X-Git-Url: https://git.whamcloud.com/?a=blobdiff_plain;f=lustre%2Fllite%2Ffile.c;h=b4f3779b671d1374dbf7c7419c0b0b32d0aaa0d8;hb=57f055f8d0df80e140724b00d1729f454222a83a;hp=dabc5d1311535c8e8b4f7205207f7d25006ae740;hpb=f2a404d1fec2287ef9ffda105727e8cd3f8e0b7b;p=fs%2Flustre-release.git diff --git a/lustre/llite/file.c b/lustre/llite/file.c index dabc5d1..b4f3779 100644 --- a/lustre/llite/file.c +++ b/lustre/llite/file.c @@ -27,7 +27,7 @@ * Copyright (c) 2002, 2010, Oracle and/or its affiliates. All rights reserved. * Use is subject to license terms. * - * Copyright (c) 2011, 2013, Intel Corporation. + * Copyright (c) 2011, 2015, Intel Corporation. */ /* * This file is part of Lustre, http://www.lustre.org/ @@ -42,15 +42,32 @@ #define DEBUG_SUBSYSTEM S_LLITE #include -#include #include #include -#include "llite_internal.h" +#include +#include +#ifdef HAVE_UIDGID_HEADER +# include +#endif #include +#include #include "cl_object.h" -struct ll_file_data *ll_file_data_get(void) +#include "llite_internal.h" +#include "vvp_internal.h" + +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; @@ -69,153 +86,117 @@ static void ll_file_data_put(struct ll_file_data *fd) OBD_SLAB_FREE_PTR(fd, ll_file_data_slab); } -void ll_pack_inode2opdata(struct inode *inode, struct md_op_data *op_data, - struct lustre_handle *fh) -{ - op_data->op_fid1 = ll_i2info(inode)->lli_fid; - op_data->op_attr.ia_mode = inode->i_mode; - op_data->op_attr.ia_atime = inode->i_atime; - op_data->op_attr.ia_mtime = inode->i_mtime; - op_data->op_attr.ia_ctime = inode->i_ctime; - op_data->op_attr.ia_size = i_size_read(inode); - op_data->op_attr_blocks = inode->i_blocks; - ((struct ll_iattr *)&op_data->op_attr)->ia_attr_flags = - ll_inode_to_ext_flags(inode->i_flags); - op_data->op_ioepoch = ll_i2info(inode)->lli_ioepoch; - if (fh) - op_data->op_handle = *fh; - op_data->op_capa1 = ll_mdscapa_get(inode); - - if (LLIF_DATA_MODIFIED & ll_i2info(inode)->lli_flags) - op_data->op_bias |= MDS_DATA_MODIFIED; -} - /** - * Closes the IO epoch and packs all the attributes into @op_data for - * the CLOSE rpc. + * Packs all the attributes into @op_data for the CLOSE rpc. */ static void ll_prepare_close(struct inode *inode, struct md_op_data *op_data, struct obd_client_handle *och) { - ENTRY; - - op_data->op_attr.ia_valid = ATTR_MODE | ATTR_ATIME | ATTR_ATIME_SET | - ATTR_MTIME | ATTR_MTIME_SET | - ATTR_CTIME | ATTR_CTIME_SET; - - if (!(och->och_flags & FMODE_WRITE)) - goto out; + ENTRY; - if (!exp_connect_som(ll_i2mdexp(inode)) || !S_ISREG(inode->i_mode)) - op_data->op_attr.ia_valid |= ATTR_SIZE | ATTR_BLOCKS; - else - ll_ioepoch_close(inode, op_data, &och, 0); + ll_prep_md_op_data(op_data, inode, NULL, NULL, + 0, 0, LUSTRE_OPC_ANY, NULL); + + op_data->op_attr.ia_mode = inode->i_mode; + op_data->op_attr.ia_atime = inode->i_atime; + op_data->op_attr.ia_mtime = inode->i_mtime; + op_data->op_attr.ia_ctime = inode->i_ctime; + op_data->op_attr.ia_size = i_size_read(inode); + op_data->op_attr.ia_valid |= ATTR_MODE | ATTR_ATIME | ATTR_ATIME_SET | + ATTR_MTIME | ATTR_MTIME_SET | + ATTR_CTIME | ATTR_CTIME_SET; + op_data->op_attr_blocks = inode->i_blocks; + op_data->op_attr_flags = ll_inode_to_ext_flags(inode->i_flags); + op_data->op_handle = och->och_fh; + + if (och->och_flags & FMODE_WRITE && + ll_file_test_and_clear_flag(ll_i2info(inode), LLIF_DATA_MODIFIED)) + /* For HSM: if inode data has been modified, pack it so that + * MDT can set data dirty flag in the archive. */ + op_data->op_bias |= MDS_DATA_MODIFIED; -out: - ll_pack_inode2opdata(inode, op_data, &och->och_fh); - ll_prep_md_op_data(op_data, inode, NULL, NULL, - 0, 0, LUSTRE_OPC_ANY, NULL); - EXIT; + EXIT; } -static int ll_close_inode_openhandle(struct obd_export *md_exp, - struct inode *inode, +/** + * Perform a close, possibly with a bias. + * The meaning of "data" depends on the value of "bias". + * + * If \a bias is MDS_HSM_RELEASE then \a data is a pointer to the data version. + * If \a bias is MDS_CLOSE_LAYOUT_SWAP then \a data is a pointer to the inode to + * swap layouts with. + */ +static int ll_close_inode_openhandle(struct inode *inode, struct obd_client_handle *och, - const __u64 *data_version) + enum mds_op_bias bias, void *data) { - struct obd_export *exp = ll_i2mdexp(inode); - struct md_op_data *op_data; - struct ptlrpc_request *req = NULL; - struct obd_device *obd = class_exp2obd(exp); - int epoch_close = 1; - int rc; - ENTRY; + struct obd_export *md_exp = ll_i2mdexp(inode); + const struct ll_inode_info *lli = ll_i2info(inode); + struct md_op_data *op_data; + struct ptlrpc_request *req = NULL; + int rc; + ENTRY; - if (obd == NULL) { - /* - * XXX: in case of LMV, is this correct to access - * ->exp_handle? - */ - CERROR("Invalid MDC connection handle "LPX64"\n", - ll_i2mdexp(inode)->exp_handle.h_cookie); - GOTO(out, rc = 0); - } + if (class_exp2obd(md_exp) == NULL) { + CERROR("%s: invalid MDC connection handle closing "DFID"\n", + ll_get_fsname(inode->i_sb, NULL, 0), + PFID(&lli->lli_fid)); + GOTO(out, rc = 0); + } - OBD_ALLOC_PTR(op_data); - if (op_data == NULL) - GOTO(out, rc = -ENOMEM); // XXX We leak openhandle and request here. + OBD_ALLOC_PTR(op_data); + /* We leak openhandle and request here on error, but not much to be + * done in OOM case since app won't retry close on error either. */ + if (op_data == NULL) + GOTO(out, rc = -ENOMEM); ll_prepare_close(inode, op_data, och); - if (data_version != NULL) { - /* Pass in data_version implies release. */ + switch (bias) { + case MDS_CLOSE_LAYOUT_SWAP: + LASSERT(data != NULL); + op_data->op_bias |= MDS_CLOSE_LAYOUT_SWAP; + op_data->op_data_version = 0; + op_data->op_lease_handle = och->och_lease_handle; + op_data->op_fid2 = *ll_inode2fid(data); + break; + + case MDS_HSM_RELEASE: + LASSERT(data != NULL); op_data->op_bias |= MDS_HSM_RELEASE; - op_data->op_data_version = *data_version; + op_data->op_data_version = *(__u64 *)data; 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) { - /* This close must have the epoch closed. */ - LASSERT(epoch_close); - /* MDS has instructed us to obtain Size-on-MDS attribute from - * OSTs and send setattr to back to MDS. */ - rc = ll_som_update(inode, op_data); - if (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. */ - if (rc == 0 && (op_data->op_bias & MDS_DATA_MODIFIED)) { - struct ll_inode_info *lli = ll_i2info(inode); + break; - spin_lock(&lli->lli_lock); - lli->lli_flags &= ~LLIF_DATA_MODIFIED; - spin_unlock(&lli->lli_lock); + default: + LASSERT(data == NULL); + break; } - if (rc == 0) { - rc = ll_objects_destroy(req, inode); - if (rc) - CERROR("%s: inode "DFID - " ll_objects destroy: rc = %d\n", - ll_i2mdexp(inode)->exp_obd->obd_name, - PFID(ll_inode2fid(inode)), rc); - } + rc = md_close(md_exp, op_data, och->och_mod, &req); + if (rc != 0 && rc != -EINTR) + CERROR("%s: inode "DFID" mdc close failed: rc = %d\n", + md_exp->exp_obd->obd_name, PFID(&lli->lli_fid), rc); - if (rc == 0 && op_data->op_bias & MDS_HSM_RELEASE) { + if (rc == 0 && + op_data->op_bias & (MDS_HSM_RELEASE | MDS_CLOSE_LAYOUT_SWAP)) { struct mdt_body *body; + body = req_capsule_server_get(&req->rq_pill, &RMF_MDT_BODY); - if (!(body->valid & OBD_MD_FLRELEASED)) + if (!(body->mbo_valid & OBD_MD_CLOSE_INTENT_EXECED)) rc = -EBUSY; } - ll_finish_md_op_data(op_data); - EXIT; + ll_finish_md_op_data(op_data); + EXIT; out: - if (exp_connect_som(exp) && !epoch_close && - S_ISREG(inode->i_mode) && (och->och_flags & FMODE_WRITE)) { - ll_queue_done_writing(inode, LLIF_DONE_WRITING); - } else { - md_clear_open_replay_data(md_exp, och); - /* Free @och if it is not waiting for DONE_WRITING. */ - och->och_fh.cookie = DEAD_HANDLE_MAGIC; - OBD_FREE_PTR(och); - } - if (req) /* This is close request */ - ptlrpc_req_finished(req); - return rc; + md_clear_open_replay_data(md_exp, och); + och->och_fh.cookie = DEAD_HANDLE_MAGIC; + OBD_FREE_PTR(och); + + ptlrpc_req_finished(req); /* This is close request */ + return rc; } int ll_md_real_close(struct inode *inode, fmode_t fmode) @@ -254,24 +235,28 @@ int ll_md_real_close(struct inode *inode, fmode_t fmode) 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); + rc = ll_close_inode_openhandle(inode, och, 0, NULL); } RETURN(rc); } -int ll_md_close(struct obd_export *md_exp, struct inode *inode, - struct file *file) +static int ll_md_close(struct inode *inode, struct file *file) { - struct ll_file_data *fd = LUSTRE_FPRIVATE(file); - struct ll_inode_info *lli = ll_i2info(inode); - int rc = 0; - ENTRY; + union ldlm_policy_data policy = { + .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_inode_info *lli = ll_i2info(inode); + struct lustre_handle lockh; + enum ldlm_mode lockmode; + int rc = 0; + ENTRY; - /* clear group lock, if present */ - if (unlikely(fd->fd_flags & LL_FILE_GROUP_LOCKED)) - ll_put_grouplock(inode, file, fd->fd_grouplock.cg_gid); + /* clear group lock, if present */ + if (unlikely(fd->fd_flags & LL_FILE_GROUP_LOCKED)) + ll_put_grouplock(inode, file, fd->fd_grouplock.lg_gid); if (fd->fd_lease_och != NULL) { bool lease_broken; @@ -286,51 +271,36 @@ int ll_md_close(struct obd_export *md_exp, struct inode *inode, } if (fd->fd_och != NULL) { - rc = ll_close_inode_openhandle(md_exp, inode, fd->fd_och, NULL); + rc = ll_close_inode_openhandle(inode, fd->fd_och, 0, 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; - __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}}; - - mutex_lock(&lli->lli_och_mutex); - if (fd->fd_omode & FMODE_WRITE) { - lockmode = LCK_CW; - LASSERT(lli->lli_open_fd_write_count); - lli->lli_open_fd_write_count--; - } else if (fd->fd_omode & FMODE_EXEC) { - lockmode = LCK_PR; - LASSERT(lli->lli_open_fd_exec_count); - lli->lli_open_fd_exec_count--; - } else { - lockmode = LCK_CR; - LASSERT(lli->lli_open_fd_read_count); - lli->lli_open_fd_read_count--; - } - mutex_unlock(&lli->lli_och_mutex); + mutex_lock(&lli->lli_och_mutex); + if (fd->fd_omode & FMODE_WRITE) { + lockmode = LCK_CW; + LASSERT(lli->lli_open_fd_write_count); + lli->lli_open_fd_write_count--; + } else if (fd->fd_omode & FMODE_EXEC) { + lockmode = LCK_PR; + LASSERT(lli->lli_open_fd_exec_count); + lli->lli_open_fd_exec_count--; + } else { + lockmode = LCK_CR; + LASSERT(lli->lli_open_fd_read_count); + lli->lli_open_fd_read_count--; + } + mutex_unlock(&lli->lli_och_mutex); - if (!md_lock_match(md_exp, flags, ll_inode2fid(inode), - LDLM_IBITS, &policy, lockmode, - &lockh)) { - rc = ll_md_real_close(file->f_dentry->d_inode, - fd->fd_omode); - } - } else { - CERROR("Releasing a file %p with negative dentry %p. Name %s", - file, file->f_dentry, file->f_dentry->d_name.name); - } + if (!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; ll_file_data_put(fd); - ll_capa_close(inode); RETURN(rc); } @@ -365,122 +335,99 @@ int ll_file_release(struct inode *inode, struct file *file) } #endif - if (inode->i_sb->s_root != file->f_dentry) + if (inode->i_sb->s_root != file->f_path.dentry) ll_stats_ops_tally(sbi, LPROC_LL_RELEASE, 1); fd = LUSTRE_FPRIVATE(file); LASSERT(fd != NULL); - /* The last ref on @file, maybe not the the owner pid of statahead. - * Different processes can open the same dir, "ll_opendir_key" means: - * it is me that should stop the statahead thread. */ - if (S_ISDIR(inode->i_mode) && lli->lli_opendir_key == fd && - lli->lli_opendir_pid != 0) - ll_stop_statahead(inode, lli->lli_opendir_key); + /* The last ref on @file, maybe not the the owner pid of statahead, + * because parent and child process can share the same file handle. */ + if (S_ISDIR(inode->i_mode) && lli->lli_opendir_key == fd) + ll_deauthorize_statahead(inode, fd); - if (inode->i_sb->s_root == file->f_dentry) { - LUSTRE_FPRIVATE(file) = NULL; - ll_file_data_put(fd); - RETURN(0); - } + if (inode->i_sb->s_root == file->f_path.dentry) { + LUSTRE_FPRIVATE(file) = NULL; + ll_file_data_put(fd); + RETURN(0); + } - if (!S_ISDIR(inode->i_mode)) { - lov_read_and_clear_async_rc(lli->lli_clob); - lli->lli_async_rc = 0; - } + if (!S_ISDIR(inode->i_mode)) { + if (lli->lli_clob != NULL) + lov_read_and_clear_async_rc(lli->lli_clob); + lli->lli_async_rc = 0; + } - rc = ll_md_close(sbi->ll_md_exp, inode, file); + rc = ll_md_close(inode, file); - if (CFS_FAIL_TIMEOUT_MS(OBD_FAIL_PTLRPC_DUMP_LOG, cfs_fail_val)) - libcfs_debug_dumplog(); + if (CFS_FAIL_TIMEOUT_MS(OBD_FAIL_PTLRPC_DUMP_LOG, cfs_fail_val)) + libcfs_debug_dumplog(); - RETURN(rc); + RETURN(rc); } -static int ll_intent_file_open(struct file *file, void *lmm, - int lmmsize, struct lookup_intent *itp) +static int ll_intent_file_open(struct file *file, void *lmm, int lmmsize, + struct lookup_intent *itp) { - struct ll_sb_info *sbi = ll_i2sbi(file->f_dentry->d_inode); - struct dentry *parent = file->f_dentry->d_parent; - const char *name = file->f_dentry->d_name.name; - const int len = file->f_dentry->d_name.len; - struct md_op_data *op_data; - struct ptlrpc_request *req; - __u32 opc = LUSTRE_OPC_ANY; - int rc; - ENTRY; + struct dentry *de = file->f_path.dentry; + struct ll_sb_info *sbi = ll_i2sbi(de->d_inode); + struct dentry *parent = de->d_parent; + const char *name = NULL; + int len = 0; + struct md_op_data *op_data; + struct ptlrpc_request *req = NULL; + int rc; + ENTRY; - if (!parent) - RETURN(-ENOENT); - - /* Usually we come here only for NFSD, and we want open lock. - But we can also get here with pre 2.6.15 patchless kernels, and in - that case that lock is also ok */ - /* We can also get here if there was cached open handle in revalidate_it - * but it disappeared while we were getting from there to ll_file_open. - * But this means this file was closed and immediatelly opened which - * makes a good candidate for using OPEN lock */ - /* If lmmsize & lmm are not 0, we are just setting stripe info - * parameters. No need for the open lock */ - if (lmm == NULL && lmmsize == 0) { - itp->it_flags |= MDS_OPEN_LOCK; - if (itp->it_flags & FMODE_WRITE) - opc = LUSTRE_OPC_CREATE; - } + LASSERT(parent != NULL); + LASSERT(itp->it_flags & MDS_OPEN_BY_FID); - op_data = ll_prep_md_op_data(NULL, parent->d_inode, - file->f_dentry->d_inode, name, len, - O_RDWR, opc, NULL); - if (IS_ERR(op_data)) - RETURN(PTR_ERR(op_data)); + /* if server supports open-by-fid, or file name is invalid, don't pack + * name in open request */ + if (!(exp_connect_flags(sbi->ll_md_exp) & OBD_CONNECT_OPEN_BY_FID) && + lu_name_is_valid_2(de->d_name.name, de->d_name.len)) { + name = de->d_name.name; + len = de->d_name.len; + } - itp->it_flags |= MDS_OPEN_BY_FID; - rc = md_intent_lock(sbi->ll_md_exp, op_data, lmm, lmmsize, itp, - 0 /*unused */, &req, ll_md_blocking_ast, 0); - ll_finish_md_op_data(op_data); - if (rc == -ESTALE) { - /* reason for keep own exit path - don`t flood log - * with messages with -ESTALE errors. - */ - if (!it_disposition(itp, DISP_OPEN_OPEN) || - it_open_error(DISP_OPEN_OPEN, itp)) - GOTO(out, rc); - ll_release_openhandle(file->f_dentry, itp); - GOTO(out, rc); - } + op_data = ll_prep_md_op_data(NULL, parent->d_inode, de->d_inode, + name, len, 0, LUSTRE_OPC_ANY, NULL); + if (IS_ERR(op_data)) + RETURN(PTR_ERR(op_data)); + op_data->op_data = lmm; + op_data->op_data_size = lmmsize; - if (it_disposition(itp, DISP_LOOKUP_NEG)) - GOTO(out, rc = -ENOENT); + rc = md_intent_lock(sbi->ll_md_exp, op_data, itp, &req, + &ll_md_blocking_ast, 0); + ll_finish_md_op_data(op_data); + if (rc == -ESTALE) { + /* reason for keep own exit path - don`t flood log + * with messages with -ESTALE errors. + */ + if (!it_disposition(itp, DISP_OPEN_OPEN) || + it_open_error(DISP_OPEN_OPEN, itp)) + GOTO(out, rc); + ll_release_openhandle(de, itp); + GOTO(out, rc); + } - if (rc != 0 || it_open_error(DISP_OPEN_OPEN, itp)) { - rc = rc ? rc : it_open_error(DISP_OPEN_OPEN, itp); - CDEBUG(D_VFSTRACE, "lock enqueue: err: %d\n", rc); - GOTO(out, rc); - } + if (it_disposition(itp, DISP_LOOKUP_NEG)) + GOTO(out, rc = -ENOENT); + + if (rc != 0 || it_open_error(DISP_OPEN_OPEN, itp)) { + rc = rc ? rc : it_open_error(DISP_OPEN_OPEN, itp); + CDEBUG(D_VFSTRACE, "lock enqueue: err: %d\n", rc); + GOTO(out, rc); + } - rc = ll_prep_inode(&file->f_dentry->d_inode, req, NULL, itp); - if (!rc && itp->d.lustre.it_lock_mode) - ll_set_lock_data(sbi->ll_md_exp, file->f_dentry->d_inode, - itp, NULL); + rc = ll_prep_inode(&de->d_inode, req, NULL, itp); + if (!rc && itp->d.lustre.it_lock_mode) + ll_set_lock_data(sbi->ll_md_exp, de->d_inode, itp, NULL); out: ptlrpc_req_finished(req); - ll_intent_drop_lock(itp); - - RETURN(rc); -} + ll_intent_drop_lock(itp); -/** - * Assign an obtained @ioepoch to client's inode. No lock is needed, MDS does - * not believe attributes if a few ioepoch holders exist. Attributes for - * previous ioepoch if new one is opened are also skipped by MDS. - */ -void ll_ioepoch_open(struct ll_inode_info *lli, __u64 ioepoch) -{ - if (ioepoch && lli->lli_ioepoch != ioepoch) { - lli->lli_ioepoch = ioepoch; - CDEBUG(D_INODE, "Epoch "LPU64" opened on "DFID"\n", - ioepoch, PFID(&lli->lli_fid)); - } + RETURN(rc); } static int ll_och_fill(struct obd_export *md_exp, struct lookup_intent *it, @@ -490,8 +437,8 @@ static int ll_och_fill(struct obd_export *md_exp, struct lookup_intent *it, struct mdt_body *body; body = req_capsule_server_get(&req->rq_pill, &RMF_MDT_BODY); - och->och_fh = body->handle; - och->och_fid = body->fid1; + och->och_fh = body->mbo_handle; + och->och_fid = body->mbo_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; @@ -499,34 +446,32 @@ static int ll_och_fill(struct obd_export *md_exp, struct lookup_intent *it, 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); - ENTRY; + struct inode *inode = file->f_path.dentry->d_inode; + ENTRY; - LASSERT(!LUSTRE_FPRIVATE(file)); + LASSERT(!LUSTRE_FPRIVATE(file)); - LASSERT(fd != NULL); + LASSERT(fd != NULL); - if (och) { - struct ptlrpc_request *req = it->d.lustre.it_data; - struct mdt_body *body; - int rc; + if (och) { + int 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); - 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); + /* ll_cl_context initialize */ + rwlock_init(&fd->fd_lock); + INIT_LIST_HEAD(&fd->fd_lccs); + RETURN(0); } @@ -545,38 +490,30 @@ int ll_local_open(struct file *file, struct lookup_intent *it, */ int ll_file_open(struct inode *inode, struct file *file) { - struct ll_inode_info *lli = ll_i2info(inode); - struct lookup_intent *it, oit = { .it_op = IT_OPEN, - .it_flags = file->f_flags }; - struct obd_client_handle **och_p = NULL; - __u64 *och_usecount = NULL; - struct ll_file_data *fd; - int rc = 0, opendir_set = 0; - ENTRY; + struct ll_inode_info *lli = ll_i2info(inode); + struct lookup_intent *it, oit = { .it_op = IT_OPEN, + .it_flags = file->f_flags }; + struct obd_client_handle **och_p = NULL; + __u64 *och_usecount = NULL; + struct ll_file_data *fd; + int rc = 0; + ENTRY; 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 */ + it = file->private_data; /* XXX: compat macro */ + file->private_data = NULL; /* prevent ll_local_open assertion */ fd = ll_file_data_get(); if (fd == NULL) GOTO(out_openerr, rc = -ENOMEM); fd->fd_file = file; - if (S_ISDIR(inode->i_mode)) { - spin_lock(&lli->lli_sa_lock); - if (lli->lli_opendir_key == NULL && lli->lli_sai == NULL && - lli->lli_opendir_pid == 0) { - lli->lli_opendir_key = fd; - lli->lli_opendir_pid = current_pid(); - opendir_set = 1; - } - spin_unlock(&lli->lli_sa_lock); - } + if (S_ISDIR(inode->i_mode)) + ll_authorize_statahead(inode, fd); - if (inode->i_sb->s_root == file->f_dentry) { + if (inode->i_sb->s_root == file->f_path.dentry) { LUSTRE_FPRIVATE(file) = fd; RETURN(0); } @@ -634,7 +571,7 @@ restart: GOTO(out_openerr, rc); } - ll_release_openhandle(file->f_dentry, it); + ll_release_openhandle(file->f_path.dentry, it); } (*och_usecount)++; @@ -653,9 +590,20 @@ restart: would attempt to grab och_mutex as well, that would result in a deadlock */ mutex_unlock(&lli->lli_och_mutex); - it->it_create_mode |= M_CHECK_STALE; + /* + * Normally called under two situations: + * 1. NFS export. + * 2. A race/condition on MDS resulting in no open + * handle to be returned from LOOKUP|OPEN request, + * for example if the target entry was a symlink. + * + * Always fetch MDS_OPEN_LOCK if this is not setstripe. + * + * Always specify MDS_OPEN_BY_FID because we don't want + * to get file with different fid. + */ + it->it_flags |= MDS_OPEN_LOCK | MDS_OPEN_BY_FID; rc = ll_intent_file_open(file, NULL, 0, it); - it->it_create_mode &= ~M_CHECK_STALE; if (rc) GOTO(out_openerr, rc); @@ -693,14 +641,6 @@ restart: if (!S_ISREG(inode->i_mode)) GOTO(out_och_free, rc); - ll_capa_open(inode); - - 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); @@ -714,10 +654,10 @@ out_och_free: mutex_unlock(&lli->lli_och_mutex); out_openerr: - if (opendir_set != 0) - ll_stop_statahead(inode, lli->lli_opendir_key); - if (fd != NULL) - ll_file_data_put(fd); + if (lli->lli_opendir_key == fd) + ll_deauthorize_statahead(inode, fd); + if (fd != NULL) + ll_file_data_put(fd); } else { ll_stats_ops_tally(ll_i2sbi(inode), LPROC_LL_OPEN, 1); } @@ -756,13 +696,14 @@ static int ll_md_blocking_lease_ast(struct ldlm_lock *lock, /** * Acquire a lease and open the file. */ -struct obd_client_handle *ll_lease_open(struct inode *inode, struct file *file, - fmode_t fmode, __u64 open_flags) +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 ptlrpc_request *req = NULL; struct lustre_handle old_handle = { 0 }; struct obd_client_handle *och = NULL; int rc; @@ -828,15 +769,15 @@ struct obd_client_handle *ll_lease_open(struct inode *inode, struct file *file, 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, + rc = md_intent_lock(sbi->ll_md_exp, op_data, &it, &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); + LDLM_FL_NO_LRU | LDLM_FL_EXCL); ll_finish_md_op_data(op_data); ptlrpc_req_finished(req); if (rc < 0) @@ -860,7 +801,7 @@ struct obd_client_handle *ll_lease_open(struct inode *inode, struct file *file, 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", + CERROR(DFID "lease granted but no open lock, %d/"LPU64".\n", PFID(ll_inode2fid(inode)), it.d.lustre.it_lock_mode, it.d.lustre.it_lock_bits); GOTO(out_close, rc = -EPROTO); @@ -877,7 +818,7 @@ out_close: 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); + rc2 = ll_close_inode_openhandle(inode, och, 0, NULL); if (rc2 < 0) CERROR("%s: error closing file "DFID": %d\n", ll_get_fsname(inode->i_sb, NULL, 0), @@ -890,14 +831,75 @@ out: OBD_FREE_PTR(och); RETURN(ERR_PTR(rc)); } -EXPORT_SYMBOL(ll_lease_open); + +/** + * Check whether a layout swap can be done between two inodes. + * + * \param[in] inode1 First inode to check + * \param[in] inode2 Second inode to check + * + * \retval 0 on success, layout swap can be performed between both inodes + * \retval negative error code if requirements are not met + */ +static int ll_check_swap_layouts_validity(struct inode *inode1, + struct inode *inode2) +{ + if (!S_ISREG(inode1->i_mode) || !S_ISREG(inode2->i_mode)) + return -EINVAL; + + if (inode_permission(inode1, MAY_WRITE) || + inode_permission(inode2, MAY_WRITE)) + return -EPERM; + + if (inode1->i_sb != inode2->i_sb) + return -EXDEV; + + return 0; +} + +static int ll_swap_layouts_close(struct obd_client_handle *och, + struct inode *inode, struct inode *inode2) +{ + const struct lu_fid *fid1 = ll_inode2fid(inode); + const struct lu_fid *fid2; + int rc; + ENTRY; + + CDEBUG(D_INODE, "%s: biased close of file "DFID"\n", + ll_get_fsname(inode->i_sb, NULL, 0), PFID(fid1)); + + rc = ll_check_swap_layouts_validity(inode, inode2); + if (rc < 0) + GOTO(out_free_och, rc); + + /* We now know that inode2 is a lustre inode */ + fid2 = ll_inode2fid(inode2); + + rc = lu_fid_cmp(fid1, fid2); + if (rc == 0) + GOTO(out_free_och, rc = -EINVAL); + + /* Close the file and swap layouts between inode & inode2. + * NB: lease lock handle is released in mdc_close_layout_swap_pack() + * because we still need it to pack l_remote_handle to MDT. */ + rc = ll_close_inode_openhandle(inode, och, MDS_CLOSE_LAYOUT_SWAP, + inode2); + + och = NULL; /* freed in ll_close_inode_openhandle() */ + +out_free_och: + if (och != NULL) + OBD_FREE_PTR(och); + + RETURN(rc); +} /** * Release lease and close the file. * It will check if the lease has ever broken. */ -int ll_lease_close(struct obd_client_handle *och, struct inode *inode, - bool *lease_broken) +static int ll_lease_close(struct obd_client_handle *och, struct inode *inode, + bool *lease_broken) { struct ldlm_lock *lock; bool cancelled = true; @@ -920,156 +922,64 @@ int ll_lease_close(struct obd_client_handle *och, struct inode *inode, if (lease_broken != NULL) *lease_broken = cancelled; - rc = ll_close_inode_openhandle(ll_i2sbi(inode)->ll_md_exp, inode, och, - NULL); - RETURN(rc); -} -EXPORT_SYMBOL(ll_lease_close); - -/* 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 dv_flags) -{ - struct ptlrpc_request_set *set; - struct obd_info oinfo = { { { 0 } } }; - int rc; - - ENTRY; - - LASSERT(lsm != NULL); - - oinfo.oi_md = lsm; - oinfo.oi_oa = obdo; - oinfo.oi_oa->o_oi = lsm->lsm_oi; - oinfo.oi_oa->o_mode = S_IFREG; - oinfo.oi_oa->o_ioepoch = ioepoch; - oinfo.oi_oa->o_valid = OBD_MD_FLID | OBD_MD_FLTYPE | - OBD_MD_FLSIZE | OBD_MD_FLBLOCKS | - OBD_MD_FLBLKSZ | OBD_MD_FLATIME | - OBD_MD_FLMTIME | OBD_MD_FLCTIME | - OBD_MD_FLGROUP | OBD_MD_FLEPOCH | - OBD_MD_FLDATAVERSION; - oinfo.oi_capa = capa; - 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) { - CERROR("can't allocate ptlrpc set\n"); - rc = -ENOMEM; - } else { - rc = obd_getattr_async(exp, &oinfo, set); - if (rc == 0) - 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 | 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); -} - -/** - * Performs the getattr on the inode and updates its fields. - * If @sync != 0, perform the getattr under the server-side lock. - */ -int ll_inode_getattr(struct inode *inode, struct obdo *obdo, - __u64 ioepoch, int sync) -{ - struct obd_capa *capa = ll_mdscapa_get(inode); - struct lov_stripe_md *lsm; - int rc; - ENTRY; - - lsm = ccc_inode_lsm_get(inode); - rc = ll_lsm_getattr(lsm, ll_i2dtexp(inode), - 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; - - obdo_refresh_inode(inode, obdo, obdo->o_valid); - CDEBUG(D_INODE, "objid "DOSTID" size %llu, blocks %llu," - " blksize %lu\n", POSTID(oi), i_size_read(inode), - (unsigned long long)inode->i_blocks, - (unsigned long)ll_inode_blksize(inode)); - } - ccc_inode_lsm_put(inode, lsm); + rc = ll_close_inode_openhandle(inode, och, 0, NULL); RETURN(rc); } -int ll_merge_lvb(const struct lu_env *env, struct inode *inode) +int ll_merge_attr(const struct lu_env *env, struct inode *inode) { struct ll_inode_info *lli = ll_i2info(inode); struct cl_object *obj = lli->lli_clob; - struct cl_attr *attr = ccc_env_thread_attr(env); - struct ost_lvb lvb; + struct cl_attr *attr = vvp_env_thread_attr(env); + s64 atime; + s64 mtime; + s64 ctime; int rc = 0; ENTRY; ll_inode_size_lock(inode); + /* merge timestamps the most recently obtained from mds with timestamps obtained from osts */ - LTIME_S(inode->i_atime) = lli->lli_lvb.lvb_atime; - LTIME_S(inode->i_mtime) = lli->lli_lvb.lvb_mtime; - LTIME_S(inode->i_ctime) = lli->lli_lvb.lvb_ctime; - inode_init_lvb(inode, &lvb); + LTIME_S(inode->i_atime) = lli->lli_atime; + LTIME_S(inode->i_mtime) = lli->lli_mtime; + LTIME_S(inode->i_ctime) = lli->lli_ctime; + + atime = LTIME_S(inode->i_atime); + mtime = LTIME_S(inode->i_mtime); + ctime = LTIME_S(inode->i_ctime); cl_object_attr_lock(obj); rc = cl_object_attr_get(env, obj, attr); cl_object_attr_unlock(obj); - if (rc == 0) { - if (lvb.lvb_atime < attr->cat_atime) - lvb.lvb_atime = attr->cat_atime; - if (lvb.lvb_ctime < attr->cat_ctime) - lvb.lvb_ctime = attr->cat_ctime; - if (lvb.lvb_mtime < attr->cat_mtime) - lvb.lvb_mtime = attr->cat_mtime; + if (rc != 0) + GOTO(out_size_unlock, rc); - CDEBUG(D_VFSTRACE, DFID" updating i_size "LPU64"\n", - PFID(&lli->lli_fid), attr->cat_size); - cl_isize_write_nolock(inode, attr->cat_size); + if (atime < attr->cat_atime) + atime = attr->cat_atime; - inode->i_blocks = attr->cat_blocks; + if (ctime < attr->cat_ctime) + ctime = attr->cat_ctime; - LTIME_S(inode->i_mtime) = lvb.lvb_mtime; - LTIME_S(inode->i_atime) = lvb.lvb_atime; - LTIME_S(inode->i_ctime) = lvb.lvb_ctime; - } - ll_inode_size_unlock(inode); + if (mtime < attr->cat_mtime) + mtime = attr->cat_mtime; - RETURN(rc); -} + CDEBUG(D_VFSTRACE, DFID" updating i_size "LPU64"\n", + PFID(&lli->lli_fid), attr->cat_size); -int ll_glimpse_ioctl(struct ll_sb_info *sbi, struct lov_stripe_md *lsm, - lstat_t *st) -{ - struct obdo obdo = { 0 }; - int rc; + i_size_write(inode, attr->cat_size); + inode->i_blocks = attr->cat_blocks; - rc = ll_lsm_getattr(lsm, sbi->ll_dt_exp, NULL, &obdo, 0, 0); - if (rc == 0) { - st->st_size = obdo.o_size; - st->st_blocks = obdo.o_blocks; - st->st_mtime = obdo.o_mtime; - st->st_atime = obdo.o_atime; - st->st_ctime = obdo.o_ctime; - } - return rc; + LTIME_S(inode->i_atime) = atime; + LTIME_S(inode->i_mtime) = mtime; + LTIME_S(inode->i_ctime) = ctime; + +out_size_unlock: + ll_inode_size_unlock(inode); + + RETURN(rc); } static bool file_is_noatime(const struct file *file) @@ -1099,9 +1009,9 @@ static bool file_is_noatime(const struct file *file) return false; } -void ll_io_init(struct cl_io *io, const struct file *file, int write) +static void ll_io_init(struct cl_io *io, const struct file *file, int write) { - struct inode *inode = file->f_dentry->d_inode; + struct inode *inode = file->f_path.dentry->d_inode; io->u.ci_rw.crw_nonblock = file->f_flags & O_NONBLOCK; if (write) { @@ -1127,191 +1037,294 @@ ll_file_io_generic(const struct lu_env *env, struct vvp_io_args *args, struct file *file, enum cl_io_type iot, loff_t *ppos, size_t count) { - 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; + struct vvp_io *vio = vvp_env_io(env); + struct inode *inode = file->f_path.dentry->d_inode; + struct ll_inode_info *lli = ll_i2info(inode); + struct ll_file_data *fd = LUSTRE_FPRIVATE(file); + struct cl_io *io; + ssize_t result = 0; + int rc = 0; + struct range_lock range; + ENTRY; - CDEBUG(D_VFSTRACE, "file: %s, type: %d ppos: "LPU64", count: %zd\n", - file->f_dentry->d_name.name, iot, *ppos, count); + CDEBUG(D_VFSTRACE, "file: %s, type: %d ppos: "LPU64", count: %zu\n", + file->f_path.dentry->d_name.name, iot, *ppos, count); restart: - io = ccc_env_thread_io(env); - ll_io_init(io, file, iot == CIT_WRITE); - - if (cl_io_rw_init(env, io, iot, *ppos, count) == 0) { - struct vvp_io *vio = vvp_env_io(env); - struct ccc_io *cio = ccc_env_io(env); - int write_mutex_locked = 0; - - cio->cui_fd = LUSTRE_FPRIVATE(file); - vio->cui_io_subtype = args->via_io_subtype; - - switch (vio->cui_io_subtype) { - case IO_NORMAL: - cio->cui_iov = args->u.normal.via_iov; - cio->cui_nrsegs = args->u.normal.via_nrsegs; - cio->cui_tot_nrsegs = cio->cui_nrsegs; - cio->cui_iocb = args->u.normal.via_iocb; - 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; + io = vvp_env_thread_io(env); + ll_io_init(io, file, iot == CIT_WRITE); + + if (cl_io_rw_init(env, io, iot, *ppos, count) == 0) { + bool range_locked = false; + + if (file->f_flags & O_APPEND) + range_lock_init(&range, 0, LUSTRE_EOF); + else + range_lock_init(&range, *ppos, *ppos + count - 1); + + vio->vui_fd = LUSTRE_FPRIVATE(file); + vio->vui_io_subtype = args->via_io_subtype; + + switch (vio->vui_io_subtype) { + case IO_NORMAL: + vio->vui_iter = args->u.normal.via_iter; +#ifndef HAVE_FILE_OPERATIONS_READ_WRITE_ITER + vio->vui_tot_nrsegs = vio->vui_iter->nr_segs; +#endif /* !HAVE_FILE_OPERATIONS_READ_WRITE_ITER */ + vio->vui_iocb = args->u.normal.via_iocb; + /* Direct IO reads must also take range lock, + * or multiple reads will try to work on the same pages + * See LU-6227 for details. */ + if (((iot == CIT_WRITE) || + (iot == CIT_READ && (file->f_flags & O_DIRECT))) && + !(vio->vui_fd->fd_flags & LL_FILE_GROUP_LOCKED)) { + CDEBUG(D_VFSTRACE, "Range lock "RL_FMT"\n", + RL_PARA(&range)); + rc = range_lock(&lli->lli_write_tree, &range); + if (rc < 0) + GOTO(out, rc); + + range_locked = true; } - 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; - break; - case IO_SPLICE: - vio->u.splice.cui_pipe = args->u.splice.via_pipe; - vio->u.splice.cui_flags = args->u.splice.via_flags; - break; - default: - CERROR("Unknow IO type - %u\n", vio->cui_io_subtype); - LBUG(); - } - result = cl_io_loop(env, io); - 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; - } + break; + case IO_SPLICE: + vio->u.splice.vui_pipe = args->u.splice.via_pipe; + vio->u.splice.vui_flags = args->u.splice.via_flags; + break; + default: + CERROR("unknown IO subtype %u\n", vio->vui_io_subtype); + LBUG(); + } - if (io->ci_nob > 0) { - result = io->ci_nob; - *ppos = io->u.ci_wr.wr.crw_pos; - } - GOTO(out, result); + ll_cl_add(file, env, io); + rc = cl_io_loop(env, io); + ll_cl_remove(file, env); + + if (range_locked) { + CDEBUG(D_VFSTRACE, "Range unlock "RL_FMT"\n", + RL_PARA(&range)); + range_unlock(&lli->lli_write_tree, &range); + } + } else { + /* cl_io_rw_init() handled IO */ + rc = io->ci_result; + } + + if (io->ci_nob > 0) { + result += io->ci_nob; + count -= io->ci_nob; + *ppos = io->u.ci_wr.wr.crw_pos; /* for splice */ + + /* prepare IO restart */ + if (count > 0 && args->via_io_subtype == IO_NORMAL) { + args->u.normal.via_iter = vio->vui_iter; +#ifndef HAVE_FILE_OPERATIONS_READ_WRITE_ITER + args->u.normal.via_iter->nr_segs = vio->vui_tot_nrsegs; +#endif /* !HAVE_FILE_OPERATIONS_READ_WRITE_ITER */ + } + } + GOTO(out, rc); 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 || result == -ENODATA) && io->ci_need_restart) { - CDEBUG(D_VFSTRACE, "Restart %s on %s from %lld, count:%zd\n", + cl_io_fini(env, io); + + if ((rc == 0 || rc == -ENODATA) && count > 0 && io->ci_need_restart) { + CDEBUG(D_VFSTRACE, + "%s: restart %s from %lld, count:%zu, result: %zd\n", + file->f_path.dentry->d_name.name, iot == CIT_READ ? "read" : "write", - file->f_dentry->d_name.name, *ppos, count); - LASSERTF(io->ci_nob == 0, "%zd", io->ci_nob); + *ppos, count, result); goto restart; } - if (iot == CIT_READ) { - if (result >= 0) - ll_stats_ops_tally(ll_i2sbi(file->f_dentry->d_inode), - LPROC_LL_READ_BYTES, result); - } else if (iot == CIT_WRITE) { - if (result >= 0) { - ll_stats_ops_tally(ll_i2sbi(file->f_dentry->d_inode), - LPROC_LL_WRITE_BYTES, result); + if (iot == CIT_READ) { + if (result > 0) + ll_stats_ops_tally(ll_i2sbi(inode), + LPROC_LL_READ_BYTES, result); + } else if (iot == CIT_WRITE) { + if (result > 0) { + ll_stats_ops_tally(ll_i2sbi(inode), + LPROC_LL_WRITE_BYTES, result); fd->fd_write_failed = false; - } else if (result != -ERESTARTSYS) { + } else if (result == 0 && rc == 0) { + rc = io->ci_result; + if (rc < 0) + fd->fd_write_failed = true; + else + fd->fd_write_failed = false; + } else if (rc != -ERESTARTSYS) { fd->fd_write_failed = true; } } + CDEBUG(D_VFSTRACE, "iot: %d, result: %zd\n", iot, result); - return result; + return result > 0 ? result : rc; } - /* - * XXX: exact copy from kernel code (__generic_file_aio_write_nolock) + * Read from a file (through the page cache). */ -static int ll_file_get_iov_count(const struct iovec *iov, - unsigned long *nr_segs, size_t *count) +static ssize_t ll_file_read_iter(struct kiocb *iocb, struct iov_iter *to) { - size_t cnt = 0; - unsigned long seg; + struct vvp_io_args *args; + struct lu_env *env; + ssize_t result; + __u16 refcheck; + + env = cl_env_get(&refcheck); + if (IS_ERR(env)) + return PTR_ERR(env); - for (seg = 0; seg < *nr_segs; seg++) { - const struct iovec *iv = &iov[seg]; + args = ll_env_args(env, IO_NORMAL); + args->u.normal.via_iter = to; + args->u.normal.via_iocb = iocb; - /* - * If any segment has a negative length, or the cumulative - * length ever wraps negative then return -EINVAL. - */ - 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)) - continue; - if (seg == 0) - return -EFAULT; - *nr_segs = seg; - cnt -= iv->iov_len; /* This segment is no good */ - break; - } - *count = cnt; - return 0; + result = ll_file_io_generic(env, args, iocb->ki_filp, CIT_READ, + &iocb->ki_pos, iov_iter_count(to)); + cl_env_put(env, &refcheck); + return result; } -static ssize_t ll_file_aio_read(struct kiocb *iocb, const struct iovec *iov, - unsigned long nr_segs, loff_t pos) +/* + * Write to a file (through the page cache). + */ +static ssize_t ll_file_write_iter(struct kiocb *iocb, struct iov_iter *from) { - struct lu_env *env; - struct vvp_io_args *args; - size_t count; - ssize_t result; - int refcheck; - ENTRY; + struct vvp_io_args *args; + struct lu_env *env; + ssize_t result; + __u16 refcheck; - 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); - env = cl_env_get(&refcheck); - if (IS_ERR(env)) - RETURN(PTR_ERR(env)); + args = ll_env_args(env, IO_NORMAL); + args->u.normal.via_iter = from; + args->u.normal.via_iocb = iocb; - args = vvp_env_args(env, IO_NORMAL); - args->u.normal.via_iov = (struct iovec *)iov; - args->u.normal.via_nrsegs = nr_segs; - args->u.normal.via_iocb = iocb; + result = ll_file_io_generic(env, args, iocb->ki_filp, CIT_WRITE, + &iocb->ki_pos, iov_iter_count(from)); + cl_env_put(env, &refcheck); + return result; +} - result = ll_file_io_generic(env, args, iocb->ki_filp, CIT_READ, - &iocb->ki_pos, count); - cl_env_put(env, &refcheck); - RETURN(result); +#ifndef HAVE_FILE_OPERATIONS_READ_WRITE_ITER +/* + * 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) +{ + size_t cnt = 0; + unsigned long seg; + + for (seg = 0; seg < *nr_segs; seg++) { + const struct iovec *iv = &iov[seg]; + + /* + * If any segment has a negative length, or the cumulative + * length ever wraps negative then return -EINVAL. + */ + 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)) + continue; + if (seg == 0) + return -EFAULT; + *nr_segs = seg; + cnt -= iv->iov_len; /* This segment is no good */ + break; + } + *count = cnt; + return 0; } -static ssize_t ll_file_read(struct file *file, char *buf, size_t count, - loff_t *ppos) +static ssize_t ll_file_aio_read(struct kiocb *iocb, const struct iovec *iov, + unsigned long nr_segs, loff_t pos) +{ + struct iovec *local_iov; + struct iov_iter *to; + size_t iov_count; + ssize_t result; + struct lu_env *env = NULL; + __u16 refcheck; + ENTRY; + + result = ll_file_get_iov_count(iov, &nr_segs, &iov_count); + if (result) + RETURN(result); + + if (nr_segs == 1) { + + env = cl_env_get(&refcheck); + if (IS_ERR(env)) + RETURN(PTR_ERR(env)); + + local_iov = &ll_env_info(env)->lti_local_iov; + *local_iov = *iov; + + } else { + OBD_ALLOC(local_iov, sizeof(*iov) * nr_segs); + if (local_iov == NULL) + RETURN(-ENOMEM); + + memcpy(local_iov, iov, sizeof(*iov) * nr_segs); + } + + OBD_ALLOC_PTR(to); + if (to == NULL) { + result = -ENOMEM; + goto out; + } +# ifdef HAVE_IOV_ITER_INIT_DIRECTION + iov_iter_init(to, READ, local_iov, nr_segs, iov_count); +# else /* !HAVE_IOV_ITER_INIT_DIRECTION */ + iov_iter_init(to, local_iov, nr_segs, iov_count, 0); +# endif /* HAVE_IOV_ITER_INIT_DIRECTION */ + + result = ll_file_read_iter(iocb, to); + + OBD_FREE_PTR(to); +out: + if (nr_segs == 1) + cl_env_put(env, &refcheck); + else + OBD_FREE(local_iov, sizeof(*iov) * nr_segs); + + RETURN(result); +} + +static ssize_t ll_file_read(struct file *file, char __user *buf, size_t count, + loff_t *ppos) { - struct lu_env *env; - struct iovec *local_iov; + struct lu_env *env; + struct iovec iov = { .iov_base = buf, .iov_len = count }; struct kiocb *kiocb; ssize_t result; - int refcheck; + __u16 refcheck; ENTRY; env = cl_env_get(&refcheck); if (IS_ERR(env)) RETURN(PTR_ERR(env)); - local_iov = &vvp_env_info(env)->vti_local_iov; - kiocb = &vvp_env_info(env)->vti_kiocb; - local_iov->iov_base = (void __user *)buf; - local_iov->iov_len = count; + kiocb = &ll_env_info(env)->lti_kiocb; init_sync_kiocb(kiocb, file); kiocb->ki_pos = *ppos; #ifdef HAVE_KIOCB_KI_LEFT - kiocb->ki_left = count; -#else - kiocb->ki_nbytes = count; + kiocb->ki_left = count; +#elif defined(HAVE_KI_NBYTES) + kiocb->ki_nbytes = count; #endif - result = ll_file_aio_read(kiocb, local_iov, 1, kiocb->ki_pos); - *ppos = kiocb->ki_pos; + result = ll_file_aio_read(kiocb, &iov, 1, kiocb->ki_pos); + *ppos = kiocb->ki_pos; - cl_env_put(env, &refcheck); - RETURN(result); + cl_env_put(env, &refcheck); + RETURN(result); } /* @@ -1319,66 +1332,89 @@ static ssize_t ll_file_read(struct file *file, char *buf, size_t count, * AIO stuff */ static ssize_t ll_file_aio_write(struct kiocb *iocb, const struct iovec *iov, - unsigned long nr_segs, loff_t pos) -{ - struct lu_env *env; - struct vvp_io_args *args; - size_t count; - ssize_t result; - int refcheck; - ENTRY; + unsigned long nr_segs, loff_t pos) +{ + struct iovec *local_iov; + struct iov_iter *from; + size_t iov_count; + ssize_t result; + struct lu_env *env = NULL; + __u16 refcheck; + ENTRY; - result = ll_file_get_iov_count(iov, &nr_segs, &count); - if (result) - RETURN(result); + result = ll_file_get_iov_count(iov, &nr_segs, &iov_count); + if (result) + RETURN(result); - env = cl_env_get(&refcheck); - if (IS_ERR(env)) - RETURN(PTR_ERR(env)); + if (nr_segs == 1) { + 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; - args->u.normal.via_iocb = iocb; + local_iov = &ll_env_info(env)->lti_local_iov; + *local_iov = *iov; + } else { + OBD_ALLOC(local_iov, sizeof(*iov) * nr_segs); + if (local_iov == NULL) + RETURN(-ENOMEM); - result = ll_file_io_generic(env, args, iocb->ki_filp, CIT_WRITE, - &iocb->ki_pos, count); - cl_env_put(env, &refcheck); - RETURN(result); + memcpy(local_iov, iov, sizeof(*iov) * nr_segs); + } + + OBD_ALLOC_PTR(from); + if (from == NULL) { + result = -ENOMEM; + goto out; + } +# ifdef HAVE_IOV_ITER_INIT_DIRECTION + iov_iter_init(from, WRITE, local_iov, nr_segs, iov_count); +# else /* !HAVE_IOV_ITER_INIT_DIRECTION */ + iov_iter_init(from, local_iov, nr_segs, iov_count, 0); +# endif /* HAVE_IOV_ITER_INIT_DIRECTION */ + + result = ll_file_write_iter(iocb, from); + + OBD_FREE_PTR(from); +out: + if (nr_segs == 1) + cl_env_put(env, &refcheck); + else + OBD_FREE(local_iov, sizeof(*iov) * nr_segs); + + RETURN(result); } -static ssize_t ll_file_write(struct file *file, const char *buf, size_t count, - loff_t *ppos) +static ssize_t ll_file_write(struct file *file, const char __user *buf, + size_t count, loff_t *ppos) { - struct lu_env *env; - struct iovec *local_iov; + struct lu_env *env; + struct iovec iov = { .iov_base = (void __user *)buf, + .iov_len = count }; struct kiocb *kiocb; ssize_t result; - int refcheck; + __u16 refcheck; ENTRY; env = cl_env_get(&refcheck); if (IS_ERR(env)) RETURN(PTR_ERR(env)); - local_iov = &vvp_env_info(env)->vti_local_iov; - kiocb = &vvp_env_info(env)->vti_kiocb; - local_iov->iov_base = (void __user *)buf; - local_iov->iov_len = count; + kiocb = &ll_env_info(env)->lti_kiocb; init_sync_kiocb(kiocb, file); kiocb->ki_pos = *ppos; #ifdef HAVE_KIOCB_KI_LEFT - kiocb->ki_left = count; -#else - kiocb->ki_nbytes = count; + kiocb->ki_left = count; +#elif defined(HAVE_KI_NBYTES) + kiocb->ki_nbytes = count; #endif - result = ll_file_aio_write(kiocb, local_iov, 1, kiocb->ki_pos); - *ppos = kiocb->ki_pos; + result = ll_file_aio_write(kiocb, &iov, 1, kiocb->ki_pos); + *ppos = kiocb->ki_pos; - cl_env_put(env, &refcheck); - RETURN(result); + cl_env_put(env, &refcheck); + RETURN(result); } +#endif /* !HAVE_FILE_OPERATIONS_READ_WRITE_ITER */ /* * Send file content (through pagecache) somewhere with helper @@ -1390,14 +1426,14 @@ static ssize_t ll_file_splice_read(struct file *in_file, loff_t *ppos, struct lu_env *env; struct vvp_io_args *args; ssize_t result; - int refcheck; + __u16 refcheck; ENTRY; env = cl_env_get(&refcheck); if (IS_ERR(env)) RETURN(PTR_ERR(env)); - args = vvp_env_args(env, IO_SPLICE); + args = ll_env_args(env, IO_SPLICE); args->u.splice.via_pipe = pipe; args->u.splice.via_flags = flags; @@ -1406,125 +1442,30 @@ static ssize_t ll_file_splice_read(struct file *in_file, loff_t *ppos, RETURN(result); } -static int ll_lov_recreate(struct inode *inode, struct ost_id *oi, - obd_count ost_idx) -{ - struct obd_export *exp = ll_i2dtexp(inode); - struct obd_trans_info oti = { 0 }; - struct obdo *oa = NULL; - int lsm_size; - int rc = 0; - struct lov_stripe_md *lsm = NULL, *lsm2; - ENTRY; - - OBDO_ALLOC(oa); - if (oa == NULL) - RETURN(-ENOMEM); - - lsm = ccc_inode_lsm_get(inode); - if (!lsm_has_objects(lsm)) - GOTO(out, rc = -ENOENT); - - lsm_size = sizeof(*lsm) + (sizeof(struct lov_oinfo) * - (lsm->lsm_stripe_count)); - - OBD_ALLOC_LARGE(lsm2, lsm_size); - if (lsm2 == NULL) - GOTO(out, rc = -ENOMEM); - - oa->o_oi = *oi; - oa->o_nlink = ost_idx; - oa->o_flags |= OBD_FL_RECREATE_OBJS; - oa->o_valid = OBD_MD_FLID | OBD_MD_FLFLAGS | OBD_MD_FLGROUP; - obdo_from_inode(oa, inode, OBD_MD_FLTYPE | OBD_MD_FLATIME | - OBD_MD_FLMTIME | OBD_MD_FLCTIME); - obdo_set_parent_fid(oa, &ll_i2info(inode)->lli_fid); - memcpy(lsm2, lsm, lsm_size); - ll_inode_size_lock(inode); - rc = obd_create(NULL, exp, oa, &lsm2, &oti); - ll_inode_size_unlock(inode); - - OBD_FREE_LARGE(lsm2, lsm_size); - GOTO(out, rc); -out: - ccc_inode_lsm_put(inode, lsm); - OBDO_FREE(oa); - return rc; -} - -static int ll_lov_recreate_obj(struct inode *inode, unsigned long arg) -{ - struct ll_recreate_obj ucreat; - struct ost_id oi; - ENTRY; - - if (!cfs_capable(CFS_CAP_SYS_ADMIN)) - RETURN(-EPERM); - - if (copy_from_user(&ucreat, (struct ll_recreate_obj *)arg, - sizeof(ucreat))) - RETURN(-EFAULT); - - ostid_set_seq_mdt0(&oi); - ostid_set_id(&oi, ucreat.lrc_id); - RETURN(ll_lov_recreate(inode, &oi, ucreat.lrc_ost_idx)); -} - -static int ll_lov_recreate_fid(struct inode *inode, unsigned long arg) -{ - struct lu_fid fid; - struct ost_id oi; - obd_count ost_idx; - ENTRY; - - if (!cfs_capable(CFS_CAP_SYS_ADMIN)) - RETURN(-EPERM); - - if (copy_from_user(&fid, (struct lu_fid *)arg, sizeof(fid))) - RETURN(-EFAULT); - - fid_to_ostid(&fid, &oi); - ost_idx = (fid_seq(&fid) >> 16) & 0xffff; - RETURN(ll_lov_recreate(inode, &oi, ost_idx)); -} - int ll_lov_setstripe_ea_info(struct inode *inode, struct file *file, __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}; - int rc = 0; + struct lookup_intent oit = { + .it_op = IT_OPEN, + .it_flags = flags | MDS_OPEN_BY_FID, + }; + int rc; ENTRY; - lsm = ccc_inode_lsm_get(inode); - if (lsm != NULL) { - ccc_inode_lsm_put(inode, lsm); - 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_unlock, rc); - rc = oit.d.lustre.it_status; if (rc < 0) - GOTO(out_req_free, rc); + GOTO(out_unlock, rc); - ll_release_openhandle(file->f_dentry, &oit); + ll_release_openhandle(file->f_path.dentry, &oit); 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); - goto out; } int ll_lov_getstripe_ea_info(struct inode *inode, const char *filename, @@ -1538,9 +1479,9 @@ 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, @@ -1560,9 +1501,9 @@ int ll_lov_getstripe_ea_info(struct inode *inode, const char *filename, body = req_capsule_server_get(&req->rq_pill, &RMF_MDT_BODY); LASSERT(body != NULL); /* checked by mdc_getattr_name */ - lmmsize = body->eadatasize; + lmmsize = body->mbo_eadatasize; - if (!(body->valid & (OBD_MD_FLEASIZE | OBD_MD_FLDIREA)) || + if (!(body->mbo_valid & (OBD_MD_FLEASIZE | OBD_MD_FLDIREA)) || lmmsize == 0) { GOTO(out, rc = -ENODATA); } @@ -1591,13 +1532,14 @@ int ll_lov_getstripe_ea_info(struct inode *inode, const char *filename, * 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->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->mode)) + 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); @@ -1628,7 +1570,7 @@ static int ll_lov_setea(struct inode *inode, struct file *file, if (lump == NULL) RETURN(-ENOMEM); - if (copy_from_user(lump, (struct lov_user_md *)arg, lum_size)) { + if (copy_from_user(lump, (struct lov_user_md __user *)arg, lum_size)) { OBD_FREE_LARGE(lump, lum_size); RETURN(-EFAULT); } @@ -1639,80 +1581,79 @@ static int ll_lov_setea(struct inode *inode, struct file *file, RETURN(rc); } +static int ll_file_getstripe(struct inode *inode, + struct lov_user_md __user *lum) +{ + struct lu_env *env; + __u16 refcheck; + int rc; + ENTRY; + + env = cl_env_get(&refcheck); + if (IS_ERR(env)) + RETURN(PTR_ERR(env)); + + rc = cl_object_getstripe(env, ll_i2info(inode)->lli_clob, lum); + cl_env_put(env, &refcheck); + RETURN(rc); +} + static int ll_lov_setstripe(struct inode *inode, struct file *file, unsigned long arg) { - struct lov_user_md_v3 lumv3; - struct lov_user_md_v1 *lumv1 = (struct lov_user_md_v1 *)&lumv3; - struct lov_user_md_v1 *lumv1p = (struct lov_user_md_v1 *)arg; - struct lov_user_md_v3 *lumv3p = (struct lov_user_md_v3 *)arg; - int lum_size, rc; - __u64 flags = FMODE_WRITE; + struct lov_user_md __user *lum = (struct lov_user_md __user *)arg; + struct lov_user_md *klum; + int lum_size, rc; + __u64 flags = FMODE_WRITE; ENTRY; - /* first try with v1 which is smaller than v3 */ - lum_size = sizeof(struct lov_user_md_v1); - if (copy_from_user(lumv1, lumv1p, lum_size)) - RETURN(-EFAULT); - - if (lumv1->lmm_magic == LOV_USER_MAGIC_V3) { - lum_size = sizeof(struct lov_user_md_v3); - if (copy_from_user(&lumv3, lumv3p, lum_size)) - RETURN(-EFAULT); - } + rc = ll_copy_user_md(lum, &klum); + if (rc < 0) + RETURN(rc); - rc = ll_lov_setstripe_ea_info(inode, file, flags, lumv1, lum_size); + lum_size = rc; + rc = ll_lov_setstripe_ea_info(inode, file, flags, klum, lum_size); if (rc == 0) { - struct lov_stripe_md *lsm; __u32 gen; - put_user(0, &lumv1p->lmm_stripe_count); + put_user(0, &lum->lmm_stripe_count); ll_layout_refresh(inode, &gen); - lsm = ccc_inode_lsm_get(inode); - rc = obd_iocontrol(LL_IOC_LOV_GETSTRIPE, ll_i2dtexp(inode), - 0, lsm, (void *)arg); - ccc_inode_lsm_put(inode, lsm); + rc = ll_file_getstripe(inode, (struct lov_user_md __user *)arg); } - RETURN(rc); -} - -static int ll_lov_getstripe(struct inode *inode, unsigned long arg) -{ - struct lov_stripe_md *lsm; - int rc = -ENODATA; - ENTRY; - lsm = ccc_inode_lsm_get(inode); - if (lsm != NULL) - rc = obd_iocontrol(LL_IOC_LOV_GETSTRIPE, ll_i2dtexp(inode), 0, - lsm, (void *)arg); - ccc_inode_lsm_put(inode, lsm); + OBD_FREE(klum, lum_size); 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); - struct ccc_grouplock grouplock; + struct ll_grouplock grouplock; int rc; ENTRY; + if (arg == 0) { + CWARN("group id for group lock must not be 0\n"); + RETURN(-EINVAL); + } + if (ll_file_nolock(file)) RETURN(-EOPNOTSUPP); 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.cg_gid); + fd->fd_grouplock.lg_gid); spin_unlock(&lli->lli_lock); RETURN(-EINVAL); } - LASSERT(fd->fd_grouplock.cg_lock == NULL); + LASSERT(fd->fd_grouplock.lg_lock == NULL); spin_unlock(&lli->lli_lock); - rc = cl_get_grouplock(cl_i2info(inode)->lli_clob, + rc = cl_get_grouplock(ll_i2info(inode)->lli_clob, arg, (file->f_flags & O_NONBLOCK), &grouplock); if (rc) RETURN(rc); @@ -1733,11 +1674,12 @@ int ll_get_grouplock(struct inode *inode, struct file *file, unsigned long arg) RETURN(0); } -int ll_put_grouplock(struct inode *inode, struct file *file, unsigned long arg) +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 ccc_grouplock grouplock; + struct ll_grouplock grouplock; ENTRY; spin_lock(&lli->lli_lock); @@ -1746,11 +1688,12 @@ int ll_put_grouplock(struct inode *inode, struct file *file, unsigned long arg) CWARN("no group lock held\n"); RETURN(-EINVAL); } - LASSERT(fd->fd_grouplock.cg_lock != NULL); - if (fd->fd_grouplock.cg_gid != arg) { - CWARN("group lock %lu doesn't match current id %lu\n", - arg, fd->fd_grouplock.cg_gid); + 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); } @@ -1799,8 +1742,7 @@ int ll_release_openhandle(struct dentry *dentry, struct lookup_intent *it) ll_och_fill(ll_i2sbi(inode)->ll_md_exp, it, och); - rc = ll_close_inode_openhandle(ll_i2sbi(inode)->ll_md_exp, - inode, och, NULL); + rc = ll_close_inode_openhandle(inode, och, 0, NULL); out: /* this one is in place of ll_file_open */ if (it_disposition(it, DISP_ENQ_OPEN_REF)) { @@ -1813,96 +1755,93 @@ out: /** * Get size for inode for which FIEMAP mapping is requested. * Make the FIEMAP get_info call and returns the result. + * \param fiemap kernel buffer to hold extens + * \param num_bytes kernel buffer size */ -int ll_do_fiemap(struct inode *inode, struct ll_user_fiemap *fiemap, - int num_bytes) +static int ll_do_fiemap(struct inode *inode, struct fiemap *fiemap, + size_t num_bytes) { - struct obd_export *exp = ll_i2dtexp(inode); - struct lov_stripe_md *lsm = NULL; - struct ll_fiemap_info_key fm_key = { .name = KEY_FIEMAP, }; - int vallen = num_bytes; - int rc; - ENTRY; + struct lu_env *env; + __u16 refcheck; + int rc = 0; + struct ll_fiemap_info_key fmkey = { .lfik_name = KEY_FIEMAP, }; + ENTRY; - /* Checks for fiemap flags */ - if (fiemap->fm_flags & ~LUSTRE_FIEMAP_FLAGS_COMPAT) { - fiemap->fm_flags &= ~LUSTRE_FIEMAP_FLAGS_COMPAT; - return -EBADR; - } + /* Checks for fiemap flags */ + if (fiemap->fm_flags & ~LUSTRE_FIEMAP_FLAGS_COMPAT) { + fiemap->fm_flags &= ~LUSTRE_FIEMAP_FLAGS_COMPAT; + return -EBADR; + } - /* Check for FIEMAP_FLAG_SYNC */ - if (fiemap->fm_flags & FIEMAP_FLAG_SYNC) { - rc = filemap_fdatawrite(inode->i_mapping); - if (rc) - return rc; - } + /* Check for FIEMAP_FLAG_SYNC */ + if (fiemap->fm_flags & FIEMAP_FLAG_SYNC) { + rc = filemap_fdatawrite(inode->i_mapping); + if (rc) + return rc; + } - lsm = ccc_inode_lsm_get(inode); - if (lsm == NULL) - return -ENOENT; - - /* If the stripe_count > 1 and the application does not understand - * DEVICE_ORDER flag, then it cannot interpret the extents correctly. - */ - if (lsm->lsm_stripe_count > 1 && - !(fiemap->fm_flags & FIEMAP_FLAG_DEVICE_ORDER)) - GOTO(out, rc = -EOPNOTSUPP); - - fm_key.oa.o_oi = lsm->lsm_oi; - fm_key.oa.o_valid = OBD_MD_FLID | OBD_MD_FLGROUP; - - obdo_from_inode(&fm_key.oa, inode, OBD_MD_FLSIZE); - obdo_set_parent_fid(&fm_key.oa, &ll_i2info(inode)->lli_fid); - /* If filesize is 0, then there would be no objects for mapping */ - if (fm_key.oa.o_size == 0) { - fiemap->fm_mapped_extents = 0; - GOTO(out, rc = 0); - } + env = cl_env_get(&refcheck); + if (IS_ERR(env)) + RETURN(PTR_ERR(env)); - memcpy(&fm_key.fiemap, fiemap, sizeof(*fiemap)); + if (i_size_read(inode) == 0) { + rc = ll_glimpse_size(inode); + if (rc) + GOTO(out, rc); + } + + fmkey.lfik_oa.o_valid = OBD_MD_FLID | OBD_MD_FLGROUP; + obdo_from_inode(&fmkey.lfik_oa, inode, OBD_MD_FLSIZE); + obdo_set_parent_fid(&fmkey.lfik_oa, &ll_i2info(inode)->lli_fid); + + /* If filesize is 0, then there would be no objects for mapping */ + if (fmkey.lfik_oa.o_size == 0) { + fiemap->fm_mapped_extents = 0; + GOTO(out, rc = 0); + } - rc = obd_get_info(NULL, exp, sizeof(fm_key), &fm_key, &vallen, - fiemap, lsm); - if (rc) - CERROR("obd_get_info failed: rc = %d\n", rc); + fmkey.lfik_fiemap = *fiemap; + rc = cl_object_fiemap(env, ll_i2info(inode)->lli_clob, + &fmkey, fiemap, &num_bytes); out: - ccc_inode_lsm_put(inode, lsm); + cl_env_put(env, &refcheck); RETURN(rc); } -int ll_fid2path(struct inode *inode, void *arg) +int ll_fid2path(struct inode *inode, void __user *arg) { struct obd_export *exp = ll_i2mdexp(inode); - struct getinfo_fid2path *gfout, *gfin; - int outsize, rc; + const struct getinfo_fid2path __user *gfin = arg; + __u32 pathlen; + struct getinfo_fid2path *gfout; + size_t outsize; + int rc; + ENTRY; if (!cfs_capable(CFS_CAP_DAC_READ_SEARCH) && !(ll_i2sbi(inode)->ll_flags & LL_SBI_USER_FID2PATH)) RETURN(-EPERM); - /* Need to get the buflen */ - OBD_ALLOC_PTR(gfin); - if (gfin == NULL) - RETURN(-ENOMEM); - if (copy_from_user(gfin, arg, sizeof(*gfin))) { - OBD_FREE_PTR(gfin); + /* Only need to get the buflen */ + if (get_user(pathlen, &gfin->gf_pathlen)) RETURN(-EFAULT); - } - outsize = sizeof(*gfout) + gfin->gf_pathlen; + if (pathlen > PATH_MAX) + RETURN(-EINVAL); + + outsize = sizeof(*gfout) + pathlen; OBD_ALLOC(gfout, outsize); - if (gfout == NULL) { - OBD_FREE_PTR(gfin); + if (gfout == NULL) RETURN(-ENOMEM); - } - memcpy(gfout, gfin, sizeof(*gfout)); - OBD_FREE_PTR(gfin); + + if (copy_from_user(gfout, arg, sizeof(*gfout))) + GOTO(gf_free, rc = -EFAULT); /* Call mdc_iocontrol */ rc = obd_iocontrol(OBD_IOC_FID2PATH, exp, outsize, gfout, NULL); - if (rc) + if (rc != 0) GOTO(gf_free, rc); if (copy_to_user(arg, gfout, outsize)) @@ -1913,102 +1852,58 @@ gf_free: RETURN(rc); } -static int ll_ioctl_fiemap(struct inode *inode, unsigned long arg) -{ - struct ll_user_fiemap *fiemap_s; - size_t num_bytes, ret_bytes; - unsigned int extent_count; - int rc = 0; - - /* Get the extent count so we can calculate the size of - * required fiemap buffer */ - if (get_user(extent_count, - &((struct ll_user_fiemap __user *)arg)->fm_extent_count)) - RETURN(-EFAULT); - num_bytes = sizeof(*fiemap_s) + (extent_count * - sizeof(struct ll_fiemap_extent)); - - OBD_ALLOC_LARGE(fiemap_s, num_bytes); - if (fiemap_s == NULL) - RETURN(-ENOMEM); - - /* get the fiemap value */ - if (copy_from_user(fiemap_s, (struct ll_user_fiemap __user *)arg, - sizeof(*fiemap_s))) - GOTO(error, rc = -EFAULT); - - /* If fm_extent_count is non-zero, read the first extent since - * it is used to calculate end_offset and device from previous - * fiemap call. */ - if (extent_count) { - if (copy_from_user(&fiemap_s->fm_extents[0], - (char __user *)arg + sizeof(*fiemap_s), - sizeof(struct ll_fiemap_extent))) - GOTO(error, rc = -EFAULT); - } - - rc = ll_do_fiemap(inode, fiemap_s, num_bytes); - if (rc) - GOTO(error, rc); - - ret_bytes = sizeof(struct ll_user_fiemap); - - if (extent_count != 0) - ret_bytes += (fiemap_s->fm_mapped_extents * - sizeof(struct ll_fiemap_extent)); - - if (copy_to_user((void *)arg, fiemap_s, ret_bytes)) - rc = -EFAULT; - -error: - OBD_FREE_LARGE(fiemap_s, num_bytes); - RETURN(rc); -} - /* * Read the data_version for inode. * * This value is computed using stripe object version on OST. * Version is computed using server side locking. * - * @param sync if do sync on the OST side; + * @param flags 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 flags) { - struct lov_stripe_md *lsm = NULL; - struct ll_sb_info *sbi = ll_i2sbi(inode); - struct obdo *obdo = NULL; - int rc; + struct cl_object *obj = ll_i2info(inode)->lli_clob; + struct lu_env *env; + struct cl_io *io; + __u16 refcheck; + int result; + ENTRY; - /* If no stripe, we consider version is 0. */ - lsm = ccc_inode_lsm_get(inode); - if (!lsm_has_objects(lsm)) { + /* If no file object initialized, we consider its version is 0. */ + if (obj == NULL) { *data_version = 0; - CDEBUG(D_INODE, "No object for inode\n"); - GOTO(out, rc = 0); + RETURN(0); } - OBD_ALLOC_PTR(obdo); - if (obdo == NULL) - GOTO(out, rc = -ENOMEM); + env = cl_env_get(&refcheck); + if (IS_ERR(env)) + RETURN(PTR_ERR(env)); - 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; - } + io = vvp_env_thread_io(env); + io->ci_obj = obj; + io->u.ci_data_version.dv_data_version = 0; + io->u.ci_data_version.dv_flags = flags; - OBD_FREE_PTR(obdo); - EXIT; -out: - ccc_inode_lsm_put(inode, lsm); - RETURN(rc); +restart: + if (cl_io_init(env, io, CIT_DATA_VERSION, io->ci_obj) == 0) + result = cl_io_loop(env, io); + else + result = io->ci_result; + + *data_version = io->u.ci_data_version.dv_data_version; + + cl_io_fini(env, io); + + if (unlikely(io->ci_need_restart)) + goto restart; + + cl_env_put(env, &refcheck); + + RETURN(result); } /* @@ -2040,13 +1935,13 @@ int ll_hsm_release(struct inode *inode) if (IS_ERR(env)) GOTO(out, rc = PTR_ERR(env)); - ll_merge_lvb(env, inode); + ll_merge_attr(env, inode); cl_env_nested_put(&nest, env); /* 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, + rc = ll_close_inode_openhandle(inode, och, MDS_HSM_RELEASE, &data_version); och = NULL; @@ -2059,10 +1954,12 @@ out: } struct ll_swap_stack { - struct iattr ia1, ia2; - __u64 dv1, dv2; - struct inode *inode1, *inode2; - bool check_dv1, check_dv2; + __u64 dv1; + __u64 dv2; + struct inode *inode1; + struct inode *inode2; + bool check_dv1; + bool check_dv2; }; static int ll_swap_layouts(struct file *file1, struct file *file2, @@ -2079,18 +1976,12 @@ static int ll_swap_layouts(struct file *file1, struct file *file2, if (llss == NULL) RETURN(-ENOMEM); - llss->inode1 = file1->f_dentry->d_inode; - llss->inode2 = file2->f_dentry->d_inode; - - if (!S_ISREG(llss->inode2->i_mode)) - GOTO(free, rc = -EINVAL); + llss->inode1 = file1->f_path.dentry->d_inode; + llss->inode2 = file2->f_path.dentry->d_inode; - 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) - GOTO(free, rc = -EXDEV); + rc = ll_check_swap_layouts_validity(llss->inode1, llss->inode2); + if (rc < 0) + GOTO(free, rc); /* we use 2 bool because it is easier to swap than 2 bits */ if (lsl->sl_flags & SWAP_LAYOUTS_CHECK_DV1) @@ -2105,7 +1996,7 @@ static int ll_swap_layouts(struct file *file1, struct file *file2, rc = lu_fid_cmp(ll_inode2fid(llss->inode1), ll_inode2fid(llss->inode2)); if (rc == 0) /* same file, done! */ - GOTO(free, rc = 0); + GOTO(free, rc); if (rc < 0) { /* sequentialize it */ swap(llss->inode1, llss->inode2); @@ -2127,18 +2018,6 @@ static int ll_swap_layouts(struct file *file1, struct file *file2, } } - /* to be able to restore mtime and atime after swap - * we need to first save them */ - if (lsl->sl_flags & - (SWAP_LAYOUTS_KEEP_MTIME | SWAP_LAYOUTS_KEEP_ATIME)) { - llss->ia1.ia_mtime = llss->inode1->i_mtime; - llss->ia1.ia_atime = llss->inode1->i_atime; - llss->ia1.ia_valid = ATTR_MTIME | ATTR_ATIME; - llss->ia2.ia_mtime = llss->inode2->i_mtime; - llss->ia2.ia_atime = llss->inode2->i_atime; - llss->ia2.ia_valid = ATTR_MTIME | ATTR_ATIME; - } - /* ultimate check, before swaping the layouts we check if * dataversion has changed (if requested) */ if (llss->check_dv1) { @@ -2173,45 +2052,15 @@ static int ll_swap_layouts(struct file *file1, struct file *file2, sizeof(*op_data), op_data, NULL); ll_finish_md_op_data(op_data); + if (rc < 0) + GOTO(putgl, rc); + putgl: if (gid != 0) { ll_put_grouplock(llss->inode2, file2, gid); ll_put_grouplock(llss->inode1, file1, gid); } - /* rc can be set from obd_iocontrol() or from a GOTO(putgl, ...) */ - if (rc != 0) - GOTO(free, rc); - - /* clear useless flags */ - if (!(lsl->sl_flags & SWAP_LAYOUTS_KEEP_MTIME)) { - llss->ia1.ia_valid &= ~ATTR_MTIME; - llss->ia2.ia_valid &= ~ATTR_MTIME; - } - - if (!(lsl->sl_flags & SWAP_LAYOUTS_KEEP_ATIME)) { - llss->ia1.ia_valid &= ~ATTR_ATIME; - llss->ia2.ia_valid &= ~ATTR_ATIME; - } - - /* update time if requested */ - rc = 0; - if (llss->ia2.ia_valid != 0) { - mutex_lock(&llss->inode1->i_mutex); - rc = ll_setattr(file1->f_dentry, &llss->ia2); - mutex_unlock(&llss->inode1->i_mutex); - } - - if (llss->ia1.ia_valid != 0) { - int rc1; - - mutex_lock(&llss->inode2->i_mutex); - rc1 = ll_setattr(file2->f_dentry, &llss->ia1); - mutex_unlock(&llss->inode2->i_mutex); - if (rc == 0) - rc = rc1; - } - free: if (llss != NULL) OBD_FREE_PTR(llss); @@ -2219,10 +2068,15 @@ free: RETURN(rc); } -static int ll_hsm_state_set(struct inode *inode, struct hsm_state_set *hss) +int ll_hsm_state_set(struct inode *inode, struct hsm_state_set *hss) { struct md_op_data *op_data; int rc; + ENTRY; + + /* Detect out-of range masks */ + if ((hss->hss_setmask | hss->hss_clearmask) & ~HSM_FLAGS_MASK) + RETURN(-EINVAL); /* Non-root users are forbidden to set or clear flags which are * NOT defined in HSM_USER_MASK. */ @@ -2230,6 +2084,11 @@ static int ll_hsm_state_set(struct inode *inode, struct hsm_state_set *hss) !cfs_capable(CFS_CAP_SYS_ADMIN)) RETURN(-EPERM); + /* Detect out-of range archive id */ + if ((hss->hss_valid & HSS_ARCHIVE_ID) && + (hss->hss_archive_id > LL_HSM_MAX_ARCHIVE)) + RETURN(-EINVAL); + op_data = ll_prep_md_op_data(NULL, inode, NULL, NULL, 0, 0, LUSTRE_OPC_ANY, hss); if (IS_ERR(op_data)) @@ -2272,8 +2131,8 @@ static int ll_hsm_import(struct inode *inode, struct file *file, attr->ia_mode = hui->hui_mode & (S_IRWXU | S_IRWXG | S_IRWXO); attr->ia_mode |= S_IFREG; - attr->ia_uid = hui->hui_uid; - attr->ia_gid = hui->hui_gid; + 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; @@ -2285,10 +2144,14 @@ static int ll_hsm_import(struct inode *inode, struct file *file, ATTR_MTIME | ATTR_MTIME_SET | ATTR_ATIME | ATTR_ATIME_SET; - rc = ll_setattr_raw(file->f_dentry, attr, true); + mutex_lock(&inode->i_mutex); + + rc = ll_setattr_raw(file->f_path.dentry, attr, true); if (rc == -ENODATA) rc = 0; + mutex_unlock(&inode->i_mutex); + out: if (hss != NULL) OBD_FREE_PTR(hss); @@ -2299,9 +2162,52 @@ out: RETURN(rc); } -long ll_file_ioctl(struct file *file, unsigned int cmd, unsigned long arg) +static inline long ll_lease_type_from_fmode(fmode_t fmode) +{ + return ((fmode & FMODE_READ) ? LL_LEASE_RDLCK : 0) | + ((fmode & FMODE_WRITE) ? LL_LEASE_WRLCK : 0); +} + +static int ll_file_futimes_3(struct file *file, const struct ll_futimes_3 *lfu) { - struct inode *inode = file->f_dentry->d_inode; + struct inode *inode = file->f_path.dentry->d_inode; + struct iattr ia = { + .ia_valid = ATTR_ATIME | ATTR_ATIME_SET | + ATTR_MTIME | ATTR_MTIME_SET | + ATTR_CTIME | ATTR_CTIME_SET, + .ia_atime = { + .tv_sec = lfu->lfu_atime_sec, + .tv_nsec = lfu->lfu_atime_nsec, + }, + .ia_mtime = { + .tv_sec = lfu->lfu_mtime_sec, + .tv_nsec = lfu->lfu_mtime_nsec, + }, + .ia_ctime = { + .tv_sec = lfu->lfu_ctime_sec, + .tv_nsec = lfu->lfu_ctime_nsec, + }, + }; + int rc; + ENTRY; + + if (!capable(CAP_SYS_ADMIN)) + RETURN(-EPERM); + + if (!S_ISREG(inode->i_mode)) + RETURN(-EINVAL); + + mutex_lock(&inode->i_mutex); + rc = ll_setattr_raw(file->f_path.dentry, &ia, false); + mutex_unlock(&inode->i_mutex); + + RETURN(rc); +} + +static long +ll_file_ioctl(struct file *file, unsigned int cmd, unsigned long arg) +{ + struct inode *inode = file->f_path.dentry->d_inode; struct ll_file_data *fd = LUSTRE_FPRIVATE(file); int flags, rc; ENTRY; @@ -2317,14 +2223,14 @@ long ll_file_ioctl(struct file *file, unsigned int cmd, unsigned long arg) switch(cmd) { case LL_IOC_GETFLAGS: /* Get the current value of the file flags */ - return put_user(fd->fd_flags, (int *)arg); + return put_user(fd->fd_flags, (int __user *)arg); case LL_IOC_SETFLAGS: case LL_IOC_CLRFLAGS: /* Set or clear specific file flags */ /* XXX This probably needs checks to ensure the flags are * not abused, and to handle any flag side effects. */ - if (get_user(flags, (int *) arg)) + if (get_user(flags, (int __user *) arg)) RETURN(-EFAULT); if (cmd == LL_IOC_SETFLAGS) { @@ -2348,43 +2254,62 @@ long ll_file_ioctl(struct file *file, unsigned int cmd, unsigned long arg) struct file *file2; struct lustre_swap_layouts lsl; - if (copy_from_user(&lsl, (char *)arg, + if (copy_from_user(&lsl, (char __user *)arg, sizeof(struct lustre_swap_layouts))) RETURN(-EFAULT); - if ((file->f_flags & O_ACCMODE) == 0) /* O_RDONLY */ + if ((file->f_flags & O_ACCMODE) == O_RDONLY) RETURN(-EPERM); file2 = fget(lsl.sl_fd); if (file2 == NULL) RETURN(-EBADF); - rc = -EPERM; - if ((file2->f_flags & O_ACCMODE) != 0) /* O_WRONLY or O_RDWR */ + /* O_WRONLY or O_RDWR */ + if ((file2->f_flags & O_ACCMODE) == O_RDONLY) + GOTO(out, rc = -EPERM); + + if (lsl.sl_flags & SWAP_LAYOUTS_CLOSE) { + struct inode *inode2; + struct ll_inode_info *lli; + struct obd_client_handle *och = NULL; + + if (lsl.sl_flags != SWAP_LAYOUTS_CLOSE) + GOTO(out, rc = -EINVAL); + + lli = ll_i2info(inode); + 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) + GOTO(out, rc = -ENOLCK); + inode2 = file2->f_path.dentry->d_inode; + rc = ll_swap_layouts_close(och, inode, inode2); + } else { rc = ll_swap_layouts(file, file2, &lsl); + } +out: fput(file2); RETURN(rc); } - case LL_IOC_LOV_GETSTRIPE: - RETURN(ll_lov_getstripe(inode, arg)); - case LL_IOC_RECREATE_OBJ: - RETURN(ll_lov_recreate_obj(inode, arg)); - case LL_IOC_RECREATE_FID: - RETURN(ll_lov_recreate_fid(inode, arg)); - case FSFILT_IOC_FIEMAP: - RETURN(ll_ioctl_fiemap(inode, arg)); + case LL_IOC_LOV_GETSTRIPE: + RETURN(ll_file_getstripe(inode, + (struct lov_user_md __user *)arg)); case FSFILT_IOC_GETFLAGS: case FSFILT_IOC_SETFLAGS: RETURN(ll_iocontrol(inode, file, cmd, arg)); case FSFILT_IOC_GETVERSION_OLD: case FSFILT_IOC_GETVERSION: - RETURN(put_user(inode->i_generation, (int *)arg)); + RETURN(put_user(inode->i_generation, (int __user *)arg)); case LL_IOC_GROUP_LOCK: RETURN(ll_get_grouplock(inode, file, arg)); case LL_IOC_GROUP_UNLOCK: RETURN(ll_put_grouplock(inode, file, arg)); case IOC_OBD_STATFS: - RETURN(ll_obd_statfs(inode, (void *)arg)); + RETURN(ll_obd_statfs(inode, (void __user *)arg)); /* We need to special case any other ioctls we want to handle, * to send them to the MDS/OST as appropriate and to properly @@ -2395,25 +2320,29 @@ long ll_file_ioctl(struct file *file, unsigned int cmd, unsigned long arg) case LL_IOC_FLUSHCTX: RETURN(ll_flush_ctx(inode)); case LL_IOC_PATH2FID: { - if (copy_to_user((void *)arg, ll_inode2fid(inode), + if (copy_to_user((void __user *)arg, ll_inode2fid(inode), sizeof(struct lu_fid))) RETURN(-EFAULT); RETURN(0); } + case LL_IOC_GETPARENT: + RETURN(ll_getparent(file, (struct getparent __user *)arg)); + case OBD_IOC_FID2PATH: - RETURN(ll_fid2path(inode, (void *)arg)); + RETURN(ll_fid2path(inode, (void __user *)arg)); case LL_IOC_DATA_VERSION: { struct ioc_data_version idv; int rc; - if (copy_from_user(&idv, (char *)arg, sizeof(idv))) + if (copy_from_user(&idv, (char __user *)arg, sizeof(idv))) RETURN(-EFAULT); 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))) + if (rc == 0 && + copy_to_user((char __user *)arg, &idv, sizeof(idv))) RETURN(-EFAULT); RETURN(rc); @@ -2426,7 +2355,7 @@ long ll_file_ioctl(struct file *file, unsigned int cmd, unsigned long arg) if (mdtidx < 0) RETURN(mdtidx); - if (put_user((int)mdtidx, (int*)arg)) + if (put_user((int)mdtidx, (int __user *)arg)) RETURN(-EFAULT); RETURN(0); @@ -2453,7 +2382,7 @@ 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 (copy_to_user((void *)arg, hus, sizeof(*hus))) + if (copy_to_user((void __user *)arg, hus, sizeof(*hus))) rc = -EFAULT; ll_finish_md_op_data(op_data); @@ -2468,7 +2397,7 @@ long ll_file_ioctl(struct file *file, unsigned int cmd, unsigned long arg) if (hss == NULL) RETURN(-ENOMEM); - if (copy_from_user(hss, (char *)arg, sizeof(*hss))) { + if (copy_from_user(hss, (char __user *)arg, sizeof(*hss))) { OBD_FREE_PTR(hss); RETURN(-EFAULT); } @@ -2497,7 +2426,7 @@ 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 (copy_to_user((char *)arg, hca, sizeof(*hca))) + if (copy_to_user((char __user *)arg, hca, sizeof(*hca))) rc = -EFAULT; ll_finish_md_op_data(op_data); @@ -2508,20 +2437,20 @@ long ll_file_ioctl(struct file *file, unsigned int cmd, unsigned long arg) struct ll_inode_info *lli = ll_i2info(inode); struct obd_client_handle *och = NULL; bool lease_broken; - fmode_t mode = 0; + fmode_t fmode; switch (arg) { - case F_WRLCK: + case LL_LEASE_WRLCK: if (!(file->f_mode & FMODE_WRITE)) RETURN(-EPERM); - mode = FMODE_WRITE; + fmode = FMODE_WRITE; break; - case F_RDLCK: + case LL_LEASE_RDLCK: if (!(file->f_mode & FMODE_READ)) RETURN(-EPERM); - mode = FMODE_READ; + fmode = FMODE_READ; break; - case F_UNLCK: + case LL_LEASE_UNLCK: mutex_lock(&lli->lli_och_mutex); if (fd->fd_lease_och != NULL) { och = fd->fd_lease_och; @@ -2529,25 +2458,26 @@ long ll_file_ioctl(struct file *file, unsigned int cmd, unsigned long arg) } 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; - } + if (och == NULL) + RETURN(-ENOLCK); + + fmode = och->och_flags; + rc = ll_lease_close(och, inode, &lease_broken); + if (rc < 0) + RETURN(rc); + + if (lease_broken) + fmode = 0; - /* return the type of lease or error */ - RETURN(rc < 0 ? rc : (int)mode); + RETURN(ll_lease_type_from_fmode(fmode)); default: RETURN(-EINVAL); } - CDEBUG(D_INODE, "Set lease with mode %d\n", mode); + CDEBUG(D_INODE, "Set lease with mode %u\n", fmode); /* apply for lease */ - och = ll_lease_open(inode, file, mode, 0); + och = ll_lease_open(inode, file, fmode, 0); if (IS_ERR(och)) RETURN(PTR_ERR(och)); @@ -2568,8 +2498,8 @@ long ll_file_ioctl(struct file *file, unsigned int cmd, unsigned long arg) case LL_IOC_GET_LEASE: { struct ll_inode_info *lli = ll_i2info(inode); struct ldlm_lock *lock = NULL; + fmode_t fmode = 0; - rc = 0; mutex_lock(&lli->lli_och_mutex); if (fd->fd_lease_och != NULL) { struct obd_client_handle *och = fd->fd_lease_och; @@ -2578,14 +2508,15 @@ long ll_file_ioctl(struct file *file, unsigned int cmd, unsigned long arg) if (lock != NULL) { lock_res_and_lock(lock); if (!ldlm_is_cancel(lock)) - rc = och->och_flags & - (FMODE_READ | FMODE_WRITE); + fmode = och->och_flags; + unlock_res_and_lock(lock); LDLM_LOCK_PUT(lock); } } mutex_unlock(&lli->lli_och_mutex); - RETURN(rc); + + RETURN(ll_lease_type_from_fmode(fmode)); } case LL_IOC_HSM_IMPORT: { struct hsm_user_import *hui; @@ -2594,7 +2525,7 @@ long ll_file_ioctl(struct file *file, unsigned int cmd, unsigned long arg) if (hui == NULL) RETURN(-ENOMEM); - if (copy_from_user(hui, (void *)arg, sizeof(*hui))) { + if (copy_from_user(hui, (void __user *)arg, sizeof(*hui))) { OBD_FREE_PTR(hui); RETURN(-EFAULT); } @@ -2604,6 +2535,16 @@ long ll_file_ioctl(struct file *file, unsigned int cmd, unsigned long arg) OBD_FREE_PTR(hui); RETURN(rc); } + case LL_IOC_FUTIMES_3: { + struct ll_futimes_3 lfu; + + if (copy_from_user(&lfu, + (const struct ll_futimes_3 __user *)arg, + sizeof(lfu))) + RETURN(-EFAULT); + + RETURN(ll_file_futimes_3(file, &lfu)); + } default: { int err; @@ -2612,7 +2553,7 @@ long ll_file_ioctl(struct file *file, unsigned int cmd, unsigned long arg) RETURN(err); RETURN(obd_iocontrol(cmd, ll_i2dtexp(inode), 0, NULL, - (void *)arg)); + (void __user *)arg)); } } } @@ -2637,7 +2578,7 @@ 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->f_dentry->d_inode; + struct inode *inode = file->f_path.dentry->d_inode; switch (origin) { case SEEK_END: @@ -2684,9 +2625,9 @@ 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; + struct inode *inode = file->f_path.dentry->d_inode; loff_t retval, eof = 0; ENTRY; @@ -2709,9 +2650,9 @@ 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 inode *inode = file->f_path.dentry->d_inode; struct ll_inode_info *lli = ll_i2info(inode); struct ll_file_data *fd = LUSTRE_FPRIVATE(file); int rc, err; @@ -2722,9 +2663,11 @@ 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. */ @@ -2735,7 +2678,7 @@ int ll_flush(struct file *file, fl_owner_t id) /** * Called to make sure a portion of file has been written out. - * if @local_only is not true, it will send OST_SYNC RPCs to ost. + * if @mode is not CL_FSYNC_LOCAL, it will send OST_SYNC RPCs to OST. * * Return how many pages have been written. */ @@ -2745,7 +2688,6 @@ int cl_sync_file_range(struct inode *inode, loff_t start, loff_t end, struct cl_env_nest nest; struct lu_env *env; struct cl_io *io; - struct obd_capa *capa = NULL; struct cl_fsync_io *fio; int result; ENTRY; @@ -2758,15 +2700,12 @@ int cl_sync_file_range(struct inode *inode, loff_t start, loff_t end, if (IS_ERR(env)) RETURN(PTR_ERR(env)); - capa = ll_osscapa_get(inode, CAPA_OPC_OSS_WRITE); - - io = ccc_env_thread_io(env); - io->ci_obj = cl_i2info(inode)->lli_clob; + io = vvp_env_thread_io(env); + io->ci_obj = ll_i2info(inode)->lli_clob; io->ci_ignore_layout = ignore_layout; /* initialize parameters for sync */ fio = &io->u.ci_fsync; - fio->fi_capa = capa; fio->fi_start = start; fio->fi_end = end; fio->fi_fid = ll_inode2fid(inode); @@ -2782,75 +2721,72 @@ int cl_sync_file_range(struct inode *inode, loff_t start, loff_t end, cl_io_fini(env, io); cl_env_nested_put(&nest, env); - capa_put(capa); - RETURN(result); } /* - * When dentry is provided (the 'else' case), *file->f_dentry may be + * When dentry is provided (the 'else' case), *file->f_path.dentry may be * null and dentry must be used directly rather than pulled from - * *file->f_dentry as is done otherwise. + * *file->f_path.dentry as is done otherwise. */ #ifdef HAVE_FILE_FSYNC_4ARGS int ll_fsync(struct file *file, loff_t start, loff_t end, int datasync) { - struct dentry *dentry = file->f_dentry; + struct dentry *dentry = file->f_path.dentry; #elif defined(HAVE_FILE_FSYNC_2ARGS) int ll_fsync(struct file *file, int datasync) { - struct dentry *dentry = file->f_dentry; + struct dentry *dentry = file->f_path.dentry; + loff_t start = 0; + loff_t end = LLONG_MAX; #else int ll_fsync(struct file *file, struct dentry *dentry, int datasync) { + loff_t start = 0; + loff_t end = LLONG_MAX; #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; + int rc, err; + ENTRY; 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); + 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_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); + err = md_fsync(ll_i2sbi(inode)->ll_md_exp, ll_inode2fid(inode), &req); + 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, - CL_FSYNC_ALL, 0); + err = cl_sync_file_range(inode, start, end, CL_FSYNC_ALL, 0); if (rc == 0 && err < 0) rc = err; if (rc < 0) @@ -2865,9 +2801,10 @@ 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 inode *inode = file->f_path.dentry->d_inode; struct ll_sb_info *sbi = ll_i2sbi(inode); struct ldlm_enqueue_info einfo = { .ei_type = LDLM_FLOCK, @@ -2875,8 +2812,9 @@ int ll_file_flock(struct file *file, int cmd, struct file_lock *file_lock) .ei_cbdata = file_lock, }; struct md_op_data *op_data; - struct lustre_handle lockh = {0}; - ldlm_policy_data_t flock = {{0}}; + struct lustre_handle lockh = { 0 }; + union ldlm_policy_data flock = { { 0 } }; + int fl_type = file_lock->fl_type; __u64 flags = 0; int rc; int rc2 = 0; @@ -2912,7 +2850,7 @@ int ll_file_flock(struct file *file, int cmd, struct file_lock *file_lock) if (file_lock->fl_lmops && file_lock->fl_lmops->lm_compare_owner) flock.l_flock.owner = (unsigned long)file_lock->fl_pid; - switch (file_lock->fl_type) { + switch (fl_type) { case F_RDLCK: einfo.ei_mode = LCK_PR; break; @@ -2931,8 +2869,7 @@ int ll_file_flock(struct file *file, int cmd, struct file_lock *file_lock) einfo.ei_mode = LCK_PW; break; default: - CDEBUG(D_INFO, "Unknown fcntl lock type: %d\n", - file_lock->fl_type); + CDEBUG(D_INFO, "Unknown fcntl lock type: %d\n", fl_type); RETURN (-ENOTSUPP); } @@ -2954,15 +2891,16 @@ int ll_file_flock(struct file *file, int cmd, struct file_lock *file_lock) case F_GETLK64: #endif flags = LDLM_FL_TEST_LOCK; - /* Save the old mode so that if the mode in the lock changes we - * can decrement the appropriate reader or writer refcount. */ - file_lock->fl_type = einfo.ei_mode; break; default: CERROR("unknown fcntl lock command: %d\n", cmd); RETURN (-EINVAL); } + /* Save the old mode so that if the mode in the lock changes we + * can decrement the appropriate reader or writer refcount. */ + file_lock->fl_type = einfo.ei_mode; + op_data = ll_prep_md_op_data(NULL, inode, NULL, NULL, 0, 0, LUSTRE_OPC_ANY, NULL); if (IS_ERR(op_data)) @@ -2973,8 +2911,12 @@ int ll_file_flock(struct file *file, int cmd, struct file_lock *file_lock) 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); + rc = md_enqueue(sbi->ll_md_exp, &einfo, &flock, NULL, op_data, &lockh, + flags); + + /* Restore the file lock type if not TEST lock. */ + if (!(flags & LDLM_FL_TEST_LOCK)) + file_lock->fl_type = fl_type; if ((file_lock->fl_flags & FL_FLOCK) && (rc == 0 || file_lock->fl_type == F_UNLCK)) @@ -2986,8 +2928,8 @@ int ll_file_flock(struct file *file, int cmd, struct file_lock *file_lock) if (rc2 && file_lock->fl_type != F_UNLCK) { einfo.ei_mode = LCK_NL; - md_enqueue(sbi->ll_md_exp, &einfo, NULL, - op_data, &lockh, &flock, 0, NULL /* req */, flags); + md_enqueue(sbi->ll_md_exp, &einfo, &flock, NULL, op_data, + &lockh, flags); rc = rc2; } @@ -2996,7 +2938,170 @@ 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 inode **inode) +{ + 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 | OBD_MD_FLTYPE; + 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->mbo_fid1; + + if (inode != NULL) + rc = ll_prep_inode(inode, req, parent->i_sb, NULL); +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 obd_client_handle *och = NULL; + struct qstr qstr; + struct mdt_body *body; + int rc; + __u64 data_version = 0; + 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_path.dentry, &qstr); + if (dchild != NULL) { + if (dchild->d_inode != NULL) + child_inode = igrab(dchild->d_inode); + dput(dchild); + } + + if (child_inode == NULL) { + rc = ll_get_fid_by_name(parent, name, namelen, + &op_data->op_fid3, &child_inode); + if (rc != 0) + GOTO(out_free, rc); + } + + if (child_inode == NULL) + GOTO(out_free, rc = -EINVAL); + + mutex_lock(&child_inode->i_mutex); + op_data->op_fid3 = *ll_inode2fid(child_inode); + 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_unlock, rc = -EINVAL); + } + + rc = ll_get_mdt_idx_by_fid(ll_i2sbi(parent), &op_data->op_fid3); + if (rc < 0) + GOTO(out_unlock, rc); + + if (rc == mdtidx) { + CDEBUG(D_INFO, "%s: "DFID" is already on MDT%04x\n", name, + PFID(&op_data->op_fid3), mdtidx); + GOTO(out_unlock, rc = 0); + } +again: + if (S_ISREG(child_inode->i_mode)) { + och = ll_lease_open(child_inode, NULL, FMODE_WRITE, 0); + if (IS_ERR(och)) { + rc = PTR_ERR(och); + och = NULL; + GOTO(out_unlock, rc); + } + + rc = ll_data_version(child_inode, &data_version, + LL_DV_WR_FLUSH); + if (rc != 0) + GOTO(out_close, rc); + + op_data->op_handle = och->och_fh; + op_data->op_data = och->och_mod; + op_data->op_data_version = data_version; + op_data->op_lease_handle = och->och_lease_handle; + op_data->op_bias |= MDS_RENAME_MIGRATE; + } + + 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); + + if (request != NULL) { + body = req_capsule_server_get(&request->rq_pill, &RMF_MDT_BODY); + if (body == NULL) { + ptlrpc_req_finished(request); + GOTO(out_close, rc = -EPROTO); + } + + /* If the server does release layout lock, then we cleanup + * the client och here, otherwise release it in out_close: */ + if (och != NULL && + body->mbo_valid & OBD_MD_CLOSE_INTENT_EXECED) { + obd_mod_put(och->och_mod); + md_clear_open_replay_data(ll_i2sbi(parent)->ll_md_exp, + och); + och->och_fh.cookie = DEAD_HANDLE_MAGIC; + OBD_FREE_PTR(och); + och = NULL; + } + ptlrpc_req_finished(request); + } + + /* Try again if the file layout has changed. */ + if (rc == -EAGAIN && S_ISREG(child_inode->i_mode)) { + request = NULL; + goto again; + } +out_close: + if (och != NULL) /* close the file */ + ll_lease_close(och, child_inode, NULL); + if (rc == 0) + clear_nlink(child_inode); +out_unlock: + mutex_unlock(&child_inode->i_mutex); + iput(child_inode); +out_free: + 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; @@ -3013,16 +3118,16 @@ int ll_file_noflock(struct file *file, int cmd, struct file_lock *file_lock) * \param l_req_mode [IN] searched lock mode * \retval boolean, true iff all bits are found */ -int ll_have_md_lock(struct inode *inode, __u64 *bits, ldlm_mode_t l_req_mode) +int ll_have_md_lock(struct inode *inode, __u64 *bits, enum ldlm_mode l_req_mode) { - struct lustre_handle lockh; - ldlm_policy_data_t policy; - ldlm_mode_t mode = (l_req_mode == LCK_MINMODE) ? - (LCK_CR|LCK_CW|LCK_PR|LCK_PW) : l_req_mode; - struct lu_fid *fid; + struct lustre_handle lockh; + union ldlm_policy_data policy; + enum ldlm_mode mode = (l_req_mode == LCK_MINMODE) ? + (LCK_CR | LCK_CW | LCK_PR | LCK_PW) : l_req_mode; + struct lu_fid *fid; __u64 flags; - int i; - ENTRY; + int i; + ENTRY; if (!inode) RETURN(0); @@ -3054,17 +3159,17 @@ int ll_have_md_lock(struct inode *inode, __u64 *bits, ldlm_mode_t l_req_mode) RETURN(*bits == 0); } -ldlm_mode_t ll_take_md_lock(struct inode *inode, __u64 bits, - struct lustre_handle *lockh, __u64 flags, - ldlm_mode_t mode) +enum ldlm_mode ll_take_md_lock(struct inode *inode, __u64 bits, + struct lustre_handle *lockh, __u64 flags, + enum ldlm_mode mode) { - ldlm_policy_data_t policy = { .l_inodebits = {bits}}; - struct lu_fid *fid; - ldlm_mode_t rc; - ENTRY; + union ldlm_policy_data policy = { .l_inodebits = { bits } }; + struct lu_fid *fid; + enum ldlm_mode rc; + ENTRY; - fid = &ll_i2info(inode)->lli_fid; - CDEBUG(D_INFO, "trying to match res "DFID"\n", PFID(fid)); + 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, mode, lockh); @@ -3077,22 +3182,29 @@ static int ll_inode_revalidate_fini(struct inode *inode, int rc) /* Already unlinked. Just update nlink and return success */ if (rc == -ENOENT) { clear_nlink(inode); + /* If it is striped directory, and there is bad stripe + * Let's revalidate the dentry again, instead of returning + * error */ + if (S_ISDIR(inode->i_mode) && + ll_i2info(inode)->lli_lsm_md != NULL) + return 0; + /* This path cannot be hit for regular files unless in * case of obscure races, so no need to to validate * size. */ 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; @@ -3124,14 +3236,9 @@ int __ll_inode_revalidate_it(struct dentry *dentry, struct lookup_intent *it, if (IS_ERR(op_data)) RETURN(PTR_ERR(op_data)); - oit.it_create_mode |= M_CHECK_STALE; - rc = md_intent_lock(exp, op_data, NULL, 0, - /* we are not interested in name - based lookup */ - &oit, 0, &req, - ll_md_blocking_ast, 0); + rc = md_intent_lock(exp, op_data, &oit, &req, + &ll_md_blocking_ast, 0); ll_finish_md_op_data(op_data); - oit.it_create_mode &= ~M_CHECK_STALE; if (rc < 0) { rc = ll_inode_revalidate_fini(inode, rc); GOTO (out, rc); @@ -3152,17 +3259,17 @@ int __ll_inode_revalidate_it(struct dentry *dentry, struct lookup_intent *it, ll_lookup_finish_locks(&oit, dentry); } else if (!ll_have_md_lock(dentry->d_inode, &ibits, LCK_MINMODE)) { - struct ll_sb_info *sbi = ll_i2sbi(dentry->d_inode); - obd_valid valid = OBD_MD_FLGETATTR; - 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; - } + struct ll_sb_info *sbi = ll_i2sbi(dentry->d_inode); + u64 valid = OBD_MD_FLGETATTR; + struct md_op_data *op_data; + int ealen = 0; + + 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, @@ -3171,9 +3278,6 @@ int __ll_inode_revalidate_it(struct dentry *dentry, struct lookup_intent *it, RETURN(PTR_ERR(op_data)); op_data->op_valid = valid; - /* Once OBD_CONNECT_ATTRFID is not supported, we can't find one - * capa for this inode. Because we only keep capas of dirs - * fresh. */ rc = md_getattr(sbi->ll_md_exp, op_data, &req); ll_finish_md_op_data(op_data); if (rc) { @@ -3195,28 +3299,29 @@ static int ll_merge_md_attr(struct inode *inode) LASSERT(ll_i2info(inode)->lli_lsm_md != NULL); rc = md_merge_attr(ll_i2mdexp(inode), ll_i2info(inode)->lli_lsm_md, - &attr); + &attr, ll_md_blocking_ast); 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; + set_nlink(inode, attr.cat_nlink); + inode->i_blocks = attr.cat_blocks; + i_size_write(inode, attr.cat_size); - 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; + ll_i2info(inode)->lli_atime = attr.cat_atime; + ll_i2info(inode)->lli_mtime = attr.cat_mtime; + ll_i2info(inode)->lli_ctime = attr.cat_ctime; RETURN(0); } -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; int rc; ENTRY; - rc = __ll_inode_revalidate_it(dentry, it, ibits); + rc = __ll_inode_revalidate(dentry, ibits); if (rc != 0) RETURN(rc); @@ -3229,9 +3334,9 @@ int ll_inode_revalidate_it(struct dentry *dentry, struct lookup_intent *it, 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; + LTIME_S(inode->i_atime) = ll_i2info(inode)->lli_atime; + LTIME_S(inode->i_mtime) = ll_i2info(inode)->lli_mtime; + LTIME_S(inode->i_ctime) = ll_i2info(inode)->lli_ctime; } else { /* In case of restore, the MDT has the right size and has * already send it back without granting the layout lock, @@ -3240,22 +3345,21 @@ int ll_inode_revalidate_it(struct dentry *dentry, struct lookup_intent *it, * 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)) + if (!ll_file_test_flag(ll_i2info(inode), LLIF_FILE_RESTORING)) rc = ll_glimpse_size(inode); } 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) @@ -3274,60 +3378,53 @@ int ll_getattr_it(struct vfsmount *mnt, struct dentry *de, stat->mtime = inode->i_mtime; stat->ctime = inode->i_ctime; stat->blksize = 1 << inode->i_blkbits; - 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); - } + stat->nlink = inode->i_nlink; + stat->size = i_size_read(inode); + stat->blocks = inode->i_blocks; 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); -} -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; - struct ll_user_fiemap *fiemap; - unsigned int extent_count = fieinfo->fi_extents_max; + int rc; + size_t num_bytes; + struct fiemap *fiemap; + unsigned int extent_count = fieinfo->fi_extents_max; - num_bytes = sizeof(*fiemap) + (extent_count * - sizeof(struct ll_fiemap_extent)); - OBD_ALLOC_LARGE(fiemap, num_bytes); - - if (fiemap == NULL) - RETURN(-ENOMEM); + num_bytes = sizeof(*fiemap) + (extent_count * + sizeof(struct fiemap_extent)); + OBD_ALLOC_LARGE(fiemap, num_bytes); - fiemap->fm_flags = fieinfo->fi_flags; - fiemap->fm_extent_count = fieinfo->fi_extents_max; - fiemap->fm_start = start; - fiemap->fm_length = len; - memcpy(&fiemap->fm_extents[0], fieinfo->fi_extents_start, - sizeof(struct ll_fiemap_extent)); + if (fiemap == NULL) + RETURN(-ENOMEM); - rc = ll_do_fiemap(inode, fiemap, num_bytes); + fiemap->fm_flags = fieinfo->fi_flags; + fiemap->fm_extent_count = fieinfo->fi_extents_max; + fiemap->fm_start = start; + fiemap->fm_length = len; + if (extent_count > 0 && + copy_from_user(&fiemap->fm_extents[0], fieinfo->fi_extents_start, + sizeof(struct fiemap_extent)) != 0) + GOTO(out, rc = -EFAULT); - fieinfo->fi_flags = fiemap->fm_flags; - fieinfo->fi_extents_mapped = fiemap->fm_mapped_extents; - memcpy(fieinfo->fi_extents_start, &fiemap->fm_extents[0], - fiemap->fm_mapped_extents * sizeof(struct ll_fiemap_extent)); + rc = ll_do_fiemap(inode, fiemap, num_bytes); - OBD_FREE_LARGE(fiemap, num_bytes); - return rc; + fieinfo->fi_flags = fiemap->fm_flags; + fieinfo->fi_extents_mapped = fiemap->fm_mapped_extents; + if (extent_count > 0 && + copy_to_user(fieinfo->fi_extents_start, &fiemap->fm_extents[0], + fiemap->fm_mapped_extents * + sizeof(struct fiemap_extent)) != 0) + GOTO(out, rc = -EFAULT); +out: + OBD_FREE_LARGE(fiemap, num_bytes); + return rc; } -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; @@ -3383,8 +3480,14 @@ int ll_inode_permission(struct inode *inode, int mask, struct nameidata *nd) # endif #endif { - int rc = 0; - ENTRY; + int rc = 0; + struct ll_sb_info *sbi; + struct root_squash_info *squash; + struct cred *cred = NULL; + const struct cred *old_cred = NULL; + cfs_cap_t cap; + bool squash_id = false; + ENTRY; #ifdef MAY_NOT_BLOCK if (mask & MAY_NOT_BLOCK) @@ -3398,10 +3501,8 @@ 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); } @@ -3409,64 +3510,126 @@ int ll_inode_permission(struct inode *inode, int mask, struct nameidata *nd) 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); + /* squash fsuid/fsgid if needed */ + sbi = ll_i2sbi(inode); + squash = &sbi->ll_squash; + if (unlikely(squash->rsi_uid != 0 && + uid_eq(current_fsuid(), GLOBAL_ROOT_UID) && + !(sbi->ll_flags & LL_SBI_NOROOTSQUASH))) { + squash_id = true; + } + if (squash_id) { + CDEBUG(D_OTHER, "squash creds (%d:%d)=>(%d:%d)\n", + __kuid_val(current_fsuid()), __kgid_val(current_fsgid()), + squash->rsi_uid, squash->rsi_gid); + + /* update current process's credentials + * and FS capability */ + cred = prepare_creds(); + if (cred == NULL) + RETURN(-ENOMEM); + + 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) + cap_lower(cred->cap_effective, cap); + } + old_cred = override_creds(cred); + } + + ll_stats_ops_tally(sbi, LPROC_LL_INODE_PERM, 1); - ll_stats_ops_tally(ll_i2sbi(inode), LPROC_LL_INODE_PERM, 1); - rc = ll_generic_permission(inode, mask, flags, ll_check_acl); + if (sbi->ll_flags & LL_SBI_RMT_CLIENT) + rc = lustre_check_remote_perm(inode, mask); + else + rc = ll_generic_permission(inode, mask, flags, ll_check_acl); + + /* restore current process's credentials and FS capability */ + if (squash_id) { + revert_creds(old_cred); + put_cred(cred); + } RETURN(rc); } /* -o localflock - only provides locally consistent flock locks */ struct file_operations ll_file_operations = { - .read = ll_file_read, - .aio_read = ll_file_aio_read, - .write = ll_file_write, - .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, - .splice_read = ll_file_splice_read, - .fsync = ll_fsync, - .flush = ll_flush +#ifdef HAVE_FILE_OPERATIONS_READ_WRITE_ITER +# ifdef HAVE_SYNC_READ_WRITE + .read = new_sync_read, + .write = new_sync_write, +# endif + .read_iter = ll_file_read_iter, + .write_iter = ll_file_write_iter, +#else /* !HAVE_FILE_OPERATIONS_READ_WRITE_ITER */ + .read = ll_file_read, + .aio_read = ll_file_aio_read, + .write = ll_file_write, + .aio_write = ll_file_aio_write, +#endif /* HAVE_FILE_OPERATIONS_READ_WRITE_ITER */ + .unlocked_ioctl = ll_file_ioctl, + .open = ll_file_open, + .release = ll_file_release, + .mmap = ll_file_mmap, + .llseek = ll_file_seek, + .splice_read = ll_file_splice_read, + .fsync = ll_fsync, + .flush = ll_flush }; struct file_operations ll_file_operations_flock = { - .read = ll_file_read, - .aio_read = ll_file_aio_read, - .write = ll_file_write, - .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, - .splice_read = ll_file_splice_read, - .fsync = ll_fsync, - .flush = ll_flush, - .flock = ll_file_flock, - .lock = ll_file_flock +#ifdef HAVE_FILE_OPERATIONS_READ_WRITE_ITER +# ifdef HAVE_SYNC_READ_WRITE + .read = new_sync_read, + .write = new_sync_write, +# endif /* HAVE_SYNC_READ_WRITE */ + .read_iter = ll_file_read_iter, + .write_iter = ll_file_write_iter, +#else /* !HAVE_FILE_OPERATIONS_READ_WRITE_ITER */ + .read = ll_file_read, + .aio_read = ll_file_aio_read, + .write = ll_file_write, + .aio_write = ll_file_aio_write, +#endif /* HAVE_FILE_OPERATIONS_READ_WRITE_ITER */ + .unlocked_ioctl = ll_file_ioctl, + .open = ll_file_open, + .release = ll_file_release, + .mmap = ll_file_mmap, + .llseek = ll_file_seek, + .splice_read = ll_file_splice_read, + .fsync = ll_fsync, + .flush = ll_flush, + .flock = ll_file_flock, + .lock = ll_file_flock }; /* These are for -o noflock - to return ENOSYS on flock calls */ struct file_operations ll_file_operations_noflock = { - .read = ll_file_read, - .aio_read = ll_file_aio_read, - .write = ll_file_write, - .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, - .splice_read = ll_file_splice_read, - .fsync = ll_fsync, - .flush = ll_flush, - .flock = ll_file_noflock, - .lock = ll_file_noflock +#ifdef HAVE_FILE_OPERATIONS_READ_WRITE_ITER +# ifdef HAVE_SYNC_READ_WRITE + .read = new_sync_read, + .write = new_sync_write, +# endif /* HAVE_SYNC_READ_WRITE */ + .read_iter = ll_file_read_iter, + .write_iter = ll_file_write_iter, +#else /* !HAVE_FILE_OPERATIONS_READ_WRITE_ITER */ + .read = ll_file_read, + .aio_read = ll_file_aio_read, + .write = ll_file_write, + .aio_write = ll_file_aio_write, +#endif /* HAVE_FILE_OPERATIONS_READ_WRITE_ITER */ + .unlocked_ioctl = ll_file_ioctl, + .open = ll_file_open, + .release = ll_file_release, + .mmap = ll_file_mmap, + .llseek = ll_file_seek, + .splice_read = ll_file_splice_read, + .fsync = ll_fsync, + .flush = ll_flush, + .flock = ll_file_noflock, + .lock = ll_file_noflock }; struct inode_operations ll_file_inode_operations = { @@ -3486,15 +3649,15 @@ struct inode_operations ll_file_inode_operations = { /* dynamic ioctl number support routins */ static struct llioc_ctl_data { struct rw_semaphore ioc_sem; - cfs_list_t ioc_head; + struct list_head ioc_head; } llioc = { - __RWSEM_INITIALIZER(llioc.ioc_sem), - CFS_LIST_HEAD_INIT(llioc.ioc_head) + __RWSEM_INITIALIZER(llioc.ioc_sem), + LIST_HEAD_INIT(llioc.ioc_head) }; struct llioc_data { - cfs_list_t iocd_list; + struct list_head iocd_list; unsigned int iocd_size; llioc_callback_t iocd_cb; unsigned int iocd_count; @@ -3523,7 +3686,7 @@ void *ll_iocontrol_register(llioc_callback_t cb, int count, unsigned int *cmd) memcpy(in_data->iocd_cmd, cmd, sizeof(unsigned int) * count); down_write(&llioc.ioc_sem); - cfs_list_add_tail(&in_data->iocd_list, &llioc.ioc_head); + list_add_tail(&in_data->iocd_list, &llioc.ioc_head); up_write(&llioc.ioc_sem); RETURN(in_data); @@ -3537,11 +3700,11 @@ void ll_iocontrol_unregister(void *magic) return; down_write(&llioc.ioc_sem); - cfs_list_for_each_entry(tmp, &llioc.ioc_head, iocd_list) { + list_for_each_entry(tmp, &llioc.ioc_head, iocd_list) { if (tmp == magic) { unsigned int size = tmp->iocd_size; - cfs_list_del(&tmp->iocd_list); + list_del(&tmp->iocd_list); up_write(&llioc.ioc_sem); OBD_FREE(tmp, size); @@ -3556,15 +3719,16 @@ 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; int rc = -EINVAL, i; down_read(&llioc.ioc_sem); - cfs_list_for_each_entry(data, &llioc.ioc_head, iocd_list) { + list_for_each_entry(data, &llioc.ioc_head, iocd_list) { for (i = 0; i < data->iocd_count; i++) { if (cmd != data->iocd_cmd[i]) continue; @@ -3586,35 +3750,53 @@ enum llioc_iter ll_iocontrol_call(struct inode *inode, struct file *file, int ll_layout_conf(struct inode *inode, const struct cl_object_conf *conf) { struct ll_inode_info *lli = ll_i2info(inode); + struct cl_object *obj = lli->lli_clob; struct cl_env_nest nest; struct lu_env *env; - int result; + int rc; ENTRY; - if (lli->lli_clob == NULL) + if (obj == NULL) RETURN(0); env = cl_env_nested_get(&nest); if (IS_ERR(env)) RETURN(PTR_ERR(env)); - result = cl_conf_set(env, lli->lli_clob, conf); - cl_env_nested_put(&nest, env); + rc = cl_conf_set(env, lli->lli_clob, conf); + if (rc < 0) + GOTO(out, rc); if (conf->coc_opc == OBJECT_CONF_SET) { struct ldlm_lock *lock = conf->coc_lock; + struct cl_layout cl = { + .cl_layout_gen = 0, + }; LASSERT(lock != NULL); LASSERT(ldlm_has_layout(lock)); - if (result == 0) { - /* 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); - } + + /* 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); + + rc = cl_object_layout_get(env, obj, &cl); + if (rc < 0) + GOTO(out, rc); + + CDEBUG(D_VFSTRACE, + DFID": layout version change: %u -> %u\n", + PFID(&lli->lli_fid), ll_layout_version_get(lli), + cl.cl_layout_gen); + ll_layout_version_set(lli, cl.cl_layout_gen); } - RETURN(result); + +out: + cl_env_nested_put(&nest, env); + + RETURN(rc); } /* Fetch layout from MDT with getxattr request, if it's not ready yet */ @@ -3622,7 +3804,6 @@ static int ll_layout_fetch(struct inode *inode, struct ldlm_lock *lock) { struct ll_sb_info *sbi = ll_i2sbi(inode); - struct obd_capa *oc; struct ptlrpc_request *req; struct mdt_body *body; void *lvbdata; @@ -3635,7 +3816,7 @@ static int ll_layout_fetch(struct inode *inode, struct ldlm_lock *lock) 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)) + if (lock->l_lvb_data != NULL) RETURN(0); /* if layout lock was granted right away, the layout is returned @@ -3643,21 +3824,19 @@ static int ll_layout_fetch(struct inode *inode, struct ldlm_lock *lock) * blocked and then granted via completion ast, we have to fetch * 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, + rc = md_getxattr(sbi->ll_md_exp, ll_inode2fid(inode), OBD_MD_FLXATTR, XATTR_NAME_LOV, NULL, 0, lmmsize, 0, &req); - capa_put(oc); if (rc < 0) 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; + lmmsize = body->mbo_eadatasize; if (lmmsize == 0) /* empty layout */ GOTO(out, rc = 0); @@ -3671,13 +3850,17 @@ 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) - OBD_FREE_LARGE(lock->l_lvb_data, lock->l_lvb_len); - - lock->l_lvb_data = lvbdata; - lock->l_lvb_len = lmmsize; + if (unlikely(lock->l_lvb_data == NULL)) { + lock->l_lvb_type = LVB_T_LAYOUT; + lock->l_lvb_data = lvbdata; + lock->l_lvb_len = lmmsize; + lvbdata = NULL; + } unlock_res_and_lock(lock); + if (lvbdata != NULL) + OBD_FREE_LARGE(lvbdata, lmmsize); + EXIT; out: @@ -3689,13 +3872,12 @@ out: * Apply the layout to the inode. Layout lock is held and will be released * in this function. */ -static int ll_layout_lock_set(struct lustre_handle *lockh, ldlm_mode_t mode, - struct inode *inode, __u32 *gen, bool reconf) +static int ll_layout_lock_set(struct lustre_handle *lockh, enum ldlm_mode mode, + struct inode *inode) { struct ll_inode_info *lli = ll_i2info(inode); struct ll_sb_info *sbi = ll_i2sbi(inode); struct ldlm_lock *lock; - struct lustre_md md = { NULL }; struct cl_object_conf conf; int rc = 0; bool lvb_ready; @@ -3708,8 +3890,8 @@ 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 "DFID"(%p) being reconfigured: %d\n", - PFID(&lli->lli_fid), inode, reconf); + LDLM_DEBUG(lock, "file "DFID"(%p) being reconfigured", + PFID(&lli->lli_fid), inode); /* in case this is a caching lock and reinstate with new inode */ md_set_lock_data(sbi->ll_md_exp, &lockh->cookie, inode, NULL); @@ -3720,54 +3902,27 @@ static int ll_layout_lock_set(struct lustre_handle *lockh, ldlm_mode_t mode, /* 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; - rc = 0; - } - GOTO(out, rc); - } + if (lvb_ready) + GOTO(out, rc = 0); rc = ll_layout_fetch(inode, lock); if (rc < 0) GOTO(out, rc); - /* for layout lock, lmm is returned in lock's lvb. + /* for layout lock, lmm is stored in lock's lvb. * lvb_data is immutable if the lock is held so it's safe to access it - * without res lock. See the description in ldlm_lock_decref_internal() - * for the condition to free lvb_data of layout lock */ - if (lock->l_lvb_data != NULL) { - rc = obd_unpackmd(sbi->ll_dt_exp, &md.lsm, - lock->l_lvb_data, lock->l_lvb_len); - if (rc >= 0) { - *gen = LL_LAYOUT_GEN_EMPTY; - if (md.lsm != NULL) - *gen = md.lsm->lsm_layout_gen; - rc = 0; - } else { - CERROR("%s: file "DFID" unpackmd error: %d\n", - ll_get_fsname(inode->i_sb, NULL, 0), - PFID(&lli->lli_fid), rc); - } - } - if (rc < 0) - GOTO(out, rc); - - /* set layout to file. Unlikely this will fail as old layout was + * without res lock. + * + * set layout to file. Unlikely this will fail as old layout was * surely eliminated */ memset(&conf, 0, sizeof conf); conf.coc_opc = OBJECT_CONF_SET; conf.coc_inode = inode; conf.coc_lock = lock; - conf.u.coc_md = &md; + conf.u.coc_layout.lb_buf = lock->l_lvb_data; + conf.u.coc_layout.lb_len = lock->l_lvb_len; rc = ll_layout_conf(inode, &conf); - if (md.lsm != NULL) - obd_free_memmd(sbi->ll_dt_exp, &md.lsm); - /* refresh layout failed, need to wait */ wait_layout = rc == -EBUSY; EXIT; @@ -3796,91 +3951,51 @@ out: RETURN(rc); } -/** - * This function checks if there exists a LAYOUT lock on the client side, - * or enqueues it if it doesn't have one in cache. - * - * This function will not hold layout lock so it may be revoked any time after - * this function returns. Any operations depend on layout should be redone - * in that case. - * - * This function should be called before lov_io_init() to get an uptodate - * layout version, the caller should save the version number and after IO - * is finished, this function should be called again to verify that layout - * is not changed during IO time. - */ -int ll_layout_refresh(struct inode *inode, __u32 *gen) +static int ll_layout_refresh_locked(struct inode *inode) { struct ll_inode_info *lli = ll_i2info(inode); struct ll_sb_info *sbi = ll_i2sbi(inode); struct md_op_data *op_data; - struct lookup_intent it; - struct lustre_handle lockh; - ldlm_mode_t mode; + struct lookup_intent it; + struct lustre_handle lockh; + enum ldlm_mode mode; struct ldlm_enqueue_info einfo = { .ei_type = LDLM_IBITS, .ei_mode = LCK_CR, - .ei_cb_bl = ll_md_blocking_ast, - .ei_cb_cp = ldlm_completion_ast, + .ei_cb_bl = &ll_md_blocking_ast, + .ei_cb_cp = &ldlm_completion_ast, }; int rc; ENTRY; - *gen = lli->lli_layout_gen; - if (!(sbi->ll_flags & LL_SBI_LAYOUT_LOCK)) - RETURN(0); - - /* sanity checks */ - LASSERT(fid_is_sane(ll_inode2fid(inode))); - LASSERT(S_ISREG(inode->i_mode)); - +again: /* 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, 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, - LCK_CR | LCK_CW | LCK_PR | LCK_PW); - if (mode != 0) { /* hit cached lock */ - rc = ll_layout_lock_set(&lockh, mode, inode, gen, true); + rc = ll_layout_lock_set(&lockh, mode, inode); if (rc == -EAGAIN) goto again; - mutex_unlock(&lli->lli_layout_mutex); RETURN(rc); } op_data = ll_prep_md_op_data(NULL, inode, inode, NULL, 0, 0, LUSTRE_OPC_ANY, NULL); - if (IS_ERR(op_data)) { - mutex_unlock(&lli->lli_layout_mutex); + if (IS_ERR(op_data)) RETURN(PTR_ERR(op_data)); - } /* have to enqueue one */ memset(&it, 0, sizeof(it)); it.it_op = IT_LAYOUT; lockh.cookie = 0ULL; - LDLM_DEBUG_NOLOCK("%s: requeue layout lock for file "DFID"(%p)\n", + LDLM_DEBUG_NOLOCK("%s: requeue layout lock for file "DFID"(%p)", 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); + rc = md_enqueue(sbi->ll_md_exp, &einfo, NULL, &it, op_data, &lockh, 0); if (it.d.lustre.it_data != NULL) ptlrpc_req_finished(it.d.lustre.it_data); it.d.lustre.it_data = NULL; @@ -3894,10 +4009,51 @@ again: if (rc == 0) { /* set lock data in case this is a new lock */ ll_set_lock_data(sbi->ll_md_exp, inode, &it, NULL); - rc = ll_layout_lock_set(&lockh, mode, inode, gen, true); + rc = ll_layout_lock_set(&lockh, mode, inode); if (rc == -EAGAIN) goto again; } + + RETURN(rc); +} + +/** + * This function checks if there exists a LAYOUT lock on the client side, + * or enqueues it if it doesn't have one in cache. + * + * This function will not hold layout lock so it may be revoked any time after + * this function returns. Any operations depend on layout should be redone + * in that case. + * + * This function should be called before lov_io_init() to get an uptodate + * layout version, the caller should save the version number and after IO + * is finished, this function should be called again to verify that layout + * is not changed during IO time. + */ +int ll_layout_refresh(struct inode *inode, __u32 *gen) +{ + struct ll_inode_info *lli = ll_i2info(inode); + struct ll_sb_info *sbi = ll_i2sbi(inode); + int rc; + ENTRY; + + *gen = ll_layout_version_get(lli); + if (!(sbi->ll_flags & LL_SBI_LAYOUT_LOCK) || *gen != CL_LAYOUT_GEN_NONE) + RETURN(0); + + /* sanity checks */ + LASSERT(fid_is_sane(ll_inode2fid(inode))); + LASSERT(S_ISREG(inode->i_mode)); + + /* take layout lock mutex to enqueue layout lock exclusively. */ + mutex_lock(&lli->lli_layout_mutex); + + rc = ll_layout_refresh_locked(inode); + if (rc < 0) + GOTO(out, rc); + + *gen = ll_layout_version_get(lli); +out: mutex_unlock(&lli->lli_layout_mutex); RETURN(rc); @@ -3926,9 +4082,8 @@ int ll_layout_restore(struct inode *inode, loff_t offset, __u64 length) 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, + rc = obd_iocontrol(LL_IOC_HSM_REQUEST, ll_i2sbi(inode)->ll_md_exp, len, hur, NULL); OBD_FREE(hur, len); RETURN(rc); } -