From: Philipp Reisner Date: Mon, 31 May 2010 08:14:17 +0000 (+0200) Subject: drbd: Finished the "on-no-data-accessible suspend-io;" functionality X-Git-Url: https://git.karo-electronics.de/?a=commitdiff_plain;h=265be2d09853d425ad14a61cda0ca63345613d0c;p=linux-beck.git drbd: Finished the "on-no-data-accessible suspend-io;" functionality When no data is accessible (no connection to the peer, nor a local disk) allow the user to select to freeze all IO operations instead of getting IO errors. Signed-off-by: Philipp Reisner Signed-off-by: Lars Ellenberg --- diff --git a/drivers/block/drbd/drbd_int.h b/drivers/block/drbd/drbd_int.h index bef9138f1975..03cc975b9e6c 100644 --- a/drivers/block/drbd/drbd_int.h +++ b/drivers/block/drbd/drbd_int.h @@ -1469,6 +1469,7 @@ extern int w_send_barrier(struct drbd_conf *, struct drbd_work *, int); extern int w_send_read_req(struct drbd_conf *, struct drbd_work *, int); extern int w_prev_work_done(struct drbd_conf *, struct drbd_work *, int); extern int w_e_reissue(struct drbd_conf *, struct drbd_work *, int); +extern int w_restart_disk_io(struct drbd_conf *, struct drbd_work *, int); extern void resync_timer_fn(unsigned long data); diff --git a/drivers/block/drbd/drbd_main.c b/drivers/block/drbd/drbd_main.c index 7d359863ae32..106b9abdc430 100644 --- a/drivers/block/drbd/drbd_main.c +++ b/drivers/block/drbd/drbd_main.c @@ -925,7 +925,12 @@ static union drbd_state sanitize_state(struct drbd_conf *mdev, union drbd_state if (fp == FP_STONITH && (ns.role == R_PRIMARY && ns.conn < C_CONNECTED && ns.pdsk > D_OUTDATED) && !(os.role == R_PRIMARY && os.conn < C_CONNECTED && os.pdsk > D_OUTDATED)) - ns.susp = 1; + ns.susp = 1; /* Suspend IO while fence-peer handler runs (peer lost) */ + + if (mdev->sync_conf.on_no_data == OND_SUSPEND_IO && + (ns.role == R_PRIMARY && ns.disk < D_UP_TO_DATE && ns.pdsk < D_UP_TO_DATE) && + !(os.role == R_PRIMARY && os.disk < D_UP_TO_DATE && os.pdsk < D_UP_TO_DATE)) + ns.susp = 1; /* Suspend IO while no data available (no accessible data available) */ if (ns.aftr_isp || ns.peer_isp || ns.user_isp) { if (ns.conn == C_SYNC_SOURCE) @@ -1236,6 +1241,25 @@ static void after_state_ch(struct drbd_conf *mdev, union drbd_state os, /* Here we have the actions that are performed after a state change. This function might sleep */ + if (os.susp && ns.susp && mdev->sync_conf.on_no_data == OND_SUSPEND_IO) { + if (os.conn < C_CONNECTED && ns.conn >= C_CONNECTED) { + if (ns.conn == C_CONNECTED) { + spin_lock_irq(&mdev->req_lock); + _tl_restart(mdev, resend); + _drbd_set_state(_NS(mdev, susp, 0), CS_VERBOSE, NULL); + spin_unlock_irq(&mdev->req_lock); + } else /* ns.conn > C_CONNECTED */ + dev_err(DEV, "Unexpected Resynd going on!\n"); + } + + if (os.disk == D_ATTACHING && ns.disk > D_ATTACHING) { + spin_lock_irq(&mdev->req_lock); + _tl_restart(mdev, restart_frozen_disk_io); + _drbd_set_state(_NS(mdev, susp, 0), CS_VERBOSE, NULL); + spin_unlock_irq(&mdev->req_lock); + } + } + if (fp == FP_STONITH && ns.susp) { /* case1: The outdate peer handler is successful: * case2: The connection was established again: */ diff --git a/drivers/block/drbd/drbd_nl.c b/drivers/block/drbd/drbd_nl.c index 73131c5ae339..563a6ade0179 100644 --- a/drivers/block/drbd/drbd_nl.c +++ b/drivers/block/drbd/drbd_nl.c @@ -33,6 +33,7 @@ #include #include #include "drbd_int.h" +#include "drbd_req.h" #include "drbd_wrappers.h" #include #include @@ -494,6 +495,8 @@ char *ppsize(char *buf, unsigned long long size) void drbd_suspend_io(struct drbd_conf *mdev) { set_bit(SUSPEND_IO, &mdev->flags); + if (mdev->state.susp) + return; wait_event(mdev->misc_wait, !atomic_read(&mdev->ap_bio_cnt)); } @@ -1557,6 +1560,7 @@ static int drbd_nl_syncer_conf(struct drbd_conf *mdev, struct drbd_nl_cfg_req *n sc.rate = DRBD_RATE_DEF; sc.after = DRBD_AFTER_DEF; sc.al_extents = DRBD_AL_EXTENTS_DEF; + sc.on_no_data = DRBD_ON_NO_DATA_DEF; } else memcpy(&sc, &mdev->sync_conf, sizeof(struct syncer_conf)); @@ -1765,7 +1769,16 @@ static int drbd_nl_suspend_io(struct drbd_conf *mdev, struct drbd_nl_cfg_req *nl static int drbd_nl_resume_io(struct drbd_conf *mdev, struct drbd_nl_cfg_req *nlp, struct drbd_nl_cfg_reply *reply) { + drbd_suspend_io(mdev); reply->ret_code = drbd_request_state(mdev, NS(susp, 0)); + if (reply->ret_code == SS_SUCCESS) { + if (mdev->state.conn < C_CONNECTED) + tl_clear(mdev); + if (mdev->state.disk == D_DISKLESS || mdev->state.disk == D_FAILED) + tl_restart(mdev, fail_frozen_disk_io); + } + drbd_resume_io(mdev); + return 0; } diff --git a/drivers/block/drbd/drbd_req.c b/drivers/block/drbd/drbd_req.c index 48647589aa0d..8259d4f77285 100644 --- a/drivers/block/drbd/drbd_req.c +++ b/drivers/block/drbd/drbd_req.c @@ -226,6 +226,8 @@ void _req_may_be_done(struct drbd_request *req, struct bio_and_error *m) return; if (s & RQ_LOCAL_PENDING) return; + if (mdev->state.susp) + return; if (req->master_bio) { /* this is data_received (remote read) @@ -634,6 +636,28 @@ int __req_mod(struct drbd_request *req, enum drbd_req_event what, /* else: done by handed_over_to_network */ break; + case fail_frozen_disk_io: + if (!(req->rq_state & RQ_LOCAL_COMPLETED)) + break; + + _req_may_be_done(req, m); + break; + + case restart_frozen_disk_io: + if (!(req->rq_state & RQ_LOCAL_COMPLETED)) + break; + + req->rq_state &= ~RQ_LOCAL_COMPLETED; + + rv = MR_READ; + if (bio_data_dir(req->master_bio) == WRITE) + rv = MR_WRITE; + + get_ldev(mdev); + req->w.cb = w_restart_disk_io; + drbd_queue_work(&mdev->data.work, &req->w); + break; + case resend: /* If RQ_NET_OK is already set, we got a P_WRITE_ACK or P_RECV_ACK before the connection loss; only P_BARRIER_ACK was missing. diff --git a/drivers/block/drbd/drbd_req.h b/drivers/block/drbd/drbd_req.h index 07cb3b12edb4..f2e45aaa2cd5 100644 --- a/drivers/block/drbd/drbd_req.h +++ b/drivers/block/drbd/drbd_req.h @@ -105,6 +105,8 @@ enum drbd_req_event { write_completed_with_error, completed_ok, resend, + fail_frozen_disk_io, + restart_frozen_disk_io, nothing, /* for tracing only */ }; diff --git a/drivers/block/drbd/drbd_worker.c b/drivers/block/drbd/drbd_worker.c index ca4a16cea2d8..3c1e88480d37 100644 --- a/drivers/block/drbd/drbd_worker.c +++ b/drivers/block/drbd/drbd_worker.c @@ -1173,6 +1173,24 @@ int w_send_read_req(struct drbd_conf *mdev, struct drbd_work *w, int cancel) return ok; } +int w_restart_disk_io(struct drbd_conf *mdev, struct drbd_work *w, int cancel) +{ + struct drbd_request *req = container_of(w, struct drbd_request, w); + + if (bio_data_dir(req->master_bio) == WRITE) + drbd_al_begin_io(mdev, req->sector); + /* Calling drbd_al_begin_io() out of the worker might deadlocks + theoretically. Practically it can not deadlock, since this is + only used when unfreezing IOs. All the extents of the requests + that made it into the TL are already active */ + + drbd_req_make_private_bio(req, req->master_bio); + req->private_bio->bi_bdev = mdev->ldev->backing_bdev; + generic_make_request(req->private_bio); + + return 1; +} + static int _drbd_may_sync_now(struct drbd_conf *mdev) { struct drbd_conf *odev = mdev; diff --git a/include/linux/drbd.h b/include/linux/drbd.h index 479ee3a1d901..7be069fcca57 100644 --- a/include/linux/drbd.h +++ b/include/linux/drbd.h @@ -91,6 +91,11 @@ enum drbd_after_sb_p { ASB_VIOLENTLY }; +enum drbd_on_no_data { + OND_IO_ERROR, + OND_SUSPEND_IO +}; + /* KEEP the order, do not delete or insert. Only append. */ enum drbd_ret_codes { ERR_CODE_BASE = 100, diff --git a/include/linux/drbd_limits.h b/include/linux/drbd_limits.h index 440b42e38e89..7eb1e98009ec 100644 --- a/include/linux/drbd_limits.h +++ b/include/linux/drbd_limits.h @@ -128,6 +128,7 @@ #define DRBD_AFTER_SB_1P_DEF ASB_DISCONNECT #define DRBD_AFTER_SB_2P_DEF ASB_DISCONNECT #define DRBD_RR_CONFLICT_DEF ASB_DISCONNECT +#define DRBD_ON_NO_DATA_DEF OND_IO_ERROR #define DRBD_MAX_BIO_BVECS_MIN 0 #define DRBD_MAX_BIO_BVECS_MAX 128 diff --git a/include/linux/drbd_nl.h b/include/linux/drbd_nl.h index 5f042810a56c..9aebd0d80a5d 100644 --- a/include/linux/drbd_nl.h +++ b/include/linux/drbd_nl.h @@ -87,6 +87,7 @@ NL_PACKET(syncer_conf, 8, NL_STRING( 51, T_MAY_IGNORE, cpu_mask, 32) NL_STRING( 64, T_MAY_IGNORE, csums_alg, SHARED_SECRET_MAX) NL_BIT( 65, T_MAY_IGNORE, use_rle) + NL_INTEGER( 75, T_MAY_IGNORE, on_no_data) ) NL_PACKET(invalidate, 9, )