpgs, npages); + + ll_pagevec_init(&lru_pvec, 0); + for (i = 1; i < npages; i++) { + unsigned long offset; + int ret; + + page = page_pool[i]; + + if (rc < 0 || i >= nrdpgs) { + page_cache_release(page); + continue; } + SetPageUptodate(page); + + dp = cfs_kmap(page); + hash = le64_to_cpu(dp->ldp_hash_start); + cfs_kunmap(page); + + offset = hash_x_index(hash, hash64); + + prefetchw(&page->flags); + ret = ll_add_to_page_cache_lru(page, inode->i_mapping, offset, + GFP_KERNEL); + if (ret == 0) { + unlock_page(page); + page_cache_get(page); + if (ll_pagevec_add(&lru_pvec, page) == 0) + ll_pagevec_lru_add_file(&lru_pvec); + } else { + CDEBUG(D_VFSTRACE, "page %lu add to page cache failed:" + " %d\n", offset, ret); + } + page_cache_release(page); } - ptlrpc_req_finished(request); + ll_pagevec_lru_add_file(&lru_pvec); - unlock_page(page); + if (page_pool != &page0) + OBD_FREE(page_pool, sizeof(struct page *) * max_pages); EXIT; return rc; } +#ifndef MS_HAS_NEW_AOPS struct address_space_operations ll_dir_aops = { .readpage = ll_dir_readpage, }; - -static inline unsigned long dir_pages(struct inode *inode) -{ - return (i_size_read(inode) + CFS_PAGE_SIZE - 1) >> CFS_PAGE_SHIFT; -} - -static inline unsigned ll_chunk_size(struct inode *inode) -{ - return inode->i_sb->s_blocksize; -} +#else +struct address_space_operations_ext ll_dir_aops = { + .orig_aops.readpage = ll_dir_readpage, +}; +#endif static void ll_check_page(struct inode *dir, struct page *page) { @@ -191,35 +279,42 @@ static void ll_check_page(struct inode *dir, struct page *page) SetPageChecked(page); } -static inline void ll_put_page(struct page *page) +void ll_release_page(struct page *page, int remove) { kunmap(page); + if (remove) { + lock_page(page); + if (likely(page->mapping != NULL)) + truncate_complete_page(page->mapping, page); + unlock_page(page); + } page_cache_release(page); } /* * Find, kmap and return page that contains given hash. */ -static struct page *ll_dir_page_locate(struct inode *dir, unsigned long hash, +static struct page *ll_dir_page_locate(struct inode *dir, __u64 *hash, __u64 *start, __u64 *end) { + int hash64 = ll_i2sbi(dir)->ll_flags & LL_SBI_64BIT_HASH; struct address_space *mapping = dir->i_mapping; /* * Complement of hash is used as an index so that * radix_tree_gang_lookup() can be used to find a page with starting * hash _smaller_ than one we are looking for. */ - unsigned long offset = hash_x_index(hash); + unsigned long offset = hash_x_index(*hash, hash64); struct page *page; int found; TREE_READ_LOCK_IRQ(mapping); - found = radix_tree_gang_lookup(&mapping->page_tree, + found = radix_tree_gang_lookup(&mapping->page_tree, (void **)&page, offset, 1); - if (found > 0) { + if (found > 0) { struct lu_dirpage *dp; - page_cache_get(page); + page_cache_get(page); TREE_READ_UNLOCK_IRQ(mapping); /* * In contrast to find_lock_page() we are sure that directory @@ -231,16 +326,31 @@ static struct page *ll_dir_page_locate(struct inode *dir, unsigned long hash, */ wait_on_page(page); if (PageUptodate(page)) { - dp = kmap(page); - *start = le64_to_cpu(dp->ldp_hash_start); - *end = le64_to_cpu(dp->ldp_hash_end); - LASSERT(*start <= hash); - if (hash > *end || (*end != *start && hash == *end)) { - kunmap(page); - lock_page(page); - ll_truncate_complete_page(page); - unlock_page(page); - page_cache_release(page); + dp = cfs_kmap(page); + if (BITS_PER_LONG == 32 && hash64) { + *start = le64_to_cpu(dp->ldp_hash_start) >> 32; + *end = le64_to_cpu(dp->ldp_hash_end) >> 32; + *hash = *hash >> 32; + } else { + *start = le64_to_cpu(dp->ldp_hash_start); + *end = le64_to_cpu(dp->ldp_hash_end); + } + LASSERTF(*start <= *hash, "start = "LPX64",end = " + LPX64",hash = "LPX64"\n", *start, *end, *hash); + CDEBUG(D_VFSTRACE, "page %lu [%llu %llu], hash "LPU64"\n", + offset, *start, *end, *hash); + if (*hash > *end) { + ll_release_page(page, 0); + page = NULL; + } else if (*end != *start && *hash == *end) { + /* + * upon hash collision, remove this page, + * otherwise put page reference, and + * ll_get_dir_page() will issue RPC to fetch + * the page we want. + */ + ll_release_page(page, + le32_to_cpu(dp->ldp_flags) & LDF_COLLIDE); page = NULL; } } else { @@ -248,15 +358,15 @@ static struct page *ll_dir_page_locate(struct inode *dir, unsigned long hash, page = ERR_PTR(-EIO); } - } else { + } else { TREE_READ_UNLOCK_IRQ(mapping); page = NULL; } return page; } -static struct page *ll_get_dir_page(struct inode *dir, __u64 hash, int exact, - struct ll_dir_chain *chain) +struct page *ll_get_dir_page(struct file *filp, struct inode *dir, __u64 hash, + struct ll_dir_chain *chain) { ldlm_policy_data_t policy = {.l_inodebits = {MDS_INODELOCK_UPDATE} }; struct address_space *mapping = dir->i_mapping; @@ -265,26 +375,30 @@ static struct page *ll_get_dir_page(struct inode *dir, __u64 hash, int exact, struct page *page; ldlm_mode_t mode; int rc; - __u64 start; - __u64 end; + __u64 start = 0; + __u64 end = 0; + __u64 lhash = hash; + struct ll_inode_info *lli = ll_i2info(dir); + int hash64 = ll_i2sbi(dir)->ll_flags & LL_SBI_64BIT_HASH; mode = LCK_PR; rc = md_lock_match(ll_i2sbi(dir)->ll_md_exp, LDLM_FL_BLOCK_GRANTED, ll_inode2fid(dir), LDLM_IBITS, &policy, mode, &lockh); if (!rc) { struct ldlm_enqueue_info einfo = { LDLM_IBITS, mode, - ll_md_blocking_ast, ldlm_completion_ast, NULL, dir }; + ll_md_blocking_ast, ldlm_completion_ast, + NULL, NULL, dir }; struct lookup_intent it = { .it_op = IT_READDIR }; struct ptlrpc_request *request; struct md_op_data *op_data; - op_data = ll_prep_md_op_data(NULL, dir, NULL, NULL, 0, 0, + op_data = ll_prep_md_op_data(NULL, dir, NULL, NULL, 0, 0, LUSTRE_OPC_ANY, NULL); if (IS_ERR(op_data)) return (void *)op_data; rc = md_enqueue(ll_i2sbi(dir)->ll_md_exp, &einfo, &it, - op_data, &lockh, NULL, 0, 0); + op_data, &lockh, NULL, 0, NULL, 0); ll_finish_md_op_data(op_data); @@ -292,21 +406,25 @@ static struct page *ll_get_dir_page(struct inode *dir, __u64 hash, int exact, if (request) ptlrpc_req_finished(request); if (rc < 0) { - CERROR("lock enqueue: rc: %d\n", rc); + CERROR("lock enqueue: "DFID" at "LPU64": rc %d\n", + PFID(ll_inode2fid(dir)), hash, rc); return ERR_PTR(rc); } } else { /* for cross-ref object, l_ast_data of the lock may not be set, * we reset it here */ - md_set_lock_data(ll_i2sbi(dir)->ll_md_exp, &lockh.cookie, dir); + md_set_lock_data(ll_i2sbi(dir)->ll_md_exp, &lockh.cookie, + dir, NULL); } ldlm_lock_dump_handle(D_OTHER, &lockh); - page = ll_dir_page_locate(dir, hash, &start, &end); - if (IS_ERR(page)) + cfs_down(&lli->lli_readdir_sem); + page = ll_dir_page_locate(dir, &lhash, &start, &end); + if (IS_ERR(page)) { + CERROR("dir page locate: "DFID" at "LPU64": rc %ld\n", + PFID(ll_inode2fid(dir)), lhash, PTR_ERR(page)); GOTO(out_unlock, page); - - if (page != NULL) { + } else if (page != NULL) { /* * XXX nikita: not entirely correct handling of a corner case: * suppose hash chain of entries with hash value HASH crosses @@ -321,44 +439,50 @@ static struct page *ll_get_dir_page(struct inode *dir, __u64 hash, int exact, * it as an "overflow" page. 1. invalidate all pages at * once. 2. use HASH|1 as an index for P1. */ - if (exact && hash != start) { - /* - * readdir asked for a page starting _exactly_ from - * given hash, but cache contains stale page, with - * entries with smaller hash values. Stale page should - * be invalidated, and new one fetched. - */ - CWARN("Stale readpage page %p: %#lx != %#lx\n", page, - (unsigned long)hash, (unsigned long)start); - lock_page(page); - ll_truncate_complete_page(page); - unlock_page(page); - page_cache_release(page); - } else - GOTO(hash_collision, page); + GOTO(hash_collision, page); } - page = read_cache_page(mapping, hash_x_index(hash), - (filler_t*)mapping->a_ops->readpage, NULL); - if (IS_ERR(page)) + page = read_cache_page(mapping, hash_x_index(hash, hash64), + (filler_t*)mapping->a_ops->readpage, filp); + if (IS_ERR(page)) { + CERROR("read cache page: "DFID" at "LPU64": rc %ld\n", + PFID(ll_inode2fid(dir)), hash, PTR_ERR(page)); GOTO(out_unlock, page); + } wait_on_page(page); (void)kmap(page); - if (!PageUptodate(page)) + if (!PageUptodate(page)) { + CERROR("page not updated: "DFID" at "LPU64": rc %d\n", + PFID(ll_inode2fid(dir)), hash, -5); goto fail; + } if (!PageChecked(page)) ll_check_page(dir, page); - if (PageError(page)) + if (PageError(page)) { + CERROR("page error: "DFID" at "LPU64": rc %d\n", + PFID(ll_inode2fid(dir)), hash, -5); goto fail; + } hash_collision: dp = page_address(page); - - start = le64_to_cpu(dp->ldp_hash_start); - end = le64_to_cpu(dp->ldp_hash_end); + if (BITS_PER_LONG == 32 && hash64) { + start = le64_to_cpu(dp->ldp_hash_start) >> 32; + end = le64_to_cpu(dp->ldp_hash_end) >> 32; + lhash = hash >> 32; + } else { + start = le64_to_cpu(dp->ldp_hash_start); + end = le64_to_cpu(dp->ldp_hash_end); + lhash = hash; + } if (end == start) { - LASSERT(start == hash); - CWARN("Page-wide hash collision: %#lx\n", (unsigned long)end); + LASSERT(start == lhash); + CWARN("Page-wide hash collision: "LPU64"\n", end); + if (BITS_PER_LONG == 32 && hash64) + CWARN("Real page-wide hash collision at ["LPU64" "LPU64 + "] with hash "LPU64"\n", + le64_to_cpu(dp->ldp_hash_start), + le64_to_cpu(dp->ldp_hash_end), hash); /* * Fetch whole overflow chain... * @@ -367,33 +491,37 @@ hash_collision: goto fail; } out_unlock: + cfs_up(&lli->lli_readdir_sem); ldlm_lock_decref(&lockh, mode); return page; fail: - ll_put_page(page); + ll_release_page(page, 1); page = ERR_PTR(-EIO); goto out_unlock; } int ll_readdir(struct file *filp, void *cookie, filldir_t filldir) { - struct inode *inode = filp->f_dentry->d_inode; - struct ll_inode_info *info = ll_i2info(inode); - struct ll_sb_info *sbi = ll_i2sbi(inode); - __u64 pos = filp->f_pos; + struct inode *inode = filp->f_dentry->d_inode; + struct ll_inode_info *info = ll_i2info(inode); + struct ll_sb_info *sbi = ll_i2sbi(inode); + struct ll_file_data *fd = LUSTRE_FPRIVATE(filp); + __u64 pos = fd->fd_dir.lfd_pos; + int api32 = ll_need_32bit_api(sbi); + int hash64 = sbi->ll_flags & LL_SBI_64BIT_HASH; struct page *page; struct ll_dir_chain chain; - int rc; - int done; - int shift; + int done; + int shift; + int rc; ENTRY; - CDEBUG(D_VFSTRACE, "VFS Op:inode=%lu/%u(%p) pos %lu/%llu\n", + CDEBUG(D_VFSTRACE, "VFS Op:inode=%lu/%u(%p) pos %lu/%llu 32bit_api %d\n", inode->i_ino, inode->i_generation, inode, - (unsigned long)pos, i_size_read(inode)); + (unsigned long)pos, i_size_read(inode), api32); - if (pos == DIR_END_OFF) + if (pos == MDS_DIR_END_OFF) /* * end-of-file. */ @@ -404,35 +532,35 @@ int ll_readdir(struct file *filp, void *cookie, filldir_t filldir) shift = 0; ll_dir_chain_init(&chain); - page = ll_get_dir_page(inode, pos, 0, &chain); + fd->fd_dir.lfd_next = pos; + page = ll_get_dir_page(filp, inode, pos, &chain); while (rc == 0 && !done) { struct lu_dirpage *dp; struct lu_dirent *ent; if (!IS_ERR(page)) { - /* - * If page is empty (end of directoryis reached), - * use this value. + /* + * If page is empty (end of directory is reached), + * use this value. */ - __u64 hash = DIR_END_OFF; + __u64 hash = MDS_DIR_END_OFF; __u64 next; dp = page_address(page); for (ent = lu_dirent_start(dp); ent != NULL && !done; ent = lu_dirent_next(ent)) { - char *name; + __u16 type; int namelen; struct lu_fid fid; - ino_t ino; + __u64 lhash; + __u64 ino; /* * XXX: implement correct swabbing here. */ - hash = le64_to_cpu(ent->lde_hash); - namelen = le16_to_cpu(ent->lde_namelen); - + hash = le64_to_cpu(ent->lde_hash); if (hash < pos) /* * Skip until we find target hash @@ -440,43 +568,59 @@ int ll_readdir(struct file *filp, void *cookie, filldir_t filldir) */ continue; + namelen = le16_to_cpu(ent->lde_namelen); if (namelen == 0) /* * Skip dummy record. */ continue; - fid = ent->lde_fid; - name = ent->lde_name; - fid_le_to_cpu(&fid, &fid); - ino = ll_fid_build_ino(sbi, &fid); - - done = filldir(cookie, name, namelen, - (loff_t)hash, ino, DT_UNKNOWN); + if (api32 && hash64) + lhash = hash >> 32; + else + lhash = hash; + fid_le_to_cpu(&fid, &ent->lde_fid); + ino = cl_fid_build_ino(&fid, api32); + type = ll_dirent_type_get(ent); + /* For 'll_nfs_get_name_filldir()', it will try + * to access the 'ent' through its 'lde_name', + * so the parameter 'name' for 'filldir()' must + * be part of the 'ent'. */ + done = filldir(cookie, ent->lde_name, namelen, + lhash, ino, type); } next = le64_to_cpu(dp->ldp_hash_end); - ll_put_page(page); if (!done) { pos = next; - if (pos == DIR_END_OFF) + if (pos == MDS_DIR_END_OFF) { /* * End of directory reached. */ done = 1; - else if (1 /* chain is exhausted*/) + ll_release_page(page, 0); + } else if (1 /* chain is exhausted*/) { /* * Normal case: continue to the next * page. */ - page = ll_get_dir_page(inode, pos, 1, + ll_release_page(page, + le32_to_cpu(dp->ldp_flags) & + LDF_COLLIDE); + fd->fd_dir.lfd_next = pos; + page = ll_get_dir_page(filp, inode, pos, &chain); - else { + } else { /* * go into overflow page. */ + LASSERT(le32_to_cpu(dp->ldp_flags) & + LDF_COLLIDE); + ll_release_page(page, 1); } - } else + } else { pos = hash; + ll_release_page(page, 0); + } } else { rc = PTR_ERR(page); CERROR("error reading dir "DFID" at %lu: rc %d\n", @@ -484,7 +628,18 @@ int ll_readdir(struct file *filp, void *cookie, filldir_t filldir) } } - filp->f_pos = (loff_t)(__s32)pos; + fd->fd_dir.lfd_pos = pos; + if (pos == MDS_DIR_END_OFF) { + if (api32) + filp->f_pos = LL_DIR_END_OFF_32BIT; + else + filp->f_pos = LL_DIR_END_OFF; + } else { + if (api32 && hash64) + filp->f_pos = pos >> 32; + else + filp->f_pos = pos; + } filp->f_version = inode->i_version; touch_atime(filp->f_vfsmnt, filp->f_dentry); @@ -493,16 +648,6 @@ int ll_readdir(struct file *filp, void *cookie, filldir_t filldir) RETURN(rc); } -#define QCTL_COPY(out, in) \ -do { \ - Q_COPY(out, in, qc_cmd); \ - Q_COPY(out, in, qc_type); \ - Q_COPY(out, in, qc_id); \ - Q_COPY(out, in, qc_stat); \ - Q_COPY(out, in, qc_dqinfo); \ - Q_COPY(out, in, qc_dqblk); \ -} while (0) - int ll_send_mgc_param(struct obd_export *mgc, char *string) { struct mgs_send_param *msp; @@ -513,7 +658,7 @@ int ll_send_mgc_param(struct obd_export *mgc, char *string) return -ENOMEM; strncpy(msp->mgs_param, string, MGS_PARAM_MAXLEN); - rc = obd_set_info_async(mgc, strlen(KEY_SET_INFO), KEY_SET_INFO, + rc = obd_set_info_async(mgc, sizeof(KEY_SET_INFO), KEY_SET_INFO, sizeof(struct mgs_send_param), msp, NULL); if (rc) CERROR("Failed to set parameter: %d\n", rc); @@ -549,17 +694,39 @@ int ll_dir_setstripe(struct inode *inode, struct lov_user_md *lump, struct lustre_sb_info *lsi = s2lsi(inode->i_sb); struct obd_device *mgc = lsi->lsi_mgc; char *fsname = NULL, *param = NULL; + int lum_size; - /* - * This is coming from userspace, so should be in - * local endian. But the MDS would like it in little - * endian, so we swab it before we send it. - */ - if (lump->lmm_magic != LOV_USER_MAGIC) - RETURN(-EINVAL); - - if (lump->lmm_magic != cpu_to_le32(LOV_USER_MAGIC)) - lustre_swab_lov_user_md(lump); + if (lump != NULL) { + /* + * This is coming from userspace, so should be in + * local endian. But the MDS would like it in little + * endian, so we swab it before we send it. + */ + switch (lump->lmm_magic) { + case LOV_USER_MAGIC_V1: { + if (lump->lmm_magic != cpu_to_le32(LOV_USER_MAGIC_V1)) + lustre_swab_lov_user_md_v1(lump); + lum_size = sizeof(struct lov_user_md_v1); + break; + } + case LOV_USER_MAGIC_V3: { + if (lump->lmm_magic != cpu_to_le32(LOV_USER_MAGIC_V3)) + lustre_swab_lov_user_md_v3( + (struct lov_user_md_v3 *)lump); + lum_size = sizeof(struct lov_user_md_v3); + break; + } + default: { + CDEBUG(D_IOCTL, "bad userland LOV MAGIC:" + " %#08x != %#08x nor %#08x\n", + lump->lmm_magic, LOV_USER_MAGIC_V1, + LOV_USER_MAGIC_V3); + RETURN(-EINVAL); + } + } + } else { + lum_size = sizeof(struct lov_user_md_v1); + } op_data = ll_prep_md_op_data(NULL, inode, NULL, NULL, 0, 0, LUSTRE_OPC_ANY, NULL); @@ -567,7 +734,7 @@ int ll_dir_setstripe(struct inode *inode, struct lov_user_md *lump, RETURN(PTR_ERR(op_data)); /* swabbing is done in lov_setstripe() on server side */ - rc = md_setattr(sbi->ll_md_exp, op_data, lump, sizeof(*lump), + rc = md_setattr(sbi->ll_md_exp, op_data, lump, lum_size, NULL, 0, &req, NULL); ll_finish_md_op_data(op_data); ptlrpc_req_finished(req); @@ -576,6 +743,9 @@ int ll_dir_setstripe(struct inode *inode, struct lov_user_md *lump, CERROR("mdc_setattr fails: rc = %d\n", rc); } + /* In the following we use the fact that LOV_USER_MAGIC_V1 and + LOV_USER_MAGIC_V3 have the same initial fields so we do not + need the make the distiction between the 2 versions */ if (set_default && mgc->u.cli.cl_mgc_mgsexp) { OBD_ALLOC(param, MGS_PARAM_MAXLEN); @@ -583,21 +753,22 @@ int ll_dir_setstripe(struct inode *inode, struct lov_user_md *lump, fsname = ll_get_fsname(inode); /* Set root stripesize */ sprintf(param, "%s-MDT0000.lov.stripesize=%u", fsname, - lump->lmm_stripe_size); + lump ? le32_to_cpu(lump->lmm_stripe_size) : 0); rc = ll_send_mgc_param(mgc->u.cli.cl_mgc_mgsexp, param); if (rc) goto end; /* Set root stripecount */ - sprintf(param, "%s-MDT0000.lov.stripecount=%u", fsname, - lump->lmm_stripe_count); + sprintf(param, "%s-MDT0000.lov.stripecount=%hd", fsname, + lump ? le16_to_cpu(lump->lmm_stripe_count) : 0); rc = ll_send_mgc_param(mgc->u.cli.cl_mgc_mgsexp, param); if (rc) goto end; /* Set root stripeoffset */ - sprintf(param, "%s-MDT0000.lov.stripeoffset=%u", fsname, - lump->lmm_stripe_offset); + sprintf(param, "%s-MDT0000.lov.stripeoffset=%hd", fsname, + lump ? le16_to_cpu(lump->lmm_stripe_offset) : + (typeof(lump->lmm_stripe_offset))(-1)); rc = ll_send_mgc_param(mgc->u.cli.cl_mgc_mgsexp, param); if (rc) goto end; @@ -610,25 +781,29 @@ end: return rc; } -int ll_dir_getstripe(struct inode *inode, struct lov_mds_md **lmmp, - int *lmm_size, struct ptlrpc_request **request) +int ll_dir_getstripe(struct inode *inode, struct lov_mds_md **lmmp, + int *lmm_size, struct ptlrpc_request **request) { struct ll_sb_info *sbi = ll_i2sbi(inode); struct mdt_body *body; struct lov_mds_md *lmm = NULL; struct ptlrpc_request *req = NULL; int rc, lmmsize; - struct obd_capa *oc; - + struct md_op_data *op_data; + rc = ll_get_max_mdsize(sbi, &lmmsize); if (rc) RETURN(rc); - oc = ll_mdscapa_get(inode); - rc = md_getattr(sbi->ll_md_exp, ll_inode2fid(inode), - oc, OBD_MD_FLEASIZE | OBD_MD_FLDIREA, - lmmsize, &req); - capa_put(oc); + op_data = ll_prep_md_op_data(NULL, inode, NULL, NULL, + 0, lmmsize, LUSTRE_OPC_ANY, + NULL); + if (op_data == NULL) + RETURN(-ENOMEM); + + op_data->op_valid = OBD_MD_FLEASIZE | OBD_MD_FLDIREA; + rc = md_getattr(sbi->ll_md_exp, op_data, &req); + ll_finish_md_op_data(op_data); if (rc < 0) { CDEBUG(D_INFO, "md_getattr failed on inode " "%lu/%u: rc %d\n", inode->i_ino, @@ -655,9 +830,19 @@ int ll_dir_getstripe(struct inode *inode, struct lov_mds_md **lmmp, * little endian. We convert it to host endian before * passing it to userspace. */ - if (lmm->lmm_magic == __swab32(LOV_MAGIC)) { - lustre_swab_lov_user_md((struct lov_user_md *)lmm); - lustre_swab_lov_user_md_objects((struct lov_user_md *)lmm); + /* We don't swab objects for directories */ + switch (le32_to_cpu(lmm->lmm_magic)) { + case LOV_MAGIC_V1: + if (LOV_MAGIC != cpu_to_le32(LOV_MAGIC)) + lustre_swab_lov_user_md_v1((struct lov_user_md_v1 *)lmm); + break; + case LOV_MAGIC_V3: + if (LOV_MAGIC != cpu_to_le32(LOV_MAGIC)) + lustre_swab_lov_user_md_v3((struct lov_user_md_v3 *)lmm); + break; + default: + CERROR("unknown magic: %lX\n", (unsigned long)lmm->lmm_magic); + rc = -EPROTO; } out: *lmmp = lmm; @@ -666,11 +851,151 @@ out: return rc; } +/* + * Get MDT index for the inode. + */ +int ll_get_mdt_idx(struct inode *inode) +{ + struct ll_sb_info *sbi = ll_i2sbi(inode); + struct md_op_data *op_data; + int rc, mdtidx; + ENTRY; + + op_data = ll_prep_md_op_data(NULL, inode, NULL, NULL, 0, + 0, LUSTRE_OPC_ANY, NULL); + if (op_data == NULL) + RETURN(-ENOMEM); + + op_data->op_valid |= OBD_MD_MDTIDX; + rc = md_getattr(sbi->ll_md_exp, op_data, NULL); + mdtidx = op_data->op_mds; + ll_finish_md_op_data(op_data); + if (rc < 0) { + CDEBUG(D_INFO, "md_getattr_name: %d\n", rc); + RETURN(rc); + } + return mdtidx; +} + +static int copy_and_ioctl(int cmd, struct obd_export *exp, void *data, int len) +{ + void *ptr; + int rc; + + OBD_ALLOC(ptr, len); + if (ptr == NULL) + return -ENOMEM; + if (cfs_copy_from_user(ptr, data, len)) { + OBD_FREE(ptr, len); + return -EFAULT; + } + rc = obd_iocontrol(cmd, exp, len, data, NULL); + OBD_FREE(ptr, len); + return rc; +} + +static int quotactl_ioctl(struct ll_sb_info *sbi, struct if_quotactl *qctl) +{ + int cmd = qctl->qc_cmd; + int type = qctl->qc_type; + int id = qctl->qc_id; + int valid = qctl->qc_valid; + int rc = 0; + ENTRY; + + switch (cmd) { + case LUSTRE_Q_INVALIDATE: + case LUSTRE_Q_FINVALIDATE: + case Q_QUOTAON: + case Q_QUOTAOFF: + case Q_SETQUOTA: + case Q_SETINFO: + if (!cfs_capable(CFS_CAP_SYS_ADMIN) || + sbi->ll_flags & LL_SBI_RMT_CLIENT) + RETURN(-EPERM); + break; + case Q_GETQUOTA: + if (((type == USRQUOTA && cfs_curproc_euid() != id) || + (type == GRPQUOTA && !in_egroup_p(id))) && + (!cfs_capable(CFS_CAP_SYS_ADMIN) || + sbi->ll_flags & LL_SBI_RMT_CLIENT)) + RETURN(-EPERM); + break; + case Q_GETINFO: + break; + default: + CERROR("unsupported quotactl op: %#x\n", cmd); + RETURN(-ENOTTY); + } + + if (valid != QC_GENERAL) { + if (sbi->ll_flags & LL_SBI_RMT_CLIENT) + RETURN(-EOPNOTSUPP); + + if (cmd == Q_GETINFO) + qctl->qc_cmd = Q_GETOINFO; + else if (cmd == Q_GETQUOTA) + qctl->qc_cmd = Q_GETOQUOTA; + else + RETURN(-EINVAL); + + switch (valid) { + case QC_MDTIDX: + rc = obd_iocontrol(OBD_IOC_QUOTACTL, sbi->ll_md_exp, + sizeof(*qctl), qctl, NULL); + break; + case QC_OSTIDX: + rc = obd_iocontrol(OBD_IOC_QUOTACTL, sbi->ll_dt_exp, + sizeof(*qctl), qctl, NULL); + break; + case QC_UUID: + rc = obd_iocontrol(OBD_IOC_QUOTACTL, sbi->ll_md_exp, + sizeof(*qctl), qctl, NULL); + if (rc == -EAGAIN) + rc = obd_iocontrol(OBD_IOC_QUOTACTL, + sbi->ll_dt_exp, + sizeof(*qctl), qctl, NULL); + break; + default: + rc = -EINVAL; + break; + } + + if (rc) + RETURN(rc); + + qctl->qc_cmd = cmd; + } else { + struct obd_quotactl *oqctl; + + OBD_ALLOC_PTR(oqctl); + if (oqctl == NULL) + RETURN(-ENOMEM); + + QCTL_COPY(oqctl, qctl); + rc = obd_quotactl(sbi->ll_md_exp, oqctl); + if (rc) { + if (rc != -EALREADY && cmd == Q_QUOTAON) { + oqctl->qc_cmd = Q_QUOTAOFF; + obd_quotactl(sbi->ll_md_exp, oqctl); + } + OBD_FREE_PTR(oqctl); + RETURN(rc); + } + + QCTL_COPY(qctl, oqctl); + OBD_FREE_PTR(oqctl); + } + + RETURN(rc); +} + static int ll_dir_ioctl(struct inode *inode, struct file *file, unsigned int cmd, unsigned long arg) { struct ll_sb_info *sbi = ll_i2sbi(inode); struct obd_ioctl_data *data; + int rc = 0; ENTRY; CDEBUG(D_VFSTRACE, "VFS Op:inode=%lu/%u(%p), cmd=%#x\n", @@ -682,24 +1007,36 @@ static int ll_dir_ioctl(struct inode *inode, struct file *file, ll_stats_ops_tally(ll_i2sbi(inode), LPROC_LL_IOCTL, 1); switch(cmd) { - case EXT3_IOC_GETFLAGS: - case EXT3_IOC_SETFLAGS: + case FSFILT_IOC_GETFLAGS: + case FSFILT_IOC_SETFLAGS: RETURN(ll_iocontrol(inode, file, cmd, arg)); - case EXT3_IOC_GETVERSION_OLD: - case EXT3_IOC_GETVERSION: + case FSFILT_IOC_GETVERSION_OLD: + case FSFILT_IOC_GETVERSION: RETURN(put_user(inode->i_generation, (int *)arg)); /* We need to special case any other ioctls we want to handle, * to send them to the MDS/OST as appropriate and to properly * network encode the arg field. - case EXT3_IOC_SETVERSION_OLD: - case EXT3_IOC_SETVERSION: + case FSFILT_IOC_SETVERSION_OLD: + case FSFILT_IOC_SETVERSION: */ + case LL_IOC_GET_MDTIDX: { + int mdtidx; + + mdtidx = ll_get_mdt_idx(inode); + if (mdtidx < 0) + RETURN(mdtidx); + + if (put_user((int)mdtidx, (int*)arg)) + RETURN(-EFAULT); + + return 0; + } case IOC_MDC_LOOKUP: { struct ptlrpc_request *request = NULL; - int namelen, rc, len = 0; + int namelen, len = 0; char *buf = NULL; char *filename; - struct obd_capa *oc; + struct md_op_data *op_data; rc = obd_ioctl_getdata(&buf, &len, (void *)arg); if (rc) @@ -707,46 +1044,56 @@ static int ll_dir_ioctl(struct inode *inode, struct file *file, data = (void *)buf; filename = data->ioc_inlbuf1; - namelen = data->ioc_inllen1; + namelen = strlen(filename); if (namelen < 1) { CDEBUG(D_INFO, "IOC_MDC_LOOKUP missing filename\n"); - GOTO(out, rc = -EINVAL); + GOTO(out_free, rc = -EINVAL); } - oc = ll_mdscapa_get(inode); - rc = md_getattr_name(sbi->ll_md_exp, ll_inode2fid(inode), oc, - filename, namelen, OBD_MD_FLID, 0, - ll_i2suppgid(inode), &request); - capa_put(oc); + op_data = ll_prep_md_op_data(NULL, inode, NULL, filename, namelen, + 0, LUSTRE_OPC_ANY, NULL); + if (op_data == NULL) + GOTO(out_free, rc = -ENOMEM); + + op_data->op_valid = OBD_MD_FLID; + rc = md_getattr_name(sbi->ll_md_exp, op_data, &request); + ll_finish_md_op_data(op_data); if (rc < 0) { CDEBUG(D_INFO, "md_getattr_name: %d\n", rc); - GOTO(out, rc); + GOTO(out_free, rc); } - ptlrpc_req_finished(request); - EXIT; - out: +out_free: obd_ioctl_freedata(buf, len); return rc; } case LL_IOC_LOV_SETSTRIPE: { - struct lov_user_md lum, *lump = (struct lov_user_md *)arg; - int rc = 0; + struct lov_user_md_v3 lumv3; + struct lov_user_md_v1 *lumv1 = (struct lov_user_md_v1 *)&lumv3; + struct lov_user_md_v1 *lumv1p = (struct lov_user_md_v1 *)arg; + struct lov_user_md_v3 *lumv3p = (struct lov_user_md_v3 *)arg; + int set_default = 0; - LASSERT(sizeof(lum) == sizeof(*lump)); - LASSERT(sizeof(lum.lmm_objects[0]) == - sizeof(lump->lmm_objects[0])); - rc = copy_from_user(&lum, lump, sizeof(lum)); - if (rc) + LASSERT(sizeof(lumv3) == sizeof(*lumv3p)); + LASSERT(sizeof(lumv3.lmm_objects[0]) == + sizeof(lumv3p->lmm_objects[0])); + /* first try with v1 which is smaller than v3 */ + if (cfs_copy_from_user(lumv1, lumv1p, sizeof(*lumv1))) RETURN(-EFAULT); + if (lumv1->lmm_magic == LOV_USER_MAGIC_V3) { + if (cfs_copy_from_user(&lumv3, lumv3p, sizeof(lumv3))) + RETURN(-EFAULT); + } + if (inode->i_sb->s_root == file->f_dentry) set_default = 1; - rc = ll_dir_setstripe(inode, &lum, set_default); + /* in v1 and v3 cases lumv1 points to data */ + rc = ll_dir_setstripe(inode, lumv1, set_default); RETURN(rc); } @@ -761,7 +1108,7 @@ static int ll_dir_ioctl(struct inode *inode, struct file *file, struct lov_mds_md *lmm = NULL; struct mdt_body *body; char *filename = NULL; - int rc, lmmsize; + int lmmsize; if (cmd == IOC_MDC_GETFILEINFO || cmd == IOC_MDC_GETFILESTRIPE) { @@ -769,7 +1116,7 @@ static int ll_dir_ioctl(struct inode *inode, struct file *file, if (IS_ERR(filename)) RETURN(PTR_ERR(filename)); - rc = ll_lov_getstripe_ea_info(inode, filename, &lmm, + rc = ll_lov_getstripe_ea_info(inode, filename, &lmm, &lmmsize, &request); } else { rc = ll_dir_getstripe(inode, &lmm, &lmmsize, &request); @@ -784,7 +1131,7 @@ static int ll_dir_ioctl(struct inode *inode, struct file *file, } if (rc < 0) { - if (rc == -ENODATA && (cmd == IOC_MDC_GETFILEINFO || + if (rc == -ENODATA && (cmd == IOC_MDC_GETFILEINFO || cmd == LL_IOC_MDC_GETINFO)) GOTO(skip_lmm, rc = 0); else @@ -799,9 +1146,11 @@ static int ll_dir_ioctl(struct inode *inode, struct file *file, lmdp = (struct lov_user_mds_data *)arg; lump = &lmdp->lmd_lmm; } - rc = copy_to_user(lump, lmm, lmmsize); - if (rc) - GOTO(out_lmm, rc = -EFAULT); + if (cfs_copy_to_user(lump, lmm, lmmsize)) { + if (cfs_copy_to_user(lump, lmm, sizeof(*lump))) + GOTO(out_req, rc = -EFAULT); + rc = -EOVERFLOW; + } skip_lmm: if (cmd == IOC_MDC_GETFILEINFO || cmd == LL_IOC_MDC_GETINFO) { struct lov_user_mds_data *lmdp; @@ -822,15 +1171,11 @@ static int ll_dir_ioctl(struct inode *inode, struct file *file, st.st_ino = inode->i_ino; lmdp = (struct lov_user_mds_data *)arg; - rc = copy_to_user(&lmdp->lmd_st, &st, sizeof(st)); - if (rc) - GOTO(out_lmm, rc = -EFAULT); + if (cfs_copy_to_user(&lmdp->lmd_st, &st, sizeof(st))) + GOTO(out_req, rc = -EFAULT); } EXIT; - out_lmm: - if (lmm && lmm->lmm_magic == LOV_MAGIC_JOIN) - OBD_FREE(lmm, lmmsize); out_req: ptlrpc_req_finished(request); if (filename) @@ -844,7 +1189,6 @@ static int ll_dir_ioctl(struct inode *inode, struct file *file, struct lov_mds_md *lmm; int lmmsize; lstat_t st; - int rc; lumd = (struct lov_user_mds_data *)arg; lum = &lumd->lmd_lmm; @@ -853,19 +1197,37 @@ static int ll_dir_ioctl(struct inode *inode, struct file *file, if (rc) RETURN(rc); - OBD_ALLOC(lmm, lmmsize); - rc = copy_from_user(lmm, lum, lmmsize); - if (rc) + OBD_ALLOC_LARGE(lmm, lmmsize); + if (cfs_copy_from_user(lmm, lum, lmmsize)) GOTO(free_lmm, rc = -EFAULT); + switch (lmm->lmm_magic) { + case LOV_USER_MAGIC_V1: + if (LOV_USER_MAGIC_V1 == cpu_to_le32(LOV_USER_MAGIC_V1)) + break; + /* swab objects first so that stripes num will be sane */ + lustre_swab_lov_user_md_objects( + ((struct lov_user_md_v1 *)lmm)->lmm_objects, + ((struct lov_user_md_v1 *)lmm)->lmm_stripe_count); + lustre_swab_lov_user_md_v1((struct lov_user_md_v1 *)lmm); + break; + case LOV_USER_MAGIC_V3: + if (LOV_USER_MAGIC_V3 == cpu_to_le32(LOV_USER_MAGIC_V3)) + break; + /* swab objects first so that stripes num will be sane */ + lustre_swab_lov_user_md_objects( + ((struct lov_user_md_v3 *)lmm)->lmm_objects, + ((struct lov_user_md_v3 *)lmm)->lmm_stripe_count); + lustre_swab_lov_user_md_v3((struct lov_user_md_v3 *)lmm); + break; + default: + GOTO(free_lmm, rc = -EINVAL); + } + rc = obd_unpackmd(sbi->ll_dt_exp, &lsm, lmm, lmmsize); if (rc < 0) GOTO(free_lmm, rc = -ENOMEM); - rc = obd_checkmd(sbi->ll_dt_exp, sbi->ll_md_exp, lsm); - if (rc) - GOTO(free_lsm, rc); - /* Perform glimpse_size operation. */ memset(&st, 0, sizeof(st)); @@ -873,15 +1235,14 @@ static int ll_dir_ioctl(struct inode *inode, struct file *file, if (rc) GOTO(free_lsm, rc); - rc = copy_to_user(&lumd->lmd_st, &st, sizeof(st)); - if (rc) + if (cfs_copy_to_user(&lumd->lmd_st, &st, sizeof(st))) GOTO(free_lsm, rc = -EFAULT); EXIT; free_lsm: obd_free_memmd(sbi->ll_dt_exp, &lsm); free_lmm: - OBD_FREE(lmm, lmmsize); + OBD_FREE_LARGE(lmm, lmmsize); return rc; } case OBD_IOC_LLOG_CATINFO: { @@ -889,7 +1250,6 @@ static int ll_dir_ioctl(struct inode *inode, struct file *file, char *buf = NULL; char *str; int len = 0; - int rc; rc = obd_ioctl_getdata(&buf, &len, (void *)arg); if (rc) @@ -933,7 +1293,9 @@ static int ll_dir_ioctl(struct inode *inode, struct file *file, if (!rc) { str = req_capsule_server_get(&req->rq_pill, &RMF_STRING); - rc = copy_to_user(data->ioc_pbuf1, str, data->ioc_plen1); + if (cfs_copy_to_user(data->ioc_pbuf1, str, + data->ioc_plen1)) + rc = -EFAULT; } ptlrpc_req_finished(req); out_catinfo: @@ -942,9 +1304,10 @@ static int ll_dir_ioctl(struct inode *inode, struct file *file, } case OBD_IOC_QUOTACHECK: { struct obd_quotactl *oqctl; - int rc, error = 0; + int error = 0; - if (!capable(CAP_SYS_ADMIN)) + if (!cfs_capable(CFS_CAP_SYS_ADMIN) || + sbi->ll_flags & LL_SBI_RMT_CLIENT) RETURN(-EPERM); OBD_ALLOC_PTR(oqctl); @@ -966,9 +1329,9 @@ static int ll_dir_ioctl(struct inode *inode, struct file *file, } case OBD_IOC_POLL_QUOTACHECK: { struct if_quotacheck *check; - int rc; - if (!capable(CAP_SYS_ADMIN)) + if (!cfs_capable(CFS_CAP_SYS_ADMIN) || + sbi->ll_flags & LL_SBI_RMT_CLIENT) RETURN(-EPERM); OBD_ALLOC_PTR(check); @@ -979,8 +1342,9 @@ static int ll_dir_ioctl(struct inode *inode, struct file *file, NULL); if (rc) { CDEBUG(D_QUOTA, "mdc ioctl %d failed: %d\n", cmd, rc); - if (copy_to_user((void *)arg, check, sizeof(*check))) - rc = -EFAULT; + if (cfs_copy_to_user((void *)arg, check, + sizeof(*check))) + CDEBUG(D_QUOTA, "cfs_copy_to_user failed\n"); GOTO(out_poll, rc); } @@ -988,132 +1352,94 @@ static int ll_dir_ioctl(struct inode *inode, struct file *file, NULL); if (rc) { CDEBUG(D_QUOTA, "osc ioctl %d failed: %d\n", cmd, rc); - if (copy_to_user((void *)arg, check, sizeof(*check))) - rc = -EFAULT; + if (cfs_copy_to_user((void *)arg, check, + sizeof(*check))) + CDEBUG(D_QUOTA, "cfs_copy_to_user failed\n"); GOTO(out_poll, rc); } out_poll: OBD_FREE_PTR(check); RETURN(rc); } -#ifdef HAVE_QUOTA_SUPPORT - case OBD_IOC_QUOTACTL: { - struct if_quotactl *qctl; - struct obd_quotactl *oqctl; - - int cmd, type, id, rc = 0; - - OBD_ALLOC_PTR(qctl); - if (!qctl) +#if LUSTRE_VERSION_CODE < OBD_OCD_VERSION(2,7,50,0) + case LL_IOC_QUOTACTL_18: { + /* copy the old 1.x quota struct for internal use, then copy + * back into old format struct. For 1.8 compatibility. */ + struct if_quotactl_18 *qctl_18; + struct if_quotactl *qctl_20; + + OBD_ALLOC_PTR(qctl_18); + if (!qctl_18) RETURN(-ENOMEM); - OBD_ALLOC_PTR(oqctl); - if (!oqctl) { - OBD_FREE_PTR(qctl); - RETURN(-ENOMEM); - } - if (copy_from_user(qctl, (void *)arg, sizeof(*qctl))) - GOTO(out_quotactl, rc = -EFAULT); - - cmd = qctl->qc_cmd; - type = qctl->qc_type; - id = qctl->qc_id; - switch (cmd) { - case Q_QUOTAON: - case Q_QUOTAOFF: - case Q_SETQUOTA: - case Q_SETINFO: - if (!capable(CAP_SYS_ADMIN)) - GOTO(out_quotactl, rc = -EPERM); - break; - case Q_GETQUOTA: - if (((type == USRQUOTA && current->euid != id) || - (type == GRPQUOTA && !in_egroup_p(id))) && - !capable(CAP_SYS_ADMIN)) - GOTO(out_quotactl, rc = -EPERM); - - /* XXX: dqb_valid is borrowed as a flag to mark that - * only mds quota is wanted */ - if (qctl->qc_dqblk.dqb_valid) - qctl->obd_uuid = sbi->ll_md_exp->exp_obd-> - u.cli.cl_target_uuid; - break; - case Q_GETINFO: - break; - default: - CERROR("unsupported quotactl op: %#x\n", cmd); - GOTO(out_quotactl, rc = -ENOTTY); + OBD_ALLOC_PTR(qctl_20); + if (!qctl_20) + GOTO(out_quotactl_18, rc = -ENOMEM); + + if (cfs_copy_from_user(qctl_18, (void *)arg, sizeof(*qctl_18))) + GOTO(out_quotactl_20, rc = -ENOMEM); + + QCTL_COPY(qctl_20, qctl_18); + qctl_20->qc_idx = 0; + + /* XXX: dqb_valid was borrowed as a flag to mark that + * only mds quota is wanted */ + if (qctl_18->qc_cmd == Q_GETQUOTA && + qctl_18->qc_dqblk.dqb_valid) { + qctl_20->qc_valid = QC_MDTIDX; + qctl_20->qc_dqblk.dqb_valid = 0; + } else if (qctl_18->obd_uuid.uuid[0] != '\0') { + qctl_20->qc_valid = QC_UUID; + qctl_20->obd_uuid = qctl_18->obd_uuid; + } else { + qctl_20->qc_valid = QC_GENERAL; } - QCTL_COPY(oqctl, qctl); - - if (qctl->obd_uuid.uuid[0]) { - struct obd_device *obd; - struct obd_uuid *uuid = &qctl->obd_uuid; + rc = quotactl_ioctl(sbi, qctl_20); - obd = class_find_client_notype(uuid, - &sbi->ll_dt_exp->exp_obd->obd_uuid); - if (!obd) - GOTO(out_quotactl, rc = -ENOENT); + if (rc == 0) { + QCTL_COPY(qctl_18, qctl_20); + qctl_18->obd_uuid = qctl_20->obd_uuid; - if (cmd == Q_GETINFO) - oqctl->qc_cmd = Q_GETOINFO; - else if (cmd == Q_GETQUOTA) - oqctl->qc_cmd = Q_GETOQUOTA; - else - GOTO(out_quotactl, rc = -EINVAL); - - if (sbi->ll_md_exp->exp_obd == obd) { - rc = obd_quotactl(sbi->ll_md_exp, oqctl); - } else { - int i; - struct obd_export *exp; - struct lov_obd *lov = &sbi->ll_dt_exp-> - exp_obd->u.lov; - - for (i = 0; i < lov->desc.ld_tgt_count; i++) { - if (!lov->lov_tgts[i] || - !lov->lov_tgts[i]->ltd_active) - continue; - exp = lov->lov_tgts[i]->ltd_exp; - if (exp->exp_obd == obd) { - rc = obd_quotactl(exp, oqctl); - break; - } - } - } - - oqctl->qc_cmd = cmd; - QCTL_COPY(qctl, oqctl); - - if (copy_to_user((void *)arg, qctl, sizeof(*qctl))) + if (cfs_copy_to_user((void *)arg, qctl_18, + sizeof(*qctl_18))) rc = -EFAULT; - - GOTO(out_quotactl, rc); } - rc = obd_quotactl(sbi->ll_md_exp, oqctl); - if (rc && rc != -EBUSY && cmd == Q_QUOTAON) { - oqctl->qc_cmd = Q_QUOTAOFF; - obd_quotactl(sbi->ll_md_exp, oqctl); - } + out_quotactl_20: + OBD_FREE_PTR(qctl_20); + out_quotactl_18: + OBD_FREE_PTR(qctl_18); + RETURN(rc); + } +#else +#warning "remove old LL_IOC_QUOTACTL_18 compatibility code" +#endif /* LUSTRE_VERSION_CODE < OBD_OCD_VERSION(2,7,50,0) */ + case LL_IOC_QUOTACTL: { + struct if_quotactl *qctl; - QCTL_COPY(qctl, oqctl); + OBD_ALLOC_PTR(qctl); + if (!qctl) + RETURN(-ENOMEM); - if (copy_to_user((void *)arg, qctl, sizeof(*qctl))) + if (cfs_copy_from_user(qctl, (void *)arg, sizeof(*qctl))) + GOTO(out_quotactl, rc = -EFAULT); + + rc = quotactl_ioctl(sbi, qctl); + + if (rc == 0 && cfs_copy_to_user((void *)arg,qctl,sizeof(*qctl))) rc = -EFAULT; + out_quotactl: OBD_FREE_PTR(qctl); - OBD_FREE_PTR(oqctl); RETURN(rc); } -#endif /* HAVE_QUOTA_SUPPORT */ case OBD_IOC_GETNAME: { struct obd_device *obd = class_exp2obd(sbi->ll_dt_exp); if (!obd) RETURN(-EFAULT); - if (copy_to_user((void *)arg, obd->obd_name, - strlen(obd->obd_name) + 1)) + if (cfs_copy_to_user((void *)arg, obd->obd_name, + strlen(obd->obd_name) + 1)) RETURN (-EFAULT); RETURN(0); } @@ -1124,7 +1450,6 @@ static int ll_dir_ioctl(struct inode *inode, struct file *file, if (sbi->ll_flags & LL_SBI_RMT_CLIENT && inode == inode->i_sb->s_root->d_inode) { struct ll_file_data *fd = LUSTRE_FPRIVATE(file); - int rc; LASSERT(fd != NULL); rc = rct_add(&sbi->ll_rct, cfs_curproc_pid(), arg); @@ -1135,11 +1460,104 @@ static int ll_dir_ioctl(struct inode *inode, struct file *file, RETURN(0); } #endif + case LL_IOC_GETOBDCOUNT: { + int count, vallen; + struct obd_export *exp; + + if (cfs_copy_from_user(&count, (int *)arg, sizeof(int))) + RETURN(-EFAULT); + + /* get ost count when count is zero, get mdt count otherwise */ + exp = count ? sbi->ll_md_exp : sbi->ll_dt_exp; + vallen = sizeof(count); + rc = obd_get_info(exp, sizeof(KEY_TGT_COUNT), KEY_TGT_COUNT, + &vallen, &count, NULL); + if (rc) { + CERROR("get target count failed: %d\n", rc); + RETURN(rc); + } + + if (cfs_copy_to_user((int *)arg, &count, sizeof(int))) + RETURN(-EFAULT); + + RETURN(0); + } + case LL_IOC_PATH2FID: + if (cfs_copy_to_user((void *)arg, ll_inode2fid(inode), + sizeof(struct lu_fid))) + RETURN(-EFAULT); + RETURN(0); + case LL_IOC_GET_CONNECT_FLAGS: { + RETURN(obd_iocontrol(cmd, sbi->ll_md_exp, 0, NULL, (void*)arg)); + } + case OBD_IOC_CHANGELOG_SEND: + case OBD_IOC_CHANGELOG_CLEAR: + rc = copy_and_ioctl(cmd, sbi->ll_md_exp, (void *)arg, + sizeof(struct ioc_changelog)); + RETURN(rc); + case OBD_IOC_FID2PATH: + RETURN(ll_fid2path(ll_i2mdexp(inode), (void *)arg)); + case LL_IOC_HSM_CT_START: + rc = copy_and_ioctl(cmd, sbi->ll_md_exp, (void *)arg, + sizeof(struct lustre_kernelcomm)); + RETURN(rc); + default: RETURN(obd_iocontrol(cmd, sbi->ll_dt_exp,0,NULL,(void *)arg)); } } +static loff_t ll_dir_seek(struct file *file, loff_t offset, int origin) +{ + struct inode *inode = file->f_mapping->host; + struct ll_file_data *fd = LUSTRE_FPRIVATE(file); + struct ll_sb_info *sbi = ll_i2sbi(inode); + int api32 = ll_need_32bit_api(sbi); + loff_t ret = -EINVAL; + ENTRY; + + cfs_mutex_lock(&inode->i_mutex); + switch (origin) { + case SEEK_SET: + break; + case SEEK_CUR: + offset += file->f_pos; + break; + case SEEK_END: + if (offset > 0) + GOTO(out, ret); + if (api32) + offset += LL_DIR_END_OFF_32BIT; + else + offset += LL_DIR_END_OFF; + break; + default: + GOTO(out, ret); + } + + if (offset >= 0 && + ((api32 && offset <= LL_DIR_END_OFF_32BIT) || + (!api32 && offset <= LL_DIR_END_OFF))) { + if (offset != file->f_pos) { + if ((api32 && offset == LL_DIR_END_OFF_32BIT) || + (!api32 && offset == LL_DIR_END_OFF)) + fd->fd_dir.lfd_pos = MDS_DIR_END_OFF; + else if (api32 && sbi->ll_flags & LL_SBI_64BIT_HASH) + fd->fd_dir.lfd_pos = offset << 32; + else + fd->fd_dir.lfd_pos = offset; + file->f_pos = offset; + file->f_version = 0; + } + ret = offset; + } + GOTO(out, ret); + +out: + cfs_mutex_unlock(&inode->i_mutex); + return ret; +} + int ll_dir_open(struct inode *inode, struct file *file) { ENTRY; @@ -1153,10 +1571,11 @@ int ll_dir_release(struct inode *inode, struct file *file) } struct file_operations ll_dir_operations = { + .llseek = ll_dir_seek, .open = ll_dir_open, .release = ll_dir_release, .read = generic_read_dir, .readdir = ll_readdir, - .ioctl = ll_dir_ioctl + .ioctl = ll_dir_ioctl, + .fsync = ll_fsync }; -