X-Git-Url: https://git.whamcloud.com/?p=fs%2Flustre-release.git;a=blobdiff_plain;f=lustre%2Fllite%2Fdir.c;h=27edd96877725655af3b8f59723d5bf4cf5674dd;hp=6f26358f3b6af72f02285907173acb73577fd466;hb=322968acf183ab16d952cd3026f6580957b31259;hpb=1cbf25caab01ac561bb35053ca34fb6eb7336b3d;ds=sidebyside diff --git a/lustre/llite/dir.c b/lustre/llite/dir.c index 6f26358..27edd96 100644 --- a/lustre/llite/dir.c +++ b/lustre/llite/dir.c @@ -30,6 +30,9 @@ * 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. * @@ -45,6 +48,7 @@ #include #include #include // for wait_on_buffer +#include #define DEBUG_SUBSYSTEM S_LLITE @@ -138,20 +142,35 @@ * * 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; @@ -160,42 +179,99 @@ static int ll_dir_readpage(struct file *file, struct page *page) hash = fd->fd_dir.lfd_next; } else { - struct ll_inode_info *lli = ll_i2info(inode); + 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); - cfs_spin_lock(&lli->lli_sa_lock); - if (lli->lli_sai) - LASSERT(lli->lli_sai->sai_pid == cfs_curproc_pid()); - else - LASSERT(lli->lli_opendir_pid == cfs_curproc_pid()); - hash = lli->lli_sa_pos; - cfs_spin_unlock(&lli->lli_sa_lock); + 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; } - CDEBUG(D_VFSTRACE, "VFS Op:inode=%lu/%u(%p) off %lu\n", - inode->i_ino, inode->i_generation, inode, (unsigned long)hash); 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) cl_isize_write(inode, body->size); - SetPageUptodate(page); + + nrdpgs = (request->rq_bulk->bd_nob_transferred+CFS_PAGE_SIZE-1) + >> CFS_PAGE_SHIFT; + SetPageUptodate(page0); } + unlock_page(page0); ptlrpc_req_finished(request); - unlock_page(page); + 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); + } + ll_pagevec_lru_add_file(&lru_pvec); + + 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, }; +#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) { @@ -203,19 +279,14 @@ static void ll_check_page(struct inode *dir, struct page *page) SetPageChecked(page); } -static void ll_release_page(struct page *page, __u64 hash, - __u64 start, __u64 end) +void ll_release_page(struct page *page, int remove) { kunmap(page); - lock_page(page); - if (likely(page->mapping != NULL)) { - ll_truncate_complete_page(page); - unlock_page(page); - } else { + if (remove) { + lock_page(page); + if (likely(page->mapping != NULL)) + truncate_complete_page(page->mapping, page); unlock_page(page); - CWARN("NULL mapping page %p, truncated by others: " - "hash("LPX64") | start("LPX64") | end("LPX64")\n", - page, hash, start, end); } page_cache_release(page); } @@ -226,13 +297,14 @@ static void ll_release_page(struct page *page, __u64 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; @@ -254,19 +326,31 @@ static struct page *ll_dir_page_locate(struct inode *dir, __u64 *hash, */ wait_on_page(page); if (PageUptodate(page)) { - dp = kmap(page); -#if BITS_PER_LONG == 32 - *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); -#endif + 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); - if (*hash > *end || (*end != *start && *hash == *end)) { - ll_release_page(page, *hash, *start, *end); + 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 { @@ -282,7 +366,7 @@ static struct page *ll_dir_page_locate(struct inode *dir, __u64 *hash, } struct page *ll_get_dir_page(struct file *filp, struct inode *dir, __u64 hash, - int exact, struct ll_dir_chain *chain) + struct ll_dir_chain *chain) { ldlm_policy_data_t policy = {.l_inodebits = {MDS_INODELOCK_UPDATE} }; struct address_space *mapping = dir->i_mapping; @@ -295,6 +379,7 @@ struct page *ll_get_dir_page(struct file *filp, struct inode *dir, __u64 hash, __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, @@ -339,9 +424,7 @@ struct page *ll_get_dir_page(struct file *filp, struct inode *dir, __u64 hash, 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 @@ -356,23 +439,10 @@ struct page *ll_get_dir_page(struct file *filp, struct inode *dir, __u64 hash, * it as an "overflow" page. 1. invalidate all pages at * once. 2. use HASH|1 as an index for P1. */ - if (exact && lhash != 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. - */ - CDEBUG(D_OTHER, "Stale readpage page %p: " - "start = "LPX64",end = "LPX64"hash ="LPX64"\n", - page, start, end, lhash); - ll_release_page(page, lhash, start, end); - } else { - GOTO(hash_collision, page); - } + GOTO(hash_collision, page); } - page = read_cache_page(mapping, hash_x_index(hash), + 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", @@ -396,23 +466,23 @@ struct page *ll_get_dir_page(struct file *filp, struct inode *dir, __u64 hash, } hash_collision: dp = page_address(page); -#if BITS_PER_LONG == 32 - 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; -#endif + 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 == lhash); CWARN("Page-wide hash collision: "LPU64"\n", end); -#if BITS_PER_LONG == 32 - 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); -#endif + 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... * @@ -426,32 +496,31 @@ out_unlock: 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); - struct ll_file_data *fd = LUSTRE_FPRIVATE(filp); - __u64 pos = fd->fd_dir.lfd_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, need_32bit; - int done; - int shift; - __u16 type; + int done; + int rc; ENTRY; - need_32bit = ll_need_32bit_api(sbi); 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), need_32bit); + (unsigned long)pos, i_size_read(inode), api32); - if (pos == DIR_END_OFF) + if (pos == MDS_DIR_END_OFF) /* * end-of-file. */ @@ -459,11 +528,10 @@ int ll_readdir(struct file *filp, void *cookie, filldir_t filldir) rc = 0; done = 0; - shift = 0; ll_dir_chain_init(&chain); fd->fd_dir.lfd_next = pos; - page = ll_get_dir_page(filp, inode, pos, 0, &chain); + page = ll_get_dir_page(filp, inode, pos, &chain); while (rc == 0 && !done) { struct lu_dirpage *dp; @@ -474,17 +542,17 @@ int ll_readdir(struct file *filp, void *cookie, filldir_t filldir) * 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; - __u64 ino; __u64 lhash; + __u64 ino; /* * XXX: implement correct swabbing here. @@ -505,43 +573,51 @@ int ll_readdir(struct file *filp, void *cookie, filldir_t filldir) */ continue; - name = ent->lde_name; - fid_le_to_cpu(&fid, &ent->lde_fid); - if (need_32bit) { + if (api32 && hash64) lhash = hash >> 32; - ino = cl_fid_build_ino32(&fid); - } else { + else lhash = hash; - ino = cl_fid_build_ino(&fid); - } + fid_le_to_cpu(&fid, &ent->lde_fid); + ino = cl_fid_build_ino(&fid, api32); type = ll_dirent_type_get(ent); - done = filldir(cookie, name, namelen, + /* 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; + ll_release_page(page, 0); } else if (1 /* chain is exhausted*/) { /* * Normal case: continue to the next * page. */ + 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, - 1, &chain); + &chain); } else { /* * go into overflow page. */ + LASSERT(le32_to_cpu(dp->ldp_flags) & + LDF_COLLIDE); + ll_release_page(page, 1); } } else { pos = hash; + ll_release_page(page, 0); } } else { rc = PTR_ERR(page); @@ -551,10 +627,17 @@ int ll_readdir(struct file *filp, void *cookie, filldir_t filldir) } fd->fd_dir.lfd_pos = pos; - if (need_32bit) - filp->f_pos = pos >> 32; - else - filp->f_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); @@ -631,7 +714,7 @@ int ll_dir_setstripe(struct inode *inode, struct lov_user_md *lump, lum_size = sizeof(struct lov_user_md_v3); break; } - default: { + default: { CDEBUG(D_IOCTL, "bad userland LOV MAGIC:" " %#08x != %#08x nor %#08x\n", lump->lmm_magic, LOV_USER_MAGIC_V1, @@ -713,8 +796,8 @@ int ll_dir_getstripe(struct inode *inode, struct lov_mds_md **lmmp, op_data = ll_prep_md_op_data(NULL, inode, NULL, NULL, 0, lmmsize, LUSTRE_OPC_ANY, NULL); - if (op_data == NULL) - RETURN(-ENOMEM); + if (IS_ERR(op_data)) + RETURN(PTR_ERR(op_data)); op_data->op_valid = OBD_MD_FLEASIZE | OBD_MD_FLDIREA; rc = md_getattr(sbi->ll_md_exp, op_data, &req); @@ -778,8 +861,8 @@ int ll_get_mdt_idx(struct inode *inode) op_data = ll_prep_md_op_data(NULL, inode, NULL, NULL, 0, 0, LUSTRE_OPC_ANY, NULL); - if (op_data == NULL) - RETURN(-ENOMEM); + if (IS_ERR(op_data)) + RETURN(PTR_ERR(op_data)); op_data->op_valid |= OBD_MD_MDTIDX; rc = md_getattr(sbi->ll_md_exp, op_data, NULL); @@ -809,9 +892,150 @@ static int copy_and_ioctl(int cmd, struct obd_export *exp, void *data, int len) return rc; } -static int ll_dir_ioctl(struct inode *inode, struct file *file, +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); + } + /* If QIF_SPACE is not set, client should collect the + * space usage from OSSs by itself */ + if (cmd == Q_GETQUOTA && + !(oqctl->qc_dqblk.dqb_valid & QIF_SPACE) && + !oqctl->qc_dqblk.dqb_curspace) { + struct obd_quotactl *oqctl_tmp; + + OBD_ALLOC_PTR(oqctl_tmp); + if (oqctl_tmp == NULL) + GOTO(out, rc = -ENOMEM); + + oqctl_tmp->qc_cmd = Q_GETOQUOTA; + oqctl_tmp->qc_id = oqctl->qc_id; + oqctl_tmp->qc_type = oqctl->qc_type; + + /* collect space usage from OSTs */ + oqctl_tmp->qc_dqblk.dqb_curspace = 0; + rc = obd_quotactl(sbi->ll_dt_exp, oqctl_tmp); + if (!rc || rc == -EREMOTEIO) { + oqctl->qc_dqblk.dqb_curspace = + oqctl_tmp->qc_dqblk.dqb_curspace; + oqctl->qc_dqblk.dqb_valid |= QIF_SPACE; + } + + /* collect space & inode usage from MDTs */ + oqctl_tmp->qc_dqblk.dqb_curspace = 0; + oqctl_tmp->qc_dqblk.dqb_curinodes = 0; + rc = obd_quotactl(sbi->ll_md_exp, oqctl_tmp); + if (!rc || rc == -EREMOTEIO) { + oqctl->qc_dqblk.dqb_curspace += + oqctl_tmp->qc_dqblk.dqb_curspace; + oqctl->qc_dqblk.dqb_curinodes = + oqctl_tmp->qc_dqblk.dqb_curinodes; + oqctl->qc_dqblk.dqb_valid |= QIF_INODES; + } else { + oqctl->qc_dqblk.dqb_valid &= ~QIF_SPACE; + } + + OBD_FREE_PTR(oqctl_tmp); + } +out: + QCTL_COPY(qctl, oqctl); + OBD_FREE_PTR(oqctl); + } + + RETURN(rc); +} + +#ifdef HAVE_UNLOCKED_IOCTL +static long ll_dir_ioctl(struct file *file, unsigned int cmd, unsigned long arg) +#else +static int ll_dir_ioctl(struct inode *unuse, struct file *file, unsigned int cmd, unsigned long arg) +#endif { + struct inode *inode = file->f_dentry->d_inode; struct ll_sb_info *sbi = ll_i2sbi(inode); struct obd_ioctl_data *data; int rc = 0; @@ -872,8 +1096,8 @@ static int ll_dir_ioctl(struct inode *inode, struct file *file, 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); + if (IS_ERR(op_data)) + GOTO(out_free, rc = PTR_ERR(op_data)); op_data->op_valid = OBD_MD_FLID; rc = md_getattr_name(sbi->ll_md_exp, op_data, &request); @@ -1180,9 +1404,62 @@ out_free: OBD_FREE_PTR(check); RETURN(rc); } - case OBD_IOC_QUOTACTL: { +#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(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; + } + + rc = quotactl_ioctl(sbi, qctl_20); + + if (rc == 0) { + QCTL_COPY(qctl_18, qctl_20); + qctl_18->obd_uuid = qctl_20->obd_uuid; + + if (cfs_copy_to_user((void *)arg, qctl_18, + sizeof(*qctl_18))) + rc = -EFAULT; + } + + 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; - int cmd, type, id, valid; OBD_ALLOC_PTR(qctl); if (!qctl) @@ -1191,115 +1468,18 @@ out_free: if (cfs_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; - valid = qctl->qc_valid; - - 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) - GOTO(out_quotactl, rc = -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)) - GOTO(out_quotactl, rc = -EPERM); - break; - case Q_GETINFO: - break; - default: - CERROR("unsupported quotactl op: %#x\n", cmd); - GOTO(out_quotactl, rc = -ENOTTY); - } - - if (valid != QC_GENERAL) { - if (sbi->ll_flags & LL_SBI_RMT_CLIENT) - GOTO(out_quotactl, rc = -EOPNOTSUPP); - - if (cmd == Q_GETINFO) - qctl->qc_cmd = Q_GETOINFO; - else if (cmd == Q_GETQUOTA) - qctl->qc_cmd = Q_GETOQUOTA; - else - GOTO(out_quotactl, rc = -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) - GOTO(out_quotactl, rc); - else - qctl->qc_cmd = cmd; - } else { - struct obd_quotactl *oqctl; - - OBD_ALLOC_PTR(oqctl); - if (!oqctl) - GOTO(out_quotactl, rc = -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); - GOTO(out_quotactl, rc); - } else { - QCTL_COPY(qctl, oqctl); - OBD_FREE_PTR(oqctl); - } - } + rc = quotactl_ioctl(sbi, qctl); - if (cfs_copy_to_user((void *)arg, qctl, sizeof(*qctl))) + if (rc == 0 && cfs_copy_to_user((void *)arg,qctl,sizeof(*qctl))) rc = -EFAULT; out_quotactl: OBD_FREE_PTR(qctl); RETURN(rc); } - case OBD_IOC_GETNAME: { - struct obd_device *obd = class_exp2obd(sbi->ll_dt_exp); - if (!obd) - RETURN(-EFAULT); - if (cfs_copy_to_user((void *)arg, obd->obd_name, - strlen(obd->obd_name) + 1)) - RETURN (-EFAULT); - RETURN(0); - } + case OBD_IOC_GETDTNAME: + case OBD_IOC_GETMDNAME: + RETURN(ll_get_obd_name(inode, cmd, arg)); case LL_IOC_FLUSHCTX: RETURN(ll_flush_ctx(inode)); #ifdef CONFIG_FS_POSIX_ACL @@ -1318,19 +1498,20 @@ out_free: } #endif case LL_IOC_GETOBDCOUNT: { - int count; + int count, vallen; + struct obd_export *exp; if (cfs_copy_from_user(&count, (int *)arg, sizeof(int))) RETURN(-EFAULT); - if (!count) { - /* get ost count */ - struct lov_obd *lov = &sbi->ll_dt_exp->exp_obd->u.lov; - count = lov->desc.ld_tgt_count; - } else { - /* get mdt count */ - struct lmv_obd *lmv = &sbi->ll_md_exp->exp_obd->u.lmv; - count = lmv->desc.ld_tgt_count; + /* 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))) @@ -1365,33 +1546,53 @@ out_free: 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); - loff_t pos = file->f_pos; - loff_t ret; + struct ll_sb_info *sbi = ll_i2sbi(inode); + int api32 = ll_need_32bit_api(sbi); + loff_t ret = -EINVAL; ENTRY; - if (origin == 1 && offset >= 0 && file->f_pos == DIR_END_OFF) { - CWARN("end of dir hash, DIR_END_OFF(-2) is returned\n"); - RETURN(DIR_END_OFF); + 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); } - ret = default_llseek(file, offset, origin); - if (ret >= 0) { - struct ll_sb_info *sbi = ll_i2sbi(file->f_dentry->d_inode); - - if (ll_need_32bit_api(sbi)) { - if (file->f_pos >> 32) { - /* hash overflow, simple revert */ - file->f_pos = pos; - RETURN(-EOVERFLOW); - } else { - fd->fd_dir.lfd_pos = file->f_pos << 32; - } - } else { - fd->fd_dir.lfd_pos = file->f_pos; + 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; } - RETURN(ret); + GOTO(out, ret); + +out: + cfs_mutex_unlock(&inode->i_mutex); + return ret; } int ll_dir_open(struct inode *inode, struct file *file) @@ -1412,5 +1613,10 @@ struct file_operations ll_dir_operations = { .release = ll_dir_release, .read = generic_read_dir, .readdir = ll_readdir, - .ioctl = ll_dir_ioctl +#ifdef HAVE_UNLOCKED_IOCTL + .unlocked_ioctl = ll_dir_ioctl, +#else + .ioctl = ll_dir_ioctl, +#endif + .fsync = ll_fsync, };