X-Git-Url: https://git.whamcloud.com/?a=blobdiff_plain;ds=sidebyside;f=lustre%2Fllite%2Ffile.c;h=ca7dd438a897f87e8b2dfe01f705ceb837c9faa4;hb=dd0adebec5edade78562c0168169ec024bf33a44;hp=492152c5cc156b9397ca086ea3ac3a5313f4b659;hpb=2841be335687840cf98961e6c6cde6ee9312e4d7;p=fs%2Flustre-release.git diff --git a/lustre/llite/file.c b/lustre/llite/file.c index 492152c..ca7dd43 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, 2012, Intel Corporation. + * Copyright (c) 2011, 2013, Intel Corporation. */ /* * This file is part of Lustre, http://www.lustre.org/ @@ -54,8 +54,12 @@ struct ll_file_data *ll_file_data_get(void) { struct ll_file_data *fd; - OBD_SLAB_ALLOC_PTR_GFP(fd, ll_file_data_slab, CFS_ALLOC_IO); + OBD_SLAB_ALLOC_PTR_GFP(fd, ll_file_data_slab, __GFP_IO); + if (fd == NULL) + return NULL; + fd->fd_write_failed = false; + return fd; } @@ -95,8 +99,9 @@ static void ll_prepare_close(struct inode *inode, struct md_op_data *op_data, { ENTRY; - op_data->op_attr.ia_valid = ATTR_MODE | ATTR_ATIME_SET | - ATTR_MTIME_SET | ATTR_CTIME_SET; + 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; @@ -438,24 +443,19 @@ void ll_ioepoch_open(struct ll_inode_info *lli, __u64 ioepoch) } } -static int ll_och_fill(struct obd_export *md_exp, struct ll_inode_info *lli, - struct lookup_intent *it, struct obd_client_handle *och) +static int ll_och_fill(struct obd_export *md_exp, struct lookup_intent *it, + struct obd_client_handle *och) { - struct ptlrpc_request *req = it->d.lustre.it_data; - struct mdt_body *body; - - LASSERT(och); + struct ptlrpc_request *req = it->d.lustre.it_data; + struct mdt_body *body; - body = req_capsule_server_get(&req->rq_pill, &RMF_MDT_BODY); - LASSERT(body != NULL); /* reply already checked out */ - - memcpy(&och->och_fh, &body->handle, sizeof(body->handle)); - och->och_magic = OBD_CLIENT_HANDLE_MAGIC; - och->och_fid = lli->lli_fid; - och->och_flags = it->it_flags; - ll_ioepoch_open(lli, body->ioepoch); + body = req_capsule_server_get(&req->rq_pill, &RMF_MDT_BODY); + och->och_fh = body->handle; + och->och_fid = body->fid1; + och->och_magic = OBD_CLIENT_HANDLE_MAGIC; + och->och_flags = it->it_flags; - return md_set_open_replay_data(md_exp, och, req); + return md_set_open_replay_data(md_exp, och, req); } int ll_local_open(struct file *file, struct lookup_intent *it, @@ -474,21 +474,19 @@ int ll_local_open(struct file *file, struct lookup_intent *it, struct mdt_body *body; int rc; - rc = ll_och_fill(ll_i2sbi(inode)->ll_md_exp, lli, it, och); - if (rc) - RETURN(rc); + rc = ll_och_fill(ll_i2sbi(inode)->ll_md_exp, it, och); + if (rc != 0) + RETURN(rc); - body = req_capsule_server_get(&req->rq_pill, &RMF_MDT_BODY); - if ((it->it_flags & FMODE_WRITE) && - (body->valid & OBD_MD_FLSIZE)) - CDEBUG(D_INODE, "Epoch "LPU64" opened on "DFID"\n", - lli->lli_ioepoch, PFID(&lli->lli_fid)); - } + body = req_capsule_server_get(&req->rq_pill, &RMF_MDT_BODY); + ll_ioepoch_open(lli, body->ioepoch); + } + + LUSTRE_FPRIVATE(file) = fd; + ll_readahead_init(inode, &fd->fd_ras); + fd->fd_omode = it->it_flags; - LUSTRE_FPRIVATE(file) = fd; - ll_readahead_init(inode, &fd->fd_ras); - fd->fd_omode = it->it_flags; - RETURN(0); + RETURN(0); } /* Open a file, and (for the very first open) create objects on the OSTs at @@ -521,12 +519,12 @@ int ll_file_open(struct inode *inode, struct file *file) it = file->private_data; /* XXX: compat macro */ file->private_data = NULL; /* prevent ll_local_open assertion */ - fd = ll_file_data_get(); - if (fd == NULL) - GOTO(out_och_free, rc = -ENOMEM); + fd = ll_file_data_get(); + if (fd == NULL) + GOTO(out_openerr, rc = -ENOMEM); - fd->fd_file = file; - if (S_ISDIR(inode->i_mode)) { + 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) { @@ -535,7 +533,7 @@ int ll_file_open(struct inode *inode, struct file *file) opendir_set = 1; } spin_unlock(&lli->lli_sa_lock); - } + } if (inode->i_sb->s_root == file->f_dentry) { LUSTRE_FPRIVATE(file) = fd; @@ -705,8 +703,7 @@ static int ll_lsm_getattr(struct lov_stripe_md *lsm, struct obd_export *exp, oinfo.oi_md = lsm; oinfo.oi_oa = obdo; - oinfo.oi_oa->o_id = lsm->lsm_object_id; - oinfo.oi_oa->o_seq = lsm->lsm_object_seq; + 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 | @@ -756,10 +753,11 @@ int ll_inode_getattr(struct inode *inode, struct obdo *obdo, capa, obdo, ioepoch, sync); 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 "LPX64" size %llu, blocks %llu, blksize %lu\n", - lsm ? lsm->lsm_object_id : 0, i_size_read(inode), + 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)); } @@ -922,12 +920,11 @@ out: cl_io_fini(env, io); /* If any bit been read/written (result != 0), we just return * short read/write instead of restart io. */ - if (result == 0 && io->ci_need_restart) { + if ((result == 0 || result == -ENODATA) && io->ci_need_restart) { CDEBUG(D_VFSTRACE, "Restart %s on %s from %lld, count:%zd\n", iot == CIT_READ ? "read" : "write", file->f_dentry->d_name.name, *ppos, count); - LASSERTF(io->u.ci_rw.crw_count == count, "%zd != %zd\n", - io->u.ci_rw.crw_count, count); + LASSERTF(io->ci_nob == 0, "%zd", io->ci_nob); goto restart; } @@ -1202,35 +1199,6 @@ static ssize_t ll_file_write(struct file *file, const char *buf, size_t count, } #endif - -#ifdef HAVE_KERNEL_SENDFILE -/* - * Send file content (through pagecache) somewhere with helper - */ -static ssize_t ll_file_sendfile(struct file *in_file, loff_t *ppos,size_t count, - read_actor_t actor, void *target) -{ - struct lu_env *env; - struct vvp_io_args *args; - ssize_t result; - int refcheck; - ENTRY; - - env = cl_env_get(&refcheck); - if (IS_ERR(env)) - RETURN(PTR_ERR(env)); - - args = vvp_env_args(env, IO_SENDFILE); - args->u.sendfile.via_target = target; - args->u.sendfile.via_actor = actor; - - result = ll_file_io_generic(env, args, in_file, CIT_READ, ppos, count); - cl_env_put(env, &refcheck); - RETURN(result); -} -#endif - -#ifdef HAVE_KERNEL_SPLICE_READ /* * Send file content (through pagecache) somewhere with helper */ @@ -1256,9 +1224,8 @@ static ssize_t ll_file_splice_read(struct file *in_file, loff_t *ppos, cl_env_put(env, &refcheck); RETURN(result); } -#endif -static int ll_lov_recreate(struct inode *inode, obd_id id, obd_seq seq, +static int ll_lov_recreate(struct inode *inode, struct ost_id *oi, obd_count ost_idx) { struct obd_export *exp = ll_i2dtexp(inode); @@ -1274,7 +1241,7 @@ static int ll_lov_recreate(struct inode *inode, obd_id id, obd_seq seq, RETURN(-ENOMEM); lsm = ccc_inode_lsm_get(inode); - if (lsm == NULL) + if (!lsm_has_objects(lsm)) GOTO(out, rc = -ENOENT); lsm_size = sizeof(*lsm) + (sizeof(struct lov_oinfo) * @@ -1284,8 +1251,7 @@ static int ll_lov_recreate(struct inode *inode, obd_id id, obd_seq seq, if (lsm2 == NULL) GOTO(out, rc = -ENOMEM); - oa->o_id = id; - oa->o_seq = seq; + 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; @@ -1308,6 +1274,7 @@ out: 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)) @@ -1317,14 +1284,15 @@ static int ll_lov_recreate_obj(struct inode *inode, unsigned long arg) sizeof(ucreat))) RETURN(-EFAULT); - RETURN(ll_lov_recreate(inode, ucreat.lrc_id, 0, - ucreat.lrc_ost_idx)); + 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; - obd_id id; + struct ost_id oi; obd_count ost_idx; ENTRY; @@ -1334,9 +1302,9 @@ static int ll_lov_recreate_fid(struct inode *inode, unsigned long arg) if (copy_from_user(&fid, (struct lu_fid *)arg, sizeof(fid))) RETURN(-EFAULT); - id = fid_oid(&fid) | ((fid_seq(&fid) & 0xffff) << 32); + fid_to_ostid(&fid, &oi); ost_idx = (fid_seq(&fid) >> 16) & 0xffff; - RETURN(ll_lov_recreate(inode, id, 0, ost_idx)); + RETURN(ll_lov_recreate(inode, &oi, ost_idx)); } int ll_lov_setstripe_ea_info(struct inode *inode, struct file *file, @@ -1429,6 +1397,12 @@ int ll_lov_getstripe_ea_info(struct inode *inode, const char *filename, * passing it to userspace. */ if (LOV_MAGIC != cpu_to_le32(LOV_MAGIC)) { + int stripe_count; + + stripe_count = le16_to_cpu(lmm->lmm_stripe_count); + if (le32_to_cpu(lmm->lmm_pattern) & LOV_PATTERN_F_RELEASED) + stripe_count = 0; + /* if function called for directory - we should * avoid swab not existent lsm objects */ if (lmm->lmm_magic == cpu_to_le32(LOV_MAGIC_V1)) { @@ -1436,13 +1410,13 @@ int ll_lov_getstripe_ea_info(struct inode *inode, const char *filename, if (S_ISREG(body->mode)) lustre_swab_lov_user_md_objects( ((struct lov_user_md_v1 *)lmm)->lmm_objects, - ((struct lov_user_md_v1 *)lmm)->lmm_stripe_count); + stripe_count); } else if (lmm->lmm_magic == cpu_to_le32(LOV_MAGIC_V3)) { lustre_swab_lov_user_md_v3((struct lov_user_md_v3 *)lmm); if (S_ISREG(body->mode)) lustre_swab_lov_user_md_objects( ((struct lov_user_md_v3 *)lmm)->lmm_objects, - ((struct lov_user_md_v3 *)lmm)->lmm_stripe_count); + stripe_count); } } @@ -1639,8 +1613,7 @@ int ll_release_openhandle(struct dentry *dentry, struct lookup_intent *it) if (!och) GOTO(out, rc = -ENOMEM); - ll_och_fill(ll_i2sbi(inode)->ll_md_exp, - ll_i2info(inode), it, och); + ll_och_fill(ll_i2sbi(inode)->ll_md_exp, it, och); rc = ll_close_inode_openhandle(ll_i2sbi(inode)->ll_md_exp, inode, och); @@ -1691,8 +1664,7 @@ int ll_do_fiemap(struct inode *inode, struct ll_user_fiemap *fiemap, !(fiemap->fm_flags & FIEMAP_FLAG_DEVICE_ORDER)) GOTO(out, rc = -EOPNOTSUPP); - fm_key.oa.o_id = lsm->lsm_object_id; - fm_key.oa.o_seq = lsm->lsm_object_seq; + 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); @@ -1748,6 +1720,7 @@ int ll_fid2path(struct inode *inode, void *arg) rc = obd_iocontrol(OBD_IOC_FID2PATH, exp, outsize, gfout, NULL); if (rc) GOTO(gf_free, rc); + if (copy_to_user(arg, gfout, outsize)) rc = -EFAULT; @@ -1828,20 +1801,18 @@ int ll_data_version(struct inode *inode, __u64 *data_version, /* If no stripe, we consider version is 0. */ lsm = ccc_inode_lsm_get(inode); - if (lsm == NULL) { + if (!lsm_has_objects(lsm)) { *data_version = 0; CDEBUG(D_INODE, "No object for inode\n"); - RETURN(0); + GOTO(out, rc = 0); } OBD_ALLOC_PTR(obdo); - if (obdo == NULL) { - ccc_inode_lsm_put(inode, lsm); - RETURN(-ENOMEM); - } + if (obdo == NULL) + GOTO(out, rc = -ENOMEM); rc = ll_lsm_getattr(lsm, sbi->ll_dt_exp, NULL, obdo, 0, extent_lock); - if (!rc) { + if (rc == 0) { if (!(obdo->o_valid & OBD_MD_FLDATAVERSION)) rc = -EOPNOTSUPP; else @@ -1849,70 +1820,170 @@ int ll_data_version(struct inode *inode, __u64 *data_version, } OBD_FREE_PTR(obdo); + EXIT; +out: ccc_inode_lsm_put(inode, lsm); - RETURN(rc); } -static int ll_swap_layout(struct file *file, struct file *file2, - struct lustre_swap_layouts *lsl) +struct ll_swap_stack { + struct iattr ia1, ia2; + __u64 dv1, dv2; + struct inode *inode1, *inode2; + bool check_dv1, check_dv2; +}; + +static int ll_swap_layouts(struct file *file1, struct file *file2, + struct lustre_swap_layouts *lsl) { - struct mdc_swap_layouts msl = { .msl_flags = lsl->sl_flags }; - struct md_op_data *op_data; - struct inode *inode = file->f_dentry->d_inode; - struct inode *inode2 = file2->f_dentry->d_inode; - __u32 gid; - int rc; + struct mdc_swap_layouts msl; + struct md_op_data *op_data; + __u32 gid; + __u64 dv; + struct ll_swap_stack *llss = NULL; + int rc; - if (!S_ISREG(inode2->i_mode)) - RETURN(-EINVAL); + OBD_ALLOC_PTR(llss); + if (llss == NULL) + RETURN(-ENOMEM); - if (inode_permission(inode, MAY_WRITE) || - inode_permission(inode2, MAY_WRITE)) - RETURN(-EPERM); + 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); + + 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); + + /* we use 2 bool because it is easier to swap than 2 bits */ + if (lsl->sl_flags & SWAP_LAYOUTS_CHECK_DV1) + llss->check_dv1 = true; + + if (lsl->sl_flags & SWAP_LAYOUTS_CHECK_DV2) + llss->check_dv2 = true; - if (inode2->i_sb != inode->i_sb) - RETURN(-EXDEV); + /* we cannot use lsl->sl_dvX directly because we may swap them */ + llss->dv1 = lsl->sl_dv1; + llss->dv2 = lsl->sl_dv2; - rc = lu_fid_cmp(ll_inode2fid(inode), ll_inode2fid(inode2)); + rc = lu_fid_cmp(ll_inode2fid(llss->inode1), ll_inode2fid(llss->inode2)); if (rc == 0) /* same file, done! */ - RETURN(0); + GOTO(free, rc = 0); if (rc < 0) { /* sequentialize it */ - swap(inode, inode2); - swap(file, file2); + swap(llss->inode1, llss->inode2); + swap(file1, file2); + swap(llss->dv1, llss->dv2); + swap(llss->check_dv1, llss->check_dv2); } gid = lsl->sl_gid; if (gid != 0) { /* application asks to flush dirty cache */ - rc = ll_get_grouplock(inode, file, gid); + rc = ll_get_grouplock(llss->inode1, file1, gid); if (rc < 0) - RETURN(rc); + GOTO(free, rc); - rc = ll_get_grouplock(inode2, file2, gid); + rc = ll_get_grouplock(llss->inode2, file2, gid); if (rc < 0) { - ll_put_grouplock(inode, file, gid); - RETURN(rc); + ll_put_grouplock(llss->inode1, file1, gid); + GOTO(free, rc); } } + /* 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) { + rc = ll_data_version(llss->inode1, &dv, 0); + if (rc) + GOTO(putgl, rc); + if (dv != llss->dv1) + GOTO(putgl, rc = -EAGAIN); + } + + if (llss->check_dv2) { + rc = ll_data_version(llss->inode2, &dv, 0); + if (rc) + GOTO(putgl, rc); + if (dv != llss->dv2) + GOTO(putgl, rc = -EAGAIN); + } + /* struct md_op_data is used to send the swap args to the mdt * only flags is missing, so we use struct mdc_swap_layouts * through the md_op_data->op_data */ + /* flags from user space have to be converted before they are send to + * server, no flag is sent today, they are only used on the client */ + msl.msl_flags = 0; rc = -ENOMEM; - op_data = ll_prep_md_op_data(NULL, inode, inode2, NULL, 0, 0, - LUSTRE_OPC_ANY, &msl); - if (op_data != NULL) { - rc = obd_iocontrol(LL_IOC_LOV_SWAP_LAYOUTS, ll_i2mdexp(inode), - sizeof(*op_data), op_data, NULL); - ll_finish_md_op_data(op_data); - } + op_data = ll_prep_md_op_data(NULL, llss->inode1, llss->inode2, NULL, 0, + 0, LUSTRE_OPC_ANY, &msl); + if (IS_ERR(op_data)) + GOTO(free, rc = PTR_ERR(op_data)); + rc = obd_iocontrol(LL_IOC_LOV_SWAP_LAYOUTS, ll_i2mdexp(llss->inode1), + sizeof(*op_data), op_data, NULL); + ll_finish_md_op_data(op_data); + +putgl: if (gid != 0) { - ll_put_grouplock(inode2, file2, gid); - ll_put_grouplock(inode, file, gid); + 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); + RETURN(rc); } @@ -1965,7 +2036,7 @@ long ll_file_ioctl(struct file *file, unsigned int cmd, unsigned long arg) struct file *file2; struct lustre_swap_layouts lsl; - if (cfs_copy_from_user(&lsl, (char *)arg, + if (copy_from_user(&lsl, (char *)arg, sizeof(struct lustre_swap_layouts))) RETURN(-EFAULT); @@ -1978,7 +2049,7 @@ long ll_file_ioctl(struct file *file, unsigned int cmd, unsigned long arg) rc = -EPERM; if ((file2->f_flags & O_ACCMODE) != 0) /* O_WRONLY or O_RDWR */ - rc = ll_swap_layout(file, file2, &lsl); + rc = ll_swap_layouts(file, file2, &lsl); fput(file2); RETURN(rc); } @@ -2062,9 +2133,9 @@ long ll_file_ioctl(struct file *file, unsigned int cmd, unsigned long arg) op_data = ll_prep_md_op_data(NULL, inode, NULL, NULL, 0, 0, LUSTRE_OPC_ANY, hus); - if (op_data == NULL) { + if (IS_ERR(op_data)) { OBD_FREE_PTR(hus); - RETURN(-ENOMEM); + RETURN(PTR_ERR(op_data)); } rc = obd_iocontrol(cmd, ll_i2mdexp(inode), sizeof(*op_data), @@ -2100,9 +2171,9 @@ long ll_file_ioctl(struct file *file, unsigned int cmd, unsigned long arg) op_data = ll_prep_md_op_data(NULL, inode, NULL, NULL, 0, 0, LUSTRE_OPC_ANY, hss); - if (op_data == NULL) { + if (IS_ERR(op_data)) { OBD_FREE_PTR(hss); - RETURN(-ENOMEM); + RETURN(PTR_ERR(op_data)); } rc = obd_iocontrol(cmd, ll_i2mdexp(inode), sizeof(*op_data), @@ -2124,15 +2195,15 @@ long ll_file_ioctl(struct file *file, unsigned int cmd, unsigned long arg) op_data = ll_prep_md_op_data(NULL, inode, NULL, NULL, 0, 0, LUSTRE_OPC_ANY, hca); - if (op_data == NULL) { + if (IS_ERR(op_data)) { OBD_FREE_PTR(hca); - RETURN(-ENOMEM); + RETURN(PTR_ERR(op_data)); } rc = obd_iocontrol(cmd, ll_i2mdexp(inode), sizeof(*op_data), op_data, NULL); - if (cfs_copy_to_user((char *)arg, hca, sizeof(*hca))) + if (copy_to_user((char *)arg, hca, sizeof(*hca))) rc = -EFAULT; ll_finish_md_op_data(op_data); @@ -2275,7 +2346,7 @@ int ll_flush(struct file *file, fl_owner_t id) * Return how many pages have been written. */ int cl_sync_file_range(struct inode *inode, loff_t start, loff_t end, - enum cl_fsync_mode mode) + enum cl_fsync_mode mode, int ignore_layout) { struct cl_env_nest nest; struct lu_env *env; @@ -2297,7 +2368,7 @@ int cl_sync_file_range(struct inode *inode, loff_t start, loff_t end, io = ccc_env_thread_io(env); io->ci_obj = cl_i2info(inode)->lli_clob; - io->ci_ignore_layout = 1; + io->ci_ignore_layout = ignore_layout; /* initialize parameters for sync */ fio = &io->u.ci_fsync; @@ -2322,15 +2393,25 @@ int cl_sync_file_range(struct inode *inode, loff_t start, loff_t end, RETURN(result); } +/* + * When dentry is provided (the 'else' case), *file->f_dentry may be + * null and dentry must be used directly rather than pulled from + * *file->f_dentry as is done otherwise. + */ + #ifdef HAVE_FILE_FSYNC_4ARGS -int ll_fsync(struct file *file, loff_t start, loff_t end, int data) +int ll_fsync(struct file *file, loff_t start, loff_t end, int datasync) +{ + struct dentry *dentry = file->f_dentry; #elif defined(HAVE_FILE_FSYNC_2ARGS) -int ll_fsync(struct file *file, int data) +int ll_fsync(struct file *file, int datasync) +{ + struct dentry *dentry = file->f_dentry; #else -int ll_fsync(struct file *file, struct dentry *dentry, int data) -#endif +int ll_fsync(struct file *file, struct dentry *dentry, int datasync) { - struct inode *inode = file->f_dentry->d_inode; +#endif + struct inode *inode = dentry->d_inode; struct ll_inode_info *lli = ll_i2info(inode); struct ptlrpc_request *req; struct obd_capa *oc; @@ -2371,11 +2452,11 @@ int ll_fsync(struct file *file, struct dentry *dentry, int data) if (!err) ptlrpc_req_finished(req); - if (data) { + if (datasync && 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); + CL_FSYNC_ALL, 0); if (rc == 0 && err < 0) rc = err; if (rc < 0) @@ -2392,18 +2473,20 @@ int ll_fsync(struct file *file, struct dentry *dentry, int data) int ll_file_flock(struct file *file, int cmd, struct file_lock *file_lock) { - struct inode *inode = file->f_dentry->d_inode; - struct ll_sb_info *sbi = ll_i2sbi(inode); - struct ldlm_enqueue_info einfo = { .ei_type = LDLM_FLOCK, - .ei_cb_cp =ldlm_flock_completion_ast, - .ei_cbdata = file_lock }; - struct md_op_data *op_data; - struct lustre_handle lockh = {0}; - ldlm_policy_data_t flock = {{0}}; - int flags = 0; - int rc; + struct inode *inode = file->f_dentry->d_inode; + struct ll_sb_info *sbi = ll_i2sbi(inode); + struct ldlm_enqueue_info einfo = { + .ei_type = LDLM_FLOCK, + .ei_cb_cp = ldlm_flock_completion_ast, + .ei_cbdata = file_lock, + }; + struct md_op_data *op_data; + struct lustre_handle lockh = {0}; + ldlm_policy_data_t flock = {{0}}; + int flags = 0; + int rc; int rc2 = 0; - ENTRY; + ENTRY; CDEBUG(D_VFSTRACE, "VFS Op:inode=%lu file_lock=%p\n", inode->i_ino, file_lock); @@ -2553,11 +2636,11 @@ int ll_have_md_lock(struct inode *inode, __u64 *bits, ldlm_mode_t l_req_mode) CDEBUG(D_INFO, "trying to match res "DFID" mode %s\n", PFID(fid), ldlm_lockname[mode]); - flags = LDLM_FL_BLOCK_GRANTED | LDLM_FL_CBPENDING | LDLM_FL_TEST_LOCK; - for (i = 0; i < MDS_INODELOCK_MAXSHIFT && *bits != 0; i++) { - policy.l_inodebits.bits = *bits & (1 << i); - if (policy.l_inodebits.bits == 0) - continue; + flags = LDLM_FL_BLOCK_GRANTED | LDLM_FL_CBPENDING | LDLM_FL_TEST_LOCK; + for (i = 0; i <= MDS_INODELOCK_MAXSHIFT && *bits != 0; i++) { + policy.l_inodebits.bits = *bits & (1 << i); + if (policy.l_inodebits.bits == 0) + continue; if (md_lock_match(ll_i2mdexp(inode), flags, fid, LDLM_IBITS, &policy, mode, &lockh)) { @@ -2669,7 +2752,7 @@ int __ll_inode_revalidate_it(struct dentry *dentry, struct lookup_intent *it, here to preserve get_cwd functionality on 2.6. Bug 10503 */ if (!dentry->d_inode->i_nlink) - d_lustre_invalidate(dentry); + d_lustre_invalidate(dentry, 0); ll_lookup_finish_locks(&oit, dentry); } else if (!ll_have_md_lock(dentry->d_inode, &ibits, LCK_MINMODE)) { @@ -2710,13 +2793,13 @@ out: } int ll_inode_revalidate_it(struct dentry *dentry, struct lookup_intent *it, - __u64 ibits) + __u64 ibits) { - struct inode *inode = dentry->d_inode; - int rc; - ENTRY; + struct inode *inode = dentry->d_inode; + int rc; + ENTRY; - rc = __ll_inode_revalidate_it(dentry, it, ibits); + rc = __ll_inode_revalidate_it(dentry, it, ibits); if (rc != 0) RETURN(rc); @@ -2726,9 +2809,17 @@ int ll_inode_revalidate_it(struct dentry *dentry, struct lookup_intent *it, LTIME_S(inode->i_mtime) = ll_i2info(inode)->lli_lvb.lvb_mtime; LTIME_S(inode->i_ctime) = ll_i2info(inode)->lli_lvb.lvb_ctime; } else { - rc = ll_glimpse_size(inode); + /* In case of restore, the MDT has the right size and has + * already send it back without granting the layout lock, + * inode is up-to-date so glimpse is useless. + * Also to glimpse we need the layout, in case of a running + * restore the MDT holds the layout lock so the glimpse will + * block up to the end of restore (getattr will block) + */ + if (!(ll_i2info(inode)->lli_flags & LLIF_FILE_RESTORING)) + rc = ll_glimpse_size(inode); } - RETURN(rc); + RETURN(rc); } int ll_getattr_it(struct vfsmount *mnt, struct dentry *de, @@ -2922,12 +3013,7 @@ struct file_operations ll_file_operations = { .release = ll_file_release, .mmap = ll_file_mmap, .llseek = ll_file_seek, -#ifdef HAVE_KERNEL_SENDFILE - .sendfile = ll_file_sendfile, -#endif -#ifdef HAVE_KERNEL_SPLICE_READ .splice_read = ll_file_splice_read, -#endif .fsync = ll_fsync, .flush = ll_flush }; @@ -2942,12 +3028,7 @@ struct file_operations ll_file_operations_flock = { .release = ll_file_release, .mmap = ll_file_mmap, .llseek = ll_file_seek, -#ifdef HAVE_KERNEL_SENDFILE - .sendfile = ll_file_sendfile, -#endif -#ifdef HAVE_KERNEL_SPLICE_READ .splice_read = ll_file_splice_read, -#endif .fsync = ll_fsync, .flush = ll_flush, .flock = ll_file_flock, @@ -2965,12 +3046,7 @@ struct file_operations ll_file_operations_noflock = { .release = ll_file_release, .mmap = ll_file_mmap, .llseek = ll_file_seek, -#ifdef HAVE_KERNEL_SENDFILE - .sendfile = ll_file_sendfile, -#endif -#ifdef HAVE_KERNEL_SPLICE_READ .splice_read = ll_file_splice_read, -#endif .fsync = ll_fsync, .flush = ll_flush, .flock = ll_file_noflock, @@ -3127,6 +3203,74 @@ int ll_layout_conf(struct inode *inode, const struct cl_object_conf *conf) RETURN(result); } +/* Fetch layout from MDT with getxattr request, if it's not ready yet */ +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; + void *lmm; + int lmmsize; + int rc; + ENTRY; + + CDEBUG(D_INODE, DFID" LVB_READY=%d l_lvb_data=%p l_lvb_len=%d\n", + PFID(ll_inode2fid(inode)), !!(lock->l_flags & LDLM_FL_LVB_READY), + lock->l_lvb_data, lock->l_lvb_len); + + if ((lock->l_lvb_data != NULL) && (lock->l_flags & LDLM_FL_LVB_READY)) + RETURN(0); + + /* if layout lock was granted right away, the layout is returned + * within DLM_LVB of dlm reply; otherwise if the lock was ever + * 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); + if (rc == 0) + rc = md_getxattr(sbi->ll_md_exp, ll_inode2fid(inode), oc, + 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) + GOTO(out, rc = -EPROTO); + + lmmsize = body->eadatasize; + if (lmmsize == 0) /* empty layout */ + GOTO(out, rc = 0); + + lmm = req_capsule_server_sized_get(&req->rq_pill, &RMF_EADATA, lmmsize); + if (lmm == NULL) + GOTO(out, rc = -EFAULT); + + OBD_ALLOC_LARGE(lvbdata, lmmsize); + if (lvbdata == NULL) + GOTO(out, rc = -ENOMEM); + + 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; + unlock_res_and_lock(lock); + + EXIT; + +out: + ptlrpc_req_finished(req); + return rc; +} + /** * Apply the layout to the inode. Layout lock is held and will be released * in this function. @@ -3141,6 +3285,7 @@ static int ll_layout_lock_set(struct lustre_handle *lockh, ldlm_mode_t mode, struct cl_object_conf conf; int rc = 0; bool lvb_ready; + bool wait_layout = false; ENTRY; LASSERT(lustre_handle_is_used(lockh)); @@ -3150,16 +3295,18 @@ static int ll_layout_lock_set(struct lustre_handle *lockh, ldlm_mode_t mode, LASSERT(ldlm_has_layout(lock)); LDLM_DEBUG(lock, "File %p/"DFID" being reconfigured: %d.\n", - inode, PFID(&lli->lli_fid), reconf); + inode, PFID(&lli->lli_fid), reconf); + + /* in case this is a caching lock and reinstate with new inode */ + md_set_lock_data(sbi->ll_md_exp, &lockh->cookie, inode, NULL); lock_res_and_lock(lock); lvb_ready = !!(lock->l_flags & LDLM_FL_LVB_READY); unlock_res_and_lock(lock); /* checking lvb_ready is racy but this is okay. The worst case is * that multi processes may configure the file on the same time. */ - if (lvb_ready || !reconf) { - LDLM_LOCK_PUT(lock); + if (lvb_ready || !reconf) { rc = -ENODATA; if (lvb_ready) { /* layout_gen must be valid if layout lock is not @@ -3167,10 +3314,13 @@ static int ll_layout_lock_set(struct lustre_handle *lockh, ldlm_mode_t mode, *gen = lli->lli_layout_gen; rc = 0; } - ldlm_lock_decref(lockh, mode); - RETURN(rc); + GOTO(out, rc); } + rc = ll_layout_fetch(inode, lock); + if (rc < 0) + GOTO(out, rc); + /* for layout lock, lmm is returned 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() @@ -3189,11 +3339,8 @@ static int ll_layout_lock_set(struct lustre_handle *lockh, ldlm_mode_t mode, PFID(&lli->lli_fid), rc); } } - if (rc < 0) { - LDLM_LOCK_PUT(lock); - ldlm_lock_decref(lockh, mode); - RETURN(rc); - } + if (rc < 0) + GOTO(out, rc); /* set layout to file. Unlikely this will fail as old layout was * surely eliminated */ @@ -3203,15 +3350,20 @@ static int ll_layout_lock_set(struct lustre_handle *lockh, ldlm_mode_t mode, conf.coc_lock = lock; conf.u.coc_md = &md; rc = ll_layout_conf(inode, &conf); - LDLM_LOCK_PUT(lock); - - ldlm_lock_decref(lockh, mode); if (md.lsm != NULL) obd_free_memmd(sbi->ll_dt_exp, &md.lsm); + /* refresh layout failed, need to wait */ + wait_layout = rc == -EBUSY; + EXIT; + +out: + LDLM_LOCK_PUT(lock); + ldlm_lock_decref(lockh, mode); + /* wait for IO to complete if it's still being used. */ - if (rc == -EBUSY) { + if (wait_layout) { CDEBUG(D_INODE, "%s: %p/"DFID" wait for layout reconf.\n", ll_get_fsname(inode->i_sb, NULL, 0), inode, PFID(&lli->lli_fid)); @@ -3226,7 +3378,6 @@ static int ll_layout_lock_set(struct lustre_handle *lockh, ldlm_mode_t mode, CDEBUG(D_INODE, "file: "DFID" waiting layout return: %d.\n", PFID(&lli->lli_fid), rc); } - RETURN(rc); } @@ -3251,15 +3402,16 @@ int ll_layout_refresh(struct inode *inode, __u32 *gen) struct lookup_intent it; struct lustre_handle lockh; ldlm_mode_t 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_cbdata = NULL }; + 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, + }; int rc; ENTRY; - *gen = LL_LAYOUT_GEN_NONE; + *gen = lli->lli_layout_gen; if (!(sbi->ll_flags & LL_SBI_LAYOUT_LOCK)) RETURN(0); @@ -3318,8 +3470,6 @@ again: ll_finish_md_op_data(op_data); - md_set_lock_data(sbi->ll_md_exp, &it.d.lustre.it_lock_handle, inode, NULL); - mode = it.d.lustre.it_lock_mode; it.d.lustre.it_lock_mode = 0; ll_intent_drop_lock(&it); @@ -3335,3 +3485,32 @@ again: RETURN(rc); } + +/** + * This function send a restore request to the MDT + */ +int ll_layout_restore(struct inode *inode) +{ + struct hsm_user_request *hur; + int len, rc; + ENTRY; + + len = sizeof(struct hsm_user_request) + + sizeof(struct hsm_user_item); + OBD_ALLOC(hur, len); + if (hur == NULL) + RETURN(-ENOMEM); + + hur->hur_request.hr_action = HUA_RESTORE; + hur->hur_request.hr_archive_id = 0; + hur->hur_request.hr_flags = 0; + memcpy(&hur->hur_user_item[0].hui_fid, &ll_i2info(inode)->lli_fid, + sizeof(hur->hur_user_item[0].hui_fid)); + hur->hur_user_item[0].hui_extent.length = -1; + hur->hur_request.hr_itemcount = 1; + rc = obd_iocontrol(LL_IOC_HSM_REQUEST, cl_i2sbi(inode)->ll_md_exp, + len, hur, NULL); + OBD_FREE(hur, len); + RETURN(rc); +} +