X-Git-Url: https://git.whamcloud.com/?a=blobdiff_plain;f=lustre%2Fobdclass%2Fllog.c;h=d3b1e04d774986385751d5a20ddaf38bf6da5b5e;hb=3442db6faf685fbdbd092bdfdc8d273e4990a141;hp=7bdd90f1c84ebd49ceae0fc3ee1c7266d35101c3;hpb=339e9adc2ce671c948c0f278f522fa62e51d8371;p=fs%2Flustre-release.git diff --git a/lustre/obdclass/llog.c b/lustre/obdclass/llog.c index 7bdd90f..d3b1e04 100644 --- a/lustre/obdclass/llog.c +++ b/lustre/obdclass/llog.c @@ -15,11 +15,7 @@ * * You should have received a copy of the GNU General Public License * version 2 along with this program; If not, see - * http://www.sun.com/software/products/lustre/docs/GPLv2.pdf - * - * Please contact Sun Microsystems, Inc., 4150 Network Circle, Santa Clara, - * CA 95054 USA or visit www.sun.com if you need additional information or - * have any questions. + * http://www.gnu.org/licenses/gpl-2.0.html * * GPL HEADER END */ @@ -27,7 +23,7 @@ * Copyright (c) 2003, 2010, Oracle and/or its affiliates. All rights reserved. * Use is subject to license terms. * - * Copyright (c) 2012, 2014, Intel Corporation. + * Copyright (c) 2012, 2016, Intel Corporation. */ /* * This file is part of Lustre, http://www.lustre.org/ @@ -47,11 +43,12 @@ #define DEBUG_SUBSYSTEM S_LOG +#include #include -#include +#include #include +#include #include "llog_internal.h" - /* * Allocate a new log or catalog handle * Used inside llog_open(). @@ -202,11 +199,13 @@ int llog_cancel_rec(const struct lu_env *env, struct llog_handle *loghandle, struct llog_log_hdr *llh = loghandle->lgh_hdr; struct thandle *th; int rc; + int rc1; + bool subtract_count = false; ENTRY; - CDEBUG(D_RPCTRACE, "Canceling %d in log "DOSTID"\n", index, - POSTID(&loghandle->lgh_id.lgl_oi)); + CDEBUG(D_RPCTRACE, "Canceling %d in log "DFID"\n", index, + PFID(&loghandle->lgh_id.lgl_oi.oi_fid)); if (index == 0) { CERROR("Can't cancel index 0 which is header\n"); @@ -227,8 +226,11 @@ int llog_cancel_rec(const struct lu_env *env, struct llog_handle *loghandle, if (rc < 0) GOTO(out_trans, rc); - if ((llh->llh_flags & LLOG_F_ZAP_WHEN_EMPTY)) + if ((llh->llh_flags & LLOG_F_ZAP_WHEN_EMPTY)) { rc = llog_declare_destroy(env, loghandle, th); + if (rc < 0) + GOTO(out_trans, rc); + } th->th_wait_submit = 1; rc = dt_trans_start_local(env, dt, th); @@ -244,32 +246,33 @@ int llog_cancel_rec(const struct lu_env *env, struct llog_handle *loghandle, } loghandle->lgh_hdr->llh_count--; + subtract_count = true; /* Pass this index to llog_osd_write_rec(), which will use the index * to only update the necesary bitmap. */ lgi->lgi_cookie.lgc_index = index; /* update header */ rc = llog_write_rec(env, loghandle, &llh->llh_hdr, &lgi->lgi_cookie, LLOG_HEADER_IDX, th); - if (rc != 0) { - loghandle->lgh_hdr->llh_count++; - ext2_set_bit(index, LLOG_HDR_BITMAP(llh)); + if (rc != 0) GOTO(out_unlock, rc); - } if ((llh->llh_flags & LLOG_F_ZAP_WHEN_EMPTY) && (llh->llh_count == 1) && - (loghandle->lgh_last_idx == LLOG_HDR_BITMAP_SIZE(llh) - 1)) { + ((loghandle->lgh_last_idx == LLOG_HDR_BITMAP_SIZE(llh) - 1) || + (loghandle->u.phd.phd_cat_handle != NULL && + loghandle->u.phd.phd_cat_handle->u.chd.chd_current_log != + loghandle))) { + /* never try to destroy it again */ + llh->llh_flags &= ~LLOG_F_ZAP_WHEN_EMPTY; rc = llog_trans_destroy(env, loghandle, th); if (rc < 0) { /* Sigh, can not destroy the final plain llog, but * the bitmap has been clearly, so the record can not * be accessed anymore, let's return 0 for now, and * the orphan will be handled by LFSCK. */ - CERROR("%s: can't destroy empty llog #"DOSTID - "#%08x: rc = %d\n", + CERROR("%s: can't destroy empty llog "DFID": rc = %d\n", loghandle->lgh_ctxt->loc_obd->obd_name, - POSTID(&loghandle->lgh_id.lgl_oi), - loghandle->lgh_id.lgl_ogen, rc); + PFID(&loghandle->lgh_id.lgl_oi.oi_fid), rc); GOTO(out_unlock, rc = 0); } rc = LLOG_DEL_PLAIN; @@ -279,16 +282,24 @@ out_unlock: mutex_unlock(&loghandle->lgh_hdr_mutex); up_write(&loghandle->lgh_lock); out_trans: - dt_trans_stop(env, dt, th); + rc1 = dt_trans_stop(env, dt, th); + if (rc == 0) + rc = rc1; + if (rc < 0 && subtract_count) { + mutex_lock(&loghandle->lgh_hdr_mutex); + loghandle->lgh_hdr->llh_count++; + ext2_set_bit(index, LLOG_HDR_BITMAP(llh)); + mutex_unlock(&loghandle->lgh_hdr_mutex); + } RETURN(rc); } -static int llog_read_header(const struct lu_env *env, - struct llog_handle *handle, - struct obd_uuid *uuid) +int llog_read_header(const struct lu_env *env, struct llog_handle *handle, + const struct obd_uuid *uuid) { struct llog_operations *lop; int rc; + ENTRY; rc = llog_handle2ops(handle, &lop); if (rc) @@ -309,18 +320,24 @@ static int llog_read_header(const struct lu_env *env, LLOG_MIN_CHUNK_SIZE); llh->llh_hdr.lrh_len = handle->lgh_ctxt->loc_chunk_size; llh->llh_hdr.lrh_index = 0; - llh->llh_timestamp = cfs_time_current_sec(); + llh->llh_timestamp = ktime_get_real_seconds(); if (uuid) memcpy(&llh->llh_tgtuuid, uuid, sizeof(llh->llh_tgtuuid)); llh->llh_bitmap_offset = offsetof(typeof(*llh), llh_bitmap); + /* Since update llog header might also call this function, + * let's reset the bitmap to 0 here */ + memset(LLOG_HDR_BITMAP(llh), 0, llh->llh_hdr.lrh_len - + llh->llh_bitmap_offset - + sizeof(llh->llh_tail)); ext2_set_bit(0, LLOG_HDR_BITMAP(llh)); LLOG_HDR_TAIL(llh)->lrt_len = llh->llh_hdr.lrh_len; LLOG_HDR_TAIL(llh)->lrt_index = llh->llh_hdr.lrh_index; rc = 0; } - return rc; + RETURN(rc); } +EXPORT_SYMBOL(llog_read_header); int llog_init_handle(const struct lu_env *env, struct llog_handle *handle, int flags, struct obd_uuid *uuid) @@ -379,6 +396,7 @@ int llog_init_handle(const struct lu_env *env, struct llog_handle *handle, LASSERT(list_empty(&handle->u.chd.chd_head)); INIT_LIST_HEAD(&handle->u.chd.chd_head); llh->llh_size = sizeof(struct llog_logid_rec); + llh->llh_flags |= LLOG_F_IS_FIXSIZE; } else if (!(flags & LLOG_F_IS_PLAIN)) { CERROR("%s: unknown flags: %#x (expected %#x or %#x)\n", handle->lgh_ctxt->loc_obd->obd_name, @@ -407,6 +425,7 @@ static int llog_process_thread(void *arg) int rc = 0, index = 1, last_index; int saved_index = 0; int last_called_index = 0; + bool repeated = false; ENTRY; @@ -434,9 +453,10 @@ static int llog_process_thread(void *arg) while (rc == 0) { struct llog_rec_hdr *rec; - off_t chunk_offset; + off_t chunk_offset = 0; unsigned int buf_offset = 0; bool partial_chunk; + int lh_last_idx; /* skip records not set in bitmap */ while (index <= last_index && @@ -453,8 +473,19 @@ static int llog_process_thread(void *arg) repeat: /* get the buf with our target record; avoid old garbage */ memset(buf, 0, chunk_size); + /* the record index for outdated chunk data */ + lh_last_idx = loghandle->lgh_last_idx + 1; rc = llog_next_block(lpi->lpi_env, loghandle, &saved_index, index, &cur_offset, buf, chunk_size); + if (repeated && rc) + CDEBUG(D_OTHER, "cur_offset %llu, chunk_offset %llu," + " buf_offset %u, rc = %d\n", cur_offset, + (__u64)chunk_offset, buf_offset, rc); + /* we`ve tried to reread the chunk, but there is no + * new records */ + if (rc == -EIO && repeated && (chunk_offset + buf_offset) == + cur_offset) + GOTO(out, rc = 0); if (rc != 0) GOTO(out, rc); @@ -463,7 +494,7 @@ repeat: * The absolute offset of the current chunk is calculated * from cur_offset value and stored in chunk_offset variable. */ - if (cur_offset % chunk_size != 0) { + if ((cur_offset & (chunk_size - 1)) != 0) { partial_chunk = true; chunk_offset = cur_offset & ~(chunk_size - 1); } else { @@ -487,14 +518,38 @@ repeat: CDEBUG(D_OTHER, "after swabbing, type=%#x idx=%d\n", rec->lrh_type, rec->lrh_index); + /* the bitmap could be changed during processing + * records from the chunk. For wrapped catalog + * it means we can read deleted record and try to + * process it. Check this case and reread the chunk. */ + /* for partial chunk the end of it is zeroed, check * for index 0 to distinguish it. */ - if (partial_chunk && rec->lrh_index == 0) { + if ((partial_chunk && rec->lrh_index == 0) || + (index == lh_last_idx && + lh_last_idx != (loghandle->lgh_last_idx + 1))) { /* concurrent llog_add() might add new records * while llog_processing, check this is not * the case and re-read the current chunk * otherwise. */ - if (index > loghandle->lgh_last_idx) + int records; + /* lgh_last_idx could be less then index + * for catalog, if catalog is wrapped */ + if ((index > loghandle->lgh_last_idx && + !(loghandle->lgh_hdr->llh_flags & + LLOG_F_IS_CAT)) || repeated || + (loghandle->lgh_obj != NULL && + dt_object_remote(loghandle->lgh_obj))) + GOTO(out, rc = 0); + /* <2 records means no more records + * if the last record we processed was + * the final one, then the underlying + * object might have been destroyed yet. + * we better don't access that.. */ + mutex_lock(&loghandle->lgh_hdr_mutex); + records = loghandle->lgh_hdr->llh_count; + mutex_unlock(&loghandle->lgh_hdr_mutex); + if (records <= 1) GOTO(out, rc = 0); CDEBUG(D_OTHER, "Re-read last llog buffer for " "new records, index %u, last %u\n", @@ -502,13 +557,20 @@ repeat: /* save offset inside buffer for the re-read */ buf_offset = (char *)rec - (char *)buf; cur_offset = chunk_offset; + repeated = true; goto repeat; } + repeated = false; + if (rec->lrh_len == 0 || rec->lrh_len > chunk_size) { - CWARN("invalid length %d in llog record for " - "index %d/%d\n", rec->lrh_len, - rec->lrh_index, index); + CWARN("%s: invalid length %d in llog "DFID + "record for index %d/%d\n", + loghandle->lgh_ctxt->loc_obd->obd_name, + rec->lrh_len, + PFID(&loghandle->lgh_id.lgl_oi.oi_fid), + rec->lrh_index, index); + GOTO(out, rc = -EINVAL); } @@ -519,9 +581,10 @@ repeat: } if (rec->lrh_index != index) { - CERROR("%s: Invalid record: index %u but " - "expected %u\n", + CERROR("%s: "DFID" Invalid record: index %u" + " but expected %u\n", loghandle->lgh_ctxt->loc_obd->obd_name, + PFID(&loghandle->lgh_id.lgl_oi.oi_fid), rec->lrh_index, index); GOTO(out, rc = -ERANGE); } @@ -549,6 +612,12 @@ repeat: } if (rc) GOTO(out, rc); + /* some stupid callbacks directly cancel records + * and delete llog. Check it and stop + * processing. */ + if (loghandle->lgh_hdr == NULL || + loghandle->lgh_hdr->llh_count == 1) + GOTO(out, rc = 0); } /* exit if the last index is reached */ if (index >= last_index) @@ -562,22 +631,42 @@ out: cd->lpcd_last_idx = last_called_index; if (unlikely(rc == -EIO && loghandle->lgh_obj != NULL)) { - /* something bad happened to the processing of a local - * llog file, probably I/O error or the log got corrupted.. - * to be able to finally release the log we discard any - * remaining bits in the header */ - CERROR("Local llog found corrupted\n"); - while (index <= last_index) { - if (ext2_test_bit(index, LLOG_HDR_BITMAP(llh)) != 0) - llog_cancel_rec(lpi->lpi_env, loghandle, index); - index++; + if (dt_object_remote(loghandle->lgh_obj)) { + /* If it is remote object, then -EIO might means + * disconnection or eviction, let's return -EAGAIN, + * so for update recovery log processing, it will + * retry until the umount or abort recovery, see + * lod_sub_recovery_thread() */ + CERROR("%s retry remote llog process\n", + loghandle->lgh_ctxt->loc_obd->obd_name); + rc = -EAGAIN; + } else { + /* something bad happened to the processing of a local + * llog file, probably I/O error or the log got + * corrupted to be able to finally release the log we + * discard any remaining bits in the header */ + CERROR("%s: Local llog found corrupted #"DOSTID":%x" + " %s index %d count %d\n", + loghandle->lgh_ctxt->loc_obd->obd_name, + POSTID(&loghandle->lgh_id.lgl_oi), + loghandle->lgh_id.lgl_ogen, + ((llh->llh_flags & LLOG_F_IS_CAT) ? "catalog" : + "plain"), index, llh->llh_count); + + while (index <= last_index) { + if (ext2_test_bit(index, + LLOG_HDR_BITMAP(llh)) != 0) + llog_cancel_rec(lpi->lpi_env, loghandle, + index); + index++; + } + rc = 0; } - rc = 0; } OBD_FREE_LARGE(buf, chunk_size); - lpi->lpi_rc = rc; - return 0; + lpi->lpi_rc = rc; + return 0; } static int llog_process_thread_daemonize(void *arg) @@ -585,6 +674,22 @@ static int llog_process_thread_daemonize(void *arg) struct llog_process_info *lpi = arg; struct lu_env env; int rc; + struct nsproxy *new_ns, *curr_ns = current->nsproxy; + + task_lock(lpi->lpi_reftask); + new_ns = lpi->lpi_reftask->nsproxy; + if (curr_ns != new_ns) { + get_nsproxy(new_ns); + + current->nsproxy = new_ns; + /* XXX: we should call put_nsproxy() instead of + * atomic_dec(&ns->count) directly. But put_nsproxy() cannot be + * used outside of the kernel itself, because it calls + * free_nsproxy() which is not exported by the kernel + * (defined in kernel/nsproxy.c) */ + atomic_dec(&curr_ns->count); + } + task_unlock(lpi->lpi_reftask); unshare_fs_struct(); @@ -611,15 +716,15 @@ int llog_process_or_fork(const struct lu_env *env, ENTRY; - OBD_ALLOC_PTR(lpi); - if (lpi == NULL) { - CERROR("cannot alloc pointer\n"); - RETURN(-ENOMEM); - } - lpi->lpi_loghandle = loghandle; - lpi->lpi_cb = cb; - lpi->lpi_cbdata = data; - lpi->lpi_catdata = catdata; + OBD_ALLOC_PTR(lpi); + if (lpi == NULL) { + CERROR("cannot alloc pointer\n"); + RETURN(-ENOMEM); + } + lpi->lpi_loghandle = loghandle; + lpi->lpi_cb = cb; + lpi->lpi_cbdata = data; + lpi->lpi_catdata = catdata; if (fork) { struct task_struct *task; @@ -628,6 +733,10 @@ int llog_process_or_fork(const struct lu_env *env, * init the new one in llog_process_thread_daemonize. */ lpi->lpi_env = NULL; init_completion(&lpi->lpi_completion); + /* take reference to current, so that + * llog_process_thread_daemonize() can use it to switch to + * namespace associated with current */ + lpi->lpi_reftask = current; task = kthread_run(llog_process_thread_daemonize, lpi, "llog_process_thread"); if (IS_ERR(task)) { @@ -860,11 +969,29 @@ int llog_write_rec(const struct lu_env *env, struct llog_handle *handle, ENTRY; + /* API sanity checks */ + if (handle == NULL) { + CERROR("loghandle is missed\n"); + RETURN(-EPROTO); + } else if (handle->lgh_obj == NULL) { + CERROR("loghandle %p with NULL object\n", + handle); + RETURN(-EPROTO); + } else if (th == NULL) { + CERROR("%s: missed transaction handle\n", + handle->lgh_obj->do_lu.lo_dev->ld_obd->obd_name); + RETURN(-EPROTO); + } else if (handle->lgh_hdr == NULL) { + CERROR("%s: loghandle %p with no header\n", + handle->lgh_obj->do_lu.lo_dev->ld_obd->obd_name, + handle); + RETURN(-EPROTO); + } + rc = llog_handle2ops(handle, &lop); if (rc) RETURN(rc); - LASSERT(lop); if (lop->lop_write_rec == NULL) RETURN(-EOPNOTSUPP); @@ -1212,3 +1339,22 @@ out_close: RETURN(rc); } EXPORT_SYMBOL(llog_backup); + +/* Get size of llog */ +__u64 llog_size(const struct lu_env *env, struct llog_handle *llh) +{ + int rc; + struct lu_attr la; + + rc = llh->lgh_obj->do_ops->do_attr_get(env, llh->lgh_obj, &la); + if (rc) { + CERROR("%s: attr_get failed for "DFID": rc = %d\n", + llh->lgh_ctxt->loc_obd->obd_name, + PFID(&llh->lgh_id.lgl_oi.oi_fid), rc); + return 0; + } + + return la.la_size; +} +EXPORT_SYMBOL(llog_size); +