X-Git-Url: https://git.whamcloud.com/?p=fs%2Flustre-release.git;a=blobdiff_plain;f=lustre%2Flmv%2Flmv_intent.c;h=9d16714f4a30e792a2e106fcb8956a37cb5c4fc8;hp=74cf0a1375b35f327c7e0d51430f13671e2994b2;hb=8a11cb6282cfbdc8617b809344e6a11223e86a38;hpb=75ee8334498f948ecd030ad5edb22bc596f300fc diff --git a/lustre/lmv/lmv_intent.c b/lustre/lmv/lmv_intent.c index 74cf0a1..9d16714 100644 --- a/lustre/lmv/lmv_intent.c +++ b/lustre/lmv/lmv_intent.c @@ -27,7 +27,7 @@ * Copyright (c) 2004, 2010, Oracle and/or its affiliates. All rights reserved. * Use is subject to license terms. * - * Copyright (c) 2011, 2013, Intel Corporation. + * Copyright (c) 2011, 2014, Intel Corporation. */ /* * This file is part of Lustre, http://www.lustre.org/ @@ -35,19 +35,15 @@ */ #define DEBUG_SUBSYSTEM S_LMV -#ifdef __KERNEL__ #include #include #include #include #include -#include +#include #include #include -#include -#else -#include -#endif +#include #include #include @@ -59,9 +55,8 @@ #include #include "lmv_internal.h" -static int lmv_intent_remote(struct obd_export *exp, void *lmm, - int lmmsize, struct lookup_intent *it, - const struct lu_fid *parent_fid, int flags, +static int lmv_intent_remote(struct obd_export *exp, struct lookup_intent *it, + const struct lu_fid *parent_fid, struct ptlrpc_request **reqp, ldlm_blocking_callback cb_blocking, __u64 extra_lock_flags) @@ -81,7 +76,7 @@ static int lmv_intent_remote(struct obd_export *exp, void *lmm, if (body == NULL) RETURN(-EPROTO); - LASSERT((body->valid & OBD_MD_MDS)); + LASSERT((body->mbo_valid & OBD_MD_MDS)); /* * Unfortunately, we have to lie to MDC/MDS to retrieve @@ -100,9 +95,9 @@ static int lmv_intent_remote(struct obd_export *exp, void *lmm, it->d.lustre.it_data = NULL; } - LASSERT(fid_is_sane(&body->fid1)); + LASSERT(fid_is_sane(&body->mbo_fid1)); - tgt = lmv_find_target(lmv, &body->fid1); + tgt = lmv_find_target(lmv, &body->mbo_fid1); if (IS_ERR(tgt)) GOTO(out, rc = PTR_ERR(tgt)); @@ -110,7 +105,7 @@ static int lmv_intent_remote(struct obd_export *exp, void *lmm, if (op_data == NULL) GOTO(out, rc = -ENOMEM); - op_data->op_fid1 = body->fid1; + op_data->op_fid1 = body->mbo_fid1; /* Sent the parent FID to the remote MDT */ if (parent_fid != NULL) { /* The parent fid is only for remote open to @@ -118,17 +113,14 @@ static int lmv_intent_remote(struct obd_export *exp, void *lmm, * see mdt_cross_open */ LASSERT(it->it_op & IT_OPEN); op_data->op_fid2 = *parent_fid; - /* Add object FID to op_fid3, in case it needs to check stale - * (M_CHECK_STALE), see mdc_finish_intent_lock */ - op_data->op_fid3 = body->fid1; } op_data->op_bias = MDS_CROSS_REF; - CDEBUG(D_INODE, "REMOTE_INTENT with fid="DFID" -> mds #%d\n", - PFID(&body->fid1), tgt->ltd_idx); + CDEBUG(D_INODE, "REMOTE_INTENT with fid="DFID" -> mds #%u\n", + PFID(&body->mbo_fid1), tgt->ltd_idx); - rc = md_intent_lock(tgt->ltd_exp, op_data, lmm, lmmsize, it, - flags, &req, cb_blocking, extra_lock_flags); + rc = md_intent_lock(tgt->ltd_exp, op_data, it, &req, cb_blocking, + extra_lock_flags); if (rc) GOTO(out_free_op_data, rc); @@ -160,21 +152,16 @@ out: return rc; } -#ifdef __KERNEL__ -int lmv_revalidate_slaves(struct obd_export *exp, struct mdt_body *mbody, - struct lmv_stripe_md *lsm, +int lmv_revalidate_slaves(struct obd_export *exp, + const struct lmv_stripe_md *lsm, ldlm_blocking_callback cb_blocking, int extra_lock_flags) { struct obd_device *obd = exp->exp_obd; struct lmv_obd *lmv = &obd->u.lmv; + struct ptlrpc_request *req = NULL; struct mdt_body *body; struct md_op_data *op_data; - unsigned long size = 0; - unsigned long nlink = 0; - obd_time atime = 0; - obd_time ctime = 0; - obd_time mtime = 0; int i; int rc = 0; @@ -184,9 +171,6 @@ int lmv_revalidate_slaves(struct obd_export *exp, struct mdt_body *mbody, * revalidate slaves has some problems, temporarily return, * we may not need that */ - if (lsm->lsm_md_stripe_count <= 1) - RETURN(0); - OBD_ALLOC_PTR(op_data); if (op_data == NULL) RETURN(-ENOMEM); @@ -198,21 +182,12 @@ int lmv_revalidate_slaves(struct obd_export *exp, struct mdt_body *mbody, for (i = 0; i < lsm->lsm_md_stripe_count; i++) { struct lu_fid fid; struct lookup_intent it = { .it_op = IT_GETATTR }; - struct ptlrpc_request *req = NULL; struct lustre_handle *lockh = NULL; struct lmv_tgt_desc *tgt = NULL; struct inode *inode; fid = lsm->lsm_md_oinfo[i].lmo_fid; inode = lsm->lsm_md_oinfo[i].lmo_root; - if (i == 0) { - if (mbody != NULL) { - body = mbody; - goto update; - } else { - goto release_lock; - } - } /* * Prepare op_data for revalidating. Note that @fid2 shluld be @@ -227,11 +202,16 @@ int lmv_revalidate_slaves(struct obd_export *exp, struct mdt_body *mbody, if (IS_ERR(tgt)) GOTO(cleanup, rc = PTR_ERR(tgt)); - CDEBUG(D_INODE, "Revalidate slave "DFID" -> mds #%d\n", + CDEBUG(D_INODE, "Revalidate slave "DFID" -> mds #%u\n", PFID(&fid), tgt->ltd_idx); - rc = md_intent_lock(tgt->ltd_exp, op_data, NULL, 0, &it, 0, - &req, cb_blocking, extra_lock_flags); + if (req != NULL) { + ptlrpc_req_finished(req); + req = NULL; + } + + rc = md_intent_lock(tgt->ltd_exp, op_data, &it, &req, + cb_blocking, extra_lock_flags); if (rc < 0) GOTO(cleanup, rc); @@ -246,16 +226,13 @@ int lmv_revalidate_slaves(struct obd_export *exp, struct mdt_body *mbody, body = req_capsule_server_get(&req->rq_pill, &RMF_MDT_BODY); LASSERT(body != NULL); -update: - if (unlikely(body->nlink < 2)) { + if (unlikely(body->mbo_nlink < 2)) { CERROR("%s: nlink %d < 2 corrupt stripe %d "DFID - ":" DFID"\n", obd->obd_name, body->nlink, - i, PFID(&lsm->lsm_md_oinfo[i].lmo_fid), + ":" DFID"\n", + obd->obd_name, body->mbo_nlink, i, + PFID(&lsm->lsm_md_oinfo[i].lmo_fid), PFID(&lsm->lsm_md_oinfo[0].lmo_fid)); - if (req != NULL) - ptlrpc_req_finished(req); - if (it.d.lustre.it_lock_mode && lockh) { ldlm_lock_decref(lockh, it.d.lustre.it_lock_mode); @@ -265,84 +242,40 @@ update: GOTO(cleanup, rc = -EIO); } - if (i != 0) - md_set_lock_data(tgt->ltd_exp, &lockh->cookie, - inode, NULL); - - i_size_write(inode, body->size); - set_nlink(inode, body->nlink); - LTIME_S(inode->i_atime) = body->atime; - LTIME_S(inode->i_ctime) = body->ctime; - LTIME_S(inode->i_mtime) = body->mtime; - if (req != NULL) - ptlrpc_req_finished(req); + i_size_write(inode, body->mbo_size); + inode->i_blocks = body->mbo_blocks; + set_nlink(inode, body->mbo_nlink); + LTIME_S(inode->i_atime) = body->mbo_atime; + LTIME_S(inode->i_ctime) = body->mbo_ctime; + LTIME_S(inode->i_mtime) = body->mbo_mtime; } -release_lock: - size += i_size_read(inode); - - if (i != 0) - nlink += inode->i_nlink - 2; - else - nlink += inode->i_nlink; - - atime = LTIME_S(inode->i_atime) > atime ? - LTIME_S(inode->i_atime) : atime; - ctime = LTIME_S(inode->i_ctime) > ctime ? - LTIME_S(inode->i_ctime) : ctime; - mtime = LTIME_S(inode->i_mtime) > mtime ? - LTIME_S(inode->i_mtime) : mtime; + md_set_lock_data(tgt->ltd_exp, &lockh->cookie, inode, NULL); if (it.d.lustre.it_lock_mode != 0 && lockh != NULL) { ldlm_lock_decref(lockh, it.d.lustre.it_lock_mode); it.d.lustre.it_lock_mode = 0; } - - CDEBUG(D_INODE, "i %d "DFID" size %llu, nlink %u, atime " - "%lu, mtime %lu, ctime %lu.\n", i, PFID(&fid), - i_size_read(inode), inode->i_nlink, - LTIME_S(inode->i_atime), LTIME_S(inode->i_mtime), - LTIME_S(inode->i_ctime)); } - /* - * update attr of master request. - */ - CDEBUG(D_INODE, "Return refreshed attrs: size = %lu nlink %lu atime " - LPU64 "ctime "LPU64" mtime "LPU64" for " DFID"\n", size, nlink, - atime, ctime, mtime, PFID(&lsm->lsm_md_oinfo[0].lmo_fid)); - - if (mbody != NULL) { - mbody->atime = atime; - mbody->ctime = ctime; - mbody->mtime = mtime; - } cleanup: + if (req != NULL) + ptlrpc_req_finished(req); + OBD_FREE_PTR(op_data); RETURN(rc); } -#else - -int lmv_revalidate_slaves(struct obd_export *exp, struct mdt_body *mbody, - struct lmv_stripe_md *lsm, - ldlm_blocking_callback cb_blocking, - int extra_lock_flags) -{ - return 0; -} - -#endif /* * IT_OPEN is intended to open (and create, possible) an object. Parent (pid) * may be split dir. */ -int lmv_intent_open(struct obd_export *exp, struct md_op_data *op_data, - void *lmm, int lmmsize, struct lookup_intent *it, - int flags, struct ptlrpc_request **reqp, - ldlm_blocking_callback cb_blocking, - __u64 extra_lock_flags) +static int lmv_intent_open(struct obd_export *exp, struct md_op_data *op_data, + struct lookup_intent *it, + struct ptlrpc_request **reqp, + ldlm_blocking_callback cb_blocking, + __u64 extra_lock_flags) { struct obd_device *obd = exp->exp_obd; struct lmv_obd *lmv = &obd->u.lmv; @@ -351,19 +284,14 @@ int lmv_intent_open(struct obd_export *exp, struct md_op_data *op_data, int rc; ENTRY; - /* Note: client might open with some random flags(sanity 33b), so we can - * not make sure op_fid2 is being initialized with BY_FID flag */ - if (it->it_flags & MDS_OPEN_BY_FID && fid_is_sane(&op_data->op_fid2)) { - if (op_data->op_mea1 != NULL) { - struct lmv_stripe_md *lsm = op_data->op_mea1; - const struct lmv_oinfo *oinfo; - - oinfo = lsm_name_to_stripe_info(lsm, op_data->op_name, - op_data->op_namelen); - if (IS_ERR(oinfo)) - RETURN(PTR_ERR(oinfo)); - op_data->op_fid1 = oinfo->lmo_fid; - } + if (it->it_flags & MDS_OPEN_BY_FID) { + LASSERT(fid_is_sane(&op_data->op_fid2)); + + /* for striped directory, we can't know parent stripe fid + * without name, but we can set it to child fid, and MDT + * will obtain it from linkea in open in such case. */ + if (op_data->op_mea1 != NULL) + op_data->op_fid1 = op_data->op_fid2; tgt = lmv_find_target(lmv, &op_data->op_fid2); if (IS_ERR(tgt)) @@ -371,6 +299,10 @@ int lmv_intent_open(struct obd_export *exp, struct md_op_data *op_data, op_data->op_mds = tgt->ltd_idx; } else { + LASSERT(fid_is_sane(&op_data->op_fid1)); + LASSERT(fid_is_zero(&op_data->op_fid2)); + LASSERT(op_data->op_name != NULL); + tgt = lmv_locate_mds(lmv, op_data, &op_data->op_fid1); if (IS_ERR(tgt)) RETURN(PTR_ERR(tgt)); @@ -378,24 +310,22 @@ int lmv_intent_open(struct obd_export *exp, struct md_op_data *op_data, /* If it is ready to open the file by FID, do not need * allocate FID at all, otherwise it will confuse MDT */ - if ((it->it_op & IT_CREAT) && - !(it->it_flags & MDS_OPEN_BY_FID)) { + if ((it->it_op & IT_CREAT) && !(it->it_flags & MDS_OPEN_BY_FID)) { /* - * For open with IT_CREATE and for IT_CREATE cases allocate new - * fid and setup FLD for it. + * For lookup(IT_CREATE) cases allocate new fid and setup FLD + * for it. */ - op_data->op_fid3 = op_data->op_fid2; - rc = lmv_fid_alloc(exp, &op_data->op_fid2, op_data); + rc = lmv_fid_alloc(NULL, exp, &op_data->op_fid2, op_data); if (rc != 0) RETURN(rc); } CDEBUG(D_INODE, "OPEN_INTENT with fid1="DFID", fid2="DFID"," - " name='%s' -> mds #%d\n", PFID(&op_data->op_fid1), + " name='%s' -> mds #%u\n", PFID(&op_data->op_fid1), PFID(&op_data->op_fid2), op_data->op_name, tgt->ltd_idx); - rc = md_intent_lock(tgt->ltd_exp, op_data, lmm, lmmsize, it, flags, - reqp, cb_blocking, extra_lock_flags); + rc = md_intent_lock(tgt->ltd_exp, op_data, it, reqp, cb_blocking, + extra_lock_flags); if (rc != 0) RETURN(rc); /* @@ -412,10 +342,9 @@ int lmv_intent_open(struct obd_export *exp, struct md_op_data *op_data, RETURN(-EPROTO); /* Not cross-ref case, just get out of here. */ - if (unlikely((body->valid & OBD_MD_MDS))) { - rc = lmv_intent_remote(exp, lmm, lmmsize, it, &op_data->op_fid1, - flags, reqp, cb_blocking, - extra_lock_flags); + if (unlikely((body->mbo_valid & OBD_MD_MDS))) { + rc = lmv_intent_remote(exp, it, &op_data->op_fid1, reqp, + cb_blocking, extra_lock_flags); if (rc != 0) RETURN(rc); @@ -430,11 +359,11 @@ int lmv_intent_open(struct obd_export *exp, struct md_op_data *op_data, /* * Handler for: getattr, lookup and revalidate cases. */ -int lmv_intent_lookup(struct obd_export *exp, struct md_op_data *op_data, - void *lmm, int lmmsize, struct lookup_intent *it, - int flags, struct ptlrpc_request **reqp, - ldlm_blocking_callback cb_blocking, - __u64 extra_lock_flags) +static int +lmv_intent_lookup(struct obd_export *exp, struct md_op_data *op_data, + struct lookup_intent *it, struct ptlrpc_request **reqp, + ldlm_blocking_callback cb_blocking, + __u64 extra_lock_flags) { struct obd_device *obd = exp->exp_obd; struct lmv_obd *lmv = &obd->u.lmv; @@ -444,23 +373,39 @@ int lmv_intent_lookup(struct obd_export *exp, struct md_op_data *op_data, int rc = 0; ENTRY; + /* If it returns ERR_PTR(-EBADFD) then it is an unknown hash type + * it will try all stripes to locate the object */ tgt = lmv_locate_mds(lmv, op_data, &op_data->op_fid1); - if (IS_ERR(tgt)) + if (IS_ERR(tgt) && (PTR_ERR(tgt) != -EBADFD)) RETURN(PTR_ERR(tgt)); + /* Both migrating dir and unknown hash dir need to try + * all of sub-stripes */ + if (lsm != NULL && !lmv_is_known_hash_type(lsm->lsm_md_hash_type)) { + struct lmv_oinfo *oinfo; + + oinfo = &lsm->lsm_md_oinfo[0]; + + op_data->op_fid1 = oinfo->lmo_fid; + op_data->op_mds = oinfo->lmo_mds; + tgt = lmv_get_target(lmv, oinfo->lmo_mds, NULL); + if (IS_ERR(tgt)) + RETURN(PTR_ERR(tgt)); + } + if (!fid_is_sane(&op_data->op_fid2)) fid_zero(&op_data->op_fid2); CDEBUG(D_INODE, "LOOKUP_INTENT with fid1="DFID", fid2="DFID - ", name='%s' -> mds #%d lsm=%p lsm_magic=%x\n", + ", name='%s' -> mds #%u lsm=%p lsm_magic=%x\n", PFID(&op_data->op_fid1), PFID(&op_data->op_fid2), op_data->op_name ? op_data->op_name : "", tgt->ltd_idx, lsm, lsm == NULL ? -1 : lsm->lsm_md_magic); op_data->op_bias &= ~MDS_CROSS_REF; - rc = md_intent_lock(tgt->ltd_exp, op_data, lmm, lmmsize, it, - flags, reqp, cb_blocking, extra_lock_flags); + rc = md_intent_lock(tgt->ltd_exp, op_data, it, reqp, cb_blocking, + extra_lock_flags); if (rc < 0) RETURN(rc); @@ -468,34 +413,46 @@ int lmv_intent_lookup(struct obd_export *exp, struct md_op_data *op_data, /* If RPC happens, lsm information will be revalidated * during update_inode process (see ll_update_lsm_md) */ if (op_data->op_mea2 != NULL) { - rc = lmv_revalidate_slaves(exp, NULL, op_data->op_mea2, + rc = lmv_revalidate_slaves(exp, op_data->op_mea2, cb_blocking, extra_lock_flags); if (rc != 0) RETURN(rc); } RETURN(rc); - } else if (it_disposition(it, DISP_LOOKUP_NEG) && - lsm != NULL && lsm->lsm_md_magic == LMV_MAGIC_MIGRATE) { - /* For migrating directory, if it can not find the child in - * the source directory(master stripe), try the targeting - * directory(stripe 1) */ - tgt = lmv_find_target(lmv, &lsm->lsm_md_oinfo[1].lmo_fid); - if (IS_ERR(tgt)) - RETURN(PTR_ERR(tgt)); - - ptlrpc_req_finished(*reqp); - it->d.lustre.it_data = NULL; - *reqp = NULL; + } else if (it_disposition(it, DISP_LOOKUP_NEG) && lsm != NULL && + lmv_need_try_all_stripes(lsm)) { + /* For migrating and unknown hash type directory, it will + * try to target the entry on other stripes */ + int stripe_index; + + for (stripe_index = 1; + stripe_index < lsm->lsm_md_stripe_count && + it_disposition(it, DISP_LOOKUP_NEG); stripe_index++) { + struct lmv_oinfo *oinfo; + + /* release the previous request */ + ptlrpc_req_finished(*reqp); + it->d.lustre.it_data = NULL; + *reqp = NULL; + + oinfo = &lsm->lsm_md_oinfo[stripe_index]; + tgt = lmv_find_target(lmv, &oinfo->lmo_fid); + if (IS_ERR(tgt)) + RETURN(PTR_ERR(tgt)); + + CDEBUG(D_INODE, "Try other stripes " DFID"\n", + PFID(&oinfo->lmo_fid)); - CDEBUG(D_INODE, "For migrating dir, try target dir "DFID"\n", - PFID(&lsm->lsm_md_oinfo[1].lmo_fid)); - - op_data->op_fid1 = lsm->lsm_md_oinfo[1].lmo_fid; - it->d.lustre.it_disposition &= ~DISP_ENQ_COMPLETE; - rc = md_intent_lock(tgt->ltd_exp, op_data, lmm, lmmsize, it, - flags, reqp, cb_blocking, extra_lock_flags); + op_data->op_fid1 = oinfo->lmo_fid; + it->d.lustre.it_disposition &= ~DISP_ENQ_COMPLETE; + rc = md_intent_lock(tgt->ltd_exp, op_data, it, reqp, + cb_blocking, extra_lock_flags); + if (rc != 0) + RETURN(rc); + } } + /* * MDS has returned success. Probably name has been resolved in * remote inode. Let's check this. @@ -505,9 +462,9 @@ int lmv_intent_lookup(struct obd_export *exp, struct md_op_data *op_data, RETURN(-EPROTO); /* Not cross-ref case, just get out of here. */ - if (unlikely((body->valid & OBD_MD_MDS))) { - rc = lmv_intent_remote(exp, lmm, lmmsize, it, NULL, flags, - reqp, cb_blocking, extra_lock_flags); + if (unlikely((body->mbo_valid & OBD_MD_MDS))) { + rc = lmv_intent_remote(exp, it, NULL, reqp, cb_blocking, + extra_lock_flags); if (rc != 0) RETURN(rc); body = req_capsule_server_get(&(*reqp)->rq_pill, &RMF_MDT_BODY); @@ -519,35 +476,56 @@ int lmv_intent_lookup(struct obd_export *exp, struct md_op_data *op_data, } int lmv_intent_lock(struct obd_export *exp, struct md_op_data *op_data, - void *lmm, int lmmsize, struct lookup_intent *it, - int flags, struct ptlrpc_request **reqp, - ldlm_blocking_callback cb_blocking, + struct lookup_intent *it, struct ptlrpc_request **reqp, + ldlm_blocking_callback cb_blocking, __u64 extra_lock_flags) { - struct obd_device *obd = exp->exp_obd; - int rc; - ENTRY; + struct obd_device *obd = exp->exp_obd; + int rc; + ENTRY; - LASSERT(it != NULL); - LASSERT(fid_is_sane(&op_data->op_fid1)); + LASSERT(it != NULL); + LASSERT(fid_is_sane(&op_data->op_fid1)); - CDEBUG(D_INODE, "INTENT LOCK '%s' for '%*s' on "DFID"\n", - LL_IT2STR(it), op_data->op_namelen, op_data->op_name, - PFID(&op_data->op_fid1)); + CDEBUG(D_INODE, "INTENT LOCK '%s' for "DFID" '%.*s' on "DFID"\n", + LL_IT2STR(it), PFID(&op_data->op_fid2), + (int)op_data->op_namelen, op_data->op_name, + PFID(&op_data->op_fid1)); - rc = lmv_check_connect(obd); - if (rc) - RETURN(rc); - - if (it->it_op & (IT_LOOKUP | IT_GETATTR | IT_LAYOUT)) - rc = lmv_intent_lookup(exp, op_data, lmm, lmmsize, it, - flags, reqp, cb_blocking, - extra_lock_flags); - else if (it->it_op & IT_OPEN) - rc = lmv_intent_open(exp, op_data, lmm, lmmsize, it, - flags, reqp, cb_blocking, - extra_lock_flags); - else - LBUG(); - RETURN(rc); + rc = lmv_check_connect(obd); + if (rc) + RETURN(rc); + + if (it->it_op & (IT_LOOKUP | IT_GETATTR | IT_LAYOUT)) + rc = lmv_intent_lookup(exp, op_data, it, reqp, cb_blocking, + extra_lock_flags); + else if (it->it_op & IT_OPEN) + rc = lmv_intent_open(exp, op_data, it, reqp, cb_blocking, + extra_lock_flags); + else + LBUG(); + + if (rc < 0) { + struct lustre_handle lock_handle; + + if (it->d.lustre.it_lock_mode != 0) { + lock_handle.cookie = it->d.lustre.it_lock_handle; + ldlm_lock_decref(&lock_handle, + it->d.lustre.it_lock_mode); + } + + it->d.lustre.it_lock_handle = 0; + it->d.lustre.it_lock_mode = 0; + + if (it->d.lustre.it_remote_lock_mode != 0) { + lock_handle.cookie = it->d.lustre.it_remote_lock_handle; + ldlm_lock_decref(&lock_handle, + it->d.lustre.it_remote_lock_mode); + } + + it->d.lustre.it_remote_lock_handle = 0; + it->d.lustre.it_remote_lock_mode = 0; + } + + RETURN(rc); }