X-Git-Url: https://git.openpandora.org/cgi-bin/gitweb.cgi?a=blobdiff_plain;f=drivers%2Fblock%2Fdrbd%2Fdrbd_receiver.c;h=e5686a81f42c0bb31740492bdece41aa0db57263;hb=62b0da3a244ac33d25a77861ef1cc0080103f2ff;hp=24487d4fb20297e6676a91e60c9525f26092e315;hpb=63e8078a12ed287f40f39082ff87e8db6621dab9;p=pandora-kernel.git diff --git a/drivers/block/drbd/drbd_receiver.c b/drivers/block/drbd/drbd_receiver.c index 24487d4fb202..e5686a81f42c 100644 --- a/drivers/block/drbd/drbd_receiver.c +++ b/drivers/block/drbd/drbd_receiver.c @@ -187,15 +187,6 @@ static struct page *drbd_pp_first_pages_or_try_alloc(struct drbd_conf *mdev, int return NULL; } -/* kick lower level device, if we have more than (arbitrary number) - * reference counts on it, which typically are locally submitted io - * requests. don't use unacked_cnt, so we speed up proto A and B, too. */ -static void maybe_kick_lo(struct drbd_conf *mdev) -{ - if (atomic_read(&mdev->local_cnt) >= mdev->net_conf->unplug_watermark) - drbd_kick_lo(mdev); -} - static void reclaim_net_ee(struct drbd_conf *mdev, struct list_head *to_be_freed) { struct drbd_epoch_entry *e; @@ -219,7 +210,6 @@ static void drbd_kick_lo_and_reclaim_net(struct drbd_conf *mdev) LIST_HEAD(reclaimed); struct drbd_epoch_entry *e, *t; - maybe_kick_lo(mdev); spin_lock_irq(&mdev->req_lock); reclaim_net_ee(mdev, &reclaimed); spin_unlock_irq(&mdev->req_lock); @@ -287,7 +277,7 @@ static void drbd_pp_free(struct drbd_conf *mdev, struct page *page, int is_net) atomic_t *a = is_net ? &mdev->pp_in_use_by_net : &mdev->pp_in_use; int i; - if (drbd_pp_vacant > (DRBD_MAX_SEGMENT_SIZE/PAGE_SIZE)*minor_count) + if (drbd_pp_vacant > (DRBD_MAX_BIO_SIZE/PAGE_SIZE)*minor_count) i = page_chain_free(page); else { struct page *tmp; @@ -329,7 +319,7 @@ struct drbd_epoch_entry *drbd_alloc_ee(struct drbd_conf *mdev, struct page *page; unsigned nr_pages = (data_size + PAGE_SIZE -1) >> PAGE_SHIFT; - if (FAULT_ACTIVE(mdev, DRBD_FAULT_AL_EE)) + if (drbd_insert_fault(mdev, DRBD_FAULT_AL_EE)) return NULL; e = mempool_alloc(drbd_ee_mempool, gfp_mask & ~__GFP_HIGHMEM); @@ -436,8 +426,7 @@ void _drbd_wait_ee_list_empty(struct drbd_conf *mdev, struct list_head *head) while (!list_empty(head)) { prepare_to_wait(&mdev->ee_wait, &wait, TASK_UNINTERRUPTIBLE); spin_unlock_irq(&mdev->req_lock); - drbd_kick_lo(mdev); - schedule(); + io_schedule(); finish_wait(&mdev->ee_wait, &wait); spin_lock_irq(&mdev->req_lock); } @@ -736,16 +725,16 @@ static int drbd_socket_okay(struct drbd_conf *mdev, struct socket **sock) char tb[4]; if (!*sock) - return FALSE; + return false; rr = drbd_recv_short(mdev, *sock, tb, 4, MSG_DONTWAIT | MSG_PEEK); if (rr > 0 || rr == -EAGAIN) { - return TRUE; + return true; } else { sock_release(*sock); *sock = NULL; - return FALSE; + return false; } } @@ -779,8 +768,7 @@ static int drbd_connect(struct drbd_conf *mdev) if (s || ++try >= 3) break; /* give the other side time to call bind() & listen() */ - __set_current_state(TASK_INTERRUPTIBLE); - schedule_timeout(HZ / 10); + schedule_timeout_interruptible(HZ / 10); } if (s) { @@ -799,8 +787,7 @@ static int drbd_connect(struct drbd_conf *mdev) } if (sock && msock) { - __set_current_state(TASK_INTERRUPTIBLE); - schedule_timeout(HZ / 10); + schedule_timeout_interruptible(HZ / 10); ok = drbd_socket_okay(mdev, &sock); ok = drbd_socket_okay(mdev, &msock) && ok; if (ok) @@ -917,7 +904,7 @@ retry: put_ldev(mdev); } - if (!drbd_send_protocol(mdev)) + if (drbd_send_protocol(mdev) == -1) return -1; drbd_send_sync_param(mdev, &mdev->sync_conf); drbd_send_sizes(mdev, 0, 0); @@ -944,7 +931,7 @@ static int drbd_recv_header(struct drbd_conf *mdev, enum drbd_packets *cmd, unsi r = drbd_recv(mdev, h, sizeof(*h)); if (unlikely(r != sizeof(*h))) { dev_err(DEV, "short read expecting header on sock: r=%d\n", r); - return FALSE; + return false; } if (likely(h->h80.magic == BE_DRBD_MAGIC)) { @@ -958,11 +945,11 @@ static int drbd_recv_header(struct drbd_conf *mdev, enum drbd_packets *cmd, unsi be32_to_cpu(h->h80.magic), be16_to_cpu(h->h80.command), be16_to_cpu(h->h80.length)); - return FALSE; + return false; } mdev->last_received = jiffies; - return TRUE; + return true; } static void drbd_flush(struct drbd_conf *mdev) @@ -1111,8 +1098,6 @@ next_bio: /* > e->sector, unless this is the first bio */ bio->bi_sector = sector; bio->bi_bdev = mdev->ldev->backing_bdev; - /* we special case some flags in the multi-bio case, see below - * (REQ_UNPLUG) */ bio->bi_rw = rw; bio->bi_private = e; bio->bi_end_io = drbd_endio_sec; @@ -1141,13 +1126,8 @@ next_bio: bios = bios->bi_next; bio->bi_next = NULL; - /* strip off REQ_UNPLUG unless it is the last bio */ - if (bios) - bio->bi_rw &= ~REQ_UNPLUG; - drbd_generic_make_request(mdev, fault_type, bio); } while (bios); - maybe_kick_lo(mdev); return 0; fail: @@ -1167,9 +1147,6 @@ static int receive_Barrier(struct drbd_conf *mdev, enum drbd_packets cmd, unsign inc_unacked(mdev); - if (mdev->net_conf->wire_protocol != DRBD_PROT_C) - drbd_kick_lo(mdev); - mdev->current_epoch->barrier_nr = p->barrier; rv = drbd_may_finish_epoch(mdev, mdev->current_epoch, EV_GOT_BARRIER_NR); @@ -1181,7 +1158,7 @@ static int receive_Barrier(struct drbd_conf *mdev, enum drbd_packets cmd, unsign switch (mdev->write_ordering) { case WO_none: if (rv == FE_RECYCLED) - return TRUE; + return true; /* receiver context, in the writeout path of the other node. * avoid potential distributed deadlock */ @@ -1209,10 +1186,10 @@ static int receive_Barrier(struct drbd_conf *mdev, enum drbd_packets cmd, unsign D_ASSERT(atomic_read(&epoch->active) == 0); D_ASSERT(epoch->flags == 0); - return TRUE; + return true; default: dev_err(DEV, "Strangeness in mdev->write_ordering %d\n", mdev->write_ordering); - return FALSE; + return false; } epoch->flags = 0; @@ -1230,7 +1207,7 @@ static int receive_Barrier(struct drbd_conf *mdev, enum drbd_packets cmd, unsign } spin_unlock(&mdev->epoch_lock); - return TRUE; + return true; } /* used from receive_RSDataReply (recv_resync_read) @@ -1260,8 +1237,9 @@ read_in_block(struct drbd_conf *mdev, u64 id, sector_t sector, int data_size) __ data_size -= dgs; + ERR_IF(data_size == 0) return NULL; ERR_IF(data_size & 0x1ff) return NULL; - ERR_IF(data_size > DRBD_MAX_SEGMENT_SIZE) return NULL; + ERR_IF(data_size > DRBD_MAX_BIO_SIZE) return NULL; /* even though we trust out peer, * we sometimes have to double check. */ @@ -1285,7 +1263,7 @@ read_in_block(struct drbd_conf *mdev, u64 id, sector_t sector, int data_size) __ unsigned len = min_t(int, ds, PAGE_SIZE); data = kmap(page); rr = drbd_recv(mdev, data, len); - if (FAULT_ACTIVE(mdev, DRBD_FAULT_RECEIVE)) { + if (drbd_insert_fault(mdev, DRBD_FAULT_RECEIVE)) { dev_err(DEV, "Fault injection: Corrupting data on receive\n"); data[0] = data[0] ^ (unsigned long)-1; } @@ -1302,7 +1280,8 @@ read_in_block(struct drbd_conf *mdev, u64 id, sector_t sector, int data_size) __ if (dgs) { drbd_csum_ee(mdev, mdev->integrity_r_tfm, e, dig_vv); if (memcmp(dig_in, dig_vv, dgs)) { - dev_err(DEV, "Digest integrity check FAILED.\n"); + dev_err(DEV, "Digest integrity check FAILED: %llus +%u\n", + (unsigned long long)sector, data_size); drbd_bcast_ee(mdev, "digest failed", dgs, dig_in, dig_vv, e); drbd_free_ee(mdev, e); @@ -1323,7 +1302,7 @@ static int drbd_drain_block(struct drbd_conf *mdev, int data_size) void *data; if (!data_size) - return TRUE; + return true; page = drbd_pp_alloc(mdev, 1, 1); @@ -1446,7 +1425,7 @@ static int recv_resync_read(struct drbd_conf *mdev, sector_t sector, int data_si atomic_add(data_size >> 9, &mdev->rs_sect_ev); if (drbd_submit_ee(mdev, e, WRITE, DRBD_FAULT_RS_WR) == 0) - return TRUE; + return true; /* drbd_submit_ee currently fails for one reason only: * not being able to allocate enough bios. @@ -1458,7 +1437,7 @@ static int recv_resync_read(struct drbd_conf *mdev, sector_t sector, int data_si drbd_free_ee(mdev, e); fail: put_ldev(mdev); - return FALSE; + return false; } static int receive_DataReply(struct drbd_conf *mdev, enum drbd_packets cmd, unsigned int data_size) @@ -1475,7 +1454,7 @@ static int receive_DataReply(struct drbd_conf *mdev, enum drbd_packets cmd, unsi spin_unlock_irq(&mdev->req_lock); if (unlikely(!req)) { dev_err(DEV, "Got a corrupt block_id/sector pair(1).\n"); - return FALSE; + return false; } /* hlist_del(&req->colision) is done in _req_may_be_done, to avoid @@ -1632,16 +1611,15 @@ static int drbd_wait_peer_seq(struct drbd_conf *mdev, const u32 packet_seq) return ret; } -static unsigned long write_flags_to_bio(struct drbd_conf *mdev, u32 dpf) +/* see also bio_flags_to_wire() + * DRBD_REQ_*, because we need to semantically map the flags to data packet + * flags and back. We may replicate to other kernel versions. */ +static unsigned long wire_flags_to_bio(struct drbd_conf *mdev, u32 dpf) { - if (mdev->agreed_pro_version >= 95) - return (dpf & DP_RW_SYNC ? REQ_SYNC : 0) | - (dpf & DP_UNPLUG ? REQ_UNPLUG : 0) | - (dpf & DP_FUA ? REQ_FUA : 0) | - (dpf & DP_FLUSH ? REQ_FUA : 0) | - (dpf & DP_DISCARD ? REQ_DISCARD : 0); - else - return dpf & DP_RW_SYNC ? (REQ_SYNC | REQ_UNPLUG) : 0; + return (dpf & DP_RW_SYNC ? REQ_SYNC : 0) | + (dpf & DP_FUA ? REQ_FUA : 0) | + (dpf & DP_FLUSH ? REQ_FLUSH : 0) | + (dpf & DP_DISCARD ? REQ_DISCARD : 0); } /* mirrored write */ @@ -1654,9 +1632,6 @@ static int receive_Data(struct drbd_conf *mdev, enum drbd_packets cmd, unsigned u32 dp_flags; if (!get_ldev(mdev)) { - if (__ratelimit(&drbd_ratelimit_state)) - dev_err(DEV, "Can not write mirrored data block " - "to local disk.\n"); spin_lock(&mdev->peer_seq_lock); if (mdev->peer_seq+1 == be32_to_cpu(p->seq_num)) mdev->peer_seq++; @@ -1676,23 +1651,23 @@ static int receive_Data(struct drbd_conf *mdev, enum drbd_packets cmd, unsigned e = read_in_block(mdev, p->block_id, sector, data_size); if (!e) { put_ldev(mdev); - return FALSE; + return false; } e->w.cb = e_end_block; + dp_flags = be32_to_cpu(p->dp_flags); + rw |= wire_flags_to_bio(mdev, dp_flags); + + if (dp_flags & DP_MAY_SET_IN_SYNC) + e->flags |= EE_MAY_SET_IN_SYNC; + spin_lock(&mdev->epoch_lock); e->epoch = mdev->current_epoch; atomic_inc(&e->epoch->epoch_size); atomic_inc(&e->epoch->active); spin_unlock(&mdev->epoch_lock); - dp_flags = be32_to_cpu(p->dp_flags); - rw |= write_flags_to_bio(mdev, dp_flags); - - if (dp_flags & DP_MAY_SET_IN_SYNC) - e->flags |= EE_MAY_SET_IN_SYNC; - /* I'm the receiver, I do hold a net_cnt reference. */ if (!mdev->net_conf->two_primaries) { spin_lock_irq(&mdev->req_lock); @@ -1795,7 +1770,7 @@ static int receive_Data(struct drbd_conf *mdev, enum drbd_packets cmd, unsigned put_ldev(mdev); wake_asender(mdev); finish_wait(&mdev->misc_wait, &wait); - return TRUE; + return true; } if (signal_pending(current)) { @@ -1851,7 +1826,7 @@ static int receive_Data(struct drbd_conf *mdev, enum drbd_packets cmd, unsigned } if (drbd_submit_ee(mdev, e, rw, DRBD_FAULT_DT_WR) == 0) - return TRUE; + return true; /* drbd_submit_ee currently fails for one reason only: * not being able to allocate enough bios. @@ -1869,7 +1844,7 @@ out_interrupted: * receive a barrier... atomic_inc(&mdev->epoch_size); */ put_ldev(mdev); drbd_free_ee(mdev, e); - return FALSE; + return false; } /* We may throttle resync, if the lower device seems to be busy, @@ -1883,10 +1858,11 @@ out_interrupted: * The current sync rate used here uses only the most recent two step marks, * to have a short time average so we can react faster. */ -int drbd_rs_should_slow_down(struct drbd_conf *mdev) +int drbd_rs_should_slow_down(struct drbd_conf *mdev, sector_t sector) { struct gendisk *disk = mdev->ldev->backing_bdev->bd_contains->bd_disk; unsigned long db, dt, dbdt; + struct lc_element *tmp; int curr_events; int throttle = 0; @@ -1894,9 +1870,22 @@ int drbd_rs_should_slow_down(struct drbd_conf *mdev) if (mdev->sync_conf.c_min_rate == 0) return 0; + spin_lock_irq(&mdev->al_lock); + tmp = lc_find(mdev->resync, BM_SECT_TO_EXT(sector)); + if (tmp) { + struct bm_extent *bm_ext = lc_entry(tmp, struct bm_extent, lce); + if (test_bit(BME_PRIORITY, &bm_ext->flags)) { + spin_unlock_irq(&mdev->al_lock); + return 0; + } + /* Do not slow down if app IO is already waiting for this extent */ + } + spin_unlock_irq(&mdev->al_lock); + curr_events = (int)part_stat_read(&disk->part0, sectors[0]) + (int)part_stat_read(&disk->part0, sectors[1]) - atomic_read(&mdev->rs_sect_ev); + if (!mdev->rs_last_events || curr_events - mdev->rs_last_events > 64) { unsigned long rs_left; int i; @@ -1905,8 +1894,12 @@ int drbd_rs_should_slow_down(struct drbd_conf *mdev) /* sync speed average over the last 2*DRBD_SYNC_MARK_STEP, * approx. */ - i = (mdev->rs_last_mark + DRBD_SYNC_MARKS-2) % DRBD_SYNC_MARKS; - rs_left = drbd_bm_total_weight(mdev) - mdev->rs_failed; + i = (mdev->rs_last_mark + DRBD_SYNC_MARKS-1) % DRBD_SYNC_MARKS; + + if (mdev->state.conn == C_VERIFY_S || mdev->state.conn == C_VERIFY_T) + rs_left = mdev->ov_left; + else + rs_left = drbd_bm_total_weight(mdev) - mdev->rs_failed; dt = ((long)jiffies - (long)mdev->rs_mark_time[i]) / HZ; if (!dt) @@ -1934,15 +1927,15 @@ static int receive_DataRequest(struct drbd_conf *mdev, enum drbd_packets cmd, un sector = be64_to_cpu(p->sector); size = be32_to_cpu(p->blksize); - if (size <= 0 || (size & 0x1ff) != 0 || size > DRBD_MAX_SEGMENT_SIZE) { + if (size <= 0 || (size & 0x1ff) != 0 || size > DRBD_MAX_BIO_SIZE) { dev_err(DEV, "%s:%d: sector: %llus, size: %u\n", __FILE__, __LINE__, (unsigned long long)sector, size); - return FALSE; + return false; } if (sector + (size>>9) > capacity) { dev_err(DEV, "%s:%d: sector: %llus, size: %u\n", __FILE__, __LINE__, (unsigned long long)sector, size); - return FALSE; + return false; } if (!get_ldev_if_state(mdev, D_UP_TO_DATE)) { @@ -1979,7 +1972,7 @@ static int receive_DataRequest(struct drbd_conf *mdev, enum drbd_packets cmd, un e = drbd_alloc_ee(mdev, p->block_id, sector, size, GFP_NOIO); if (!e) { put_ldev(mdev); - return FALSE; + return false; } switch (cmd) { @@ -1992,6 +1985,8 @@ static int receive_DataRequest(struct drbd_conf *mdev, enum drbd_packets cmd, un case P_RS_DATA_REQUEST: e->w.cb = w_e_end_rsdata_req; fault_type = DRBD_FAULT_RS_RD; + /* used in the sector offset progress display */ + mdev->bm_resync_fo = BM_SECT_TO_BIT(sector); break; case P_OV_REPLY: @@ -2013,7 +2008,11 @@ static int receive_DataRequest(struct drbd_conf *mdev, enum drbd_packets cmd, un if (cmd == P_CSUM_RS_REQUEST) { D_ASSERT(mdev->agreed_pro_version >= 89); e->w.cb = w_e_end_csum_rs_req; + /* used in the sector offset progress display */ + mdev->bm_resync_fo = BM_SECT_TO_BIT(sector); } else if (cmd == P_OV_REPLY) { + /* track progress, we may need to throttle */ + atomic_add(size >> 9, &mdev->rs_sect_in); e->w.cb = w_e_end_ov_reply; dec_rs_pending(mdev); /* drbd_rs_begin_io done when we sent this request, @@ -2025,9 +2024,16 @@ static int receive_DataRequest(struct drbd_conf *mdev, enum drbd_packets cmd, un case P_OV_REQUEST: if (mdev->ov_start_sector == ~(sector_t)0 && mdev->agreed_pro_version >= 90) { + unsigned long now = jiffies; + int i; mdev->ov_start_sector = sector; mdev->ov_position = sector; - mdev->ov_left = mdev->rs_total - BM_SECT_TO_BIT(sector); + mdev->ov_left = drbd_bm_bits(mdev) - BM_SECT_TO_BIT(sector); + mdev->rs_total = mdev->ov_left; + for (i = 0; i < DRBD_SYNC_MARKS; i++) { + mdev->rs_mark_left[i] = mdev->ov_left; + mdev->rs_mark_time[i] = now; + } dev_info(DEV, "Online Verify start sector: %llu\n", (unsigned long long)sector); } @@ -2064,9 +2070,9 @@ static int receive_DataRequest(struct drbd_conf *mdev, enum drbd_packets cmd, un * we would also throttle its application reads. * In that case, throttling is done on the SyncTarget only. */ - if (mdev->state.peer != R_PRIMARY && drbd_rs_should_slow_down(mdev)) - msleep(100); - if (drbd_rs_begin_io(mdev, e->sector)) + if (mdev->state.peer != R_PRIMARY && drbd_rs_should_slow_down(mdev, sector)) + schedule_timeout_uninterruptible(HZ/10); + if (drbd_rs_begin_io(mdev, sector)) goto out_free_e; submit_for_resync: @@ -2079,7 +2085,7 @@ submit: spin_unlock_irq(&mdev->req_lock); if (drbd_submit_ee(mdev, e, READ, fault_type) == 0) - return TRUE; + return true; /* drbd_submit_ee currently fails for one reason only: * not being able to allocate enough bios. @@ -2092,7 +2098,7 @@ submit: out_free_e: put_ldev(mdev); drbd_free_ee(mdev, e); - return FALSE; + return false; } static int drbd_asb_recover_0p(struct drbd_conf *mdev) __must_hold(local) @@ -2169,10 +2175,7 @@ static int drbd_asb_recover_0p(struct drbd_conf *mdev) __must_hold(local) static int drbd_asb_recover_1p(struct drbd_conf *mdev) __must_hold(local) { - int self, peer, hg, rv = -100; - - self = mdev->ldev->md.uuid[UI_BITMAP] & 1; - peer = mdev->p_uuid[UI_BITMAP] & 1; + int hg, rv = -100; switch (mdev->net_conf->after_sb_1p) { case ASB_DISCARD_YOUNGER_PRI: @@ -2199,12 +2202,14 @@ static int drbd_asb_recover_1p(struct drbd_conf *mdev) __must_hold(local) case ASB_CALL_HELPER: hg = drbd_asb_recover_0p(mdev); if (hg == -1 && mdev->state.role == R_PRIMARY) { - self = drbd_set_role(mdev, R_SECONDARY, 0); + enum drbd_state_rv rv2; + + drbd_set_role(mdev, R_SECONDARY, 0); /* drbd_change_state() does not sleep while in SS_IN_TRANSIENT_STATE, * we might be here in C_WF_REPORT_PARAMS which is transient. * we do not need to wait for the after state change work either. */ - self = drbd_change_state(mdev, CS_VERBOSE, NS(role, R_SECONDARY)); - if (self != SS_SUCCESS) { + rv2 = drbd_change_state(mdev, CS_VERBOSE, NS(role, R_SECONDARY)); + if (rv2 != SS_SUCCESS) { drbd_khelper(mdev, "pri-lost-after-sb"); } else { dev_warn(DEV, "Successfully gave up primary role.\n"); @@ -2219,10 +2224,7 @@ static int drbd_asb_recover_1p(struct drbd_conf *mdev) __must_hold(local) static int drbd_asb_recover_2p(struct drbd_conf *mdev) __must_hold(local) { - int self, peer, hg, rv = -100; - - self = mdev->ldev->md.uuid[UI_BITMAP] & 1; - peer = mdev->p_uuid[UI_BITMAP] & 1; + int hg, rv = -100; switch (mdev->net_conf->after_sb_2p) { case ASB_DISCARD_YOUNGER_PRI: @@ -2242,11 +2244,13 @@ static int drbd_asb_recover_2p(struct drbd_conf *mdev) __must_hold(local) case ASB_CALL_HELPER: hg = drbd_asb_recover_0p(mdev); if (hg == -1) { + enum drbd_state_rv rv2; + /* drbd_change_state() does not sleep while in SS_IN_TRANSIENT_STATE, * we might be here in C_WF_REPORT_PARAMS which is transient. * we do not need to wait for the after state change work either. */ - self = drbd_change_state(mdev, CS_VERBOSE, NS(role, R_SECONDARY)); - if (self != SS_SUCCESS) { + rv2 = drbd_change_state(mdev, CS_VERBOSE, NS(role, R_SECONDARY)); + if (rv2 != SS_SUCCESS) { drbd_khelper(mdev, "pri-lost-after-sb"); } else { dev_warn(DEV, "Successfully gave up primary role.\n"); @@ -2285,6 +2289,8 @@ static void drbd_uuid_dump(struct drbd_conf *mdev, char *text, u64 *uuid, -2 C_SYNC_TARGET set BitMap -100 after split brain, disconnect -1000 unrelated data +-1091 requires proto 91 +-1096 requires proto 96 */ static int drbd_uuid_compare(struct drbd_conf *mdev, int *rule_nr) __must_hold(local) { @@ -2314,7 +2320,7 @@ static int drbd_uuid_compare(struct drbd_conf *mdev, int *rule_nr) __must_hold(l if (mdev->p_uuid[UI_BITMAP] == (u64)0 && mdev->ldev->md.uuid[UI_BITMAP] != (u64)0) { if (mdev->agreed_pro_version < 91) - return -1001; + return -1091; if ((mdev->ldev->md.uuid[UI_BITMAP] & ~((u64)1)) == (mdev->p_uuid[UI_HISTORY_START] & ~((u64)1)) && (mdev->ldev->md.uuid[UI_HISTORY_START] & ~((u64)1)) == (mdev->p_uuid[UI_HISTORY_START + 1] & ~((u64)1))) { @@ -2335,7 +2341,7 @@ static int drbd_uuid_compare(struct drbd_conf *mdev, int *rule_nr) __must_hold(l if (mdev->ldev->md.uuid[UI_BITMAP] == (u64)0 && mdev->p_uuid[UI_BITMAP] != (u64)0) { if (mdev->agreed_pro_version < 91) - return -1001; + return -1091; if ((mdev->ldev->md.uuid[UI_HISTORY_START] & ~((u64)1)) == (mdev->p_uuid[UI_BITMAP] & ~((u64)1)) && (mdev->ldev->md.uuid[UI_HISTORY_START + 1] & ~((u64)1)) == (mdev->p_uuid[UI_HISTORY_START] & ~((u64)1))) { @@ -2380,17 +2386,22 @@ static int drbd_uuid_compare(struct drbd_conf *mdev, int *rule_nr) __must_hold(l *rule_nr = 51; peer = mdev->p_uuid[UI_HISTORY_START] & ~((u64)1); if (self == peer) { - self = mdev->ldev->md.uuid[UI_HISTORY_START] & ~((u64)1); - peer = mdev->p_uuid[UI_HISTORY_START + 1] & ~((u64)1); - if (self == peer) { + if (mdev->agreed_pro_version < 96 ? + (mdev->ldev->md.uuid[UI_HISTORY_START] & ~((u64)1)) == + (mdev->p_uuid[UI_HISTORY_START + 1] & ~((u64)1)) : + peer + UUID_NEW_BM_OFFSET == (mdev->p_uuid[UI_BITMAP] & ~((u64)1))) { /* The last P_SYNC_UUID did not get though. Undo the last start of resync as sync source modifications of the peer's UUIDs. */ if (mdev->agreed_pro_version < 91) - return -1001; + return -1091; mdev->p_uuid[UI_BITMAP] = mdev->p_uuid[UI_HISTORY_START]; mdev->p_uuid[UI_HISTORY_START] = mdev->p_uuid[UI_HISTORY_START + 1]; + + dev_info(DEV, "Did not got last syncUUID packet, corrected:\n"); + drbd_uuid_dump(mdev, "peer", mdev->p_uuid, mdev->p_uuid[UI_SIZE], mdev->p_uuid[UI_FLAGS]); + return -1; } } @@ -2412,20 +2423,20 @@ static int drbd_uuid_compare(struct drbd_conf *mdev, int *rule_nr) __must_hold(l *rule_nr = 71; self = mdev->ldev->md.uuid[UI_HISTORY_START] & ~((u64)1); if (self == peer) { - self = mdev->ldev->md.uuid[UI_HISTORY_START + 1] & ~((u64)1); - peer = mdev->p_uuid[UI_HISTORY_START] & ~((u64)1); - if (self == peer) { + if (mdev->agreed_pro_version < 96 ? + (mdev->ldev->md.uuid[UI_HISTORY_START + 1] & ~((u64)1)) == + (mdev->p_uuid[UI_HISTORY_START] & ~((u64)1)) : + self + UUID_NEW_BM_OFFSET == (mdev->ldev->md.uuid[UI_BITMAP] & ~((u64)1))) { /* The last P_SYNC_UUID did not get though. Undo the last start of resync as sync source modifications of our UUIDs. */ if (mdev->agreed_pro_version < 91) - return -1001; + return -1091; _drbd_uuid_set(mdev, UI_BITMAP, mdev->ldev->md.uuid[UI_HISTORY_START]); _drbd_uuid_set(mdev, UI_HISTORY_START, mdev->ldev->md.uuid[UI_HISTORY_START + 1]); - dev_info(DEV, "Undid last start of resync:\n"); - + dev_info(DEV, "Last syncUUID did not get through, corrected:\n"); drbd_uuid_dump(mdev, "self", mdev->ldev->md.uuid, mdev->state.disk >= D_NEGOTIATING ? drbd_bm_total_weight(mdev) : 0, 0); @@ -2488,8 +2499,8 @@ static enum drbd_conns drbd_sync_handshake(struct drbd_conf *mdev, enum drbd_rol dev_alert(DEV, "Unrelated data, aborting!\n"); return C_MASK; } - if (hg == -1001) { - dev_alert(DEV, "To resolve this both sides have to support at least protocol\n"); + if (hg < -1000) { + dev_alert(DEV, "To resolve this both sides have to support at least protocol %d\n", -hg - 1000); return C_MASK; } @@ -2682,7 +2693,7 @@ static int receive_protocol(struct drbd_conf *mdev, enum drbd_packets cmd, unsig unsigned char *my_alg = mdev->net_conf->integrity_alg; if (drbd_recv(mdev, p_integrity_alg, data_size) != data_size) - return FALSE; + return false; p_integrity_alg[SHARED_SECRET_MAX-1] = 0; if (strcmp(p_integrity_alg, my_alg)) { @@ -2693,11 +2704,11 @@ static int receive_protocol(struct drbd_conf *mdev, enum drbd_packets cmd, unsig my_alg[0] ? my_alg : (unsigned char *)""); } - return TRUE; + return true; disconnect: drbd_force_state(mdev, NS(conn, C_DISCONNECTING)); - return FALSE; + return false; } /* helper function @@ -2729,7 +2740,7 @@ struct crypto_hash *drbd_crypto_alloc_digest_safe(const struct drbd_conf *mdev, static int receive_SyncParam(struct drbd_conf *mdev, enum drbd_packets cmd, unsigned int packet_size) { - int ok = TRUE; + int ok = true; struct p_rs_param_95 *p = &mdev->data.rbuf.rs_param_95; unsigned int header_size, data_size, exp_max_sz; struct crypto_hash *verify_tfm = NULL; @@ -2747,7 +2758,7 @@ static int receive_SyncParam(struct drbd_conf *mdev, enum drbd_packets cmd, unsi if (packet_size > exp_max_sz) { dev_err(DEV, "SyncParam packet too long: received %u, expected <= %u bytes\n", packet_size, exp_max_sz); - return FALSE; + return false; } if (apv <= 88) { @@ -2767,7 +2778,7 @@ static int receive_SyncParam(struct drbd_conf *mdev, enum drbd_packets cmd, unsi memset(p->verify_alg, 0, 2 * SHARED_SECRET_MAX); if (drbd_recv(mdev, &p->head.payload, header_size) != header_size) - return FALSE; + return false; mdev->sync_conf.rate = be32_to_cpu(p->rate); @@ -2777,11 +2788,11 @@ static int receive_SyncParam(struct drbd_conf *mdev, enum drbd_packets cmd, unsi dev_err(DEV, "verify-alg too long, " "peer wants %u, accepting only %u byte\n", data_size, SHARED_SECRET_MAX); - return FALSE; + return false; } if (drbd_recv(mdev, p->verify_alg, data_size) != data_size) - return FALSE; + return false; /* we expect NUL terminated string */ /* but just in case someone tries to be evil */ @@ -2875,7 +2886,7 @@ disconnect: /* but free the verify_tfm again, if csums_tfm did not work out */ crypto_free_hash(verify_tfm); drbd_force_state(mdev, NS(conn, C_DISCONNECTING)); - return FALSE; + return false; } static void drbd_setup_order_type(struct drbd_conf *mdev, int peer) @@ -2901,7 +2912,7 @@ static int receive_sizes(struct drbd_conf *mdev, enum drbd_packets cmd, unsigned { struct p_sizes *p = &mdev->data.rbuf.sizes; enum determine_dev_size dd = unchanged; - unsigned int max_seg_s; + unsigned int max_bio_size; sector_t p_size, p_usize, my_usize; int ldsc = 0; /* local disk size changed */ enum dds_flags ddsf; @@ -2912,7 +2923,7 @@ static int receive_sizes(struct drbd_conf *mdev, enum drbd_packets cmd, unsigned if (p_size == 0 && mdev->state.disk == D_DISKLESS) { dev_err(DEV, "some backing storage is needed\n"); drbd_force_state(mdev, NS(conn, C_DISCONNECTING)); - return FALSE; + return false; } /* just store the peer's disk size for now. @@ -2949,18 +2960,17 @@ static int receive_sizes(struct drbd_conf *mdev, enum drbd_packets cmd, unsigned drbd_force_state(mdev, NS(conn, C_DISCONNECTING)); mdev->ldev->dc.disk_size = my_usize; put_ldev(mdev); - return FALSE; + return false; } put_ldev(mdev); } -#undef min_not_zero ddsf = be16_to_cpu(p->dds_flags); if (get_ldev(mdev)) { dd = drbd_determin_dev_size(mdev, ddsf); put_ldev(mdev); if (dd == dev_size_error) - return FALSE; + return false; drbd_md_sync(mdev); } else { /* I am diskless, need to accept the peer's size. */ @@ -2974,14 +2984,14 @@ static int receive_sizes(struct drbd_conf *mdev, enum drbd_packets cmd, unsigned } if (mdev->agreed_pro_version < 94) - max_seg_s = be32_to_cpu(p->max_segment_size); + max_bio_size = be32_to_cpu(p->max_bio_size); else if (mdev->agreed_pro_version == 94) - max_seg_s = DRBD_MAX_SIZE_H80_PACKET; + max_bio_size = DRBD_MAX_SIZE_H80_PACKET; else /* drbd 8.3.8 onwards */ - max_seg_s = DRBD_MAX_SEGMENT_SIZE; + max_bio_size = DRBD_MAX_BIO_SIZE; - if (max_seg_s != queue_max_segment_size(mdev->rq_queue)) - drbd_setup_queue_param(mdev, max_seg_s); + if (max_bio_size != queue_max_hw_sectors(mdev->rq_queue) << 9) + drbd_setup_queue_param(mdev, max_bio_size); drbd_setup_order_type(mdev, be16_to_cpu(p->queue_order_type)); put_ldev(mdev); @@ -3007,14 +3017,14 @@ static int receive_sizes(struct drbd_conf *mdev, enum drbd_packets cmd, unsigned } } - return TRUE; + return true; } static int receive_uuids(struct drbd_conf *mdev, enum drbd_packets cmd, unsigned int data_size) { struct p_uuids *p = &mdev->data.rbuf.uuids; u64 *p_uuid; - int i; + int i, updated_uuids = 0; p_uuid = kmalloc(sizeof(u64)*UI_EXTENDED_SIZE, GFP_NOIO); @@ -3031,7 +3041,7 @@ static int receive_uuids(struct drbd_conf *mdev, enum drbd_packets cmd, unsigned dev_err(DEV, "Can only connect to data with current UUID=%016llX\n", (unsigned long long)mdev->ed_uuid); drbd_force_state(mdev, NS(conn, C_DISCONNECTING)); - return FALSE; + return false; } if (get_ldev(mdev)) { @@ -3049,13 +3059,14 @@ static int receive_uuids(struct drbd_conf *mdev, enum drbd_packets cmd, unsigned _drbd_set_state(_NS2(mdev, disk, D_UP_TO_DATE, pdsk, D_UP_TO_DATE), CS_VERBOSE, NULL); drbd_md_sync(mdev); + updated_uuids = 1; } put_ldev(mdev); } else if (mdev->state.disk < D_INCONSISTENT && mdev->state.role == R_PRIMARY) { /* I am a diskless primary, the peer just created a new current UUID for me. */ - drbd_set_ed_uuid(mdev, p_uuid[UI_CURRENT]); + updated_uuids = drbd_set_ed_uuid(mdev, p_uuid[UI_CURRENT]); } /* Before we test for the disk state, we should wait until an eventually @@ -3064,9 +3075,12 @@ static int receive_uuids(struct drbd_conf *mdev, enum drbd_packets cmd, unsigned new disk state... */ wait_event(mdev->misc_wait, !test_bit(CLUSTER_ST_CHANGE, &mdev->flags)); if (mdev->state.conn >= C_CONNECTED && mdev->state.disk < D_INCONSISTENT) - drbd_set_ed_uuid(mdev, p_uuid[UI_CURRENT]); + updated_uuids |= drbd_set_ed_uuid(mdev, p_uuid[UI_CURRENT]); + + if (updated_uuids) + drbd_print_uuids(mdev, "receiver updated UUIDs to"); - return TRUE; + return true; } /** @@ -3103,7 +3117,7 @@ static int receive_req_state(struct drbd_conf *mdev, enum drbd_packets cmd, unsi { struct p_req_state *p = &mdev->data.rbuf.req_state; union drbd_state mask, val; - int rv; + enum drbd_state_rv rv; mask.i = be32_to_cpu(p->mask); val.i = be32_to_cpu(p->val); @@ -3111,7 +3125,7 @@ static int receive_req_state(struct drbd_conf *mdev, enum drbd_packets cmd, unsi if (test_bit(DISCARD_CONCURRENT, &mdev->flags) && test_bit(CLUSTER_ST_CHANGE, &mdev->flags)) { drbd_send_sr_reply(mdev, SS_CONCURRENT_ST_CHG); - return TRUE; + return true; } mask = convert_state(mask); @@ -3122,7 +3136,7 @@ static int receive_req_state(struct drbd_conf *mdev, enum drbd_packets cmd, unsi drbd_send_sr_reply(mdev, rv); drbd_md_sync(mdev); - return TRUE; + return true; } static int receive_state(struct drbd_conf *mdev, enum drbd_packets cmd, unsigned int data_size) @@ -3167,7 +3181,7 @@ static int receive_state(struct drbd_conf *mdev, enum drbd_packets cmd, unsigned peer_state.conn == C_CONNECTED) { if (drbd_bm_total_weight(mdev) <= mdev->rs_failed) drbd_resync_finished(mdev); - return TRUE; + return true; } } @@ -3183,6 +3197,9 @@ static int receive_state(struct drbd_conf *mdev, enum drbd_packets cmd, unsigned if (ns.conn == C_WF_REPORT_PARAMS) ns.conn = C_CONNECTED; + if (peer_state.conn == C_AHEAD) + ns.conn = C_BEHIND; + if (mdev->p_uuid && peer_state.disk >= D_NEGOTIATING && get_ldev_if_state(mdev, D_NEGOTIATING)) { int cr; /* consider resync */ @@ -3217,10 +3234,10 @@ static int receive_state(struct drbd_conf *mdev, enum drbd_packets cmd, unsigned real_peer_disk = D_DISKLESS; } else { if (test_and_clear_bit(CONN_DRY_RUN, &mdev->flags)) - return FALSE; + return false; D_ASSERT(os.conn == C_WF_REPORT_PARAMS); drbd_force_state(mdev, NS(conn, C_DISCONNECTING)); - return FALSE; + return false; } } } @@ -3245,7 +3262,7 @@ static int receive_state(struct drbd_conf *mdev, enum drbd_packets cmd, unsigned drbd_uuid_new_current(mdev); clear_bit(NEW_CUR_UUID, &mdev->flags); drbd_force_state(mdev, NS2(conn, C_PROTOCOL_ERROR, susp, 0)); - return FALSE; + return false; } rv = _drbd_set_state(mdev, ns, cs_flags, NULL); ns = mdev->state; @@ -3253,7 +3270,7 @@ static int receive_state(struct drbd_conf *mdev, enum drbd_packets cmd, unsigned if (rv < SS_SUCCESS) { drbd_force_state(mdev, NS(conn, C_DISCONNECTING)); - return FALSE; + return false; } if (os.conn > C_WF_REPORT_PARAMS) { @@ -3271,7 +3288,7 @@ static int receive_state(struct drbd_conf *mdev, enum drbd_packets cmd, unsigned drbd_md_sync(mdev); /* update connected indicator, la_size, ... */ - return TRUE; + return true; } static int receive_sync_uuid(struct drbd_conf *mdev, enum drbd_packets cmd, unsigned int data_size) @@ -3280,6 +3297,7 @@ static int receive_sync_uuid(struct drbd_conf *mdev, enum drbd_packets cmd, unsi wait_event(mdev->misc_wait, mdev->state.conn == C_WF_SYNC_UUID || + mdev->state.conn == C_BEHIND || mdev->state.conn < C_CONNECTED || mdev->state.disk < D_NEGOTIATING); @@ -3291,32 +3309,42 @@ static int receive_sync_uuid(struct drbd_conf *mdev, enum drbd_packets cmd, unsi _drbd_uuid_set(mdev, UI_CURRENT, be64_to_cpu(p->uuid)); _drbd_uuid_set(mdev, UI_BITMAP, 0UL); + drbd_print_uuids(mdev, "updated sync uuid"); drbd_start_resync(mdev, C_SYNC_TARGET); put_ldev(mdev); } else dev_err(DEV, "Ignoring SyncUUID packet!\n"); - return TRUE; + return true; } -enum receive_bitmap_ret { OK, DONE, FAILED }; - -static enum receive_bitmap_ret +/** + * receive_bitmap_plain + * + * Return 0 when done, 1 when another iteration is needed, and a negative error + * code upon failure. + */ +static int receive_bitmap_plain(struct drbd_conf *mdev, unsigned int data_size, unsigned long *buffer, struct bm_xfer_ctx *c) { unsigned num_words = min_t(size_t, BM_PACKET_WORDS, c->bm_words - c->word_offset); unsigned want = num_words * sizeof(long); + int err; if (want != data_size) { dev_err(DEV, "%s:want (%u) != data_size (%u)\n", __func__, want, data_size); - return FAILED; + return -EIO; } if (want == 0) - return DONE; - if (drbd_recv(mdev, buffer, want) != want) - return FAILED; + return 0; + err = drbd_recv(mdev, buffer, want); + if (err != want) { + if (err >= 0) + err = -EIO; + return err; + } drbd_bm_merge_lel(mdev, c->word_offset, num_words, buffer); @@ -3325,10 +3353,16 @@ receive_bitmap_plain(struct drbd_conf *mdev, unsigned int data_size, if (c->bit_offset > c->bm_bits) c->bit_offset = c->bm_bits; - return OK; + return 1; } -static enum receive_bitmap_ret +/** + * recv_bm_rle_bits + * + * Return 0 when done, 1 when another iteration is needed, and a negative error + * code upon failure. + */ +static int recv_bm_rle_bits(struct drbd_conf *mdev, struct p_compressed_bm *p, struct bm_xfer_ctx *c) @@ -3348,18 +3382,18 @@ recv_bm_rle_bits(struct drbd_conf *mdev, bits = bitstream_get_bits(&bs, &look_ahead, 64); if (bits < 0) - return FAILED; + return -EIO; for (have = bits; have > 0; s += rl, toggle = !toggle) { bits = vli_decode_bits(&rl, look_ahead); if (bits <= 0) - return FAILED; + return -EIO; if (toggle) { e = s + rl -1; if (e >= c->bm_bits) { dev_err(DEV, "bitmap overflow (e:%lu) while decoding bm RLE packet\n", e); - return FAILED; + return -EIO; } _drbd_bm_set_bits(mdev, s, e); } @@ -3369,14 +3403,14 @@ recv_bm_rle_bits(struct drbd_conf *mdev, have, bits, look_ahead, (unsigned int)(bs.cur.b - p->code), (unsigned int)bs.buf_len); - return FAILED; + return -EIO; } look_ahead >>= bits; have -= bits; bits = bitstream_get_bits(&bs, &tmp, 64 - have); if (bits < 0) - return FAILED; + return -EIO; look_ahead |= tmp << have; have += bits; } @@ -3384,10 +3418,16 @@ recv_bm_rle_bits(struct drbd_conf *mdev, c->bit_offset = s; bm_xfer_ctx_bit_to_word_offset(c); - return (s == c->bm_bits) ? DONE : OK; + return (s != c->bm_bits); } -static enum receive_bitmap_ret +/** + * decode_bitmap_c + * + * Return 0 when done, 1 when another iteration is needed, and a negative error + * code upon failure. + */ +static int decode_bitmap_c(struct drbd_conf *mdev, struct p_compressed_bm *p, struct bm_xfer_ctx *c) @@ -3401,7 +3441,7 @@ decode_bitmap_c(struct drbd_conf *mdev, dev_err(DEV, "receive_bitmap_c: unknown encoding %u\n", p->encoding); drbd_force_state(mdev, NS(conn, C_PROTOCOL_ERROR)); - return FAILED; + return -EIO; } void INFO_bm_xfer_stats(struct drbd_conf *mdev, @@ -3450,13 +3490,11 @@ static int receive_bitmap(struct drbd_conf *mdev, enum drbd_packets cmd, unsigne { struct bm_xfer_ctx c; void *buffer; - enum receive_bitmap_ret ret; - int ok = FALSE; + int err; + int ok = false; struct p_header80 *h = &mdev->data.rbuf.header.h80; - wait_event(mdev->misc_wait, !atomic_read(&mdev->ap_bio_cnt)); - - drbd_bm_lock(mdev, "receive bitmap"); + /* drbd_bm_lock(mdev, "receive bitmap"); By intention no bm_lock */ /* maybe we should use some per thread scratch page, * and allocate that during initial device creation? */ @@ -3471,9 +3509,9 @@ static int receive_bitmap(struct drbd_conf *mdev, enum drbd_packets cmd, unsigne .bm_words = drbd_bm_words(mdev), }; - do { + for(;;) { if (cmd == P_BITMAP) { - ret = receive_bitmap_plain(mdev, data_size, buffer, &c); + err = receive_bitmap_plain(mdev, data_size, buffer, &c); } else if (cmd == P_COMPRESSED_BITMAP) { /* MAYBE: sanity check that we speak proto >= 90, * and the feature is enabled! */ @@ -3490,9 +3528,9 @@ static int receive_bitmap(struct drbd_conf *mdev, enum drbd_packets cmd, unsigne goto out; if (data_size <= (sizeof(*p) - sizeof(p->head))) { dev_err(DEV, "ReportCBitmap packet too small (l:%u)\n", data_size); - return FAILED; + goto out; } - ret = decode_bitmap_c(mdev, p, &c); + err = decode_bitmap_c(mdev, p, &c); } else { dev_warn(DEV, "receive_bitmap: cmd neither ReportBitMap nor ReportCBitMap (is 0x%x)", cmd); goto out; @@ -3501,24 +3539,26 @@ static int receive_bitmap(struct drbd_conf *mdev, enum drbd_packets cmd, unsigne c.packets[cmd == P_BITMAP]++; c.bytes[cmd == P_BITMAP] += sizeof(struct p_header80) + data_size; - if (ret != OK) + if (err <= 0) { + if (err < 0) + goto out; break; - + } if (!drbd_recv_header(mdev, &cmd, &data_size)) goto out; - } while (ret == OK); - if (ret == FAILED) - goto out; + } INFO_bm_xfer_stats(mdev, "receive", &c); if (mdev->state.conn == C_WF_BITMAP_T) { + enum drbd_state_rv rv; + ok = !drbd_send_bitmap(mdev); if (!ok) goto out; /* Omit CS_ORDERED with this state transition to avoid deadlocks. */ - ok = _drbd_request_state(mdev, NS(conn, C_WF_SYNC_UUID), CS_VERBOSE); - D_ASSERT(ok == SS_SUCCESS); + rv = _drbd_request_state(mdev, NS(conn, C_WF_SYNC_UUID), CS_VERBOSE); + D_ASSERT(rv == SS_SUCCESS); } else if (mdev->state.conn != C_WF_BITMAP_S) { /* admin may have requested C_DISCONNECTING, * other threads may have noticed network errors */ @@ -3526,9 +3566,9 @@ static int receive_bitmap(struct drbd_conf *mdev, enum drbd_packets cmd, unsigne drbd_conn_str(mdev->state.conn)); } - ok = TRUE; + ok = true; out: - drbd_bm_unlock(mdev); + /* drbd_bm_unlock(mdev); by intention no lock */ if (ok && mdev->state.conn == C_WF_BITMAP_S) drbd_start_resync(mdev, C_SYNC_SOURCE); free_page((unsigned long) buffer); @@ -3556,14 +3596,30 @@ static int receive_skip(struct drbd_conf *mdev, enum drbd_packets cmd, unsigned static int receive_UnplugRemote(struct drbd_conf *mdev, enum drbd_packets cmd, unsigned int data_size) { - if (mdev->state.disk >= D_INCONSISTENT) - drbd_kick_lo(mdev); - /* Make sure we've acked all the TCP data associated * with the data requests being unplugged */ drbd_tcp_quickack(mdev->data.socket); - return TRUE; + return true; +} + +static int receive_out_of_sync(struct drbd_conf *mdev, enum drbd_packets cmd, unsigned int data_size) +{ + struct p_block_desc *p = &mdev->data.rbuf.block_desc; + + switch (mdev->state.conn) { + case C_WF_SYNC_UUID: + case C_WF_BITMAP_T: + case C_BEHIND: + break; + default: + dev_err(DEV, "ASSERT FAILED cstate = %s, expected: WFSyncUUID|WFBitMapT|Behind\n", + drbd_conn_str(mdev->state.conn)); + } + + drbd_set_out_of_sync(mdev, be64_to_cpu(p->sector), be32_to_cpu(p->blksize)); + + return true; } typedef int (*drbd_cmd_handler_f)(struct drbd_conf *, enum drbd_packets cmd, unsigned int to_receive); @@ -3596,6 +3652,7 @@ static struct data_cmd drbd_cmd_handler[] = { [P_OV_REPLY] = { 1, sizeof(struct p_block_req), receive_DataRequest }, [P_CSUM_RS_REQUEST] = { 1, sizeof(struct p_block_req), receive_DataRequest }, [P_DELAY_PROBE] = { 0, sizeof(struct p_delay_probe93), receive_skip }, + [P_OUT_OF_SYNC] = { 0, sizeof(struct p_block_desc), receive_out_of_sync }, /* anything missing from this table is in * the asender_tbl, see get_asender_cmd */ [P_MAX_CMD] = { 0, 0, NULL }, @@ -3707,9 +3764,6 @@ static void drbd_disconnect(struct drbd_conf *mdev) if (mdev->state.conn == C_STANDALONE) return; - if (mdev->state.conn >= C_WF_CONNECTION) - dev_err(DEV, "ASSERT FAILED cstate = %s, expected < WFConnection\n", - drbd_conn_str(mdev->state.conn)); /* asender does not clean up anything. it must not interfere, either */ drbd_thread_stop(&mdev->asender); @@ -3763,6 +3817,7 @@ static void drbd_disconnect(struct drbd_conf *mdev) fp = FP_DONT_CARE; if (get_ldev(mdev)) { + drbd_bitmap_io(mdev, &drbd_bm_write, "write from disconnect"); fp = mdev->ldev->dc.fencing; put_ldev(mdev); } @@ -3783,13 +3838,6 @@ static void drbd_disconnect(struct drbd_conf *mdev) if (os.conn == C_DISCONNECTING) { wait_event(mdev->net_cnt_wait, atomic_read(&mdev->net_cnt) == 0); - if (!is_susp(mdev->state)) { - /* we must not free the tl_hash - * while application io is still on the fly */ - wait_event(mdev->misc_wait, !atomic_read(&mdev->ap_bio_cnt)); - drbd_free_tl_hash(mdev); - } - crypto_free_hash(mdev->cram_hmac_tfm); mdev->cram_hmac_tfm = NULL; @@ -4099,8 +4147,7 @@ int drbdd_init(struct drbd_thread *thi) h = drbd_connect(mdev); if (h == 0) { drbd_disconnect(mdev); - __set_current_state(TASK_INTERRUPTIBLE); - schedule_timeout(HZ); + schedule_timeout_interruptible(HZ); } if (h == -1) { dev_warn(DEV, "Discarding network configuration.\n"); @@ -4138,7 +4185,7 @@ static int got_RqSReply(struct drbd_conf *mdev, struct p_header80 *h) } wake_up(&mdev->state_wait); - return TRUE; + return true; } static int got_Ping(struct drbd_conf *mdev, struct p_header80 *h) @@ -4154,7 +4201,7 @@ static int got_PingAck(struct drbd_conf *mdev, struct p_header80 *h) if (!test_and_set_bit(GOT_PING_ACK, &mdev->flags)) wake_up(&mdev->misc_wait); - return TRUE; + return true; } static int got_IsInSync(struct drbd_conf *mdev, struct p_header80 *h) @@ -4177,7 +4224,7 @@ static int got_IsInSync(struct drbd_conf *mdev, struct p_header80 *h) dec_rs_pending(mdev); atomic_add(blksize >> 9, &mdev->rs_sect_in); - return TRUE; + return true; } /* when we receive the ACK for a write request, @@ -4201,8 +4248,6 @@ static struct drbd_request *_ack_id_to_req(struct drbd_conf *mdev, return req; } } - dev_err(DEV, "_ack_id_to_req: failed to find req %p, sector %llus in list\n", - (void *)(unsigned long)id, (unsigned long long)sector); return NULL; } @@ -4220,15 +4265,17 @@ static int validate_req_change_req_state(struct drbd_conf *mdev, req = validator(mdev, id, sector); if (unlikely(!req)) { spin_unlock_irq(&mdev->req_lock); - dev_err(DEV, "%s: got a corrupt block_id/sector pair\n", func); - return FALSE; + + dev_err(DEV, "%s: failed to find req %p, sector %llus\n", func, + (void *)(unsigned long)id, (unsigned long long)sector); + return false; } __req_mod(req, what, &m); spin_unlock_irq(&mdev->req_lock); if (m.bio) complete_master_bio(mdev, &m); - return TRUE; + return true; } static int got_BlockAck(struct drbd_conf *mdev, struct p_header80 *h) @@ -4243,7 +4290,7 @@ static int got_BlockAck(struct drbd_conf *mdev, struct p_header80 *h) if (is_syncer_block_id(p->block_id)) { drbd_set_in_sync(mdev, sector, blksize); dec_rs_pending(mdev); - return TRUE; + return true; } switch (be16_to_cpu(h->command)) { case P_RS_WRITE_ACK: @@ -4264,7 +4311,7 @@ static int got_BlockAck(struct drbd_conf *mdev, struct p_header80 *h) break; default: D_ASSERT(0); - return FALSE; + return false; } return validate_req_change_req_state(mdev, p->block_id, sector, @@ -4275,20 +4322,44 @@ static int got_NegAck(struct drbd_conf *mdev, struct p_header80 *h) { struct p_block_ack *p = (struct p_block_ack *)h; sector_t sector = be64_to_cpu(p->sector); - - if (__ratelimit(&drbd_ratelimit_state)) - dev_warn(DEV, "Got NegAck packet. Peer is in troubles?\n"); + int size = be32_to_cpu(p->blksize); + struct drbd_request *req; + struct bio_and_error m; update_peer_seq(mdev, be32_to_cpu(p->seq_num)); if (is_syncer_block_id(p->block_id)) { - int size = be32_to_cpu(p->blksize); dec_rs_pending(mdev); drbd_rs_failed_io(mdev, sector, size); - return TRUE; + return true; } - return validate_req_change_req_state(mdev, p->block_id, sector, - _ack_id_to_req, __func__ , neg_acked); + + spin_lock_irq(&mdev->req_lock); + req = _ack_id_to_req(mdev, p->block_id, sector); + if (!req) { + spin_unlock_irq(&mdev->req_lock); + if (mdev->net_conf->wire_protocol == DRBD_PROT_A || + mdev->net_conf->wire_protocol == DRBD_PROT_B) { + /* Protocol A has no P_WRITE_ACKs, but has P_NEG_ACKs. + The master bio might already be completed, therefore the + request is no longer in the collision hash. + => Do not try to validate block_id as request. */ + /* In Protocol B we might already have got a P_RECV_ACK + but then get a P_NEG_ACK after wards. */ + drbd_set_out_of_sync(mdev, sector, size); + return true; + } else { + dev_err(DEV, "%s: failed to find req %p, sector %llus\n", __func__, + (void *)(unsigned long)p->block_id, (unsigned long long)sector); + return false; + } + } + __req_mod(req, neg_acked, &m); + spin_unlock_irq(&mdev->req_lock); + + if (m.bio) + complete_master_bio(mdev, &m); + return true; } static int got_NegDReply(struct drbd_conf *mdev, struct p_header80 *h) @@ -4319,11 +4390,20 @@ static int got_NegRSDReply(struct drbd_conf *mdev, struct p_header80 *h) if (get_ldev_if_state(mdev, D_FAILED)) { drbd_rs_complete_io(mdev, sector); - drbd_rs_failed_io(mdev, sector, size); + switch (be16_to_cpu(h->command)) { + case P_NEG_RS_DREPLY: + drbd_rs_failed_io(mdev, sector, size); + case P_RS_CANCEL: + break; + default: + D_ASSERT(0); + put_ldev(mdev); + return false; + } put_ldev(mdev); } - return TRUE; + return true; } static int got_BarrierAck(struct drbd_conf *mdev, struct p_header80 *h) @@ -4332,7 +4412,14 @@ static int got_BarrierAck(struct drbd_conf *mdev, struct p_header80 *h) tl_release(mdev, p->barrier, be32_to_cpu(p->set_size)); - return TRUE; + if (mdev->state.conn == C_AHEAD && + atomic_read(&mdev->ap_in_flight) == 0 && + !test_and_set_bit(AHEAD_TO_SYNC_SOURCE, &mdev->current_epoch->flags)) { + mdev->start_resync_timer.expires = jiffies + HZ; + add_timer(&mdev->start_resync_timer); + } + + return true; } static int got_OVResult(struct drbd_conf *mdev, struct p_header80 *h) @@ -4353,12 +4440,18 @@ static int got_OVResult(struct drbd_conf *mdev, struct p_header80 *h) ov_oos_print(mdev); if (!get_ldev(mdev)) - return TRUE; + return true; drbd_rs_complete_io(mdev, sector); dec_rs_pending(mdev); - if (--mdev->ov_left == 0) { + --mdev->ov_left; + + /* let's advance progress step marks only for every other megabyte */ + if ((mdev->ov_left & 0x200) == 0x200) + drbd_advance_rs_marks(mdev, mdev->ov_left); + + if (mdev->ov_left == 0) { w = kmalloc(sizeof(*w), GFP_NOIO); if (w) { w->cb = w_ov_finished; @@ -4370,12 +4463,12 @@ static int got_OVResult(struct drbd_conf *mdev, struct p_header80 *h) } } put_ldev(mdev); - return TRUE; + return true; } static int got_skip(struct drbd_conf *mdev, struct p_header80 *h) { - return TRUE; + return true; } struct asender_cmd { @@ -4403,6 +4496,7 @@ static struct asender_cmd *get_asender_cmd(int cmd) [P_STATE_CHG_REPLY] = { sizeof(struct p_req_state_reply), got_RqSReply }, [P_RS_IS_IN_SYNC] = { sizeof(struct p_block_ack), got_IsInSync }, [P_DELAY_PROBE] = { sizeof(struct p_delay_probe93), got_skip }, + [P_RS_CANCEL] = { sizeof(struct p_block_ack), got_NegRSDReply}, [P_MAX_CMD] = { 0, NULL }, }; if (cmd > P_MAX_CMD || asender_tbl[cmd].process == NULL)