X-Git-Url: https://git.whamcloud.com/?p=fs%2Flustre-release.git;a=blobdiff_plain;f=lustre%2Fllite%2Ffile.c;h=e0019c3070ba5959ef24c4bec85f744b80d1b4f4;hp=2947f4baad8e2593fb935c7ad29b6f6bdec7cefb;hb=0603cc1a773dbe72f520e856743ef40fd7dfa1af;hpb=caa55aec4ae0421db69a0b9feac5cf0880f1c098 diff --git a/lustre/llite/file.c b/lustre/llite/file.c index 2947f4b..e0019c3 100644 --- a/lustre/llite/file.c +++ b/lustre/llite/file.c @@ -50,11 +50,21 @@ #include "cl_object.h" -struct ll_file_data *ll_file_data_get(void) +static int +ll_put_grouplock(struct inode *inode, struct file *file, unsigned long arg); + +static int ll_lease_close(struct obd_client_handle *och, struct inode *inode, + bool *lease_broken); + +static enum llioc_iter +ll_iocontrol_call(struct inode *inode, struct file *file, + unsigned int cmd, unsigned long arg, int *rcp); + +static struct ll_file_data *ll_file_data_get(void) { struct ll_file_data *fd; - OBD_SLAB_ALLOC_PTR_GFP(fd, ll_file_data_slab, CFS_ALLOC_IO); + OBD_SLAB_ALLOC_PTR_GFP(fd, ll_file_data_slab, GFP_NOFS); if (fd == NULL) return NULL; @@ -119,8 +129,9 @@ out: } static int ll_close_inode_openhandle(struct obd_export *md_exp, - struct inode *inode, - struct obd_client_handle *och) + struct inode *inode, + struct obd_client_handle *och, + const __u64 *data_version) { struct obd_export *exp = ll_i2mdexp(inode); struct md_op_data *op_data; @@ -144,7 +155,14 @@ static int ll_close_inode_openhandle(struct obd_export *md_exp, if (op_data == NULL) GOTO(out, rc = -ENOMEM); // XXX We leak openhandle and request here. - ll_prepare_close(inode, op_data, och); + ll_prepare_close(inode, op_data, och); + if (data_version != NULL) { + /* Pass in data_version implies release. */ + op_data->op_bias |= MDS_HSM_RELEASE; + op_data->op_data_version = *data_version; + op_data->op_lease_handle = och->och_lease_handle; + op_data->op_attr.ia_valid |= ATTR_SIZE | ATTR_BLOCKS; + } epoch_close = (op_data->op_flags & MF_EPOCH_CLOSE); rc = md_close(md_exp, op_data, och->och_mod, &req); if (rc == -EAGAIN) { @@ -154,14 +172,17 @@ static int ll_close_inode_openhandle(struct obd_export *md_exp, * OSTs and send setattr to back to MDS. */ rc = ll_som_update(inode, op_data); if (rc) { - CERROR("inode %lu mdc Size-on-MDS update failed: " - "rc = %d\n", inode->i_ino, rc); - rc = 0; - } - } else if (rc) { - CERROR("inode %lu mdc close failed: rc = %d\n", - inode->i_ino, rc); - } + CERROR("%s: inode "DFID" mdc Size-on-MDS update" + " failed: rc = %d\n", + ll_i2mdexp(inode)->exp_obd->obd_name, + PFID(ll_inode2fid(inode)), rc); + rc = 0; + } + } else if (rc) { + CERROR("%s: inode "DFID" mdc close failed: rc = %d\n", + ll_i2mdexp(inode)->exp_obd->obd_name, + PFID(ll_inode2fid(inode)), rc); + } /* DATA_MODIFIED flag was successfully sent on close, cancel data * modification flag. */ @@ -173,15 +194,23 @@ static int ll_close_inode_openhandle(struct obd_export *md_exp, spin_unlock(&lli->lli_lock); } - ll_finish_md_op_data(op_data); - if (rc == 0) { rc = ll_objects_destroy(req, inode); if (rc) - CERROR("inode %lu ll_objects destroy: rc = %d\n", - inode->i_ino, rc); + CERROR("%s: inode "DFID + " ll_objects destroy: rc = %d\n", + ll_i2mdexp(inode)->exp_obd->obd_name, + PFID(ll_inode2fid(inode)), rc); } + if (rc == 0 && op_data->op_bias & MDS_HSM_RELEASE) { + struct mdt_body *body; + body = req_capsule_server_get(&req->rq_pill, &RMF_MDT_BODY); + if (!(body->valid & OBD_MD_FLRELEASED)) + rc = -EBUSY; + } + + ll_finish_md_op_data(op_data); EXIT; out: @@ -199,48 +228,51 @@ out: return rc; } -int ll_md_real_close(struct inode *inode, int flags) +int ll_md_real_close(struct inode *inode, fmode_t fmode) { - struct ll_inode_info *lli = ll_i2info(inode); - struct obd_client_handle **och_p; - struct obd_client_handle *och; - __u64 *och_usecount; - int rc = 0; - ENTRY; + struct ll_inode_info *lli = ll_i2info(inode); + struct obd_client_handle **och_p; + struct obd_client_handle *och; + __u64 *och_usecount; + int rc = 0; + ENTRY; - if (flags & FMODE_WRITE) { - och_p = &lli->lli_mds_write_och; - och_usecount = &lli->lli_open_fd_write_count; - } else if (flags & FMODE_EXEC) { - och_p = &lli->lli_mds_exec_och; - och_usecount = &lli->lli_open_fd_exec_count; - } else { - LASSERT(flags & FMODE_READ); - och_p = &lli->lli_mds_read_och; - och_usecount = &lli->lli_open_fd_read_count; - } + if (fmode & FMODE_WRITE) { + och_p = &lli->lli_mds_write_och; + och_usecount = &lli->lli_open_fd_write_count; + } else if (fmode & FMODE_EXEC) { + och_p = &lli->lli_mds_exec_och; + och_usecount = &lli->lli_open_fd_exec_count; + } else { + LASSERT(fmode & FMODE_READ); + och_p = &lli->lli_mds_read_och; + och_usecount = &lli->lli_open_fd_read_count; + } mutex_lock(&lli->lli_och_mutex); - if (*och_usecount) { /* There are still users of this handle, so - skip freeing it. */ + if (*och_usecount > 0) { + /* There are still users of this handle, so skip + * freeing it. */ mutex_unlock(&lli->lli_och_mutex); - RETURN(0); - } - och=*och_p; - *och_p = NULL; + RETURN(0); + } + + och = *och_p; + *och_p = NULL; mutex_unlock(&lli->lli_och_mutex); - if (och) { /* There might be a race and somebody have freed this och - already */ - rc = ll_close_inode_openhandle(ll_i2sbi(inode)->ll_md_exp, - inode, och); - } + if (och != NULL) { + /* There might be a race and this handle may already + * be closed. */ + rc = ll_close_inode_openhandle(ll_i2sbi(inode)->ll_md_exp, + inode, och, NULL); + } - RETURN(rc); + RETURN(rc); } -int ll_md_close(struct obd_export *md_exp, struct inode *inode, - struct file *file) +static int ll_md_close(struct obd_export *md_exp, struct inode *inode, + struct file *file) { struct ll_file_data *fd = LUSTRE_FPRIVATE(file); struct ll_inode_info *lli = ll_i2info(inode); @@ -251,11 +283,29 @@ int ll_md_close(struct obd_export *md_exp, struct inode *inode, if (unlikely(fd->fd_flags & LL_FILE_GROUP_LOCKED)) ll_put_grouplock(inode, file, fd->fd_grouplock.cg_gid); + if (fd->fd_lease_och != NULL) { + bool lease_broken; + + /* Usually the lease is not released when the + * application crashed, we need to release here. */ + rc = ll_lease_close(fd->fd_lease_och, inode, &lease_broken); + CDEBUG(rc ? D_ERROR : D_INODE, "Clean up lease "DFID" %d/%d\n", + PFID(&lli->lli_fid), rc, lease_broken); + + fd->fd_lease_och = NULL; + } + + if (fd->fd_och != NULL) { + rc = ll_close_inode_openhandle(md_exp, inode, fd->fd_och, NULL); + fd->fd_och = NULL; + GOTO(out, rc); + } + /* Let's see if we have good enough OPEN lock on the file and if we can skip talking to MDS */ if (file->f_dentry->d_inode) { /* Can this ever be false? */ int lockmode; - int flags = LDLM_FL_BLOCK_GRANTED | LDLM_FL_TEST_LOCK; + __u64 flags = LDLM_FL_BLOCK_GRANTED | LDLM_FL_TEST_LOCK; struct lustre_handle lockh; struct inode *inode = file->f_dentry->d_inode; ldlm_policy_data_t policy = {.l_inodebits={MDS_INODELOCK_OPEN}}; @@ -287,11 +337,12 @@ int ll_md_close(struct obd_export *md_exp, struct inode *inode, file, file->f_dentry, file->f_dentry->d_name.name); } - LUSTRE_FPRIVATE(file) = NULL; - ll_file_data_put(fd); - ll_capa_close(inode); +out: + LUSTRE_FPRIVATE(file) = NULL; + ll_file_data_put(fd); + ll_capa_close(inode); - RETURN(rc); + RETURN(rc); } /* While this returns an error code, fput() the caller does not, so we need @@ -307,21 +358,21 @@ int ll_file_release(struct inode *inode, struct file *file) int rc; ENTRY; - CDEBUG(D_VFSTRACE, "VFS Op:inode=%lu/%u(%p)\n", inode->i_ino, - inode->i_generation, inode); + CDEBUG(D_VFSTRACE, "VFS Op:inode="DFID"(%p)\n", + PFID(ll_inode2fid(inode)), inode); #ifdef CONFIG_FS_POSIX_ACL - if (sbi->ll_flags & LL_SBI_RMT_CLIENT && - inode == inode->i_sb->s_root->d_inode) { - struct ll_file_data *fd = LUSTRE_FPRIVATE(file); - - LASSERT(fd != NULL); - if (unlikely(fd->fd_flags & LL_FILE_RMTACL)) { - fd->fd_flags &= ~LL_FILE_RMTACL; - rct_del(&sbi->ll_rct, cfs_curproc_pid()); - et_search_free(&sbi->ll_et, cfs_curproc_pid()); - } - } + if (sbi->ll_flags & LL_SBI_RMT_CLIENT && + inode == inode->i_sb->s_root->d_inode) { + struct ll_file_data *fd = LUSTRE_FPRIVATE(file); + + LASSERT(fd != NULL); + if (unlikely(fd->fd_flags & LL_FILE_RMTACL)) { + fd->fd_flags &= ~LL_FILE_RMTACL; + rct_del(&sbi->ll_rct, current_pid()); + et_search_free(&sbi->ll_et, current_pid()); + } + } #endif if (inode->i_sb->s_root != file->f_dentry) @@ -343,7 +394,8 @@ int ll_file_release(struct inode *inode, struct file *file) } if (!S_ISDIR(inode->i_mode)) { - lov_read_and_clear_async_rc(lli->lli_clob); + if (lli->lli_clob != NULL) + lov_read_and_clear_async_rc(lli->lli_clob); lli->lli_async_rc = 0; } @@ -422,8 +474,7 @@ static int ll_intent_file_open(struct file *file, void *lmm, itp, NULL); out: - ptlrpc_req_finished(itp->d.lustre.it_data); - it_clear_disposition(itp, DISP_ENQ_COMPLETE); + ptlrpc_req_finished(req); ll_intent_drop_lock(itp); RETURN(rc); @@ -443,28 +494,24 @@ void ll_ioepoch_open(struct ll_inode_info *lli, __u64 ioepoch) } } -static int ll_och_fill(struct obd_export *md_exp, struct ll_inode_info *lli, - struct lookup_intent *it, struct obd_client_handle *och) +static int ll_och_fill(struct obd_export *md_exp, struct lookup_intent *it, + struct obd_client_handle *och) { - struct ptlrpc_request *req = it->d.lustre.it_data; - struct mdt_body *body; - - LASSERT(och); - - body = req_capsule_server_get(&req->rq_pill, &RMF_MDT_BODY); - LASSERT(body != NULL); /* reply already checked out */ + struct ptlrpc_request *req = it->d.lustre.it_data; + struct mdt_body *body; - memcpy(&och->och_fh, &body->handle, sizeof(body->handle)); - och->och_magic = OBD_CLIENT_HANDLE_MAGIC; - och->och_fid = lli->lli_fid; - och->och_flags = it->it_flags; - ll_ioepoch_open(lli, body->ioepoch); + body = req_capsule_server_get(&req->rq_pill, &RMF_MDT_BODY); + och->och_fh = body->handle; + och->och_fid = body->fid1; + och->och_lease_handle.cookie = it->d.lustre.it_lock_handle; + och->och_magic = OBD_CLIENT_HANDLE_MAGIC; + och->och_flags = it->it_flags; - return md_set_open_replay_data(md_exp, och, req); + return md_set_open_replay_data(md_exp, och, it); } -int ll_local_open(struct file *file, struct lookup_intent *it, - struct ll_file_data *fd, struct obd_client_handle *och) +static int ll_local_open(struct file *file, struct lookup_intent *it, + struct ll_file_data *fd, struct obd_client_handle *och) { struct inode *inode = file->f_dentry->d_inode; struct ll_inode_info *lli = ll_i2info(inode); @@ -479,21 +526,19 @@ int ll_local_open(struct file *file, struct lookup_intent *it, struct mdt_body *body; int rc; - rc = ll_och_fill(ll_i2sbi(inode)->ll_md_exp, lli, it, och); - if (rc) - RETURN(rc); + rc = ll_och_fill(ll_i2sbi(inode)->ll_md_exp, it, och); + if (rc != 0) + RETURN(rc); - body = req_capsule_server_get(&req->rq_pill, &RMF_MDT_BODY); - if ((it->it_flags & FMODE_WRITE) && - (body->valid & OBD_MD_FLSIZE)) - CDEBUG(D_INODE, "Epoch "LPU64" opened on "DFID"\n", - lli->lli_ioepoch, PFID(&lli->lli_fid)); - } + body = req_capsule_server_get(&req->rq_pill, &RMF_MDT_BODY); + ll_ioepoch_open(lli, body->ioepoch); + } + + LUSTRE_FPRIVATE(file) = fd; + ll_readahead_init(inode, &fd->fd_ras); + fd->fd_omode = it->it_flags & (FMODE_READ | FMODE_WRITE | FMODE_EXEC); - LUSTRE_FPRIVATE(file) = fd; - ll_readahead_init(inode, &fd->fd_ras); - fd->fd_omode = it->it_flags; - RETURN(0); + RETURN(0); } /* Open a file, and (for the very first open) create objects on the OSTs at @@ -520,8 +565,8 @@ int ll_file_open(struct inode *inode, struct file *file) int rc = 0, opendir_set = 0; ENTRY; - CDEBUG(D_VFSTRACE, "VFS Op:inode=%lu/%u(%p), flags %o\n", inode->i_ino, - inode->i_generation, inode, file->f_flags); + CDEBUG(D_VFSTRACE, "VFS Op:inode="DFID"(%p), flags %o\n", + PFID(ll_inode2fid(inode)), inode, file->f_flags); it = file->private_data; /* XXX: compat macro */ file->private_data = NULL; /* prevent ll_local_open assertion */ @@ -536,7 +581,7 @@ int ll_file_open(struct inode *inode, struct file *file) if (lli->lli_opendir_key == NULL && lli->lli_sai == NULL && lli->lli_opendir_pid == 0) { lli->lli_opendir_key = fd; - lli->lli_opendir_pid = cfs_curproc_pid(); + lli->lli_opendir_pid = current_pid(); opendir_set = 1; } spin_unlock(&lli->lli_sa_lock); @@ -638,16 +683,18 @@ restart: * (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) - GOTO(out_och_free, rc); + rc = it_open_error(DISP_OPEN_OPEN, it); + if (rc != 0) + GOTO(out_och_free, rc); - LASSERT(it_disposition(it, DISP_ENQ_OPEN_REF)); + LASSERTF(it_disposition(it, DISP_ENQ_OPEN_REF), + "inode %p: disposition %x, status %d\n", inode, + it_disposition(it, ~0), it->d.lustre.it_status); - rc = ll_local_open(file, it, fd, *och_p); - if (rc) - GOTO(out_och_free, rc); - } + rc = ll_local_open(file, it, fd, *och_p); + if (rc) + GOTO(out_och_free, rc); + } mutex_unlock(&lli->lli_och_mutex); fd = NULL; @@ -659,15 +706,14 @@ restart: ll_capa_open(inode); - if (!lli->lli_has_smd) { - if (file->f_flags & O_LOV_DELAY_CREATE || - !(file->f_mode & FMODE_WRITE)) { - CDEBUG(D_INODE, "object creation was delayed\n"); - GOTO(out_och_free, rc); - } - } - file->f_flags &= ~O_LOV_DELAY_CREATE; - GOTO(out_och_free, rc); + if (!lli->lli_has_smd && + (cl_is_lov_delay_create(file->f_flags) || + (file->f_mode & FMODE_WRITE) == 0)) { + CDEBUG(D_INODE, "object creation was delayed\n"); + GOTO(out_och_free, rc); + } + cl_lov_delay_create_clear(&file->f_flags); + GOTO(out_och_free, rc); out_och_free: if (rc) { @@ -695,10 +741,205 @@ out_openerr: return rc; } +static int ll_md_blocking_lease_ast(struct ldlm_lock *lock, + struct ldlm_lock_desc *desc, void *data, int flag) +{ + int rc; + struct lustre_handle lockh; + ENTRY; + + switch (flag) { + case LDLM_CB_BLOCKING: + ldlm_lock2handle(lock, &lockh); + rc = ldlm_cli_cancel(&lockh, LCF_ASYNC); + if (rc < 0) { + CDEBUG(D_INODE, "ldlm_cli_cancel: %d\n", rc); + RETURN(rc); + } + break; + case LDLM_CB_CANCELING: + /* do nothing */ + break; + } + RETURN(0); +} + +/** + * Acquire a lease and open the file. + */ +static struct obd_client_handle * +ll_lease_open(struct inode *inode, struct file *file, fmode_t fmode, + __u64 open_flags) +{ + struct lookup_intent it = { .it_op = IT_OPEN }; + struct ll_sb_info *sbi = ll_i2sbi(inode); + struct md_op_data *op_data; + struct ptlrpc_request *req; + struct lustre_handle old_handle = { 0 }; + struct obd_client_handle *och = NULL; + int rc; + int rc2; + ENTRY; + + if (fmode != FMODE_WRITE && fmode != FMODE_READ) + RETURN(ERR_PTR(-EINVAL)); + + if (file != NULL) { + struct ll_inode_info *lli = ll_i2info(inode); + struct ll_file_data *fd = LUSTRE_FPRIVATE(file); + struct obd_client_handle **och_p; + __u64 *och_usecount; + + if (!(fmode & file->f_mode) || (file->f_mode & FMODE_EXEC)) + RETURN(ERR_PTR(-EPERM)); + + /* Get the openhandle of the file */ + rc = -EBUSY; + mutex_lock(&lli->lli_och_mutex); + if (fd->fd_lease_och != NULL) { + mutex_unlock(&lli->lli_och_mutex); + RETURN(ERR_PTR(rc)); + } + + if (fd->fd_och == NULL) { + if (file->f_mode & FMODE_WRITE) { + LASSERT(lli->lli_mds_write_och != NULL); + och_p = &lli->lli_mds_write_och; + och_usecount = &lli->lli_open_fd_write_count; + } else { + LASSERT(lli->lli_mds_read_och != NULL); + och_p = &lli->lli_mds_read_och; + och_usecount = &lli->lli_open_fd_read_count; + } + if (*och_usecount == 1) { + fd->fd_och = *och_p; + *och_p = NULL; + *och_usecount = 0; + rc = 0; + } + } + mutex_unlock(&lli->lli_och_mutex); + if (rc < 0) /* more than 1 opener */ + RETURN(ERR_PTR(rc)); + + LASSERT(fd->fd_och != NULL); + old_handle = fd->fd_och->och_fh; + } + + OBD_ALLOC_PTR(och); + if (och == NULL) + RETURN(ERR_PTR(-ENOMEM)); + + op_data = ll_prep_md_op_data(NULL, inode, inode, NULL, 0, 0, + LUSTRE_OPC_ANY, NULL); + if (IS_ERR(op_data)) + GOTO(out, rc = PTR_ERR(op_data)); + + /* To tell the MDT this openhandle is from the same owner */ + op_data->op_handle = old_handle; + + it.it_flags = fmode | open_flags; + it.it_flags |= MDS_OPEN_LOCK | MDS_OPEN_BY_FID | MDS_OPEN_LEASE; + rc = md_intent_lock(sbi->ll_md_exp, op_data, NULL, 0, &it, 0, &req, + ll_md_blocking_lease_ast, + /* LDLM_FL_NO_LRU: To not put the lease lock into LRU list, otherwise + * it can be cancelled which may mislead applications that the lease is + * broken; + * LDLM_FL_EXCL: Set this flag so that it won't be matched by normal + * open in ll_md_blocking_ast(). Otherwise as ll_md_blocking_lease_ast + * doesn't deal with openhandle, so normal openhandle will be leaked. */ + LDLM_FL_NO_LRU | LDLM_FL_EXCL); + ll_finish_md_op_data(op_data); + ptlrpc_req_finished(req); + if (rc < 0) + GOTO(out_release_it, rc); + + if (it_disposition(&it, DISP_LOOKUP_NEG)) + GOTO(out_release_it, rc = -ENOENT); + + rc = it_open_error(DISP_OPEN_OPEN, &it); + if (rc) + GOTO(out_release_it, rc); + + LASSERT(it_disposition(&it, DISP_ENQ_OPEN_REF)); + ll_och_fill(sbi->ll_md_exp, &it, och); + + if (!it_disposition(&it, DISP_OPEN_LEASE)) /* old server? */ + GOTO(out_close, rc = -EOPNOTSUPP); + + /* already get lease, handle lease lock */ + ll_set_lock_data(sbi->ll_md_exp, inode, &it, NULL); + if (it.d.lustre.it_lock_mode == 0 || + it.d.lustre.it_lock_bits != MDS_INODELOCK_OPEN) { + /* open lock must return for lease */ + CERROR(DFID "lease granted but no open lock, %d/%Lu.\n", + PFID(ll_inode2fid(inode)), it.d.lustre.it_lock_mode, + it.d.lustre.it_lock_bits); + GOTO(out_close, rc = -EPROTO); + } + + ll_intent_release(&it); + RETURN(och); + +out_close: + /* Cancel open lock */ + if (it.d.lustre.it_lock_mode != 0) { + ldlm_lock_decref_and_cancel(&och->och_lease_handle, + it.d.lustre.it_lock_mode); + it.d.lustre.it_lock_mode = 0; + och->och_lease_handle.cookie = 0ULL; + } + rc2 = ll_close_inode_openhandle(sbi->ll_md_exp, inode, och, NULL); + if (rc2 < 0) + CERROR("%s: error closing file "DFID": %d\n", + ll_get_fsname(inode->i_sb, NULL, 0), + PFID(&ll_i2info(inode)->lli_fid), rc2); + och = NULL; /* och has been freed in ll_close_inode_openhandle() */ +out_release_it: + ll_intent_release(&it); +out: + if (och != NULL) + OBD_FREE_PTR(och); + RETURN(ERR_PTR(rc)); +} + +/** + * Release lease and close the file. + * It will check if the lease has ever broken. + */ +static int ll_lease_close(struct obd_client_handle *och, struct inode *inode, + bool *lease_broken) +{ + struct ldlm_lock *lock; + bool cancelled = true; + int rc; + ENTRY; + + lock = ldlm_handle2lock(&och->och_lease_handle); + if (lock != NULL) { + lock_res_and_lock(lock); + cancelled = ldlm_is_cancel(lock); + unlock_res_and_lock(lock); + LDLM_LOCK_PUT(lock); + } + + CDEBUG(D_INODE, "lease for "DFID" broken? %d\n", + PFID(&ll_i2info(inode)->lli_fid), cancelled); + + if (!cancelled) + ldlm_cli_cancel(&och->och_lease_handle, 0); + if (lease_broken != NULL) + *lease_broken = cancelled; + + rc = ll_close_inode_openhandle(ll_i2sbi(inode)->ll_md_exp, inode, och, + NULL); + RETURN(rc); +} + /* Fills the obdo with the attributes for the lsm */ static int ll_lsm_getattr(struct lov_stripe_md *lsm, struct obd_export *exp, - struct obd_capa *capa, struct obdo *obdo, - __u64 ioepoch, int sync) + struct obd_capa *capa, struct obdo *obdo, + __u64 ioepoch, int dv_flags) { struct ptlrpc_request_set *set; struct obd_info oinfo = { { { 0 } } }; @@ -720,10 +961,12 @@ static int ll_lsm_getattr(struct lov_stripe_md *lsm, struct obd_export *exp, OBD_MD_FLGROUP | OBD_MD_FLEPOCH | OBD_MD_FLDATAVERSION; oinfo.oi_capa = capa; - if (sync) { - oinfo.oi_oa->o_valid |= OBD_MD_FLFLAGS; - oinfo.oi_oa->o_flags |= OBD_FL_SRVLOCK; - } + if (dv_flags & (LL_DV_WR_FLUSH | LL_DV_RD_FLUSH)) { + oinfo.oi_oa->o_valid |= OBD_MD_FLFLAGS; + oinfo.oi_oa->o_flags |= OBD_FL_SRVLOCK; + if (dv_flags & LL_DV_WR_FLUSH) + oinfo.oi_oa->o_flags |= OBD_FL_FLUSH; + } set = ptlrpc_prep_set(); if (set == NULL) { @@ -735,12 +978,17 @@ static int ll_lsm_getattr(struct lov_stripe_md *lsm, struct obd_export *exp, rc = ptlrpc_set_wait(set); ptlrpc_set_destroy(set); } - if (rc == 0) - oinfo.oi_oa->o_valid &= (OBD_MD_FLBLOCKS | OBD_MD_FLBLKSZ | - OBD_MD_FLATIME | OBD_MD_FLMTIME | - OBD_MD_FLCTIME | OBD_MD_FLSIZE | - OBD_MD_FLDATAVERSION); - RETURN(rc); + if (rc == 0) { + oinfo.oi_oa->o_valid &= (OBD_MD_FLBLOCKS | OBD_MD_FLBLKSZ | + OBD_MD_FLATIME | OBD_MD_FLMTIME | + OBD_MD_FLCTIME | OBD_MD_FLSIZE | + OBD_MD_FLDATAVERSION | OBD_MD_FLFLAGS); + if (dv_flags & LL_DV_WR_FLUSH && + !(oinfo.oi_oa->o_valid & OBD_MD_FLFLAGS && + oinfo.oi_oa->o_flags & OBD_FL_FLUSH)) + RETURN(-ENOTSUPP); + } + RETURN(rc); } /** @@ -757,7 +1005,7 @@ int ll_inode_getattr(struct inode *inode, struct obdo *obdo, lsm = ccc_inode_lsm_get(inode); rc = ll_lsm_getattr(lsm, ll_i2dtexp(inode), - capa, obdo, ioepoch, sync); + capa, obdo, ioepoch, sync ? LL_DV_RD_FLUSH : 0); capa_put(capa); if (rc == 0) { struct ost_id *oi = lsm ? &lsm->lsm_oi : &obdo->o_oi; @@ -834,7 +1082,34 @@ int ll_glimpse_ioctl(struct ll_sb_info *sbi, struct lov_stripe_md *lsm, return rc; } -void ll_io_init(struct cl_io *io, const struct file *file, int write) +static bool file_is_noatime(const struct file *file) +{ + const struct vfsmount *mnt = file->f_path.mnt; + const struct inode *inode = file->f_path.dentry->d_inode; + + /* Adapted from file_accessed() and touch_atime().*/ + if (file->f_flags & O_NOATIME) + return true; + + if (inode->i_flags & S_NOATIME) + return true; + + if (IS_NOATIME(inode)) + return true; + + if (mnt->mnt_flags & (MNT_NOATIME | MNT_READONLY)) + return true; + + if ((mnt->mnt_flags & MNT_NODIRATIME) && S_ISDIR(inode->i_mode)) + return true; + + if ((inode->i_sb->s_flags & MS_NODIRATIME) && S_ISDIR(inode->i_mode)) + return true; + + return false; +} + +static void ll_io_init(struct cl_io *io, const struct file *file, int write) { struct inode *inode = file->f_dentry->d_inode; @@ -853,6 +1128,8 @@ void ll_io_init(struct cl_io *io, const struct file *file, int write) } else if (file->f_flags & O_APPEND) { io->ci_lockreq = CILR_MANDATORY; } + + io->ci_noatime = file_is_noatime(file); } static ssize_t @@ -862,9 +1139,12 @@ ll_file_io_generic(const struct lu_env *env, struct vvp_io_args *args, { struct ll_inode_info *lli = ll_i2info(file->f_dentry->d_inode); struct ll_file_data *fd = LUSTRE_FPRIVATE(file); - struct cl_io *io; - ssize_t result; - ENTRY; + struct cl_io *io; + ssize_t result; + ENTRY; + + CDEBUG(D_VFSTRACE, "file: %s, type: %d ppos: "LPU64", count: %zd\n", + file->f_dentry->d_name.name, iot, *ppos, count); restart: io = ccc_env_thread_io(env); @@ -883,22 +1163,15 @@ restart: cio->cui_iov = args->u.normal.via_iov; cio->cui_nrsegs = args->u.normal.via_nrsegs; cio->cui_tot_nrsegs = cio->cui_nrsegs; -#ifndef HAVE_FILE_WRITEV cio->cui_iocb = args->u.normal.via_iocb; -#endif if ((iot == CIT_WRITE) && !(cio->cui_fd->fd_flags & LL_FILE_GROUP_LOCKED)) { if (mutex_lock_interruptible(&lli-> - lli_write_mutex)) - GOTO(out, result = -ERESTARTSYS); - write_mutex_locked = 1; - } else if (iot == CIT_READ) { - down_read(&lli->lli_trunc_sem); - } - break; - case IO_SENDFILE: - vio->u.sendfile.cui_actor = args->u.sendfile.via_actor; - vio->u.sendfile.cui_target = args->u.sendfile.via_target; + lli_write_mutex)) + GOTO(out, result = -ERESTARTSYS); + write_mutex_locked = 1; + } + down_read(&lli->lli_trunc_sem); break; case IO_SPLICE: vio->u.splice.cui_pipe = args->u.splice.via_pipe; @@ -909,10 +1182,10 @@ restart: LBUG(); } result = cl_io_loop(env, io); - if (write_mutex_locked) - mutex_unlock(&lli->lli_write_mutex); - else if (args->via_io_subtype == IO_NORMAL && iot == CIT_READ) + if (args->via_io_subtype == IO_NORMAL) up_read(&lli->lli_trunc_sem); + if (write_mutex_locked) + mutex_unlock(&lli->lli_write_mutex); } else { /* cl_io_rw_init() handled IO */ result = io->ci_result; @@ -927,7 +1200,7 @@ out: cl_io_fini(env, io); /* If any bit been read/written (result != 0), we just return * short read/write instead of restart io. */ - if (result == 0 && io->ci_need_restart) { + if ((result == 0 || result == -ENODATA) && io->ci_need_restart) { CDEBUG(D_VFSTRACE, "Restart %s on %s from %lld, count:%zd\n", iot == CIT_READ ? "read" : "write", file->f_dentry->d_name.name, *ppos, count); @@ -948,6 +1221,7 @@ out: fd->fd_write_failed = true; } } + CDEBUG(D_VFSTRACE, "iot: %d, result: %zd\n", iot, result); return result; } @@ -984,56 +1258,6 @@ static int ll_file_get_iov_count(const struct iovec *iov, return 0; } -#ifdef HAVE_FILE_READV -static ssize_t ll_file_readv(struct file *file, const struct iovec *iov, - unsigned long nr_segs, loff_t *ppos) -{ - struct lu_env *env; - struct vvp_io_args *args; - size_t count; - ssize_t result; - int refcheck; - ENTRY; - - result = ll_file_get_iov_count(iov, &nr_segs, &count); - if (result) - RETURN(result); - - env = cl_env_get(&refcheck); - if (IS_ERR(env)) - RETURN(PTR_ERR(env)); - - args = vvp_env_args(env, IO_NORMAL); - args->u.normal.via_iov = (struct iovec *)iov; - args->u.normal.via_nrsegs = nr_segs; - - result = ll_file_io_generic(env, args, file, CIT_READ, ppos, count); - cl_env_put(env, &refcheck); - RETURN(result); -} - -static ssize_t ll_file_read(struct file *file, char *buf, size_t count, - loff_t *ppos) -{ - struct lu_env *env; - struct iovec *local_iov; - ssize_t result; - int refcheck; - ENTRY; - - env = cl_env_get(&refcheck); - if (IS_ERR(env)) - RETURN(PTR_ERR(env)); - - local_iov = &vvp_env_info(env)->vti_local_iov; - local_iov->iov_base = (void __user *)buf; - local_iov->iov_len = count; - result = ll_file_readv(file, local_iov, 1, ppos); - cl_env_put(env, &refcheck); - RETURN(result); -} - -#else static ssize_t ll_file_aio_read(struct kiocb *iocb, const struct iovec *iov, unsigned long nr_segs, loff_t pos) { @@ -1083,7 +1307,11 @@ static ssize_t ll_file_read(struct file *file, char *buf, size_t count, local_iov->iov_len = count; init_sync_kiocb(kiocb, file); kiocb->ki_pos = *ppos; +#ifdef HAVE_KIOCB_KI_LEFT kiocb->ki_left = count; +#else + kiocb->ki_nbytes = count; +#endif result = ll_file_aio_read(kiocb, local_iov, 1, kiocb->ki_pos); *ppos = kiocb->ki_pos; @@ -1091,62 +1319,11 @@ static ssize_t ll_file_read(struct file *file, char *buf, size_t count, cl_env_put(env, &refcheck); RETURN(result); } -#endif /* * Write to a file (through the page cache). + * AIO stuff */ -#ifdef HAVE_FILE_WRITEV -static ssize_t ll_file_writev(struct file *file, const struct iovec *iov, - unsigned long nr_segs, loff_t *ppos) -{ - struct lu_env *env; - struct vvp_io_args *args; - size_t count; - ssize_t result; - int refcheck; - ENTRY; - - result = ll_file_get_iov_count(iov, &nr_segs, &count); - if (result) - RETURN(result); - - env = cl_env_get(&refcheck); - if (IS_ERR(env)) - RETURN(PTR_ERR(env)); - - args = vvp_env_args(env, IO_NORMAL); - args->u.normal.via_iov = (struct iovec *)iov; - args->u.normal.via_nrsegs = nr_segs; - - result = ll_file_io_generic(env, args, file, CIT_WRITE, ppos, count); - cl_env_put(env, &refcheck); - RETURN(result); -} - -static ssize_t ll_file_write(struct file *file, const char *buf, size_t count, - loff_t *ppos) -{ - struct lu_env *env; - struct iovec *local_iov; - ssize_t result; - int refcheck; - ENTRY; - - env = cl_env_get(&refcheck); - if (IS_ERR(env)) - RETURN(PTR_ERR(env)); - - local_iov = &vvp_env_info(env)->vti_local_iov; - local_iov->iov_base = (void __user *)buf; - local_iov->iov_len = count; - - result = ll_file_writev(file, local_iov, 1, ppos); - cl_env_put(env, &refcheck); - RETURN(result); -} - -#else /* AIO stuff */ static ssize_t ll_file_aio_write(struct kiocb *iocb, const struct iovec *iov, unsigned long nr_segs, loff_t pos) { @@ -1196,7 +1373,11 @@ static ssize_t ll_file_write(struct file *file, const char *buf, size_t count, local_iov->iov_len = count; init_sync_kiocb(kiocb, file); kiocb->ki_pos = *ppos; +#ifdef HAVE_KIOCB_KI_LEFT kiocb->ki_left = count; +#else + kiocb->ki_nbytes = count; +#endif result = ll_file_aio_write(kiocb, local_iov, 1, kiocb->ki_pos); *ppos = kiocb->ki_pos; @@ -1204,37 +1385,7 @@ static ssize_t ll_file_write(struct file *file, const char *buf, size_t count, cl_env_put(env, &refcheck); RETURN(result); } -#endif - - -#ifdef HAVE_KERNEL_SENDFILE -/* - * Send file content (through pagecache) somewhere with helper - */ -static ssize_t ll_file_sendfile(struct file *in_file, loff_t *ppos,size_t count, - read_actor_t actor, void *target) -{ - struct lu_env *env; - struct vvp_io_args *args; - ssize_t result; - int refcheck; - ENTRY; - - env = cl_env_get(&refcheck); - if (IS_ERR(env)) - RETURN(PTR_ERR(env)); - - args = vvp_env_args(env, IO_SENDFILE); - args->u.sendfile.via_target = target; - args->u.sendfile.via_actor = actor; - - result = ll_file_io_generic(env, args, in_file, CIT_READ, ppos, count); - cl_env_put(env, &refcheck); - RETURN(result); -} -#endif -#ifdef HAVE_KERNEL_SPLICE_READ /* * Send file content (through pagecache) somewhere with helper */ @@ -1260,7 +1411,6 @@ static ssize_t ll_file_splice_read(struct file *in_file, loff_t *ppos, cl_env_put(env, &refcheck); RETURN(result); } -#endif static int ll_lov_recreate(struct inode *inode, struct ost_id *oi, obd_count ost_idx) @@ -1278,7 +1428,7 @@ static int ll_lov_recreate(struct inode *inode, struct ost_id *oi, RETURN(-ENOMEM); lsm = ccc_inode_lsm_get(inode); - if (lsm == NULL) + if (!lsm_has_objects(lsm)) GOTO(out, rc = -ENOENT); lsm_size = sizeof(*lsm) + (sizeof(struct lov_oinfo) * @@ -1345,7 +1495,8 @@ static int ll_lov_recreate_fid(struct inode *inode, unsigned long arg) } int ll_lov_setstripe_ea_info(struct inode *inode, struct file *file, - int flags, struct lov_user_md *lum, int lum_size) + __u64 flags, struct lov_user_md *lum, + int lum_size) { struct lov_stripe_md *lsm = NULL; struct lookup_intent oit = {.it_op = IT_OPEN, .it_flags = flags}; @@ -1355,25 +1506,27 @@ int ll_lov_setstripe_ea_info(struct inode *inode, struct file *file, lsm = ccc_inode_lsm_get(inode); if (lsm != NULL) { ccc_inode_lsm_put(inode, lsm); - CDEBUG(D_IOCTL, "stripe already exists for ino %lu\n", - inode->i_ino); - RETURN(-EEXIST); + CDEBUG(D_IOCTL, "stripe already exists for inode "DFID"\n", + PFID(ll_inode2fid(inode))); + GOTO(out, rc = -EEXIST); } ll_inode_size_lock(inode); - rc = ll_intent_file_open(file, lum, lum_size, &oit); - if (rc) - GOTO(out, rc); - rc = oit.d.lustre.it_status; - if (rc < 0) - GOTO(out_req_free, rc); + rc = ll_intent_file_open(file, lum, lum_size, &oit); + if (rc) + GOTO(out_unlock, rc); + rc = oit.d.lustre.it_status; + if (rc < 0) + GOTO(out_req_free, rc); - ll_release_openhandle(file->f_dentry, &oit); + ll_release_openhandle(file->f_dentry, &oit); - out: +out_unlock: ll_inode_size_unlock(inode); ll_intent_release(&oit); ccc_inode_lsm_put(inode, lsm); +out: + cl_lov_delay_create_clear(&file->f_flags); RETURN(rc); out_req_free: ptlrpc_req_finished((struct ptlrpc_request *) oit.d.lustre.it_data); @@ -1391,9 +1544,9 @@ int ll_lov_getstripe_ea_info(struct inode *inode, const char *filename, struct md_op_data *op_data; int rc, lmmsize; - rc = ll_get_max_mdsize(sbi, &lmmsize); - if (rc) - RETURN(rc); + rc = ll_get_default_mdsize(sbi, &lmmsize); + if (rc) + RETURN(rc); op_data = ll_prep_md_op_data(NULL, inode, NULL, filename, strlen(filename), lmmsize, @@ -1434,6 +1587,12 @@ int ll_lov_getstripe_ea_info(struct inode *inode, const char *filename, * passing it to userspace. */ if (LOV_MAGIC != cpu_to_le32(LOV_MAGIC)) { + int stripe_count; + + stripe_count = le16_to_cpu(lmm->lmm_stripe_count); + if (le32_to_cpu(lmm->lmm_pattern) & LOV_PATTERN_F_RELEASED) + 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)) { @@ -1441,13 +1600,13 @@ int ll_lov_getstripe_ea_info(struct inode *inode, const char *filename, if (S_ISREG(body->mode)) lustre_swab_lov_user_md_objects( ((struct lov_user_md_v1 *)lmm)->lmm_objects, - ((struct lov_user_md_v1 *)lmm)->lmm_stripe_count); + 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->mode)) lustre_swab_lov_user_md_objects( ((struct lov_user_md_v3 *)lmm)->lmm_objects, - ((struct lov_user_md_v3 *)lmm)->lmm_stripe_count); + stripe_count); } } @@ -1461,7 +1620,7 @@ out: static int ll_lov_setea(struct inode *inode, struct file *file, unsigned long arg) { - int flags = MDS_OPEN_HAS_OBJS | FMODE_WRITE; + __u64 flags = MDS_OPEN_HAS_OBJS | FMODE_WRITE; struct lov_user_md *lump; int lum_size = sizeof(struct lov_user_md) + sizeof(struct lov_user_ost_data); @@ -1494,7 +1653,7 @@ static int ll_lov_setstripe(struct inode *inode, struct file *file, 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 lum_size, rc; - int flags = FMODE_WRITE; + __u64 flags = FMODE_WRITE; ENTRY; /* first try with v1 which is smaller than v3 */ @@ -1538,7 +1697,8 @@ static int ll_lov_getstripe(struct inode *inode, unsigned long arg) RETURN(rc); } -int ll_get_grouplock(struct inode *inode, struct file *file, unsigned long arg) +static int +ll_get_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); @@ -1644,26 +1804,25 @@ 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, - ll_i2info(inode), it, och); + ll_och_fill(ll_i2sbi(inode)->ll_md_exp, it, och); rc = ll_close_inode_openhandle(ll_i2sbi(inode)->ll_md_exp, - inode, och); - out: - /* this one is in place of ll_file_open */ - if (it_disposition(it, DISP_ENQ_OPEN_REF)) { - ptlrpc_req_finished(it->d.lustre.it_data); - it_clear_disposition(it, DISP_ENQ_OPEN_REF); - } - RETURN(rc); + inode, och, NULL); +out: + /* this one is in place of ll_file_open */ + if (it_disposition(it, DISP_ENQ_OPEN_REF)) { + ptlrpc_req_finished(it->d.lustre.it_data); + it_clear_disposition(it, DISP_ENQ_OPEN_REF); + } + RETURN(rc); } /** * Get size for inode for which FIEMAP mapping is requested. * Make the FIEMAP get_info call and returns the result. */ -int ll_do_fiemap(struct inode *inode, struct ll_user_fiemap *fiemap, - int num_bytes) +static int ll_do_fiemap(struct inode *inode, struct ll_user_fiemap *fiemap, + int num_bytes) { struct obd_export *exp = ll_i2dtexp(inode); struct lov_stripe_md *lsm = NULL; @@ -1819,11 +1978,12 @@ error: * This value is computed using stripe object version on OST. * Version is computed using server side locking. * - * @param extent_lock Take extent lock. Not needed if a process is already - * holding the OST object group locks. + * @param sync if do sync on the OST side; + * 0: no sync + * LL_DV_RD_FLUSH: flush dirty pages, LCK_PR on OSTs + * LL_DV_WR_FLUSH: drop all caching pages, LCK_PW on OSTs */ -int ll_data_version(struct inode *inode, __u64 *data_version, - int extent_lock) +int ll_data_version(struct inode *inode, __u64 *data_version, int flags) { struct lov_stripe_md *lsm = NULL; struct ll_sb_info *sbi = ll_i2sbi(inode); @@ -1833,30 +1993,76 @@ int ll_data_version(struct inode *inode, __u64 *data_version, /* If no stripe, we consider version is 0. */ lsm = ccc_inode_lsm_get(inode); - if (lsm == NULL) { + if (!lsm_has_objects(lsm)) { *data_version = 0; CDEBUG(D_INODE, "No object for inode\n"); - RETURN(0); + GOTO(out, rc = 0); + } + + OBD_ALLOC_PTR(obdo); + if (obdo == NULL) + GOTO(out, rc = -ENOMEM); + + rc = ll_lsm_getattr(lsm, sbi->ll_dt_exp, NULL, obdo, 0, flags); + if (rc == 0) { + if (!(obdo->o_valid & OBD_MD_FLDATAVERSION)) + rc = -EOPNOTSUPP; + else + *data_version = obdo->o_data_version; } - OBD_ALLOC_PTR(obdo); - if (obdo == NULL) { - ccc_inode_lsm_put(inode, lsm); - RETURN(-ENOMEM); - } + OBD_FREE_PTR(obdo); + EXIT; +out: + ccc_inode_lsm_put(inode, lsm); + RETURN(rc); +} + +/* + * Trigger a HSM release request for the provided inode. + */ +int ll_hsm_release(struct inode *inode) +{ + struct cl_env_nest nest; + struct lu_env *env; + struct obd_client_handle *och = NULL; + __u64 data_version = 0; + int rc; + ENTRY; + + CDEBUG(D_INODE, "%s: Releasing file "DFID".\n", + ll_get_fsname(inode->i_sb, NULL, 0), + PFID(&ll_i2info(inode)->lli_fid)); + + och = ll_lease_open(inode, NULL, FMODE_WRITE, MDS_OPEN_RELEASE); + if (IS_ERR(och)) + GOTO(out, rc = PTR_ERR(och)); + + /* Grab latest data_version and [am]time values */ + rc = ll_data_version(inode, &data_version, LL_DV_WR_FLUSH); + if (rc != 0) + GOTO(out, rc); + + env = cl_env_nested_get(&nest); + if (IS_ERR(env)) + GOTO(out, rc = PTR_ERR(env)); - rc = ll_lsm_getattr(lsm, sbi->ll_dt_exp, NULL, obdo, 0, extent_lock); - if (!rc) { - if (!(obdo->o_valid & OBD_MD_FLDATAVERSION)) - rc = -EOPNOTSUPP; - else - *data_version = obdo->o_data_version; - } + ll_merge_lvb(env, inode); + cl_env_nested_put(&nest, env); - OBD_FREE_PTR(obdo); - ccc_inode_lsm_put(inode, lsm); + /* Release the file. + * NB: lease lock handle is released in mdc_hsm_release_pack() because + * we still need it to pack l_remote_handle to MDT. */ + rc = ll_close_inode_openhandle(ll_i2sbi(inode)->ll_md_exp, inode, och, + &data_version); + och = NULL; - RETURN(rc); + EXIT; +out: + if (och != NULL && !IS_ERR(och)) /* close the file */ + ll_lease_close(och, inode, NULL); + + return rc; } struct ll_swap_stack { @@ -1886,8 +2092,8 @@ static int ll_swap_layouts(struct file *file1, struct file *file2, if (!S_ISREG(llss->inode2->i_mode)) GOTO(free, rc = -EINVAL); - if (ll_permission(llss->inode1, MAY_WRITE, NULL) || - ll_permission(llss->inode2, MAY_WRITE, NULL)) + if (inode_permission(llss->inode1, MAY_WRITE) || + inode_permission(llss->inode2, MAY_WRITE)) GOTO(free, rc = -EPERM); if (llss->inode2->i_sb != llss->inode1->i_sb) @@ -2020,15 +2226,96 @@ free: RETURN(rc); } -long ll_file_ioctl(struct file *file, unsigned int cmd, unsigned long arg) +static int ll_hsm_state_set(struct inode *inode, struct hsm_state_set *hss) +{ + struct md_op_data *op_data; + int rc; + + /* Non-root users are forbidden to set or clear flags which are + * NOT defined in HSM_USER_MASK. */ + if (((hss->hss_setmask | hss->hss_clearmask) & ~HSM_USER_MASK) && + !cfs_capable(CFS_CAP_SYS_ADMIN)) + RETURN(-EPERM); + + op_data = ll_prep_md_op_data(NULL, inode, NULL, NULL, 0, 0, + LUSTRE_OPC_ANY, hss); + if (IS_ERR(op_data)) + RETURN(PTR_ERR(op_data)); + + rc = obd_iocontrol(LL_IOC_HSM_STATE_SET, ll_i2mdexp(inode), + sizeof(*op_data), op_data, NULL); + + ll_finish_md_op_data(op_data); + + RETURN(rc); +} + +static int ll_hsm_import(struct inode *inode, struct file *file, + struct hsm_user_import *hui) +{ + struct hsm_state_set *hss = NULL; + struct iattr *attr = NULL; + int rc; + ENTRY; + + if (!S_ISREG(inode->i_mode)) + RETURN(-EINVAL); + + /* set HSM flags */ + OBD_ALLOC_PTR(hss); + if (hss == NULL) + GOTO(out, rc = -ENOMEM); + + hss->hss_valid = HSS_SETMASK | HSS_ARCHIVE_ID; + hss->hss_archive_id = hui->hui_archive_id; + hss->hss_setmask = HS_ARCHIVED | HS_EXISTS | HS_RELEASED; + rc = ll_hsm_state_set(inode, hss); + if (rc != 0) + GOTO(out, rc); + + OBD_ALLOC_PTR(attr); + if (attr == NULL) + GOTO(out, rc = -ENOMEM); + + attr->ia_mode = hui->hui_mode & (S_IRWXU | S_IRWXG | S_IRWXO); + attr->ia_mode |= S_IFREG; + attr->ia_uid = make_kuid(&init_user_ns, hui->hui_uid); + attr->ia_gid = make_kgid(&init_user_ns, hui->hui_gid); + attr->ia_size = hui->hui_size; + attr->ia_mtime.tv_sec = hui->hui_mtime; + attr->ia_mtime.tv_nsec = hui->hui_mtime_ns; + attr->ia_atime.tv_sec = hui->hui_atime; + attr->ia_atime.tv_nsec = hui->hui_atime_ns; + + attr->ia_valid = ATTR_SIZE | ATTR_MODE | ATTR_FORCE | + ATTR_UID | ATTR_GID | + ATTR_MTIME | ATTR_MTIME_SET | + ATTR_ATIME | ATTR_ATIME_SET; + + rc = ll_setattr_raw(file->f_dentry, attr, true); + if (rc == -ENODATA) + rc = 0; + +out: + if (hss != NULL) + OBD_FREE_PTR(hss); + + if (attr != NULL) + OBD_FREE_PTR(attr); + + RETURN(rc); +} + +static long +ll_file_ioctl(struct file *file, unsigned int cmd, unsigned long arg) { struct inode *inode = file->f_dentry->d_inode; struct ll_file_data *fd = LUSTRE_FPRIVATE(file); int flags, rc; ENTRY; - CDEBUG(D_VFSTRACE, "VFS Op:inode=%lu/%u(%p),cmd=%x\n", inode->i_ino, - inode->i_generation, inode, cmd); + CDEBUG(D_VFSTRACE, "VFS Op:inode="DFID"(%p), cmd=%x\n", + PFID(ll_inode2fid(inode)), inode, cmd); ll_stats_ops_tally(ll_i2sbi(inode), LPROC_LL_IOCTL, 1); /* asm-ppc{,64} declares TCGETS, et. al. as type 't' not 'T' */ @@ -2069,7 +2356,7 @@ long ll_file_ioctl(struct file *file, unsigned int cmd, unsigned long arg) struct file *file2; struct lustre_swap_layouts lsl; - if (cfs_copy_from_user(&lsl, (char *)arg, + if (copy_from_user(&lsl, (char *)arg, sizeof(struct lustre_swap_layouts))) RETURN(-EFAULT); @@ -2126,13 +2413,13 @@ long ll_file_ioctl(struct file *file, unsigned int cmd, unsigned long arg) RETURN(ll_fid2path(inode, (void *)arg)); case LL_IOC_DATA_VERSION: { struct ioc_data_version idv; - int rc; + int rc; if (copy_from_user(&idv, (char *)arg, sizeof(idv))) RETURN(-EFAULT); - rc = ll_data_version(inode, &idv.idv_version, - !(idv.idv_flags & LL_DV_NOFLUSH)); + idv.idv_flags &= LL_DV_RD_FLUSH | LL_DV_WR_FLUSH; + rc = ll_data_version(inode, &idv.idv_version, idv.idv_flags); if (rc == 0 && copy_to_user((char *) arg, &idv, sizeof(idv))) RETURN(-EFAULT); @@ -2182,37 +2469,19 @@ long ll_file_ioctl(struct file *file, unsigned int cmd, unsigned long arg) RETURN(rc); } case LL_IOC_HSM_STATE_SET: { - struct md_op_data *op_data; struct hsm_state_set *hss; int rc; OBD_ALLOC_PTR(hss); if (hss == NULL) RETURN(-ENOMEM); + if (copy_from_user(hss, (char *)arg, sizeof(*hss))) { OBD_FREE_PTR(hss); RETURN(-EFAULT); } - /* Non-root users are forbidden to set or clear flags which are - * NOT defined in HSM_USER_MASK. */ - if (((hss->hss_setmask | hss->hss_clearmask) & ~HSM_USER_MASK) - && !cfs_capable(CFS_CAP_SYS_ADMIN)) { - OBD_FREE_PTR(hss); - RETURN(-EPERM); - } - - op_data = ll_prep_md_op_data(NULL, inode, NULL, NULL, 0, 0, - LUSTRE_OPC_ANY, hss); - if (IS_ERR(op_data)) { - OBD_FREE_PTR(hss); - RETURN(PTR_ERR(op_data)); - } - - rc = obd_iocontrol(cmd, ll_i2mdexp(inode), sizeof(*op_data), - op_data, NULL); - - ll_finish_md_op_data(op_data); + rc = ll_hsm_state_set(inode, hss); OBD_FREE_PTR(hss); RETURN(rc); @@ -2236,13 +2505,114 @@ long ll_file_ioctl(struct file *file, unsigned int cmd, unsigned long arg) rc = obd_iocontrol(cmd, ll_i2mdexp(inode), sizeof(*op_data), op_data, NULL); - if (cfs_copy_to_user((char *)arg, hca, sizeof(*hca))) + if (copy_to_user((char *)arg, hca, sizeof(*hca))) rc = -EFAULT; ll_finish_md_op_data(op_data); OBD_FREE_PTR(hca); RETURN(rc); } + case LL_IOC_SET_LEASE: { + struct ll_inode_info *lli = ll_i2info(inode); + struct obd_client_handle *och = NULL; + bool lease_broken; + fmode_t mode = 0; + + switch (arg) { + case F_WRLCK: + if (!(file->f_mode & FMODE_WRITE)) + RETURN(-EPERM); + mode = FMODE_WRITE; + break; + case F_RDLCK: + if (!(file->f_mode & FMODE_READ)) + RETURN(-EPERM); + mode = FMODE_READ; + break; + case F_UNLCK: + mutex_lock(&lli->lli_och_mutex); + if (fd->fd_lease_och != NULL) { + och = fd->fd_lease_och; + fd->fd_lease_och = NULL; + } + mutex_unlock(&lli->lli_och_mutex); + + if (och != NULL) { + mode = och->och_flags &(FMODE_READ|FMODE_WRITE); + rc = ll_lease_close(och, inode, &lease_broken); + if (rc == 0 && lease_broken) + mode = 0; + } else { + rc = -ENOLCK; + } + + /* return the type of lease or error */ + RETURN(rc < 0 ? rc : (int)mode); + default: + RETURN(-EINVAL); + } + + CDEBUG(D_INODE, "Set lease with mode %d\n", mode); + + /* apply for lease */ + och = ll_lease_open(inode, file, mode, 0); + if (IS_ERR(och)) + RETURN(PTR_ERR(och)); + + rc = 0; + mutex_lock(&lli->lli_och_mutex); + if (fd->fd_lease_och == NULL) { + fd->fd_lease_och = och; + och = NULL; + } + mutex_unlock(&lli->lli_och_mutex); + if (och != NULL) { + /* impossible now that only excl is supported for now */ + ll_lease_close(och, inode, &lease_broken); + rc = -EBUSY; + } + RETURN(rc); + } + case LL_IOC_GET_LEASE: { + struct ll_inode_info *lli = ll_i2info(inode); + struct ldlm_lock *lock = NULL; + + rc = 0; + mutex_lock(&lli->lli_och_mutex); + if (fd->fd_lease_och != NULL) { + struct obd_client_handle *och = fd->fd_lease_och; + + lock = ldlm_handle2lock(&och->och_lease_handle); + if (lock != NULL) { + lock_res_and_lock(lock); + if (!ldlm_is_cancel(lock)) + rc = och->och_flags & + (FMODE_READ | FMODE_WRITE); + unlock_res_and_lock(lock); + LDLM_LOCK_PUT(lock); + } + } + mutex_unlock(&lli->lli_och_mutex); + RETURN(rc); + } + case LL_IOC_HSM_IMPORT: { + struct hsm_user_import *hui; + + OBD_ALLOC_PTR(hui); + if (hui == NULL) + RETURN(-ENOMEM); + + if (copy_from_user(hui, (void *)arg, sizeof(*hui))) { + OBD_FREE_PTR(hui); + RETURN(-EFAULT); + } + + rc = ll_hsm_import(inode, file, hui); + + OBD_FREE_PTR(hui); + RETURN(rc); + } + default: { int err; @@ -2323,7 +2693,7 @@ generic_file_llseek_size(struct file *file, loff_t offset, int origin, } #endif -loff_t ll_file_seek(struct file *file, loff_t offset, int origin) +static loff_t ll_file_seek(struct file *file, loff_t offset, int origin) { struct inode *inode = file->f_dentry->d_inode; loff_t retval, eof = 0; @@ -2331,8 +2701,8 @@ loff_t ll_file_seek(struct file *file, loff_t offset, int origin) ENTRY; retval = offset + ((origin == SEEK_END) ? i_size_read(inode) : (origin == SEEK_CUR) ? file->f_pos : 0); - CDEBUG(D_VFSTRACE, "VFS Op:inode=%lu/%u(%p), to=%llu=%#llx(%d)\n", - inode->i_ino, inode->i_generation, inode, retval, retval, + 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); @@ -2348,7 +2718,7 @@ loff_t ll_file_seek(struct file *file, loff_t offset, int origin) RETURN(retval); } -int ll_flush(struct file *file, fl_owner_t id) +static int ll_flush(struct file *file, fl_owner_t id) { struct inode *inode = file->f_dentry->d_inode; struct ll_inode_info *lli = ll_i2info(inode); @@ -2361,9 +2731,11 @@ int ll_flush(struct file *file, fl_owner_t id) * failed for pages in this mapping. */ rc = lli->lli_async_rc; lli->lli_async_rc = 0; - err = lov_read_and_clear_async_rc(lli->lli_clob); - if (rc == 0) - rc = err; + if (lli->lli_clob != NULL) { + err = lov_read_and_clear_async_rc(lli->lli_clob); + if (rc == 0) + rc = err; + } /* The application has been told write failure already. * Do not report failure again. */ @@ -2444,48 +2816,48 @@ int ll_fsync(struct file *file, int datasync) int ll_fsync(struct file *file, struct dentry *dentry, int datasync) { #endif - struct inode *inode = dentry->d_inode; - struct ll_inode_info *lli = ll_i2info(inode); - struct ptlrpc_request *req; - struct obd_capa *oc; - int rc, err; - ENTRY; + struct inode *inode = dentry->d_inode; + struct ll_inode_info *lli = ll_i2info(inode); + struct ptlrpc_request *req; + struct obd_capa *oc; + int rc, err; + ENTRY; - CDEBUG(D_VFSTRACE, "VFS Op:inode=%lu/%u(%p)\n", inode->i_ino, - inode->i_generation, inode); - ll_stats_ops_tally(ll_i2sbi(inode), LPROC_LL_FSYNC, 1); + CDEBUG(D_VFSTRACE, "VFS Op:inode="DFID"(%p)\n", + PFID(ll_inode2fid(inode)), inode); + ll_stats_ops_tally(ll_i2sbi(inode), LPROC_LL_FSYNC, 1); #ifdef HAVE_FILE_FSYNC_4ARGS rc = filemap_write_and_wait_range(inode->i_mapping, start, end); mutex_lock(&inode->i_mutex); #else - /* 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_fdatawait(inode->i_mapping); + /* 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_fdatawait(inode->i_mapping); #endif - /* catch async errors that were recorded back when async writeback - * failed for pages in this mapping. */ - if (!S_ISDIR(inode->i_mode)) { - err = lli->lli_async_rc; - lli->lli_async_rc = 0; - if (rc == 0) - rc = err; + /* catch async errors that were recorded back when async writeback + * failed for pages in this mapping. */ + if (!S_ISDIR(inode->i_mode)) { + err = lli->lli_async_rc; + lli->lli_async_rc = 0; + if (rc == 0) + rc = err; err = lov_read_and_clear_async_rc(lli->lli_clob); if (rc == 0) rc = err; - } + } - oc = ll_mdscapa_get(inode); - err = md_sync(ll_i2sbi(inode)->ll_md_exp, ll_inode2fid(inode), oc, - &req); - capa_put(oc); - if (!rc) - rc = err; - if (!err) - ptlrpc_req_finished(req); + oc = ll_mdscapa_get(inode); + err = md_fsync(ll_i2sbi(inode)->ll_md_exp, ll_inode2fid(inode), oc, + &req); + capa_put(oc); + if (!rc) + rc = err; + if (!err) + ptlrpc_req_finished(req); - if (datasync && S_ISREG(inode->i_mode)) { + if (S_ISREG(inode->i_mode)) { struct ll_file_data *fd = LUSTRE_FPRIVATE(file); err = cl_sync_file_range(inode, 0, OBD_OBJECT_EOF, @@ -2504,7 +2876,8 @@ int ll_fsync(struct file *file, struct dentry *dentry, int datasync) RETURN(rc); } -int ll_file_flock(struct file *file, int cmd, struct file_lock *file_lock) +static int +ll_file_flock(struct file *file, int cmd, struct file_lock *file_lock) { struct inode *inode = file->f_dentry->d_inode; struct ll_sb_info *sbi = ll_i2sbi(inode); @@ -2516,13 +2889,13 @@ int ll_file_flock(struct file *file, int cmd, struct file_lock *file_lock) struct md_op_data *op_data; struct lustre_handle lockh = {0}; ldlm_policy_data_t flock = {{0}}; - int flags = 0; + __u64 flags = 0; int rc; int rc2 = 0; ENTRY; - CDEBUG(D_VFSTRACE, "VFS Op:inode=%lu file_lock=%p\n", - inode->i_ino, 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); @@ -2607,9 +2980,10 @@ int ll_file_flock(struct file *file, int cmd, struct file_lock *file_lock) if (IS_ERR(op_data)) RETURN(PTR_ERR(op_data)); - CDEBUG(D_DLMTRACE, "inode=%lu, pid=%u, flags=%#x, mode=%u, " - "start="LPU64", end="LPU64"\n", inode->i_ino, flock.l_flock.pid, - flags, einfo.ei_mode, flock.l_flock.start, flock.l_flock.end); + CDEBUG(D_DLMTRACE, "inode="DFID", pid=%u, flags="LPX64", mode=%u, " + "start="LPU64", end="LPU64"\n", PFID(ll_inode2fid(inode)), + flock.l_flock.pid, flags, einfo.ei_mode, + flock.l_flock.start, flock.l_flock.end); rc = md_enqueue(sbi->ll_md_exp, &einfo, NULL, op_data, &lockh, &flock, 0, NULL /* req */, flags); @@ -2634,7 +3008,114 @@ int ll_file_flock(struct file *file, int cmd, struct file_lock *file_lock) RETURN(rc); } -int ll_file_noflock(struct file *file, int cmd, struct file_lock *file_lock) +int ll_get_fid_by_name(struct inode *parent, const char *name, + int namelen, struct lu_fid *fid) +{ + struct md_op_data *op_data = NULL; + struct mdt_body *body; + struct ptlrpc_request *req; + int rc; + ENTRY; + + op_data = ll_prep_md_op_data(NULL, parent, NULL, name, namelen, 0, + LUSTRE_OPC_ANY, NULL); + if (IS_ERR(op_data)) + RETURN(PTR_ERR(op_data)); + + op_data->op_valid = OBD_MD_FLID; + rc = md_getattr_name(ll_i2sbi(parent)->ll_md_exp, op_data, &req); + ll_finish_md_op_data(op_data); + if (rc < 0) + RETURN(rc); + + body = req_capsule_server_get(&req->rq_pill, &RMF_MDT_BODY); + if (body == NULL) + GOTO(out_req, rc = -EFAULT); + if (fid != NULL) + *fid = body->fid1; +out_req: + ptlrpc_req_finished(req); + RETURN(rc); +} + +int ll_migrate(struct inode *parent, struct file *file, int mdtidx, + const char *name, int namelen) +{ + struct dentry *dchild = NULL; + struct inode *child_inode = NULL; + struct md_op_data *op_data; + struct ptlrpc_request *request = NULL; + struct qstr qstr; + int rc; + ENTRY; + + CDEBUG(D_VFSTRACE, "migrate %s under "DFID" to MDT%04x\n", + name, PFID(ll_inode2fid(parent)), mdtidx); + + op_data = ll_prep_md_op_data(NULL, parent, NULL, name, namelen, + 0, LUSTRE_OPC_ANY, NULL); + if (IS_ERR(op_data)) + RETURN(PTR_ERR(op_data)); + + /* Get child FID first */ + qstr.hash = full_name_hash(name, namelen); + qstr.name = name; + qstr.len = namelen; + dchild = d_lookup(file->f_dentry, &qstr); + if (dchild != NULL && dchild->d_inode != NULL) { + op_data->op_fid3 = *ll_inode2fid(dchild->d_inode); + if (dchild->d_inode != NULL) { + child_inode = igrab(dchild->d_inode); + ll_invalidate_aliases(child_inode); + } + dput(dchild); + } else { + rc = ll_get_fid_by_name(parent, name, namelen, + &op_data->op_fid3); + if (rc != 0) + GOTO(out_free, rc); + } + + if (!fid_is_sane(&op_data->op_fid3)) { + CERROR("%s: migrate %s , but fid "DFID" is insane\n", + ll_get_fsname(parent->i_sb, NULL, 0), name, + PFID(&op_data->op_fid3)); + GOTO(out_free, rc = -EINVAL); + } + + rc = ll_get_mdt_idx_by_fid(ll_i2sbi(parent), &op_data->op_fid3); + if (rc < 0) + GOTO(out_free, rc); + + if (rc == mdtidx) { + CDEBUG(D_INFO, "%s:"DFID" is already on MDT%d.\n", name, + PFID(&op_data->op_fid3), mdtidx); + GOTO(out_free, rc = 0); + } + + op_data->op_mds = mdtidx; + op_data->op_cli_flags = CLI_MIGRATE; + rc = md_rename(ll_i2sbi(parent)->ll_md_exp, op_data, name, + namelen, name, namelen, &request); + if (rc == 0) + ll_update_times(request, parent); + + ptlrpc_req_finished(request); + if (rc != 0) + GOTO(out_free, rc); + +out_free: + if (child_inode != NULL) { + clear_nlink(child_inode); + iput(child_inode); + } + + ll_finish_md_op_data(op_data); + RETURN(rc); +} + +static int +ll_file_noflock(struct file *file, int cmd, struct file_lock *file_lock) { ENTRY; @@ -2693,7 +3174,8 @@ int ll_have_md_lock(struct inode *inode, __u64 *bits, ldlm_mode_t l_req_mode) } ldlm_mode_t ll_take_md_lock(struct inode *inode, __u64 bits, - struct lustre_handle *lockh, __u64 flags) + struct lustre_handle *lockh, __u64 flags, + ldlm_mode_t mode) { ldlm_policy_data_t policy = { .l_inodebits = {bits}}; struct lu_fid *fid; @@ -2703,10 +3185,10 @@ ldlm_mode_t ll_take_md_lock(struct inode *inode, __u64 bits, fid = &ll_i2info(inode)->lli_fid; CDEBUG(D_INFO, "trying to match res "DFID"\n", PFID(fid)); - rc = md_lock_match(ll_i2mdexp(inode), LDLM_FL_BLOCK_GRANTED|flags, - fid, LDLM_IBITS, &policy, - LCK_CR|LCK_CW|LCK_PR|LCK_PW, lockh); - RETURN(rc); + rc = md_lock_match(ll_i2mdexp(inode), LDLM_FL_BLOCK_GRANTED|flags, + fid, LDLM_IBITS, &policy, mode, lockh); + + RETURN(rc); } static int ll_inode_revalidate_fini(struct inode *inode, int rc) @@ -2720,16 +3202,16 @@ static int ll_inode_revalidate_fini(struct inode *inode, int rc) if (!S_ISREG(inode->i_mode) && !S_ISDIR(inode->i_mode)) return 0; } else if (rc != 0) { - CERROR("%s: revalidate FID "DFID" error: rc = %d\n", - ll_get_fsname(inode->i_sb, NULL, 0), - PFID(ll_inode2fid(inode)), rc); + CDEBUG_LIMIT((rc == -EACCES || rc == -EIDRM) ? D_INFO : D_ERROR, + "%s: revalidate FID "DFID" error: rc = %d\n", + ll_get_fsname(inode->i_sb, NULL, 0), + PFID(ll_inode2fid(inode)), rc); } return rc; } -int __ll_inode_revalidate_it(struct dentry *dentry, struct lookup_intent *it, - __u64 ibits) +static int __ll_inode_revalidate(struct dentry *dentry, __u64 ibits) { struct inode *inode = dentry->d_inode; struct ptlrpc_request *req = NULL; @@ -2739,8 +3221,8 @@ int __ll_inode_revalidate_it(struct dentry *dentry, struct lookup_intent *it, LASSERT(inode != NULL); - CDEBUG(D_VFSTRACE, "VFS Op:inode=%lu/%u(%p),name=%s\n", - inode->i_ino, inode->i_generation, inode, dentry->d_name.name); + CDEBUG(D_VFSTRACE, "VFS Op:inode="DFID"(%p),name=%s\n", + PFID(ll_inode2fid(inode)), inode, dentry->d_name.name); exp = ll_i2mdexp(inode); @@ -2755,7 +3237,7 @@ int __ll_inode_revalidate_it(struct dentry *dentry, struct lookup_intent *it, oit.it_op = IT_LOOKUP; /* Call getattr by fid, so do not provide name at all. */ - op_data = ll_prep_md_op_data(NULL, dentry->d_parent->d_inode, + op_data = ll_prep_md_op_data(NULL, dentry->d_inode, dentry->d_inode, NULL, 0, 0, LUSTRE_OPC_ANY, NULL); if (IS_ERR(op_data)) @@ -2794,12 +3276,12 @@ int __ll_inode_revalidate_it(struct dentry *dentry, struct lookup_intent *it, struct md_op_data *op_data; int ealen = 0; - if (S_ISREG(inode->i_mode)) { - rc = ll_get_max_mdsize(sbi, &ealen); - if (rc) - RETURN(rc); - valid |= OBD_MD_FLEASIZE | OBD_MD_FLMODEASIZE; - } + if (S_ISREG(inode->i_mode)) { + rc = ll_get_default_mdsize(sbi, &ealen); + if (rc) + RETURN(rc); + valid |= OBD_MD_FLEASIZE | OBD_MD_FLMODEASIZE; + } op_data = ll_prep_md_op_data(NULL, inode, NULL, NULL, 0, ealen, LUSTRE_OPC_ANY, @@ -2825,73 +3307,107 @@ out: return rc; } -int ll_inode_revalidate_it(struct dentry *dentry, struct lookup_intent *it, - __u64 ibits) +static int ll_merge_md_attr(struct inode *inode) { - struct inode *inode = dentry->d_inode; - int rc; - ENTRY; + struct cl_attr attr = { 0 }; + int rc; + + LASSERT(ll_i2info(inode)->lli_lsm_md != NULL); + rc = md_merge_attr(ll_i2mdexp(inode), ll_i2info(inode)->lli_lsm_md, + &attr); + if (rc != 0) + RETURN(rc); + + ll_i2info(inode)->lli_stripe_dir_size = attr.cat_size; + ll_i2info(inode)->lli_stripe_dir_nlink = attr.cat_nlink; + + ll_i2info(inode)->lli_lvb.lvb_atime = attr.cat_atime; + ll_i2info(inode)->lli_lvb.lvb_mtime = attr.cat_mtime; + ll_i2info(inode)->lli_lvb.lvb_ctime = attr.cat_ctime; - rc = __ll_inode_revalidate_it(dentry, it, ibits); + RETURN(0); +} + +static int +ll_inode_revalidate(struct dentry *dentry, __u64 ibits) +{ + struct inode *inode = dentry->d_inode; + int rc; + ENTRY; + + rc = __ll_inode_revalidate(dentry, ibits); if (rc != 0) RETURN(rc); /* if object isn't regular file, don't validate size */ if (!S_ISREG(inode->i_mode)) { + if (S_ISDIR(inode->i_mode) && + ll_i2info(inode)->lli_lsm_md != NULL) { + rc = ll_merge_md_attr(inode); + if (rc != 0) + RETURN(rc); + } + LTIME_S(inode->i_atime) = ll_i2info(inode)->lli_lvb.lvb_atime; LTIME_S(inode->i_mtime) = ll_i2info(inode)->lli_lvb.lvb_mtime; LTIME_S(inode->i_ctime) = ll_i2info(inode)->lli_lvb.lvb_ctime; } else { - rc = ll_glimpse_size(inode); + /* 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. + * Also to glimpse we need the layout, in case of a running + * restore the MDT holds the layout lock so the glimpse will + * block up to the end of restore (getattr will block) + */ + if (!(ll_i2info(inode)->lli_flags & LLIF_FILE_RESTORING)) + rc = ll_glimpse_size(inode); } - RETURN(rc); + RETURN(rc); } -int ll_getattr_it(struct vfsmount *mnt, struct dentry *de, - struct lookup_intent *it, struct kstat *stat) +int ll_getattr(struct vfsmount *mnt, struct dentry *de, struct kstat *stat) { struct inode *inode = de->d_inode; struct ll_sb_info *sbi = ll_i2sbi(inode); struct ll_inode_info *lli = ll_i2info(inode); int res = 0; - res = ll_inode_revalidate_it(de, it, MDS_INODELOCK_UPDATE | - MDS_INODELOCK_LOOKUP); + res = ll_inode_revalidate(de, MDS_INODELOCK_UPDATE | + MDS_INODELOCK_LOOKUP); ll_stats_ops_tally(sbi, LPROC_LL_GETATTR, 1); if (res) return res; - stat->dev = inode->i_sb->s_dev; - if (ll_need_32bit_api(sbi)) - stat->ino = cl_fid_build_ino(&lli->lli_fid, 1); - else - stat->ino = inode->i_ino; - stat->mode = inode->i_mode; - stat->nlink = inode->i_nlink; - stat->uid = inode->i_uid; - stat->gid = inode->i_gid; + stat->dev = inode->i_sb->s_dev; + if (ll_need_32bit_api(sbi)) + stat->ino = cl_fid_build_ino(&lli->lli_fid, 1); + else + stat->ino = inode->i_ino; + stat->mode = inode->i_mode; + stat->uid = inode->i_uid; + stat->gid = inode->i_gid; stat->rdev = inode->i_rdev; - stat->atime = inode->i_atime; - stat->mtime = inode->i_mtime; - stat->ctime = inode->i_ctime; + stat->atime = inode->i_atime; + stat->mtime = inode->i_mtime; + stat->ctime = inode->i_ctime; stat->blksize = 1 << inode->i_blkbits; + stat->blocks = inode->i_blocks; - stat->size = i_size_read(inode); - stat->blocks = inode->i_blocks; + if (S_ISDIR(inode->i_mode) && + ll_i2info(inode)->lli_lsm_md != NULL) { + stat->nlink = lli->lli_stripe_dir_nlink; + stat->size = lli->lli_stripe_dir_size; + } else { + stat->nlink = inode->i_nlink; + stat->size = i_size_read(inode); + } return 0; } -int ll_getattr(struct vfsmount *mnt, struct dentry *de, struct kstat *stat) -{ - struct lookup_intent it = { .it_op = IT_GETATTR }; - - return ll_getattr_it(mnt, de, &it, stat); -} -#ifdef HAVE_LINUX_FIEMAP_H -int ll_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, - __u64 start, __u64 len) +static int ll_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, + __u64 start, __u64 len) { int rc; size_t num_bytes; @@ -2922,9 +3438,8 @@ int ll_fiemap(struct inode *inode, struct fiemap_extent_info *fieinfo, OBD_FREE_LARGE(fiemap, num_bytes); return rc; } -#endif -struct posix_acl * ll_get_acl(struct inode *inode, int type) +struct posix_acl *ll_get_acl(struct inode *inode, int type) { struct ll_inode_info *lli = ll_i2info(inode); struct posix_acl *acl = NULL; @@ -2995,16 +3510,14 @@ int ll_inode_permission(struct inode *inode, int mask, struct nameidata *nd) * need to do it before permission check. */ if (inode == inode->i_sb->s_root->d_inode) { - struct lookup_intent it = { .it_op = IT_LOOKUP }; - - rc = __ll_inode_revalidate_it(inode->i_sb->s_root, &it, - MDS_INODELOCK_LOOKUP); + rc = __ll_inode_revalidate(inode->i_sb->s_root, + MDS_INODELOCK_LOOKUP); if (rc) RETURN(rc); } - CDEBUG(D_VFSTRACE, "VFS Op:inode=%lu/%u(%p), inode mode %x mask %o\n", - inode->i_ino, inode->i_generation, inode, inode->i_mode, mask); + CDEBUG(D_VFSTRACE, "VFS Op:inode="DFID"(%p), inode mode %x mask %o\n", + PFID(ll_inode2fid(inode)), inode, inode->i_mode, mask); if (ll_i2sbi(inode)->ll_flags & LL_SBI_RMT_CLIENT) return lustre_check_remote_perm(inode, mask); @@ -3015,55 +3528,33 @@ int ll_inode_permission(struct inode *inode, int mask, struct nameidata *nd) RETURN(rc); } -#ifdef HAVE_FILE_READV -#define READ_METHOD readv -#define READ_FUNCTION ll_file_readv -#define WRITE_METHOD writev -#define WRITE_FUNCTION ll_file_writev -#else -#define READ_METHOD aio_read -#define READ_FUNCTION ll_file_aio_read -#define WRITE_METHOD aio_write -#define WRITE_FUNCTION ll_file_aio_write -#endif - /* -o localflock - only provides locally consistent flock locks */ struct file_operations ll_file_operations = { .read = ll_file_read, - .READ_METHOD = READ_FUNCTION, + .aio_read = ll_file_aio_read, .write = ll_file_write, - .WRITE_METHOD = WRITE_FUNCTION, + .aio_write = ll_file_aio_write, .unlocked_ioctl = ll_file_ioctl, .open = ll_file_open, .release = ll_file_release, .mmap = ll_file_mmap, .llseek = ll_file_seek, -#ifdef HAVE_KERNEL_SENDFILE - .sendfile = ll_file_sendfile, -#endif -#ifdef HAVE_KERNEL_SPLICE_READ .splice_read = ll_file_splice_read, -#endif .fsync = ll_fsync, .flush = ll_flush }; struct file_operations ll_file_operations_flock = { .read = ll_file_read, - .READ_METHOD = READ_FUNCTION, + .aio_read = ll_file_aio_read, .write = ll_file_write, - .WRITE_METHOD = WRITE_FUNCTION, + .aio_write = ll_file_aio_write, .unlocked_ioctl = ll_file_ioctl, .open = ll_file_open, .release = ll_file_release, .mmap = ll_file_mmap, .llseek = ll_file_seek, -#ifdef HAVE_KERNEL_SENDFILE - .sendfile = ll_file_sendfile, -#endif -#ifdef HAVE_KERNEL_SPLICE_READ .splice_read = ll_file_splice_read, -#endif .fsync = ll_fsync, .flush = ll_flush, .flock = ll_file_flock, @@ -3073,20 +3564,15 @@ struct file_operations ll_file_operations_flock = { /* These are for -o noflock - to return ENOSYS on flock calls */ struct file_operations ll_file_operations_noflock = { .read = ll_file_read, - .READ_METHOD = READ_FUNCTION, + .aio_read = ll_file_aio_read, .write = ll_file_write, - .WRITE_METHOD = WRITE_FUNCTION, + .aio_write = ll_file_aio_write, .unlocked_ioctl = ll_file_ioctl, .open = ll_file_open, .release = ll_file_release, .mmap = ll_file_mmap, .llseek = ll_file_seek, -#ifdef HAVE_KERNEL_SENDFILE - .sendfile = ll_file_sendfile, -#endif -#ifdef HAVE_KERNEL_SPLICE_READ .splice_read = ll_file_splice_read, -#endif .fsync = ll_fsync, .flush = ll_flush, .flock = ll_file_noflock, @@ -3101,9 +3587,7 @@ struct inode_operations ll_file_inode_operations = { .getxattr = ll_getxattr, .listxattr = ll_listxattr, .removexattr = ll_removexattr, -#ifdef HAVE_LINUX_FIEMAP_H .fiemap = ll_fiemap, -#endif #ifdef HAVE_IOP_GET_ACL .get_acl = ll_get_acl, #endif @@ -3182,8 +3666,9 @@ void ll_iocontrol_unregister(void *magic) EXPORT_SYMBOL(ll_iocontrol_register); EXPORT_SYMBOL(ll_iocontrol_unregister); -enum llioc_iter ll_iocontrol_call(struct inode *inode, struct file *file, - unsigned int cmd, unsigned long arg, int *rcp) +static enum llioc_iter +ll_iocontrol_call(struct inode *inode, struct file *file, + unsigned int cmd, unsigned long arg, int *rcp) { enum llioc_iter ret = LLIOC_CONT; struct llioc_data *data; @@ -3233,11 +3718,24 @@ int ll_layout_conf(struct inode *inode, const struct cl_object_conf *conf) LASSERT(lock != NULL); LASSERT(ldlm_has_layout(lock)); if (result == 0) { + struct lustre_md *md = conf->u.coc_md; + __u32 gen = LL_LAYOUT_GEN_EMPTY; + /* it can only be allowed to match after layout is * applied to inode otherwise false layout would be * seen. Applying layout shoud happen before dropping * the intent lock. */ ldlm_lock_allow_match(lock); + + lli->lli_has_smd = lsm_has_objects(md->lsm); + if (md->lsm != NULL) + gen = md->lsm->lsm_layout_gen; + + CDEBUG(D_VFSTRACE, + DFID ": layout version change: %u -> %u\n", + PFID(&lli->lli_fid), ll_layout_version_get(lli), + gen); + ll_layout_version_set(lli, gen); } } RETURN(result); @@ -3257,7 +3755,11 @@ static int ll_layout_fetch(struct inode *inode, struct ldlm_lock *lock) int rc; ENTRY; - if (lock->l_lvb_data != NULL) + CDEBUG(D_INODE, DFID" LVB_READY=%d l_lvb_data=%p l_lvb_len=%d\n", + PFID(ll_inode2fid(inode)), ldlm_is_lvb_ready(lock), + lock->l_lvb_data, lock->l_lvb_len); + + if ((lock->l_lvb_data != NULL) && ldlm_is_lvb_ready(lock)) RETURN(0); /* if layout lock was granted right away, the layout is returned @@ -3266,7 +3768,7 @@ static int ll_layout_fetch(struct inode *inode, struct ldlm_lock *lock) * layout here. Please note that we can't use the LVB buffer in * completion AST because it doesn't have a large enough buffer */ oc = ll_mdscapa_get(inode); - rc = ll_get_max_mdsize(sbi, &lmmsize); + rc = ll_get_default_mdsize(sbi, &lmmsize); if (rc == 0) rc = md_getxattr(sbi->ll_md_exp, ll_inode2fid(inode), oc, OBD_MD_FLXATTR, XATTR_NAME_LOV, NULL, 0, @@ -3276,7 +3778,7 @@ static int ll_layout_fetch(struct inode *inode, struct ldlm_lock *lock) RETURN(rc); body = req_capsule_server_get(&req->rq_pill, &RMF_MDT_BODY); - if (body == NULL || body->eadatasize > lmmsize) + if (body == NULL) GOTO(out, rc = -EPROTO); lmmsize = body->eadatasize; @@ -3293,15 +3795,13 @@ static int ll_layout_fetch(struct inode *inode, struct ldlm_lock *lock) memcpy(lvbdata, lmm, lmmsize); lock_res_and_lock(lock); - if (lock->l_lvb_data == NULL) { - lock->l_lvb_data = lvbdata; - lock->l_lvb_len = lmmsize; - lvbdata = NULL; - } + if (lock->l_lvb_data != NULL) + OBD_FREE_LARGE(lock->l_lvb_data, lock->l_lvb_len); + + lock->l_lvb_data = lvbdata; + lock->l_lvb_len = lmmsize; unlock_res_and_lock(lock); - if (lvbdata != NULL) - OBD_FREE_LARGE(lvbdata, lmmsize); EXIT; out: @@ -3332,23 +3832,24 @@ static int ll_layout_lock_set(struct lustre_handle *lockh, ldlm_mode_t mode, LASSERT(lock != NULL); LASSERT(ldlm_has_layout(lock)); - LDLM_DEBUG(lock, "File %p/"DFID" being reconfigured: %d.\n", - inode, PFID(&lli->lli_fid), reconf); + LDLM_DEBUG(lock, "file "DFID"(%p) being reconfigured: %d\n", + PFID(&lli->lli_fid), inode, reconf); /* in case this is a caching lock and reinstate with new inode */ md_set_lock_data(sbi->ll_md_exp, &lockh->cookie, inode, NULL); lock_res_and_lock(lock); - lvb_ready = !!(lock->l_flags & LDLM_FL_LVB_READY); + lvb_ready = ldlm_is_lvb_ready(lock); unlock_res_and_lock(lock); /* checking lvb_ready is racy but this is okay. The worst case is * that multi processes may configure the file on the same time. */ + if (lvb_ready || !reconf) { rc = -ENODATA; if (lvb_ready) { /* layout_gen must be valid if layout lock is not * cancelled and stripe has already set */ - *gen = lli->lli_layout_gen; + *gen = ll_layout_version_get(lli); rc = 0; } GOTO(out, rc); @@ -3401,9 +3902,9 @@ out: /* wait for IO to complete if it's still being used. */ if (wait_layout) { - CDEBUG(D_INODE, "%s: %p/"DFID" wait for layout reconf.\n", - ll_get_fsname(inode->i_sb, NULL, 0), - inode, PFID(&lli->lli_fid)); + CDEBUG(D_INODE, "%s: "DFID"(%p) wait for layout reconf\n", + ll_get_fsname(inode->i_sb, NULL, 0), + PFID(&lli->lli_fid), inode); memset(&conf, 0, sizeof conf); conf.coc_opc = OBJECT_CONF_WAIT; @@ -3412,8 +3913,9 @@ out: if (rc == 0) rc = -EAGAIN; - CDEBUG(D_INODE, "file: "DFID" waiting layout return: %d.\n", - PFID(&lli->lli_fid), rc); + CDEBUG(D_INODE, "%s file="DFID" waiting layout return: %d\n", + ll_get_fsname(inode->i_sb, NULL, 0), + PFID(&lli->lli_fid), rc); } RETURN(rc); } @@ -3448,32 +3950,22 @@ int ll_layout_refresh(struct inode *inode, __u32 *gen) int rc; ENTRY; - *gen = lli->lli_layout_gen; - if (!(sbi->ll_flags & LL_SBI_LAYOUT_LOCK)) + *gen = ll_layout_version_get(lli); + if (!(sbi->ll_flags & LL_SBI_LAYOUT_LOCK) || *gen != LL_LAYOUT_GEN_NONE) RETURN(0); /* sanity checks */ LASSERT(fid_is_sane(ll_inode2fid(inode))); LASSERT(S_ISREG(inode->i_mode)); - /* mostly layout lock is caching on the local side, so try to match - * it before grabbing layout lock mutex. */ - mode = ll_take_md_lock(inode, MDS_INODELOCK_LAYOUT, &lockh, 0); - if (mode != 0) { /* hit cached lock */ - rc = ll_layout_lock_set(&lockh, mode, inode, gen, false); - if (rc == 0) - RETURN(0); - - /* better hold lli_layout_mutex to try again otherwise - * it will have starvation problem. */ - } - /* take layout lock mutex to enqueue layout lock exclusively. */ mutex_lock(&lli->lli_layout_mutex); again: - /* try again. Maybe somebody else has done this. */ - mode = ll_take_md_lock(inode, MDS_INODELOCK_LAYOUT, &lockh, 0); + /* mostly layout lock is caching on the local side, so try to match + * it before grabbing layout lock mutex. */ + mode = ll_take_md_lock(inode, MDS_INODELOCK_LAYOUT, &lockh, 0, + LCK_CR | LCK_CW | LCK_PR | LCK_PW); if (mode != 0) { /* hit cached lock */ rc = ll_layout_lock_set(&lockh, mode, inode, gen, true); if (rc == -EAGAIN) @@ -3484,7 +3976,7 @@ again: } op_data = ll_prep_md_op_data(NULL, inode, inode, NULL, - 0, 0, LUSTRE_OPC_ANY, NULL); + 0, 0, LUSTRE_OPC_ANY, NULL); if (IS_ERR(op_data)) { mutex_unlock(&lli->lli_layout_mutex); RETURN(PTR_ERR(op_data)); @@ -3495,9 +3987,9 @@ again: it.it_op = IT_LAYOUT; lockh.cookie = 0ULL; - LDLM_DEBUG_NOLOCK("%s: requeue layout lock for file %p/"DFID".\n", - ll_get_fsname(inode->i_sb, NULL, 0), inode, - PFID(&lli->lli_fid)); + LDLM_DEBUG_NOLOCK("%s: requeue layout lock for file "DFID"(%p)\n", + ll_get_fsname(inode->i_sb, NULL, 0), + PFID(&lli->lli_fid), inode); rc = md_enqueue(sbi->ll_md_exp, &einfo, &it, op_data, &lockh, NULL, 0, NULL, 0); @@ -3522,3 +4014,33 @@ again: RETURN(rc); } + +/** + * This function send a restore request to the MDT + */ +int ll_layout_restore(struct inode *inode, loff_t offset, __u64 length) +{ + struct hsm_user_request *hur; + int len, rc; + ENTRY; + + len = sizeof(struct hsm_user_request) + + sizeof(struct hsm_user_item); + OBD_ALLOC(hur, len); + if (hur == NULL) + RETURN(-ENOMEM); + + hur->hur_request.hr_action = HUA_RESTORE; + hur->hur_request.hr_archive_id = 0; + hur->hur_request.hr_flags = 0; + memcpy(&hur->hur_user_item[0].hui_fid, &ll_i2info(inode)->lli_fid, + sizeof(hur->hur_user_item[0].hui_fid)); + hur->hur_user_item[0].hui_extent.offset = offset; + hur->hur_user_item[0].hui_extent.length = length; + hur->hur_request.hr_itemcount = 1; + rc = obd_iocontrol(LL_IOC_HSM_REQUEST, cl_i2sbi(inode)->ll_md_exp, + len, hur, NULL); + OBD_FREE(hur, len); + RETURN(rc); +} +