X-Git-Url: https://git.whamcloud.com/?p=fs%2Flustre-release.git;a=blobdiff_plain;f=lustre%2Fllite%2Ffile.c;h=5a543daf80b72a1077f5d0bddfb67a5b2019ec31;hp=e081f33dfa3f3681fafe1ba7894821bb46f0a469;hb=eecf86131d099242d2e8c1f5d6be241ec1416c9a;hpb=72a84970e6d2a2d4b3a35f2ee058511be2fda82e;ds=sidebyside diff --git a/lustre/llite/file.c b/lustre/llite/file.c index e081f33..5a543da 100644 --- a/lustre/llite/file.c +++ b/lustre/llite/file.c @@ -42,9 +42,8 @@ #include #include #include -#ifdef HAVE_UIDGID_HEADER -# include -#endif +#include +#include #include #include @@ -163,6 +162,7 @@ static int ll_close_inode_openhandle(struct inode *inode, op_data->op_attr_blocks += ((struct inode *)data)->i_blocks; op_data->op_attr.ia_valid |= ATTR_SIZE; op_data->op_xvalid |= OP_XVALID_BLOCKS; + /* fallthrough */ case MDS_CLOSE_LAYOUT_SPLIT: case MDS_CLOSE_LAYOUT_SWAP: { struct split_param *sp = data; @@ -306,7 +306,7 @@ static int ll_md_close(struct inode *inode, struct file *file) .l_inodebits = { MDS_INODELOCK_OPEN }, }; __u64 flags = LDLM_FL_BLOCK_GRANTED | LDLM_FL_TEST_LOCK; - struct ll_file_data *fd = LUSTRE_FPRIVATE(file); + struct ll_file_data *fd = file->private_data; struct ll_inode_info *lli = ll_i2info(inode); struct lustre_handle lockh; enum ldlm_mode lockmode; @@ -353,12 +353,14 @@ static int ll_md_close(struct inode *inode, struct file *file) } mutex_unlock(&lli->lli_och_mutex); - if (!md_lock_match(ll_i2mdexp(inode), flags, ll_inode2fid(inode), + /* LU-4398: do not cache write open lock if the file has exec bit */ + if ((lockmode == LCK_CW && inode->i_mode & S_IXUGO) || + !md_lock_match(ll_i2mdexp(inode), flags, ll_inode2fid(inode), LDLM_IBITS, &policy, lockmode, &lockh)) rc = ll_md_real_close(inode, fd->fd_omode); out: - LUSTRE_FPRIVATE(file) = NULL; + file->private_data = NULL; ll_file_data_put(fd); RETURN(rc); @@ -371,19 +373,19 @@ out: */ int ll_file_release(struct inode *inode, struct file *file) { - struct ll_file_data *fd; - struct ll_sb_info *sbi = ll_i2sbi(inode); - struct ll_inode_info *lli = ll_i2info(inode); - int rc; - ENTRY; + struct ll_file_data *fd; + struct ll_sb_info *sbi = ll_i2sbi(inode); + struct ll_inode_info *lli = ll_i2info(inode); + ktime_t kstart = ktime_get(); + int rc; + + ENTRY; CDEBUG(D_VFSTRACE, "VFS Op:inode="DFID"(%p)\n", PFID(ll_inode2fid(inode)), inode); - if (inode->i_sb->s_root != file_dentry(file)) - ll_stats_ops_tally(sbi, LPROC_LL_RELEASE, 1); - fd = LUSTRE_FPRIVATE(file); - LASSERT(fd != NULL); + fd = file->private_data; + LASSERT(fd != NULL); /* The last ref on @file, maybe not the the owner pid of statahead, * because parent and child process can share the same file handle. */ @@ -391,9 +393,9 @@ int ll_file_release(struct inode *inode, struct file *file) ll_deauthorize_statahead(inode, fd); if (inode->i_sb->s_root == file_dentry(file)) { - LUSTRE_FPRIVATE(file) = NULL; + file->private_data = NULL; ll_file_data_put(fd); - RETURN(0); + GOTO(out, rc = 0); } pcc_file_release(inode, file); @@ -409,6 +411,10 @@ int ll_file_release(struct inode *inode, struct file *file) if (CFS_FAIL_TIMEOUT_MS(OBD_FAIL_PTLRPC_DUMP_LOG, cfs_fail_val)) libcfs_debug_dumplog(); +out: + if (!rc && inode->i_sb->s_root != file_dentry(file)) + ll_stats_ops_tally(sbi, LPROC_LL_RELEASE, + ktime_us_delta(ktime_get(), kstart)); RETURN(rc); } @@ -417,14 +423,14 @@ static inline int ll_dom_readpage(void *data, struct page *page) struct niobuf_local *lnb = data; void *kaddr; - kaddr = ll_kmap_atomic(page, KM_USER0); + kaddr = kmap_atomic(page); memcpy(kaddr, lnb->lnb_data, lnb->lnb_len); if (lnb->lnb_len < PAGE_SIZE) memset(kaddr + lnb->lnb_len, 0, PAGE_SIZE - lnb->lnb_len); flush_dcache_page(page); SetPageUptodate(page); - ll_kunmap_atomic(kaddr, KM_USER0); + kunmap_atomic(kaddr); unlock_page(page); return 0; @@ -448,8 +454,8 @@ void ll_dom_finish_open(struct inode *inode, struct ptlrpc_request *req, if (obj == NULL) RETURN_EXIT; - if (!req_capsule_has_field(&req->rq_pill, &RMF_NIOBUF_INLINE, - RCL_SERVER)) + if (!req_capsule_field_present(&req->rq_pill, &RMF_NIOBUF_INLINE, + RCL_SERVER)) RETURN_EXIT; rnb = req_capsule_server_get(&req->rq_pill, &RMF_NIOBUF_INLINE); @@ -461,7 +467,7 @@ void ll_dom_finish_open(struct inode *inode, struct ptlrpc_request *req, * client PAGE_SIZE to be used on that client, if server's PAGE_SIZE is * smaller then offset may be not aligned and that data is just ignored. */ - if (rnb->rnb_offset % PAGE_SIZE) + if (rnb->rnb_offset & ~PAGE_MASK) RETURN_EXIT; /* Server returns whole file or just file tail if it fills in reply @@ -481,9 +487,9 @@ void ll_dom_finish_open(struct inode *inode, struct ptlrpc_request *req, data = (char *)rnb + sizeof(*rnb); lnb.lnb_file_offset = rnb->rnb_offset; - start = lnb.lnb_file_offset / PAGE_SIZE; + start = lnb.lnb_file_offset >> PAGE_SHIFT; index = 0; - LASSERT(lnb.lnb_file_offset % PAGE_SIZE == 0); + LASSERT((lnb.lnb_file_offset & ~PAGE_MASK) == 0); lnb.lnb_page_offset = 0; do { lnb.lnb_data = data + (index << PAGE_SHIFT); @@ -646,7 +652,7 @@ static int ll_local_open(struct file *file, struct lookup_intent *it, struct inode *inode = file_inode(file); ENTRY; - LASSERT(!LUSTRE_FPRIVATE(file)); + LASSERT(!file->private_data); LASSERT(fd != NULL); @@ -658,7 +664,7 @@ static int ll_local_open(struct file *file, struct lookup_intent *it, RETURN(rc); } - LUSTRE_FPRIVATE(file) = fd; + file->private_data = fd; ll_readahead_init(inode, &fd->fd_ras); fd->fd_omode = it->it_flags & (FMODE_READ | FMODE_WRITE | FMODE_EXEC); @@ -690,6 +696,7 @@ int ll_file_open(struct inode *inode, struct file *file) struct obd_client_handle **och_p = NULL; __u64 *och_usecount = NULL; struct ll_file_data *fd; + ktime_t kstart = ktime_get(); int rc = 0; ENTRY; @@ -699,6 +706,12 @@ int ll_file_open(struct inode *inode, struct file *file) it = file->private_data; /* XXX: compat macro */ file->private_data = NULL; /* prevent ll_local_open assertion */ + if (S_ISREG(inode->i_mode)) { + rc = llcrypt_file_open(inode, file); + if (rc) + GOTO(out_nofiledata, rc); + } + fd = ll_file_data_get(); if (fd == NULL) GOTO(out_nofiledata, rc = -ENOMEM); @@ -708,18 +721,18 @@ int ll_file_open(struct inode *inode, struct file *file) ll_authorize_statahead(inode, fd); if (inode->i_sb->s_root == file_dentry(file)) { - LUSTRE_FPRIVATE(file) = fd; - RETURN(0); - } + file->private_data = fd; + RETURN(0); + } if (!it || !it->it_disposition) { - /* Convert f_flags into access mode. We cannot use file->f_mode, - * because everything but O_ACCMODE mask was stripped from - * there */ - if ((oit.it_flags + 1) & O_ACCMODE) - oit.it_flags++; - if (file->f_flags & O_TRUNC) - oit.it_flags |= FMODE_WRITE; + /* Convert f_flags into access mode. We cannot use file->f_mode, + * because everything but O_ACCMODE mask was stripped from + * there */ + if ((oit.it_flags + 1) & O_ACCMODE) + oit.it_flags++; + if (file->f_flags & O_TRUNC) + oit.it_flags |= FMODE_WRITE; /* kernel only call f_op->open in dentry_open. filp_open calls * dentry_open after call to open_namei that checks permissions. @@ -729,62 +742,65 @@ int ll_file_open(struct inode *inode, struct file *file) if (oit.it_flags & (FMODE_WRITE | FMODE_READ)) oit.it_flags |= MDS_OPEN_OWNEROVERRIDE; - /* We do not want O_EXCL here, presumably we opened the file - * already? XXX - NFS implications? */ - oit.it_flags &= ~O_EXCL; + /* We do not want O_EXCL here, presumably we opened the file + * already? XXX - NFS implications? */ + oit.it_flags &= ~O_EXCL; - /* bug20584, if "it_flags" contains O_CREAT, the file will be - * created if necessary, then "IT_CREAT" should be set to keep - * consistent with it */ - if (oit.it_flags & O_CREAT) - oit.it_op |= IT_CREAT; + /* bug20584, if "it_flags" contains O_CREAT, the file will be + * created if necessary, then "IT_CREAT" should be set to keep + * consistent with it */ + if (oit.it_flags & O_CREAT) + oit.it_op |= IT_CREAT; - it = &oit; - } + it = &oit; + } restart: - /* Let's see if we have file open on MDS already. */ - if (it->it_flags & FMODE_WRITE) { - och_p = &lli->lli_mds_write_och; - och_usecount = &lli->lli_open_fd_write_count; - } else if (it->it_flags & FMODE_EXEC) { - och_p = &lli->lli_mds_exec_och; - och_usecount = &lli->lli_open_fd_exec_count; - } else { - och_p = &lli->lli_mds_read_och; - och_usecount = &lli->lli_open_fd_read_count; - } + /* Let's see if we have file open on MDS already. */ + if (it->it_flags & FMODE_WRITE) { + och_p = &lli->lli_mds_write_och; + och_usecount = &lli->lli_open_fd_write_count; + } else if (it->it_flags & FMODE_EXEC) { + och_p = &lli->lli_mds_exec_och; + och_usecount = &lli->lli_open_fd_exec_count; + } else { + och_p = &lli->lli_mds_read_och; + och_usecount = &lli->lli_open_fd_read_count; + } mutex_lock(&lli->lli_och_mutex); - if (*och_p) { /* Open handle is present */ - if (it_disposition(it, DISP_OPEN_OPEN)) { - /* Well, there's extra open request that we do not need, - let's close it somehow. This will decref request. */ - rc = it_open_error(DISP_OPEN_OPEN, it); - if (rc) { + if (*och_p) { /* Open handle is present */ + if (it_disposition(it, DISP_OPEN_OPEN)) { + /* Well, there's extra open request that we do not need, + * let's close it somehow. This will decref request. */ + rc = it_open_error(DISP_OPEN_OPEN, it); + if (rc) { mutex_unlock(&lli->lli_och_mutex); - GOTO(out_openerr, rc); - } + GOTO(out_openerr, rc); + } ll_release_openhandle(file_dentry(file), it); - } - (*och_usecount)++; + } + (*och_usecount)++; - rc = ll_local_open(file, it, fd, NULL); - if (rc) { - (*och_usecount)--; + rc = ll_local_open(file, it, fd, NULL); + if (rc) { + (*och_usecount)--; mutex_unlock(&lli->lli_och_mutex); - GOTO(out_openerr, rc); - } - } else { - LASSERT(*och_usecount == 0); + GOTO(out_openerr, rc); + } + } else { + LASSERT(*och_usecount == 0); if (!it->it_disposition) { - struct ll_dentry_data *ldd = ll_d2d(file->f_path.dentry); - /* We cannot just request lock handle now, new ELC code - means that one of other OPEN locks for this file - could be cancelled, and since blocking ast handler - would attempt to grab och_mutex as well, that would - result in a deadlock */ + struct dentry *dentry = file_dentry(file); + struct ll_dentry_data *ldd; + + /* We cannot just request lock handle now, new ELC code + * means that one of other OPEN locks for this file + * could be cancelled, and since blocking ast handler + * would attempt to grab och_mutex as well, that would + * result in a deadlock + */ mutex_unlock(&lli->lli_och_mutex); /* * Normally called under two situations: @@ -801,34 +817,37 @@ restart: * lookup path only, since ll_iget_for_nfs always calls * ll_d_init(). */ + ldd = ll_d2d(dentry); if (ldd && ldd->lld_nfs_dentry) { ldd->lld_nfs_dentry = 0; - it->it_flags |= MDS_OPEN_LOCK; + if (!filename_is_volatile(dentry->d_name.name, + dentry->d_name.len, + NULL)) + it->it_flags |= MDS_OPEN_LOCK; } - /* + /* * Always specify MDS_OPEN_BY_FID because we don't want * to get file with different fid. */ it->it_flags |= MDS_OPEN_BY_FID; - rc = ll_intent_file_open(file_dentry(file), NULL, 0, - it); - if (rc) - GOTO(out_openerr, rc); + rc = ll_intent_file_open(dentry, NULL, 0, it); + if (rc) + GOTO(out_openerr, rc); - goto restart; - } - OBD_ALLOC(*och_p, sizeof (struct obd_client_handle)); - if (!*och_p) - GOTO(out_och_free, rc = -ENOMEM); + goto restart; + } + OBD_ALLOC(*och_p, sizeof(struct obd_client_handle)); + if (!*och_p) + GOTO(out_och_free, rc = -ENOMEM); - (*och_usecount)++; + (*och_usecount)++; - /* md_intent_lock() didn't get a request ref if there was an - * open error, so don't do cleanup on the request here - * (bug 3430) */ - /* XXX (green): Should not we bail out on any error here, not - * just open error? */ + /* md_intent_lock() didn't get a request ref if there was an + * open error, so don't do cleanup on the request here + * (bug 3430) */ + /* XXX (green): Should not we bail out on any error here, not + * just open error? */ rc = it_open_error(DISP_OPEN_OPEN, it); if (rc != 0) GOTO(out_och_free, rc); @@ -847,24 +866,27 @@ restart: GOTO(out_och_free, rc); mutex_unlock(&lli->lli_och_mutex); - fd = NULL; - /* Must do this outside lli_och_mutex lock to prevent deadlock where - different kind of OPEN lock for this same inode gets cancelled - by ldlm_cancel_lru */ - if (!S_ISREG(inode->i_mode)) - GOTO(out_och_free, rc); + /* lockless for direct IO so that it can do IO in parallel */ + if (file->f_flags & O_DIRECT) + fd->fd_flags |= LL_FILE_LOCKLESS_IO; + fd = NULL; + /* Must do this outside lli_och_mutex lock to prevent deadlock where + different kind of OPEN lock for this same inode gets cancelled + by ldlm_cancel_lru */ + if (!S_ISREG(inode->i_mode)) + GOTO(out_och_free, rc); cl_lov_delay_create_clear(&file->f_flags); GOTO(out_och_free, rc); out_och_free: - if (rc) { - if (och_p && *och_p) { - OBD_FREE(*och_p, sizeof (struct obd_client_handle)); - *och_p = NULL; /* OBD_FREE writes some magic there */ - (*och_usecount)--; - } + if (rc) { + if (och_p && *och_p) { + OBD_FREE(*och_p, sizeof(struct obd_client_handle)); + *och_p = NULL; /* OBD_FREE writes some magic there */ + (*och_usecount)--; + } mutex_unlock(&lli->lli_och_mutex); out_openerr: @@ -873,9 +895,10 @@ out_openerr: if (fd != NULL) ll_file_data_put(fd); - } else { - ll_stats_ops_tally(ll_i2sbi(inode), LPROC_LL_OPEN, 1); - } + } else { + ll_stats_ops_tally(ll_i2sbi(inode), LPROC_LL_OPEN, + ktime_us_delta(ktime_get(), kstart)); + } out_nofiledata: if (it && it_disposition(it, DISP_ENQ_OPEN_REF)) { @@ -883,7 +906,7 @@ out_nofiledata: it_clear_disposition(it, DISP_ENQ_OPEN_REF); } - return rc; + return rc; } static int ll_md_blocking_lease_ast(struct ldlm_lock *lock, @@ -918,7 +941,7 @@ static int ll_lease_och_acquire(struct inode *inode, struct file *file, struct lustre_handle *old_open_handle) { struct ll_inode_info *lli = ll_i2info(inode); - struct ll_file_data *fd = LUSTRE_FPRIVATE(file); + struct ll_file_data *fd = file->private_data; struct obd_client_handle **och_p; __u64 *och_usecount; int rc = 0; @@ -962,7 +985,7 @@ out_unlock: static int ll_lease_och_release(struct inode *inode, struct file *file) { struct ll_inode_info *lli = ll_i2info(inode); - struct ll_file_data *fd = LUSTRE_FPRIVATE(file); + struct ll_file_data *fd = file->private_data; struct obd_client_handle **och_p; struct obd_client_handle *old_och = NULL; __u64 *och_usecount; @@ -1063,7 +1086,9 @@ ll_lease_open(struct inode *inode, struct file *file, fmode_t fmode, GOTO(out_release_it, rc); LASSERT(it_disposition(&it, DISP_ENQ_OPEN_REF)); - ll_och_fill(sbi->ll_md_exp, &it, och); + rc = ll_och_fill(sbi->ll_md_exp, &it, och); + if (rc) + GOTO(out_release_it, rc); if (!it_disposition(&it, DISP_OPEN_LEASE)) /* old server? */ GOTO(out_close, rc = -EOPNOTSUPP); @@ -1278,11 +1303,10 @@ int ll_merge_attr(const struct lu_env *env, struct inode *inode) * POSIX. Solving this problem needs to send an RPC to MDT for each * read, this will hurt performance. */ - if (inode->i_atime.tv_sec < lli->lli_atime || - lli->lli_update_atime) { + if (ll_file_test_and_clear_flag(lli, LLIF_UPDATE_ATIME) || + inode->i_atime.tv_sec < lli->lli_atime) inode->i_atime.tv_sec = lli->lli_atime; - lli->lli_update_atime = 0; - } + inode->i_mtime.tv_sec = lli->lli_mtime; inode->i_ctime.tv_sec = lli->lli_ctime; @@ -1333,7 +1357,7 @@ out_size_unlock: */ void ll_io_set_mirror(struct cl_io *io, const struct file *file) { - struct ll_file_data *fd = LUSTRE_FPRIVATE(file); + struct ll_file_data *fd = file->private_data; /* clear layout version for generic(non-resync) I/O in case it carries * stale layout version due to I/O restart */ @@ -1378,10 +1402,11 @@ static bool file_is_noatime(const struct file *file) return false; } -void ll_io_init(struct cl_io *io, struct file *file, enum cl_io_type iot) +void ll_io_init(struct cl_io *io, struct file *file, enum cl_io_type iot, + struct vvp_io_args *args) { struct inode *inode = file_inode(file); - struct ll_file_data *fd = LUSTRE_FPRIVATE(file); + struct ll_file_data *fd = file->private_data; io->u.ci_rw.crw_nonblock = file->f_flags & O_NONBLOCK; io->ci_lock_no_expand = fd->ll_lock_no_expand; @@ -1391,7 +1416,13 @@ void ll_io_init(struct cl_io *io, struct file *file, enum cl_io_type iot) io->u.ci_wr.wr_sync = !!(file->f_flags & O_SYNC || file->f_flags & O_DIRECT || IS_SYNC(inode)); +#ifdef HAVE_GENERIC_WRITE_SYNC_2ARGS + io->u.ci_wr.wr_sync |= !!(args && + args->via_io_subtype == IO_NORMAL && + args->u.normal.via_iocb->ki_flags & IOCB_DSYNC); +#endif } + io->ci_obj = ll_i2info(inode)->lli_clob; io->ci_lockreq = CILR_MAYBE; if (ll_file_nolock(file)) { @@ -1449,13 +1480,13 @@ ll_file_io_generic(const struct lu_env *env, struct vvp_io_args *args, struct vvp_io *vio = vvp_env_io(env); struct inode *inode = file_inode(file); struct ll_inode_info *lli = ll_i2info(inode); - struct ll_file_data *fd = LUSTRE_FPRIVATE(file); + struct ll_file_data *fd = file->private_data; struct range_lock range; struct cl_io *io; ssize_t result = 0; int rc = 0; unsigned retried = 0; - bool restarted = false; + unsigned ignore_lockless = 0; ENTRY; @@ -1465,7 +1496,8 @@ ll_file_io_generic(const struct lu_env *env, struct vvp_io_args *args, restart: io = vvp_env_thread_io(env); - ll_io_init(io, file, iot); + ll_io_init(io, file, iot, args); + io->ci_ignore_lockless = ignore_lockless; io->ci_ndelay_tried = retried; if (cl_io_rw_init(env, io, iot, *ppos, count) == 0) { @@ -1476,7 +1508,7 @@ restart: else range_lock_init(&range, *ppos, *ppos + count - 1); - vio->vui_fd = LUSTRE_FPRIVATE(file); + vio->vui_fd = file->private_data; vio->vui_io_subtype = args->via_io_subtype; switch (vio->vui_io_subtype) { @@ -1538,7 +1570,8 @@ out: file->f_path.dentry->d_name.name, iot, rc, result, io->ci_need_restart); - if ((rc == 0 || rc == -ENODATA) && count > 0 && io->ci_need_restart) { + if ((rc == 0 || rc == -ENODATA || rc == -ENOLCK) && + count > 0 && io->ci_need_restart) { CDEBUG(D_VFSTRACE, "%s: restart %s from %lld, count: %zu, ret: %zd, rc: %d\n", file_dentry(file)->d_name.name, @@ -1546,7 +1579,7 @@ out: *ppos, count, result, rc); /* preserve the tried count for FLR */ retried = io->ci_ndelay_tried; - restarted = true; + ignore_lockless = io->ci_ignore_lockless; goto restart; } @@ -1636,7 +1669,7 @@ ll_do_fast_read(struct kiocb *iocb, struct iov_iter *iter) if (result > 0) { ll_heat_add(file_inode(iocb->ki_filp), CIT_READ, result); ll_stats_ops_tally(ll_i2sbi(file_inode(iocb->ki_filp)), - LPROC_LL_READ_BYTES, result); + LPROC_LL_READ_BYTES, result); } return result; @@ -1649,9 +1682,11 @@ static ssize_t ll_file_read_iter(struct kiocb *iocb, struct iov_iter *to) { struct lu_env *env; struct vvp_io_args *args; + struct file *file = iocb->ki_filp; ssize_t result; ssize_t rc2; __u16 refcheck; + ktime_t kstart = ktime_get(); bool cached; if (!iov_iter_count(to)) @@ -1670,9 +1705,9 @@ static ssize_t ll_file_read_iter(struct kiocb *iocb, struct iov_iter *to) */ result = pcc_file_read_iter(iocb, to, &cached); if (cached) - return result; + GOTO(out, result); - ll_ras_enter(iocb->ki_filp); + ll_ras_enter(file, iocb->ki_pos, iov_iter_count(to)); result = ll_do_fast_read(iocb, to); if (result < 0 || iov_iter_count(to) == 0) @@ -1686,7 +1721,7 @@ static ssize_t ll_file_read_iter(struct kiocb *iocb, struct iov_iter *to) args->u.normal.via_iter = to; args->u.normal.via_iocb = iocb; - rc2 = ll_file_io_generic(env, args, iocb->ki_filp, CIT_READ, + rc2 = ll_file_io_generic(env, args, file, CIT_READ, &iocb->ki_pos, iov_iter_count(to)); if (rc2 > 0) result += rc2; @@ -1695,6 +1730,14 @@ static ssize_t ll_file_read_iter(struct kiocb *iocb, struct iov_iter *to) cl_env_put(env, &refcheck); out: + if (result > 0) { + ll_rw_stats_tally(ll_i2sbi(file_inode(file)), current->pid, + file->private_data, iocb->ki_pos, result, + READ); + ll_stats_ops_tally(ll_i2sbi(file_inode(file)), LPROC_LL_READ, + ktime_us_delta(ktime_get(), kstart)); + } + return result; } @@ -1765,8 +1808,10 @@ static ssize_t ll_file_write_iter(struct kiocb *iocb, struct iov_iter *from) struct vvp_io_args *args; struct lu_env *env; ssize_t rc_tiny = 0, rc_normal; + struct file *file = iocb->ki_filp; __u16 refcheck; bool cached; + ktime_t kstart = ktime_get(); int result; ENTRY; @@ -1786,15 +1831,15 @@ static ssize_t ll_file_write_iter(struct kiocb *iocb, struct iov_iter *from) */ result = pcc_file_write_iter(iocb, from, &cached); if (cached && result != -ENOSPC && result != -EDQUOT) - return result; + GOTO(out, rc_normal = result); /* NB: we can't do direct IO for tiny writes because they use the page * cache, we can't do sync writes because tiny writes can't flush * pages, and we can't do append writes because we can't guarantee the * required DLM locks are held to protect file size. */ - if (ll_sbi_has_tiny_write(ll_i2sbi(file_inode(iocb->ki_filp))) && - !(iocb->ki_filp->f_flags & (O_DIRECT | O_SYNC | O_APPEND))) + if (ll_sbi_has_tiny_write(ll_i2sbi(file_inode(file))) && + !(file->f_flags & (O_DIRECT | O_SYNC | O_APPEND))) rc_tiny = ll_do_tiny_write(iocb, from); /* In case of error, go on and try normal write - Only stop if tiny @@ -1811,8 +1856,8 @@ static ssize_t ll_file_write_iter(struct kiocb *iocb, struct iov_iter *from) args->u.normal.via_iter = from; args->u.normal.via_iocb = iocb; - rc_normal = ll_file_io_generic(env, args, iocb->ki_filp, CIT_WRITE, - &iocb->ki_pos, iov_iter_count(from)); + rc_normal = ll_file_io_generic(env, args, file, CIT_WRITE, + &iocb->ki_pos, iov_iter_count(from)); /* On success, combine bytes written. */ if (rc_tiny >= 0 && rc_normal > 0) @@ -1825,6 +1870,14 @@ static ssize_t ll_file_write_iter(struct kiocb *iocb, struct iov_iter *from) cl_env_put(env, &refcheck); out: + if (rc_normal > 0) { + ll_rw_stats_tally(ll_i2sbi(file_inode(file)), current->pid, + file->private_data, iocb->ki_pos, + rc_normal, WRITE); + ll_stats_ops_tally(ll_i2sbi(file_inode(file)), LPROC_LL_WRITE, + ktime_us_delta(ktime_get(), kstart)); + } + RETURN(rc_normal); } @@ -1833,7 +1886,8 @@ out: * XXX: exact copy from kernel code (__generic_file_aio_write_nolock) */ static int ll_file_get_iov_count(const struct iovec *iov, - unsigned long *nr_segs, size_t *count) + unsigned long *nr_segs, size_t *count, + int access_flags) { size_t cnt = 0; unsigned long seg; @@ -1848,7 +1902,7 @@ static int ll_file_get_iov_count(const struct iovec *iov, cnt += iv->iov_len; if (unlikely((ssize_t)(cnt|iv->iov_len) < 0)) return -EINVAL; - if (access_ok(VERIFY_READ, iv->iov_base, iv->iov_len)) + if (access_ok(access_flags, iv->iov_base, iv->iov_len)) continue; if (seg == 0) return -EFAULT; @@ -1868,7 +1922,7 @@ static ssize_t ll_file_aio_read(struct kiocb *iocb, const struct iovec *iov, ssize_t result; ENTRY; - result = ll_file_get_iov_count(iov, &nr_segs, &iov_count); + result = ll_file_get_iov_count(iov, &nr_segs, &iov_count, VERIFY_READ); if (result) RETURN(result); @@ -1924,7 +1978,7 @@ static ssize_t ll_file_aio_write(struct kiocb *iocb, const struct iovec *iov, ssize_t result; ENTRY; - result = ll_file_get_iov_count(iov, &nr_segs, &iov_count); + result = ll_file_get_iov_count(iov, &nr_segs, &iov_count, VERIFY_WRITE); if (result) RETURN(result); @@ -1983,26 +2037,31 @@ static ssize_t ll_file_splice_read(struct file *in_file, loff_t *ppos, __u16 refcheck; bool cached; - ENTRY; + ENTRY; result = pcc_file_splice_read(in_file, ppos, pipe, count, flags, &cached); if (cached) RETURN(result); - ll_ras_enter(in_file); + ll_ras_enter(in_file, *ppos, count); env = cl_env_get(&refcheck); - if (IS_ERR(env)) - RETURN(PTR_ERR(env)); + if (IS_ERR(env)) + RETURN(PTR_ERR(env)); args = ll_env_args(env, IO_SPLICE); - args->u.splice.via_pipe = pipe; - args->u.splice.via_flags = flags; + args->u.splice.via_pipe = pipe; + args->u.splice.via_flags = flags; + + result = ll_file_io_generic(env, args, in_file, CIT_READ, ppos, count); + cl_env_put(env, &refcheck); - result = ll_file_io_generic(env, args, in_file, CIT_READ, ppos, count); - cl_env_put(env, &refcheck); - RETURN(result); + if (result > 0) + ll_rw_stats_tally(ll_i2sbi(file_inode(in_file)), current->pid, + in_file->private_data, *ppos, result, + READ); + RETURN(result); } int ll_lov_setstripe_ea_info(struct inode *inode, struct dentry *dentry, @@ -2015,6 +2074,12 @@ int ll_lov_setstripe_ea_info(struct inode *inode, struct dentry *dentry, int rc; ENTRY; + if ((__swab32(lum->lmm_magic) & le32_to_cpu(LOV_MAGIC_MASK)) == + le32_to_cpu(LOV_MAGIC_MAGIC)) { + /* this code will only exist for big-endian systems */ + lustre_swab_lov_user_md(lum, 0); + } + ll_inode_size_lock(inode); rc = ll_intent_file_open(dentry, lum, lum_size, &oit); if (rc < 0) @@ -2078,13 +2143,14 @@ int ll_lov_getstripe_ea_info(struct inode *inode, const char *filename, lmm->lmm_magic != cpu_to_le32(LOV_MAGIC_FOREIGN)) GOTO(out, rc = -EPROTO); - /* - * This is coming from the MDS, so is probably in - * little endian. We convert it to host endian before - * passing it to userspace. - */ - if (LOV_MAGIC != cpu_to_le32(LOV_MAGIC)) { - int stripe_count; + /* + * This is coming from the MDS, so is probably in + * little endian. We convert it to host endian before + * passing it to userspace. + */ + if ((lmm->lmm_magic & __swab32(LOV_MAGIC_MAGIC)) == + __swab32(LOV_MAGIC_MAGIC)) { + int stripe_count = 0; if (lmm->lmm_magic == cpu_to_le32(LOV_MAGIC_V1) || lmm->lmm_magic == cpu_to_le32(LOV_MAGIC_V3)) { @@ -2094,36 +2160,19 @@ int ll_lov_getstripe_ea_info(struct inode *inode, const char *filename, stripe_count = 0; } - /* if function called for directory - we should - * avoid swab not existent lsm objects */ - if (lmm->lmm_magic == cpu_to_le32(LOV_MAGIC_V1)) { - lustre_swab_lov_user_md_v1( - (struct lov_user_md_v1 *)lmm); - if (S_ISREG(body->mbo_mode)) - lustre_swab_lov_user_md_objects( - ((struct lov_user_md_v1 *)lmm)->lmm_objects, - stripe_count); - } else if (lmm->lmm_magic == cpu_to_le32(LOV_MAGIC_V3)) { - lustre_swab_lov_user_md_v3( - (struct lov_user_md_v3 *)lmm); - if (S_ISREG(body->mbo_mode)) - lustre_swab_lov_user_md_objects( - ((struct lov_user_md_v3 *)lmm)->lmm_objects, - stripe_count); - } else if (lmm->lmm_magic == - cpu_to_le32(LOV_MAGIC_COMP_V1)) { - lustre_swab_lov_comp_md_v1( - (struct lov_comp_md_v1 *)lmm); - } else if (lmm->lmm_magic == - cpu_to_le32(LOV_MAGIC_FOREIGN)) { - struct lov_foreign_md *lfm; - - lfm = (struct lov_foreign_md *)lmm; - __swab32s(&lfm->lfm_magic); - __swab32s(&lfm->lfm_length); - __swab32s(&lfm->lfm_type); - __swab32s(&lfm->lfm_flags); - } + lustre_swab_lov_user_md((struct lov_user_md *)lmm, 0); + + /* if function called for directory - we should + * avoid swab not existent lsm objects */ + if (lmm->lmm_magic == LOV_MAGIC_V1 && S_ISREG(body->mbo_mode)) + lustre_swab_lov_user_md_objects( + ((struct lov_user_md_v1 *)lmm)->lmm_objects, + stripe_count); + else if (lmm->lmm_magic == LOV_MAGIC_V3 && + S_ISREG(body->mbo_mode)) + lustre_swab_lov_user_md_objects( + ((struct lov_user_md_v3 *)lmm)->lmm_objects, + stripe_count); } out: @@ -2210,16 +2259,17 @@ static int ll_lov_setstripe(struct inode *inode, struct file *file, cl_lov_delay_create_clear(&file->f_flags); out: - OBD_FREE(klum, lum_size); + OBD_FREE_LARGE(klum, lum_size); RETURN(rc); } + static int ll_get_grouplock(struct inode *inode, struct file *file, unsigned long arg) { struct ll_inode_info *lli = ll_i2info(inode); struct cl_object *obj = lli->lli_clob; - struct ll_file_data *fd = LUSTRE_FPRIVATE(file); + struct ll_file_data *fd = file->private_data; struct ll_grouplock grouplock; int rc; ENTRY; @@ -2229,18 +2279,28 @@ ll_get_grouplock(struct inode *inode, struct file *file, unsigned long arg) RETURN(-EINVAL); } - if (ll_file_nolock(file)) - RETURN(-EOPNOTSUPP); + if (ll_file_nolock(file)) + RETURN(-EOPNOTSUPP); +retry: + if (file->f_flags & O_NONBLOCK) { + if (!mutex_trylock(&lli->lli_group_mutex)) + RETURN(-EAGAIN); + } else + mutex_lock(&lli->lli_group_mutex); - spin_lock(&lli->lli_lock); if (fd->fd_flags & LL_FILE_GROUP_LOCKED) { CWARN("group lock already existed with gid %lu\n", fd->fd_grouplock.lg_gid); - spin_unlock(&lli->lli_lock); - RETURN(-EINVAL); + GOTO(out, rc = -EINVAL); + } + if (arg != lli->lli_group_gid && lli->lli_group_users != 0) { + if (file->f_flags & O_NONBLOCK) + GOTO(out, rc = -EAGAIN); + mutex_unlock(&lli->lli_group_mutex); + wait_var_event(&lli->lli_group_users, !lli->lli_group_users); + GOTO(retry, rc = 0); } LASSERT(fd->fd_grouplock.lg_lock == NULL); - spin_unlock(&lli->lli_lock); /** * XXX: group lock needs to protect all OST objects while PFL @@ -2260,7 +2320,7 @@ ll_get_grouplock(struct inode *inode, struct file *file, unsigned long arg) env = cl_env_get(&refcheck); if (IS_ERR(env)) - RETURN(PTR_ERR(env)); + GOTO(out, rc = PTR_ERR(env)); rc = cl_object_layout_get(env, obj, &cl); if (!rc && cl.cl_is_composite) @@ -2269,62 +2329,68 @@ ll_get_grouplock(struct inode *inode, struct file *file, unsigned long arg) cl_env_put(env, &refcheck); if (rc) - RETURN(rc); + GOTO(out, rc); } rc = cl_get_grouplock(ll_i2info(inode)->lli_clob, arg, (file->f_flags & O_NONBLOCK), &grouplock); - if (rc) - RETURN(rc); - spin_lock(&lli->lli_lock); - if (fd->fd_flags & LL_FILE_GROUP_LOCKED) { - spin_unlock(&lli->lli_lock); - CERROR("another thread just won the race\n"); - cl_put_grouplock(&grouplock); - RETURN(-EINVAL); - } + if (rc) + GOTO(out, rc); fd->fd_flags |= LL_FILE_GROUP_LOCKED; fd->fd_grouplock = grouplock; - spin_unlock(&lli->lli_lock); + if (lli->lli_group_users == 0) + lli->lli_group_gid = grouplock.lg_gid; + lli->lli_group_users++; CDEBUG(D_INFO, "group lock %lu obtained\n", arg); - RETURN(0); +out: + mutex_unlock(&lli->lli_group_mutex); + + RETURN(rc); } static int ll_put_grouplock(struct inode *inode, struct file *file, unsigned long arg) { struct ll_inode_info *lli = ll_i2info(inode); - struct ll_file_data *fd = LUSTRE_FPRIVATE(file); + struct ll_file_data *fd = file->private_data; struct ll_grouplock grouplock; + int rc; ENTRY; - spin_lock(&lli->lli_lock); + mutex_lock(&lli->lli_group_mutex); if (!(fd->fd_flags & LL_FILE_GROUP_LOCKED)) { - spin_unlock(&lli->lli_lock); - CWARN("no group lock held\n"); - RETURN(-EINVAL); - } + CWARN("no group lock held\n"); + GOTO(out, rc = -EINVAL); + } LASSERT(fd->fd_grouplock.lg_lock != NULL); if (fd->fd_grouplock.lg_gid != arg) { CWARN("group lock %lu doesn't match current id %lu\n", arg, fd->fd_grouplock.lg_gid); - spin_unlock(&lli->lli_lock); - RETURN(-EINVAL); + GOTO(out, rc = -EINVAL); } grouplock = fd->fd_grouplock; memset(&fd->fd_grouplock, 0, sizeof(fd->fd_grouplock)); fd->fd_flags &= ~LL_FILE_GROUP_LOCKED; - spin_unlock(&lli->lli_lock); cl_put_grouplock(&grouplock); + + lli->lli_group_users--; + if (lli->lli_group_users == 0) { + lli->lli_group_gid = 0; + wake_up_var(&lli->lli_group_users); + } CDEBUG(D_INFO, "group lock %lu released\n", arg); - RETURN(0); + GOTO(out, rc = 0); +out: + mutex_unlock(&lli->lli_group_mutex); + + RETURN(rc); } /** @@ -2359,7 +2425,9 @@ int ll_release_openhandle(struct dentry *dentry, struct lookup_intent *it) if (!och) GOTO(out, rc = -ENOMEM); - ll_och_fill(ll_i2sbi(inode)->ll_md_exp, it, och); + rc = ll_och_fill(ll_i2sbi(inode)->ll_md_exp, it, och); + if (rc) + GOTO(out, rc); rc = ll_close_inode_openhandle(inode, och, 0, NULL); out: @@ -2898,9 +2966,9 @@ int ll_file_lock_ahead(struct file *file, struct llapi_lu_ladvise *ladvise) ENTRY; - CDEBUG(D_VFSTRACE, "Lock request: file=%.*s, inode=%p, mode=%s " - "start=%llu, end=%llu\n", dentry->d_name.len, - dentry->d_name.name, dentry->d_inode, + CDEBUG(D_VFSTRACE, + "Lock request: file=%pd, inode=%p, mode=%s start=%llu, end=%llu\n", + dentry, dentry->d_inode, user_lockname[ladvise->lla_lockahead_mode], (__u64) start, (__u64) end); @@ -2976,8 +3044,8 @@ static int ll_ladvise_sanity(struct inode *inode, if (advice > LU_LADVISE_MAX || advice == LU_LADVISE_INVALID) { rc = -EINVAL; - CDEBUG(D_VFSTRACE, "%s: advice with value '%d' not recognized," - "last supported advice is %s (value '%d'): rc = %d\n", + CDEBUG(D_VFSTRACE, + "%s: advice with value '%d' not recognized, last supported advice is %s (value '%d'): rc = %d\n", sbi->ll_fsname, advice, ladvise_names[LU_LADVISE_MAX-1], LU_LADVISE_MAX-1, rc); GOTO(out, rc); @@ -3005,6 +3073,7 @@ static int ll_ladvise_sanity(struct inode *inode, ladvise_names[advice], rc); GOTO(out, rc); } + /* fallthrough */ case LU_LADVISE_WILLREAD: case LU_LADVISE_DONTNEED: default: @@ -3084,7 +3153,7 @@ static int ll_ladvise(struct inode *inode, struct file *file, __u64 flags, static int ll_lock_noexpand(struct file *file, int flags) { - struct ll_file_data *fd = LUSTRE_FPRIVATE(file); + struct ll_file_data *fd = file->private_data; fd->ll_lock_no_expand = !(flags & LF_UNSET); @@ -3178,6 +3247,11 @@ int ll_ioctl_fssetxattr(struct inode *inode, unsigned int cmd, if (obj == NULL) GOTO(out_fsxattr, rc); + /* Avoiding OST RPC if this is only project ioctl */ + if (fsxattr.fsx_xflags == 0 || + fsxattr.fsx_xflags == FS_XFLAG_PROJINHERIT) + GOTO(out_fsxattr, rc); + OBD_ALLOC_PTR(attr); if (attr == NULL) GOTO(out_fsxattr, rc = -ENOMEM); @@ -3194,7 +3268,7 @@ static long ll_file_unlock_lease(struct file *file, struct ll_ioc_lease *ioc, unsigned long arg) { struct inode *inode = file_inode(file); - struct ll_file_data *fd = LUSTRE_FPRIVATE(file); + struct ll_file_data *fd = file->private_data; struct ll_inode_info *lli = ll_i2info(inode); struct obd_client_handle *och = NULL; struct split_param sp; @@ -3357,7 +3431,7 @@ static long ll_file_set_lease(struct file *file, struct ll_ioc_lease *ioc, { struct inode *inode = file_inode(file); struct ll_inode_info *lli = ll_i2info(inode); - struct ll_file_data *fd = LUSTRE_FPRIVATE(file); + struct ll_file_data *fd = file->private_data; struct obd_client_handle *och = NULL; __u64 open_flags = 0; bool lease_broken; @@ -3458,7 +3532,7 @@ static long ll_file_ioctl(struct file *file, unsigned int cmd, unsigned long arg) { struct inode *inode = file_inode(file); - struct ll_file_data *fd = LUSTRE_FPRIVATE(file); + struct ll_file_data *fd = file->private_data; int flags, rc; ENTRY; @@ -3831,9 +3905,9 @@ out_ladvise: fd->fd_designated_mirror = (__u32)arg; RETURN(0); } - case LL_IOC_FSGETXATTR: + case FS_IOC_FSGETXATTR: RETURN(ll_ioctl_fsgetxattr(inode, cmd, arg)); - case LL_IOC_FSSETXATTR: + case FS_IOC_FSSETXATTR: RETURN(ll_ioctl_fssetxattr(inode, cmd, arg)); case BLKSSZGET: RETURN(put_user(PAGE_SIZE, (int __user *)arg)); @@ -3920,77 +3994,11 @@ out_state: } } -#ifndef HAVE_FILE_LLSEEK_SIZE -static inline loff_t -llseek_execute(struct file *file, loff_t offset, loff_t maxsize) -{ - if (offset < 0 && !(file->f_mode & FMODE_UNSIGNED_OFFSET)) - return -EINVAL; - if (offset > maxsize) - return -EINVAL; - - if (offset != file->f_pos) { - file->f_pos = offset; - file->f_version = 0; - } - return offset; -} - -static loff_t -generic_file_llseek_size(struct file *file, loff_t offset, int origin, - loff_t maxsize, loff_t eof) -{ - struct inode *inode = file_inode(file); - - switch (origin) { - case SEEK_END: - offset += eof; - break; - case SEEK_CUR: - /* - * Here we special-case the lseek(fd, 0, SEEK_CUR) - * position-querying operation. Avoid rewriting the "same" - * f_pos value back to the file because a concurrent read(), - * write() or lseek() might have altered it - */ - if (offset == 0) - return file->f_pos; - /* - * f_lock protects against read/modify/write race with other - * SEEK_CURs. Note that parallel writes and reads behave - * like SEEK_SET. - */ - inode_lock(inode); - offset = llseek_execute(file, file->f_pos + offset, maxsize); - inode_unlock(inode); - return offset; - case SEEK_DATA: - /* - * In the generic case the entire file is data, so as long as - * offset isn't at the end of the file then the offset is data. - */ - if (offset >= eof) - return -ENXIO; - break; - case SEEK_HOLE: - /* - * There is a virtual hole at the end of the file, so as long as - * offset isn't i_size or larger, return i_size. - */ - if (offset >= eof) - return -ENXIO; - offset = eof; - break; - } - - return llseek_execute(file, offset, maxsize); -} -#endif - static loff_t ll_file_seek(struct file *file, loff_t offset, int origin) { struct inode *inode = file_inode(file); loff_t retval, eof = 0; + ktime_t kstart = ktime_get(); ENTRY; retval = offset + ((origin == SEEK_END) ? i_size_read(inode) : @@ -3998,7 +4006,6 @@ static loff_t ll_file_seek(struct file *file, loff_t offset, int origin) CDEBUG(D_VFSTRACE, "VFS Op:inode="DFID"(%p), to=%llu=%#llx(%d)\n", PFID(ll_inode2fid(inode)), inode, retval, retval, origin); - ll_stats_ops_tally(ll_i2sbi(inode), LPROC_LL_LLSEEK, 1); if (origin == SEEK_END || origin == SEEK_HOLE || origin == SEEK_DATA) { retval = ll_glimpse_size(inode); @@ -4007,8 +4014,11 @@ static loff_t ll_file_seek(struct file *file, loff_t offset, int origin) eof = i_size_read(inode); } - retval = ll_generic_file_llseek_size(file, offset, origin, + retval = generic_file_llseek_size(file, offset, origin, ll_file_maxbytes(inode), eof); + if (retval >= 0) + ll_stats_ops_tally(ll_i2sbi(inode), LPROC_LL_LLSEEK, + ktime_us_delta(ktime_get(), kstart)); RETURN(retval); } @@ -4016,7 +4026,7 @@ static int ll_flush(struct file *file, fl_owner_t id) { struct inode *inode = file_inode(file); struct ll_inode_info *lli = ll_i2info(inode); - struct ll_file_data *fd = LUSTRE_FPRIVATE(file); + struct ll_file_data *fd = file->private_data; int rc, err; LASSERT(!S_ISDIR(inode->i_mode)); @@ -4098,16 +4108,15 @@ int ll_fsync(struct file *file, loff_t start, loff_t end, int datasync) struct inode *inode = dentry->d_inode; struct ll_inode_info *lli = ll_i2info(inode); struct ptlrpc_request *req; + ktime_t kstart = ktime_get(); int rc, err; ENTRY; - CDEBUG(D_VFSTRACE, "VFS Op:inode="DFID"(%p), start %lld, end %lld," - "datasync %d\n", + CDEBUG(D_VFSTRACE, + "VFS Op:inode="DFID"(%p), start %lld, end %lld, datasync %d\n", PFID(ll_inode2fid(inode)), inode, start, end, datasync); - ll_stats_ops_tally(ll_i2sbi(inode), LPROC_LL_FSYNC, 1); - /* fsync's caller has already called _fdata{sync,write}, we want * that IO to finish before calling the osc and mdc sync methods */ rc = filemap_write_and_wait_range(inode->i_mapping, start, end); @@ -4134,7 +4143,7 @@ int ll_fsync(struct file *file, loff_t start, loff_t end, int datasync) ptlrpc_req_finished(req); if (S_ISREG(inode->i_mode)) { - struct ll_file_data *fd = LUSTRE_FPRIVATE(file); + struct ll_file_data *fd = file->private_data; bool cached; /* Sync metadata on MDT first, and then sync the cached data @@ -4153,6 +4162,10 @@ int ll_fsync(struct file *file, loff_t start, loff_t end, int datasync) } inode_unlock(inode); + + if (!rc) + ll_stats_ops_tally(ll_i2sbi(inode), LPROC_LL_FSYNC, + ktime_us_delta(ktime_get(), kstart)); RETURN(rc); } @@ -4170,6 +4183,7 @@ ll_file_flock(struct file *file, int cmd, struct file_lock *file_lock) struct lustre_handle lockh = { 0 }; union ldlm_policy_data flock = { { 0 } }; int fl_type = file_lock->fl_type; + ktime_t kstart = ktime_get(); __u64 flags = 0; int rc; int rc2 = 0; @@ -4178,23 +4192,22 @@ ll_file_flock(struct file *file, int cmd, struct file_lock *file_lock) CDEBUG(D_VFSTRACE, "VFS Op:inode="DFID" file_lock=%p\n", PFID(ll_inode2fid(inode)), file_lock); - ll_stats_ops_tally(ll_i2sbi(inode), LPROC_LL_FLOCK, 1); - - if (file_lock->fl_flags & FL_FLOCK) { - LASSERT((cmd == F_SETLKW) || (cmd == F_SETLK)); - /* flocks are whole-file locks */ - flock.l_flock.end = OFFSET_MAX; - /* For flocks owner is determined by the local file desctiptor*/ - flock.l_flock.owner = (unsigned long)file_lock->fl_file; - } else if (file_lock->fl_flags & FL_POSIX) { - flock.l_flock.owner = (unsigned long)file_lock->fl_owner; - flock.l_flock.start = file_lock->fl_start; - flock.l_flock.end = file_lock->fl_end; - } else { - RETURN(-EINVAL); - } - flock.l_flock.pid = file_lock->fl_pid; + if (file_lock->fl_flags & FL_FLOCK) { + LASSERT((cmd == F_SETLKW) || (cmd == F_SETLK)); + /* flocks are whole-file locks */ + flock.l_flock.end = OFFSET_MAX; + /* For flocks owner is determined by the local file desctiptor*/ + flock.l_flock.owner = (unsigned long)file_lock->fl_file; + } else if (file_lock->fl_flags & FL_POSIX) { + flock.l_flock.owner = (unsigned long)file_lock->fl_owner; + flock.l_flock.start = file_lock->fl_start; + flock.l_flock.end = file_lock->fl_end; + } else { + RETURN(-EINVAL); + } + flock.l_flock.pid = file_lock->fl_pid; +#if defined(HAVE_LM_COMPARE_OWNER) || defined(lm_compare_owner) /* Somewhat ugly workaround for svc lockd. * lockd installs custom fl_lmops->lm_compare_owner that checks * for the fl_owner to be the same (which it always is on local node @@ -4204,6 +4217,7 @@ ll_file_flock(struct file *file, int cmd, struct file_lock *file_lock) * pointer space for current->files are not intersecting */ if (file_lock->fl_lmops && file_lock->fl_lmops->lm_compare_owner) flock.l_flock.owner = (unsigned long)file_lock->fl_pid; +#endif switch (fl_type) { case F_RDLCK: @@ -4296,7 +4310,10 @@ ll_file_flock(struct file *file, int cmd, struct file_lock *file_lock) ll_finish_md_op_data(op_data); - RETURN(rc); + if (!rc) + ll_stats_ops_tally(ll_i2sbi(inode), LPROC_LL_FLOCK, + ktime_us_delta(ktime_get(), kstart)); + RETURN(rc); } int ll_get_fid_by_name(struct inode *parent, const char *name, @@ -4486,7 +4503,7 @@ out_iput: static int ll_file_noflock(struct file *file, int cmd, struct file_lock *file_lock) { - struct ll_file_data *fd = LUSTRE_FPRIVATE(file); + struct ll_file_data *fd = file->private_data; ENTRY; /* @@ -4532,24 +4549,24 @@ int ll_have_md_lock(struct inode *inode, __u64 *bits, enum ldlm_mode l_req_mode) flags = LDLM_FL_BLOCK_GRANTED | LDLM_FL_CBPENDING | LDLM_FL_TEST_LOCK; for (i = 0; i < MDS_INODELOCK_NUMBITS && *bits != 0; i++) { - policy.l_inodebits.bits = *bits & (1 << i); + policy.l_inodebits.bits = *bits & BIT(i); if (policy.l_inodebits.bits == 0) continue; - if (md_lock_match(ll_i2mdexp(inode), flags, fid, LDLM_IBITS, - &policy, mode, &lockh)) { - struct ldlm_lock *lock; - - lock = ldlm_handle2lock(&lockh); - if (lock) { - *bits &= - ~(lock->l_policy_data.l_inodebits.bits); - LDLM_LOCK_PUT(lock); - } else { - *bits &= ~policy.l_inodebits.bits; - } - } - } + if (md_lock_match(ll_i2mdexp(inode), flags, fid, LDLM_IBITS, + &policy, mode, &lockh)) { + struct ldlm_lock *lock; + + lock = ldlm_handle2lock(&lockh); + if (lock) { + *bits &= + ~(lock->l_policy_data.l_inodebits.bits); + LDLM_LOCK_PUT(lock); + } else { + *bits &= ~policy.l_inodebits.bits; + } + } + } RETURN(*bits == 0); } @@ -4636,9 +4653,9 @@ static int ll_inode_revalidate(struct dentry *dentry, enum ldlm_intent_flags op) * here to preserve get_cwd functionality on 2.6. * Bug 10503 */ if (!dentry->d_inode->i_nlink) { - ll_lock_dcache(inode); + spin_lock(&inode->i_lock); d_lustre_invalidate(dentry, 0); - ll_unlock_dcache(inode); + spin_unlock(&inode->i_lock); } ll_lookup_finish_locks(&oit, dentry); @@ -4677,14 +4694,28 @@ static int ll_merge_md_attr(struct inode *inode) RETURN(0); } -int ll_getattr_dentry(struct dentry *de, struct kstat *stat) +int ll_getattr_dentry(struct dentry *de, struct kstat *stat, u32 request_mask, + unsigned int flags) { struct inode *inode = de->d_inode; struct ll_sb_info *sbi = ll_i2sbi(inode); struct ll_inode_info *lli = ll_i2info(inode); + struct inode *dir = de->d_parent->d_inode; + bool need_glimpse = true; + ktime_t kstart = ktime_get(); int rc; - ll_stats_ops_tally(sbi, LPROC_LL_GETATTR, 1); + /* The OST object(s) determine the file size, blocks and mtime. */ + if (!(request_mask & STATX_SIZE || request_mask & STATX_BLOCKS || + request_mask & STATX_MTIME)) + need_glimpse = false; + + if (dentry_may_statahead(dir, de)) + ll_start_statahead(dir, de, need_glimpse && + !(flags & AT_STATX_DONT_SYNC)); + + if (flags & AT_STATX_DONT_SYNC) + GOTO(fill_attr, rc = 0); rc = ll_inode_revalidate(de, IT_GETATTR); if (rc < 0) @@ -4693,10 +4724,36 @@ int ll_getattr_dentry(struct dentry *de, struct kstat *stat) if (S_ISREG(inode->i_mode)) { bool cached; - rc = pcc_inode_getattr(inode, &cached); + if (!need_glimpse) + GOTO(fill_attr, rc); + + rc = pcc_inode_getattr(inode, request_mask, flags, &cached); if (cached && rc < 0) RETURN(rc); + if (cached) + GOTO(fill_attr, rc); + + /* + * If the returned attr is masked with OBD_MD_FLSIZE & + * OBD_MD_FLBLOCKS & OBD_MD_FLMTIME, it means that the file size + * or blocks obtained from MDT is strictly correct, and the file + * is usually not being modified by clients, and the [a|m|c]time + * got from MDT is also strictly correct. + * Under this circumstance, it does not need to send glimpse + * RPCs to OSTs for file attributes such as the size and blocks. + */ + if (lli->lli_attr_valid & OBD_MD_FLSIZE && + lli->lli_attr_valid & OBD_MD_FLBLOCKS && + lli->lli_attr_valid & OBD_MD_FLMTIME) { + inode->i_mtime.tv_sec = lli->lli_mtime; + if (lli->lli_attr_valid & OBD_MD_FLATIME) + inode->i_atime.tv_sec = lli->lli_atime; + if (lli->lli_attr_valid & OBD_MD_FLCTIME) + inode->i_ctime.tv_sec = lli->lli_ctime; + GOTO(fill_attr, rc); + } + /* In case of restore, the MDT has the right size and has * already send it back without granting the layout lock, * inode is up-to-date so glimpse is useless. @@ -4704,7 +4761,7 @@ int ll_getattr_dentry(struct dentry *de, struct kstat *stat) * restore the MDT holds the layout lock so the glimpse will * block up to the end of restore (getattr will block) */ - if (!cached && !ll_file_test_flag(lli, LLIF_FILE_RESTORING)) { + if (!ll_file_test_flag(lli, LLIF_FILE_RESTORING)) { rc = ll_glimpse_size(inode); if (rc < 0) RETURN(rc); @@ -4717,11 +4774,15 @@ int ll_getattr_dentry(struct dentry *de, struct kstat *stat) RETURN(rc); } - inode->i_atime.tv_sec = lli->lli_atime; - inode->i_mtime.tv_sec = lli->lli_mtime; - inode->i_ctime.tv_sec = lli->lli_ctime; + if (lli->lli_attr_valid & OBD_MD_FLATIME) + inode->i_atime.tv_sec = lli->lli_atime; + if (lli->lli_attr_valid & OBD_MD_FLMTIME) + inode->i_mtime.tv_sec = lli->lli_mtime; + if (lli->lli_attr_valid & OBD_MD_FLCTIME) + inode->i_ctime.tv_sec = lli->lli_ctime; } +fill_attr: OBD_FAIL_TIMEOUT(OBD_FAIL_GETATTR_DELAY, 30); if (ll_need_32bit_api(sbi)) { @@ -4740,25 +4801,135 @@ int ll_getattr_dentry(struct dentry *de, struct kstat *stat) stat->atime = inode->i_atime; stat->mtime = inode->i_mtime; stat->ctime = inode->i_ctime; - stat->blksize = sbi->ll_stat_blksize ?: 1 << inode->i_blkbits; + /* stat->blksize is used to tell about preferred IO size */ + if (sbi->ll_stat_blksize) + stat->blksize = sbi->ll_stat_blksize; + else if (S_ISREG(inode->i_mode)) + stat->blksize = 1 << min(PTLRPC_MAX_BRW_BITS + 1, + LL_MAX_BLKSIZE_BITS); + else + stat->blksize = 1 << inode->i_sb->s_blocksize_bits; stat->nlink = inode->i_nlink; stat->size = i_size_read(inode); stat->blocks = inode->i_blocks; - return 0; +#ifdef HAVE_INODEOPS_ENHANCED_GETATTR + if (flags & AT_STATX_DONT_SYNC) { + if (stat->size == 0 && + lli->lli_attr_valid & OBD_MD_FLLAZYSIZE) + stat->size = lli->lli_lazysize; + if (stat->blocks == 0 && + lli->lli_attr_valid & OBD_MD_FLLAZYBLOCKS) + stat->blocks = lli->lli_lazyblocks; + } + + if (lli->lli_attr_valid & OBD_MD_FLBTIME) { + stat->result_mask |= STATX_BTIME; + stat->btime.tv_sec = lli->lli_btime; + } + + stat->attributes_mask = STATX_ATTR_IMMUTABLE | STATX_ATTR_APPEND; + stat->attributes |= ll_inode_to_ext_flags(inode->i_flags); + stat->result_mask &= request_mask; +#endif + + ll_stats_ops_tally(sbi, LPROC_LL_GETATTR, + ktime_us_delta(ktime_get(), kstart)); + + return 0; } #ifdef HAVE_INODEOPS_ENHANCED_GETATTR int ll_getattr(const struct path *path, struct kstat *stat, u32 request_mask, unsigned int flags) { - struct dentry *de = path->dentry; + return ll_getattr_dentry(path->dentry, stat, request_mask, flags); +} #else int ll_getattr(struct vfsmount *mnt, struct dentry *de, struct kstat *stat) { + return ll_getattr_dentry(de, stat, STATX_BASIC_STATS, + AT_STATX_SYNC_AS_STAT); +} #endif - return ll_getattr_dentry(de, stat); + +int cl_falloc(struct inode *inode, int mode, loff_t offset, loff_t len) +{ + struct lu_env *env; + struct cl_io *io; + __u16 refcheck; + int rc; loff_t sa_falloc_end; + loff_t size = i_size_read(inode); + + ENTRY; + + env = cl_env_get(&refcheck); + if (IS_ERR(env)) + RETURN(PTR_ERR(env)); + + io = vvp_env_thread_io(env); + io->ci_obj = ll_i2info(inode)->lli_clob; + io->ci_verify_layout = 1; + io->u.ci_setattr.sa_parent_fid = lu_object_fid(&io->ci_obj->co_lu); + io->u.ci_setattr.sa_falloc_mode = mode; + io->u.ci_setattr.sa_falloc_offset = offset; + io->u.ci_setattr.sa_falloc_len = len; + io->u.ci_setattr.sa_falloc_end = io->u.ci_setattr.sa_falloc_offset + + io->u.ci_setattr.sa_falloc_len; + io->u.ci_setattr.sa_subtype = CL_SETATTR_FALLOCATE; + sa_falloc_end = io->u.ci_setattr.sa_falloc_end; + if (sa_falloc_end > size) { + /* Check new size against VFS/VM file size limit and rlimit */ + rc = inode_newsize_ok(inode, sa_falloc_end); + if (rc) + goto out; + if (sa_falloc_end > ll_file_maxbytes(inode)) { + CDEBUG(D_INODE, "file size too large %llu > %llu\n", + (unsigned long long)(sa_falloc_end), + ll_file_maxbytes(inode)); + rc = -EFBIG; + goto out; + } + io->u.ci_setattr.sa_attr.lvb_size = sa_falloc_end; + if (!(mode & FALLOC_FL_KEEP_SIZE)) + io->u.ci_setattr.sa_avalid |= ATTR_SIZE; + } else { + io->u.ci_setattr.sa_attr.lvb_size = size; + } + +again: + if (cl_io_init(env, io, CIT_SETATTR, io->ci_obj) == 0) + rc = cl_io_loop(env, io); + else + rc = io->ci_result; + + cl_io_fini(env, io); + if (unlikely(io->ci_need_restart)) + goto again; + +out: + cl_env_put(env, &refcheck); + RETURN(rc); +} + +long ll_fallocate(struct file *filp, int mode, loff_t offset, loff_t len) +{ + struct inode *inode = filp->f_path.dentry->d_inode; + int rc; + + /* + * Only mode == 0 (which is standard prealloc) is supported now. + * Punch is not supported yet. + */ + if (mode & ~FALLOC_FL_KEEP_SIZE) + RETURN(-EOPNOTSUPP); + + ll_stats_ops_tally(ll_i2sbi(inode), LPROC_LL_FALLOCATE, 1); + + rc = cl_falloc(inode, mode, offset, len); + + RETURN(rc); } static int ll_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, @@ -4814,7 +4985,7 @@ struct posix_acl *ll_get_acl(struct inode *inode, int type) } #ifdef HAVE_IOP_SET_ACL -#ifdef CONFIG_FS_POSIX_ACL +#ifdef CONFIG_LUSTRE_FS_POSIX_ACL int ll_set_acl(struct inode *inode, struct posix_acl *acl, int type) { struct ll_sb_info *sbi = ll_i2sbi(inode); @@ -4870,50 +5041,10 @@ out: set_cached_acl(inode, type, acl); RETURN(rc); } -#endif /* CONFIG_FS_POSIX_ACL */ +#endif /* CONFIG_LUSTRE_FS_POSIX_ACL */ #endif /* HAVE_IOP_SET_ACL */ -#ifndef HAVE_GENERIC_PERMISSION_2ARGS -static int -# ifdef HAVE_GENERIC_PERMISSION_4ARGS -ll_check_acl(struct inode *inode, int mask, unsigned int flags) -# else -ll_check_acl(struct inode *inode, int mask) -# endif -{ -# ifdef CONFIG_FS_POSIX_ACL - struct posix_acl *acl; - int rc; - ENTRY; - -# ifdef HAVE_GENERIC_PERMISSION_4ARGS - if (flags & IPERM_FLAG_RCU) - return -ECHILD; -# endif - acl = ll_get_acl(inode, ACL_TYPE_ACCESS); - - if (!acl) - RETURN(-EAGAIN); - - rc = posix_acl_permission(inode, acl, mask); - posix_acl_release(acl); - - RETURN(rc); -# else /* !CONFIG_FS_POSIX_ACL */ - return -EAGAIN; -# endif /* CONFIG_FS_POSIX_ACL */ -} -#endif /* HAVE_GENERIC_PERMISSION_2ARGS */ - -#ifdef HAVE_GENERIC_PERMISSION_4ARGS -int ll_inode_permission(struct inode *inode, int mask, unsigned int flags) -#else -# ifdef HAVE_INODE_PERMISION_2ARGS int ll_inode_permission(struct inode *inode, int mask) -# else -int ll_inode_permission(struct inode *inode, int mask, struct nameidata *nd) -# endif -#endif { int rc = 0; struct ll_sb_info *sbi; @@ -4922,24 +5053,23 @@ int ll_inode_permission(struct inode *inode, int mask, struct nameidata *nd) const struct cred *old_cred = NULL; cfs_cap_t cap; bool squash_id = false; + ktime_t kstart = ktime_get(); + ENTRY; -#ifdef MAY_NOT_BLOCK if (mask & MAY_NOT_BLOCK) return -ECHILD; -#elif defined(HAVE_GENERIC_PERMISSION_4ARGS) - if (flags & IPERM_FLAG_RCU) - return -ECHILD; -#endif - /* as root inode are NOT getting validated in lookup operation, - * need to do it before permission check. */ + /* + * as root inode are NOT getting validated in lookup operation, + * need to do it before permission check. + */ - if (inode == inode->i_sb->s_root->d_inode) { + if (inode == inode->i_sb->s_root->d_inode) { rc = ll_inode_revalidate(inode->i_sb->s_root, IT_LOOKUP); - if (rc) - RETURN(rc); - } + if (rc) + RETURN(rc); + } CDEBUG(D_VFSTRACE, "VFS Op:inode="DFID"(%p), inode mode %x mask %o\n", PFID(ll_inode2fid(inode)), inode, inode->i_mode, mask); @@ -4966,20 +5096,23 @@ int ll_inode_permission(struct inode *inode, int mask, struct nameidata *nd) cred->fsuid = make_kuid(&init_user_ns, squash->rsi_uid); cred->fsgid = make_kgid(&init_user_ns, squash->rsi_gid); for (cap = 0; cap < sizeof(cfs_cap_t) * 8; cap++) { - if ((1 << cap) & CFS_CAP_FS_MASK) + if (BIT(cap) & CFS_CAP_FS_MASK) cap_lower(cred->cap_effective, cap); } old_cred = override_creds(cred); } - ll_stats_ops_tally(sbi, LPROC_LL_INODE_PERM, 1); - rc = ll_generic_permission(inode, mask, flags, ll_check_acl); + rc = generic_permission(inode, mask); /* restore current process's credentials and FS capability */ if (squash_id) { revert_creds(old_cred); put_cred(cred); } + if (!rc) + ll_stats_ops_tally(sbi, LPROC_LL_INODE_PERM, + ktime_us_delta(ktime_get(), kstart)); + RETURN(rc); } @@ -5005,7 +5138,8 @@ struct file_operations ll_file_operations = { .llseek = ll_file_seek, .splice_read = ll_file_splice_read, .fsync = ll_fsync, - .flush = ll_flush + .flush = ll_flush, + .fallocate = ll_fallocate, }; struct file_operations ll_file_operations_flock = { @@ -5031,7 +5165,8 @@ struct file_operations ll_file_operations_flock = { .fsync = ll_fsync, .flush = ll_flush, .flock = ll_file_flock, - .lock = ll_file_flock + .lock = ll_file_flock, + .fallocate = ll_fallocate, }; /* These are for -o noflock - to return ENOSYS on flock calls */ @@ -5058,7 +5193,8 @@ struct file_operations ll_file_operations_noflock = { .fsync = ll_fsync, .flush = ll_flush, .flock = ll_file_noflock, - .lock = ll_file_noflock + .lock = ll_file_noflock, + .fallocate = ll_fallocate, }; struct inode_operations ll_file_inode_operations = { @@ -5072,9 +5208,7 @@ struct inode_operations ll_file_inode_operations = { #endif .listxattr = ll_listxattr, .fiemap = ll_fiemap, -#ifdef HAVE_IOP_GET_ACL .get_acl = ll_get_acl, -#endif #ifdef HAVE_IOP_SET_ACL .set_acl = ll_set_acl, #endif