From 7eaceaccab5f40bbfda044629a6298616aeaed50 Mon Sep 17 00:00:00 2001 From: Jens Axboe Date: Thu, 10 Mar 2011 08:52:07 +0100 Subject: block: remove per-queue plugging Code has been converted over to the new explicit on-stack plugging, and delay users have been converted to use the new API for that. So lets kill off the old plugging along with aops->sync_page(). Signed-off-by: Jens Axboe --- drivers/block/drbd/drbd_worker.c | 1 - 1 file changed, 1 deletion(-) (limited to 'drivers/block/drbd/drbd_worker.c') diff --git a/drivers/block/drbd/drbd_worker.c b/drivers/block/drbd/drbd_worker.c index 34f224b018b..e027446590d 100644 --- a/drivers/block/drbd/drbd_worker.c +++ b/drivers/block/drbd/drbd_worker.c @@ -792,7 +792,6 @@ int drbd_resync_finished(struct drbd_conf *mdev) * queue (or even the read operations for those packets * is not finished by now). Retry in 100ms. */ - drbd_kick_lo(mdev); __set_current_state(TASK_INTERRUPTIBLE); schedule_timeout(HZ / 10); w = kmalloc(sizeof(struct drbd_work), GFP_ATOMIC); -- cgit v1.2.3-70-g09d2 From 26525618863afcc4aab8b2a83451d37c6f513460 Mon Sep 17 00:00:00 2001 From: Lars Ellenberg Date: Fri, 5 Nov 2010 09:56:33 +0100 Subject: drbd: only reset online-verify start sector if verify completed For network hickups during online-verify, on the next verify triggered, we by default want to resume where it left off. After any replication link interruption, there will be a (possibly empty) resync. Do not reset online-verify start sector if some resync completed, that would defeats the purpose. Only reset the start sector once a verify run is completed. Signed-off-by: Philipp Reisner Signed-off-by: Lars Ellenberg --- drivers/block/drbd/drbd_worker.c | 9 ++++++--- 1 file changed, 6 insertions(+), 3 deletions(-) (limited to 'drivers/block/drbd/drbd_worker.c') diff --git a/drivers/block/drbd/drbd_worker.c b/drivers/block/drbd/drbd_worker.c index e027446590d..6cfc5417da8 100644 --- a/drivers/block/drbd/drbd_worker.c +++ b/drivers/block/drbd/drbd_worker.c @@ -782,6 +782,7 @@ int drbd_resync_finished(struct drbd_conf *mdev) union drbd_state os, ns; struct drbd_work *w; char *khelper_cmd = NULL; + int verify_done = 0; /* Remove all elements from the resync LRU. Since future actions * might set bits in the (main) bitmap, then the entries in the @@ -818,6 +819,8 @@ int drbd_resync_finished(struct drbd_conf *mdev) spin_lock_irq(&mdev->req_lock); os = mdev->state; + verify_done = (os.conn == C_VERIFY_S || os.conn == C_VERIFY_T); + /* This protects us against multiple calls (that can happen in the presence of application IO), and against connectivity loss just before we arrive here. */ if (os.conn <= C_CONNECTED) @@ -827,8 +830,7 @@ int drbd_resync_finished(struct drbd_conf *mdev) ns.conn = C_CONNECTED; dev_info(DEV, "%s done (total %lu sec; paused %lu sec; %lu K/sec)\n", - (os.conn == C_VERIFY_S || os.conn == C_VERIFY_T) ? - "Online verify " : "Resync", + verify_done ? "Online verify " : "Resync", dt + mdev->rs_paused, mdev->rs_paused, dbdt); n_oos = drbd_bm_total_weight(mdev); @@ -905,7 +907,8 @@ out: mdev->rs_total = 0; mdev->rs_failed = 0; mdev->rs_paused = 0; - mdev->ov_start_sector = 0; + if (verify_done) + mdev->ov_start_sector = 0; drbd_md_sync(mdev); -- cgit v1.2.3-70-g09d2 From ea5442aff68c559c951373739201721185191748 Mon Sep 17 00:00:00 2001 From: Lars Ellenberg Date: Fri, 5 Nov 2010 09:48:01 +0100 Subject: drbd: advance progress step marks for online-verify Signed-off-by: Philipp Reisner Signed-off-by: Lars Ellenberg --- drivers/block/drbd/drbd_int.h | 1 + drivers/block/drbd/drbd_receiver.c | 8 +++++++- drivers/block/drbd/drbd_worker.c | 8 +++++++- 3 files changed, 15 insertions(+), 2 deletions(-) (limited to 'drivers/block/drbd/drbd_worker.c') diff --git a/drivers/block/drbd/drbd_int.h b/drivers/block/drbd/drbd_int.h index b0bd27dfc1e..6cba131011d 100644 --- a/drivers/block/drbd/drbd_int.h +++ b/drivers/block/drbd/drbd_int.h @@ -1619,6 +1619,7 @@ extern int drbd_rs_del_all(struct drbd_conf *mdev); extern void drbd_rs_failed_io(struct drbd_conf *mdev, sector_t sector, int size); extern int drbd_al_read_log(struct drbd_conf *mdev, struct drbd_backing_dev *); +extern void drbd_advance_rs_marks(struct drbd_conf *mdev, unsigned long still_to_go); extern void __drbd_set_in_sync(struct drbd_conf *mdev, sector_t sector, int size, const char *file, const unsigned int line); #define drbd_set_in_sync(mdev, sector, size) \ diff --git a/drivers/block/drbd/drbd_receiver.c b/drivers/block/drbd/drbd_receiver.c index 3b95eef0754..10db70a2337 100644 --- a/drivers/block/drbd/drbd_receiver.c +++ b/drivers/block/drbd/drbd_receiver.c @@ -4340,7 +4340,13 @@ static int got_OVResult(struct drbd_conf *mdev, struct p_header80 *h) 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; diff --git a/drivers/block/drbd/drbd_worker.c b/drivers/block/drbd/drbd_worker.c index 6cfc5417da8..a050eb0d93b 100644 --- a/drivers/block/drbd/drbd_worker.c +++ b/drivers/block/drbd/drbd_worker.c @@ -1185,7 +1185,13 @@ int w_e_end_ov_reply(struct drbd_conf *mdev, struct drbd_work *w, int cancel) drbd_free_ee(mdev, e); - 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) { ov_oos_print(mdev); drbd_resync_finished(mdev); } -- cgit v1.2.3-70-g09d2 From 9bd28d3c90c80c7ec46085de281b38f67331da41 Mon Sep 17 00:00:00 2001 From: Lars Ellenberg Date: Fri, 5 Nov 2010 09:55:18 +0100 Subject: drbd: factor out drbd_rs_controller_reset Preparation patch to be able to use the auto-throttling resync controller for online-verify requests as well. Signed-off-by: Philipp Reisner Signed-off-by: Lars Ellenberg --- drivers/block/drbd/drbd_int.h | 1 + drivers/block/drbd/drbd_worker.c | 19 ++++++++++++------- 2 files changed, 13 insertions(+), 7 deletions(-) (limited to 'drivers/block/drbd/drbd_worker.c') diff --git a/drivers/block/drbd/drbd_int.h b/drivers/block/drbd/drbd_int.h index 7131bf2af6a..85207b275e4 100644 --- a/drivers/block/drbd/drbd_int.h +++ b/drivers/block/drbd/drbd_int.h @@ -1499,6 +1499,7 @@ extern int drbd_resync_finished(struct drbd_conf *mdev); extern int drbd_md_sync_page_io(struct drbd_conf *mdev, struct drbd_backing_dev *bdev, sector_t sector, int rw); extern void drbd_ov_oos_found(struct drbd_conf*, sector_t, int); +extern void drbd_rs_controller_reset(struct drbd_conf *mdev); static inline void ov_oos_print(struct drbd_conf *mdev) { diff --git a/drivers/block/drbd/drbd_worker.c b/drivers/block/drbd/drbd_worker.c index a050eb0d93b..971e9b3b13b 100644 --- a/drivers/block/drbd/drbd_worker.c +++ b/drivers/block/drbd/drbd_worker.c @@ -1439,6 +1439,17 @@ int drbd_alter_sa(struct drbd_conf *mdev, int na) return retcode; } +void drbd_rs_controller_reset(struct drbd_conf *mdev) +{ + atomic_set(&mdev->rs_sect_in, 0); + atomic_set(&mdev->rs_sect_ev, 0); + mdev->rs_in_flight = 0; + mdev->rs_planed = 0; + spin_lock(&mdev->peer_seq_lock); + fifo_set(&mdev->rs_plan_s, 0); + spin_unlock(&mdev->peer_seq_lock); +} + /** * drbd_start_resync() - Start the resync process * @mdev: DRBD device. @@ -1556,13 +1567,7 @@ void drbd_start_resync(struct drbd_conf *mdev, enum drbd_conns side) drbd_resync_finished(mdev); } - atomic_set(&mdev->rs_sect_in, 0); - atomic_set(&mdev->rs_sect_ev, 0); - mdev->rs_in_flight = 0; - mdev->rs_planed = 0; - spin_lock(&mdev->peer_seq_lock); - fifo_set(&mdev->rs_plan_s, 0); - spin_unlock(&mdev->peer_seq_lock); + drbd_rs_controller_reset(mdev); /* ns.conn may already be != mdev->state.conn, * we may have been paused in between, or become paused until * the timer triggers. -- cgit v1.2.3-70-g09d2 From e65f440d474d7d6a6fd8a2c844e851d8c96ed9c5 Mon Sep 17 00:00:00 2001 From: Lars Ellenberg Date: Fri, 5 Nov 2010 10:04:07 +0100 Subject: drbd: factor out drbd_rs_number_requests Preparation patch to be able to use the auto-throttling resync controller for online-verify requests as well. Signed-off-by: Philipp Reisner Signed-off-by: Lars Ellenberg --- drivers/block/drbd/drbd_worker.c | 61 +++++++++++++++++++--------------------- 1 file changed, 29 insertions(+), 32 deletions(-) (limited to 'drivers/block/drbd/drbd_worker.c') diff --git a/drivers/block/drbd/drbd_worker.c b/drivers/block/drbd/drbd_worker.c index 971e9b3b13b..6d111c8515f 100644 --- a/drivers/block/drbd/drbd_worker.c +++ b/drivers/block/drbd/drbd_worker.c @@ -492,6 +492,32 @@ int drbd_rs_controller(struct drbd_conf *mdev) return req_sect; } +int drbd_rs_number_requests(struct drbd_conf *mdev) +{ + int number; + if (mdev->rs_plan_s.size) { /* mdev->sync_conf.c_plan_ahead */ + number = drbd_rs_controller(mdev) >> (BM_BLOCK_SHIFT - 9); + mdev->c_sync_rate = number * HZ * (BM_BLOCK_SIZE / 1024) / SLEEP_TIME; + } else { + mdev->c_sync_rate = mdev->sync_conf.rate; + number = SLEEP_TIME * mdev->c_sync_rate / ((BM_BLOCK_SIZE / 1024) * HZ); + } + + /* Throttle resync on lower level disk activity, which may also be + * caused by application IO on Primary/SyncTarget. + * Keep this after the call to drbd_rs_controller, as that assumes + * to be called as precisely as possible every SLEEP_TIME, + * and would be confused otherwise. */ + if (number && drbd_rs_should_slow_down(mdev)) { + mdev->c_sync_rate = 1; + number = 0; + } + + /* ignore the amount of pending requests, the resync controller should + * throttle down to incoming reply rate soon enough anyways. */ + return number; +} + int w_make_resync_request(struct drbd_conf *mdev, struct drbd_work *w, int cancel) { @@ -499,7 +525,7 @@ int w_make_resync_request(struct drbd_conf *mdev, sector_t sector; const sector_t capacity = drbd_get_capacity(mdev->this_bdev); int max_segment_size; - int number, rollback_i, size, pe, mx; + int number, rollback_i, size; int align, queued, sndbuf; int i = 0; @@ -537,39 +563,10 @@ int w_make_resync_request(struct drbd_conf *mdev, mdev->agreed_pro_version < 94 ? queue_max_segment_size(mdev->rq_queue) : mdev->agreed_pro_version < 95 ? DRBD_MAX_SIZE_H80_PACKET : DRBD_MAX_SEGMENT_SIZE; - if (mdev->rs_plan_s.size) { /* mdev->sync_conf.c_plan_ahead */ - number = drbd_rs_controller(mdev) >> (BM_BLOCK_SHIFT - 9); - mdev->c_sync_rate = number * HZ * (BM_BLOCK_SIZE / 1024) / SLEEP_TIME; - } else { - mdev->c_sync_rate = mdev->sync_conf.rate; - number = SLEEP_TIME * mdev->c_sync_rate / ((BM_BLOCK_SIZE / 1024) * HZ); - } - - /* Throttle resync on lower level disk activity, which may also be - * caused by application IO on Primary/SyncTarget. - * Keep this after the call to drbd_rs_controller, as that assumes - * to be called as precisely as possible every SLEEP_TIME, - * and would be confused otherwise. */ - if (drbd_rs_should_slow_down(mdev)) + number = drbd_rs_number_requests(mdev); + if (number == 0) goto requeue; - mutex_lock(&mdev->data.mutex); - if (mdev->data.socket) - mx = mdev->data.socket->sk->sk_rcvbuf / sizeof(struct p_block_req); - else - mx = 1; - mutex_unlock(&mdev->data.mutex); - - /* For resync rates >160MB/sec, allow more pending RS requests */ - if (number > mx) - mx = number; - - /* Limit the number of pending RS requests to no more than the peer's receive buffer */ - pe = atomic_read(&mdev->rs_pending_cnt); - if ((pe + number) > mx) { - number = mx - pe; - } - for (i = 0; i < number; i++) { /* Stop generating RS requests, when half of the send buffer is filled */ mutex_lock(&mdev->data.mutex); -- cgit v1.2.3-70-g09d2 From 2649f0809f55e4df98c333a2b85c6fc8fee04804 Mon Sep 17 00:00:00 2001 From: Lars Ellenberg Date: Fri, 5 Nov 2010 10:05:47 +0100 Subject: drbd: use the resync controller for online-verify requests as well Signed-off-by: Philipp Reisner Signed-off-by: Lars Ellenberg --- drivers/block/drbd/drbd_main.c | 2 ++ drivers/block/drbd/drbd_proc.c | 3 ++- drivers/block/drbd/drbd_receiver.c | 10 ++++++++-- drivers/block/drbd/drbd_worker.c | 8 ++------ 4 files changed, 14 insertions(+), 9 deletions(-) (limited to 'drivers/block/drbd/drbd_worker.c') diff --git a/drivers/block/drbd/drbd_main.c b/drivers/block/drbd/drbd_main.c index a1a2cb1eadf..6afb81f807b 100644 --- a/drivers/block/drbd/drbd_main.c +++ b/drivers/block/drbd/drbd_main.c @@ -1126,6 +1126,8 @@ int __drbd_set_state(struct drbd_conf *mdev, mdev->rs_mark_time[i] = now; } + drbd_rs_controller_reset(mdev); + if (ns.conn == C_VERIFY_S) { dev_info(DEV, "Starting Online Verify from sector %llu\n", (unsigned long long)mdev->ov_position); diff --git a/drivers/block/drbd/drbd_proc.c b/drivers/block/drbd/drbd_proc.c index 0ec6f4b6a4a..fab3fde3477 100644 --- a/drivers/block/drbd/drbd_proc.c +++ b/drivers/block/drbd/drbd_proc.c @@ -154,7 +154,8 @@ static void drbd_syncer_progress(struct drbd_conf *mdev, struct seq_file *seq) seq_printf_with_thousands_grouping(seq, dbdt); seq_printf(seq, ")"); - if (mdev->state.conn == C_SYNC_TARGET) { + if (mdev->state.conn == C_SYNC_TARGET || + mdev->state.conn == C_VERIFY_S) { if (mdev->c_sync_rate > 1000) seq_printf(seq, " want: %d,%03d", mdev->c_sync_rate / 1000, mdev->c_sync_rate % 1000); diff --git a/drivers/block/drbd/drbd_receiver.c b/drivers/block/drbd/drbd_receiver.c index 10db70a2337..1c56af03c38 100644 --- a/drivers/block/drbd/drbd_receiver.c +++ b/drivers/block/drbd/drbd_receiver.c @@ -1883,8 +1883,12 @@ 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) @@ -1992,6 +1996,8 @@ static int receive_DataRequest(struct drbd_conf *mdev, enum drbd_packets cmd, un D_ASSERT(mdev->agreed_pro_version >= 89); e->w.cb = w_e_end_csum_rs_req; } 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, diff --git a/drivers/block/drbd/drbd_worker.c b/drivers/block/drbd/drbd_worker.c index 6d111c8515f..af805efc94d 100644 --- a/drivers/block/drbd/drbd_worker.c +++ b/drivers/block/drbd/drbd_worker.c @@ -708,11 +708,7 @@ static int w_make_ov_request(struct drbd_conf *mdev, struct drbd_work *w, int ca return 0; } - number = SLEEP_TIME*mdev->sync_conf.rate / ((BM_BLOCK_SIZE/1024)*HZ); - if (atomic_read(&mdev->rs_pending_cnt) > number) - goto requeue; - - number -= atomic_read(&mdev->rs_pending_cnt); + number = drbd_rs_number_requests(mdev); sector = mdev->ov_position; for (i = 0; i < number; i++) { @@ -741,11 +737,11 @@ static int w_make_ov_request(struct drbd_conf *mdev, struct drbd_work *w, int ca mdev->ov_position = sector; requeue: + mdev->rs_in_flight += (i << (BM_BLOCK_SHIFT - 9)); mod_timer(&mdev->resync_timer, jiffies + SLEEP_TIME); return 1; } - int w_ov_finished(struct drbd_conf *mdev, struct drbd_work *w, int cancel) { kfree(w); -- cgit v1.2.3-70-g09d2 From 1816a2b47afae838e53a177d5d166cc7be97d6b5 Mon Sep 17 00:00:00 2001 From: Lars Ellenberg Date: Thu, 11 Nov 2010 15:19:07 +0100 Subject: drbd: properly use max_hw_sectors to limit the our bio size To ease tracking of bios in some hash tables, we want it to not cross certain boundaries (128k, used to be 32k). We limit the maximum bio size using queue parameters. Historically some defines and variables we use there have been named max_segment_size, which was misguided. Rename them to max_bio_size, and use [blk_]queue_max_hw_sectors where appropriate. Signed-off-by: Philipp Reisner Signed-off-by: Lars Ellenberg --- drivers/block/drbd/drbd_actlog.c | 8 ++++---- drivers/block/drbd/drbd_int.h | 4 ++-- drivers/block/drbd/drbd_main.c | 6 +++--- drivers/block/drbd/drbd_nl.c | 27 +++++++++++++-------------- drivers/block/drbd/drbd_receiver.c | 18 +++++++++--------- drivers/block/drbd/drbd_req.c | 8 ++++---- drivers/block/drbd/drbd_worker.c | 12 ++++++------ 7 files changed, 41 insertions(+), 42 deletions(-) (limited to 'drivers/block/drbd/drbd_worker.c') diff --git a/drivers/block/drbd/drbd_actlog.c b/drivers/block/drbd/drbd_actlog.c index b3f18545b46..b4adb58c747 100644 --- a/drivers/block/drbd/drbd_actlog.c +++ b/drivers/block/drbd/drbd_actlog.c @@ -952,7 +952,7 @@ void __drbd_set_in_sync(struct drbd_conf *mdev, sector_t sector, int size, int wake_up = 0; unsigned long flags; - 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, "drbd_set_in_sync: sector=%llus size=%d nonsense!\n", (unsigned long long)sector, size); return; @@ -1002,7 +1002,7 @@ void __drbd_set_in_sync(struct drbd_conf *mdev, sector_t sector, int size, /* * this is intended to set one request worth of data out of sync. * affects at least 1 bit, - * and at most 1+DRBD_MAX_SEGMENT_SIZE/BM_BLOCK_SIZE bits. + * and at most 1+DRBD_MAX_BIO_SIZE/BM_BLOCK_SIZE bits. * * called by tl_clear and drbd_send_dblock (==drbd_make_request). * so this can be _any_ process. @@ -1015,7 +1015,7 @@ void __drbd_set_out_of_sync(struct drbd_conf *mdev, sector_t sector, int size, unsigned int enr, count; struct lc_element *e; - 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, "sector: %llus, size: %d\n", (unsigned long long)sector, size); return; @@ -1387,7 +1387,7 @@ void drbd_rs_failed_io(struct drbd_conf *mdev, sector_t sector, int size) sector_t esector, nr_sectors; int wake_up = 0; - 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, "drbd_rs_failed_io: sector=%llus size=%d nonsense!\n", (unsigned long long)sector, size); return; diff --git a/drivers/block/drbd/drbd_int.h b/drivers/block/drbd/drbd_int.h index 85207b275e4..bcba2742cfb 100644 --- a/drivers/block/drbd/drbd_int.h +++ b/drivers/block/drbd/drbd_int.h @@ -512,7 +512,7 @@ struct p_sizes { u64 d_size; /* size of disk */ u64 u_size; /* user requested size */ u64 c_size; /* current exported size */ - u32 max_segment_size; /* Maximal size of a BIO */ + u32 max_bio_size; /* Maximal size of a BIO */ u16 queue_order_type; /* not yet implemented in DRBD*/ u16 dds_flags; /* use enum dds_flags here. */ } __packed; @@ -1398,7 +1398,7 @@ struct bm_extent { * With a value of 8 all IO in one 128K block make it to the same slot of the * hash table. */ #define HT_SHIFT 8 -#define DRBD_MAX_SEGMENT_SIZE (1U<<(9+HT_SHIFT)) +#define DRBD_MAX_BIO_SIZE (1U<<(9+HT_SHIFT)) #define DRBD_MAX_SIZE_H80_PACKET (1 << 15) /* The old header only allows packets up to 32Kib data */ diff --git a/drivers/block/drbd/drbd_main.c b/drivers/block/drbd/drbd_main.c index 451fc36a85c..9d9c2ed31e9 100644 --- a/drivers/block/drbd/drbd_main.c +++ b/drivers/block/drbd/drbd_main.c @@ -1924,7 +1924,7 @@ int drbd_send_sizes(struct drbd_conf *mdev, int trigger_reply, enum dds_flags fl p.d_size = cpu_to_be64(d_size); p.u_size = cpu_to_be64(u_size); p.c_size = cpu_to_be64(trigger_reply ? 0 : drbd_get_capacity(mdev->this_bdev)); - p.max_segment_size = cpu_to_be32(queue_max_segment_size(mdev->rq_queue)); + p.max_bio_size = cpu_to_be32(queue_max_hw_sectors(mdev->rq_queue) << 9); p.queue_order_type = cpu_to_be16(q_order_type); p.dds_flags = cpu_to_be16(flags); @@ -2952,7 +2952,7 @@ static void drbd_destroy_mempools(void) static int drbd_create_mempools(void) { struct page *page; - const int number = (DRBD_MAX_SEGMENT_SIZE/PAGE_SIZE) * minor_count; + const int number = (DRBD_MAX_BIO_SIZE/PAGE_SIZE) * minor_count; int i; /* prepare our caches and mempools */ @@ -3218,7 +3218,7 @@ struct drbd_conf *drbd_new_device(unsigned int minor) q->backing_dev_info.congested_data = mdev; blk_queue_make_request(q, drbd_make_request_26); - blk_queue_max_segment_size(q, DRBD_MAX_SEGMENT_SIZE); + blk_queue_max_hw_sectors(q, DRBD_MAX_BIO_SIZE >> 9); blk_queue_bounce_limit(q, BLK_BOUNCE_ANY); blk_queue_merge_bvec(q, drbd_merge_bvec); q->queue_lock = &mdev->req_lock; diff --git a/drivers/block/drbd/drbd_nl.c b/drivers/block/drbd/drbd_nl.c index dad559810ed..9e27d82a9a1 100644 --- a/drivers/block/drbd/drbd_nl.c +++ b/drivers/block/drbd/drbd_nl.c @@ -765,22 +765,21 @@ static int drbd_check_al_size(struct drbd_conf *mdev) return 0; } -void drbd_setup_queue_param(struct drbd_conf *mdev, unsigned int max_seg_s) __must_hold(local) +void drbd_setup_queue_param(struct drbd_conf *mdev, unsigned int max_bio_size) __must_hold(local) { struct request_queue * const q = mdev->rq_queue; struct request_queue * const b = mdev->ldev->backing_bdev->bd_disk->queue; int max_segments = mdev->ldev->dc.max_bio_bvecs; + int max_hw_sectors = min(queue_max_hw_sectors(b), max_bio_size >> 9); - max_seg_s = min(queue_max_sectors(b) * queue_logical_block_size(b), max_seg_s); - - blk_queue_max_hw_sectors(q, max_seg_s >> 9); - blk_queue_max_segments(q, max_segments ? max_segments : BLK_MAX_SEGMENTS); - blk_queue_max_segment_size(q, max_seg_s); blk_queue_logical_block_size(q, 512); - blk_queue_segment_boundary(q, PAGE_SIZE-1); - blk_stack_limits(&q->limits, &b->limits, 0); + blk_queue_max_hw_sectors(q, max_hw_sectors); + /* This is the workaround for "bio would need to, but cannot, be split" */ + blk_queue_max_segments(q, max_segments ? max_segments : BLK_MAX_SEGMENTS); + blk_queue_segment_boundary(q, PAGE_CACHE_SIZE-1); + blk_queue_stack_limits(q, b); - dev_info(DEV, "max_segment_size ( = BIO size ) = %u\n", queue_max_segment_size(q)); + dev_info(DEV, "max BIO size = %u\n", queue_max_hw_sectors(q) << 9); if (q->backing_dev_info.ra_pages != b->backing_dev_info.ra_pages) { dev_info(DEV, "Adjusting my ra_pages to backing device's (%lu -> %lu)\n", @@ -858,7 +857,7 @@ static int drbd_nl_disk_conf(struct drbd_conf *mdev, struct drbd_nl_cfg_req *nlp struct block_device *bdev; struct lru_cache *resync_lru = NULL; union drbd_state ns, os; - unsigned int max_seg_s; + unsigned int max_bio_size; int rv; int cp_discovered = 0; int logical_block_size; @@ -1109,20 +1108,20 @@ static int drbd_nl_disk_conf(struct drbd_conf *mdev, struct drbd_nl_cfg_req *nlp mdev->read_cnt = 0; mdev->writ_cnt = 0; - max_seg_s = DRBD_MAX_SEGMENT_SIZE; + max_bio_size = DRBD_MAX_BIO_SIZE; if (mdev->state.conn == C_CONNECTED) { /* We are Primary, Connected, and now attach a new local * backing store. We must not increase the user visible maximum * bio size on this device to something the peer may not be * able to handle. */ if (mdev->agreed_pro_version < 94) - max_seg_s = queue_max_segment_size(mdev->rq_queue); + max_bio_size = queue_max_hw_sectors(mdev->rq_queue) << 9; 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 8.3.9 and later, stay with default */ } - drbd_setup_queue_param(mdev, max_seg_s); + drbd_setup_queue_param(mdev, max_bio_size); /* If I am currently not R_PRIMARY, * but meta data primary indicator is set, diff --git a/drivers/block/drbd/drbd_receiver.c b/drivers/block/drbd/drbd_receiver.c index ca213c6e5f9..79e7b57006b 100644 --- a/drivers/block/drbd/drbd_receiver.c +++ b/drivers/block/drbd/drbd_receiver.c @@ -277,7 +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; @@ -1240,7 +1240,7 @@ read_in_block(struct drbd_conf *mdev, u64 id, sector_t sector, int data_size) __ data_size -= dgs; 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. */ @@ -1917,7 +1917,7 @@ 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; @@ -2897,7 +2897,7 @@ 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; @@ -2970,14 +2970,14 @@ 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); diff --git a/drivers/block/drbd/drbd_req.c b/drivers/block/drbd/drbd_req.c index ad3fc6228f2..08f53ce9b88 100644 --- a/drivers/block/drbd/drbd_req.c +++ b/drivers/block/drbd/drbd_req.c @@ -1047,7 +1047,7 @@ int drbd_make_request_26(struct request_queue *q, struct bio *bio) /* can this bio be split generically? * Maybe add our own split-arbitrary-bios function. */ - if (bio->bi_vcnt != 1 || bio->bi_idx != 0 || bio->bi_size > DRBD_MAX_SEGMENT_SIZE) { + if (bio->bi_vcnt != 1 || bio->bi_idx != 0 || bio->bi_size > DRBD_MAX_BIO_SIZE) { /* rather error out here than BUG in bio_split */ dev_err(DEV, "bio would need to, but cannot, be split: " "(vcnt=%u,idx=%u,size=%u,sector=%llu)\n", @@ -1098,7 +1098,7 @@ int drbd_make_request_26(struct request_queue *q, struct bio *bio) } /* This is called by bio_add_page(). With this function we reduce - * the number of BIOs that span over multiple DRBD_MAX_SEGMENT_SIZEs + * the number of BIOs that span over multiple DRBD_MAX_BIO_SIZEs * units (was AL_EXTENTs). * * we do the calculation within the lower 32bit of the byte offsets, @@ -1118,8 +1118,8 @@ int drbd_merge_bvec(struct request_queue *q, struct bvec_merge_data *bvm, struct unsigned int bio_size = bvm->bi_size; int limit, backing_limit; - limit = DRBD_MAX_SEGMENT_SIZE - - ((bio_offset & (DRBD_MAX_SEGMENT_SIZE-1)) + bio_size); + limit = DRBD_MAX_BIO_SIZE + - ((bio_offset & (DRBD_MAX_BIO_SIZE-1)) + bio_size); if (limit < 0) limit = 0; if (bio_size == 0) { diff --git a/drivers/block/drbd/drbd_worker.c b/drivers/block/drbd/drbd_worker.c index af805efc94d..782d87237cb 100644 --- a/drivers/block/drbd/drbd_worker.c +++ b/drivers/block/drbd/drbd_worker.c @@ -524,7 +524,7 @@ int w_make_resync_request(struct drbd_conf *mdev, unsigned long bit; sector_t sector; const sector_t capacity = drbd_get_capacity(mdev->this_bdev); - int max_segment_size; + int max_bio_size; int number, rollback_i, size; int align, queued, sndbuf; int i = 0; @@ -559,9 +559,9 @@ int w_make_resync_request(struct drbd_conf *mdev, /* starting with drbd 8.3.8, we can handle multi-bio EEs, * if it should be necessary */ - max_segment_size = - mdev->agreed_pro_version < 94 ? queue_max_segment_size(mdev->rq_queue) : - mdev->agreed_pro_version < 95 ? DRBD_MAX_SIZE_H80_PACKET : DRBD_MAX_SEGMENT_SIZE; + max_bio_size = + mdev->agreed_pro_version < 94 ? queue_max_hw_sectors(mdev->rq_queue) << 9 : + mdev->agreed_pro_version < 95 ? DRBD_MAX_SIZE_H80_PACKET : DRBD_MAX_BIO_SIZE; number = drbd_rs_number_requests(mdev); if (number == 0) @@ -605,7 +605,7 @@ next_sector: goto next_sector; } -#if DRBD_MAX_SEGMENT_SIZE > BM_BLOCK_SIZE +#if DRBD_MAX_BIO_SIZE > BM_BLOCK_SIZE /* try to find some adjacent bits. * we stop if we have already the maximum req size. * @@ -615,7 +615,7 @@ next_sector: align = 1; rollback_i = i; for (;;) { - if (size + BM_BLOCK_SIZE > max_segment_size) + if (size + BM_BLOCK_SIZE > max_bio_size) break; /* Be always aligned */ -- cgit v1.2.3-70-g09d2 From 73a01a18b9c28a0fab1131ece5b0a9bc00a879b8 Mon Sep 17 00:00:00 2001 From: Philipp Reisner Date: Wed, 27 Oct 2010 14:33:00 +0200 Subject: drbd: New packet for Ahead/Behind mode: P_OUT_OF_SYNC Signed-off-by: Philipp Reisner Signed-off-by: Lars Ellenberg --- drivers/block/drbd/drbd_actlog.c | 10 +++++---- drivers/block/drbd/drbd_int.h | 14 +++++++++++- drivers/block/drbd/drbd_main.c | 10 +++++++++ drivers/block/drbd/drbd_receiver.c | 10 +++++++++ drivers/block/drbd/drbd_req.c | 44 +++++++++++++++++++++++++++----------- drivers/block/drbd/drbd_req.h | 4 +++- drivers/block/drbd/drbd_worker.c | 16 ++++++++++++++ include/linux/drbd.h | 2 +- 8 files changed, 91 insertions(+), 19 deletions(-) (limited to 'drivers/block/drbd/drbd_worker.c') diff --git a/drivers/block/drbd/drbd_actlog.c b/drivers/block/drbd/drbd_actlog.c index b4adb58c747..33f6cc537d0 100644 --- a/drivers/block/drbd/drbd_actlog.c +++ b/drivers/block/drbd/drbd_actlog.c @@ -1007,22 +1007,22 @@ void __drbd_set_in_sync(struct drbd_conf *mdev, sector_t sector, int size, * called by tl_clear and drbd_send_dblock (==drbd_make_request). * so this can be _any_ process. */ -void __drbd_set_out_of_sync(struct drbd_conf *mdev, sector_t sector, int size, +int __drbd_set_out_of_sync(struct drbd_conf *mdev, sector_t sector, int size, const char *file, const unsigned int line) { unsigned long sbnr, ebnr, lbnr, flags; sector_t esector, nr_sectors; - unsigned int enr, count; + unsigned int enr, count = 0; struct lc_element *e; if (size <= 0 || (size & 0x1ff) != 0 || size > DRBD_MAX_BIO_SIZE) { dev_err(DEV, "sector: %llus, size: %d\n", (unsigned long long)sector, size); - return; + return 0; } if (!get_ldev(mdev)) - return; /* no disk, no metadata, no bitmap to set bits in */ + return 0; /* no disk, no metadata, no bitmap to set bits in */ nr_sectors = drbd_get_capacity(mdev->this_bdev); esector = sector + (size >> 9) - 1; @@ -1052,6 +1052,8 @@ void __drbd_set_out_of_sync(struct drbd_conf *mdev, sector_t sector, int size, out: put_ldev(mdev); + + return count; } static diff --git a/drivers/block/drbd/drbd_int.h b/drivers/block/drbd/drbd_int.h index 21b7439438c..47133123682 100644 --- a/drivers/block/drbd/drbd_int.h +++ b/drivers/block/drbd/drbd_int.h @@ -212,6 +212,7 @@ enum drbd_packets { /* P_CKPT_FENCE_REQ = 0x25, * currently reserved for protocol D */ /* P_CKPT_DISABLE_REQ = 0x26, * currently reserved for protocol D */ P_DELAY_PROBE = 0x27, /* is used on BOTH sockets */ + P_OUT_OF_SYNC = 0x28, /* Mark as out of sync (Outrunning), data socket */ P_MAX_CMD = 0x28, P_MAY_IGNORE = 0x100, /* Flag to test if (cmd > P_MAY_IGNORE) ... */ @@ -269,6 +270,7 @@ static inline const char *cmdname(enum drbd_packets cmd) [P_RS_IS_IN_SYNC] = "CsumRSIsInSync", [P_COMPRESSED_BITMAP] = "CBitmap", [P_DELAY_PROBE] = "DelayProbe", + [P_OUT_OF_SYNC] = "OutOfSync", [P_MAX_CMD] = NULL, }; @@ -550,6 +552,13 @@ struct p_discard { u32 pad; } __packed; +struct p_block_desc { + struct p_header80 head; + u64 sector; + u32 blksize; + u32 pad; /* to multiple of 8 Byte */ +} __packed; + /* Valid values for the encoding field. * Bump proto version when changing this. */ enum drbd_bitmap_code { @@ -647,6 +656,7 @@ union p_polymorph { struct p_block_req block_req; struct p_delay_probe93 delay_probe93; struct p_rs_uuid rs_uuid; + struct p_block_desc block_desc; } __packed; /**********************************************************************/ @@ -1221,6 +1231,7 @@ extern int drbd_send_ack_dp(struct drbd_conf *mdev, enum drbd_packets cmd, struct p_data *dp, int data_size); extern int drbd_send_ack_ex(struct drbd_conf *mdev, enum drbd_packets cmd, sector_t sector, int blksize, u64 block_id); +extern int drbd_send_oos(struct drbd_conf *mdev, struct drbd_request *req); extern int drbd_send_block(struct drbd_conf *mdev, enum drbd_packets cmd, struct drbd_epoch_entry *e); extern int drbd_send_dblock(struct drbd_conf *mdev, struct drbd_request *req); @@ -1534,6 +1545,7 @@ 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 int w_send_oos(struct drbd_conf *, struct drbd_work *, int); extern void resync_timer_fn(unsigned long data); @@ -1626,7 +1638,7 @@ extern void __drbd_set_in_sync(struct drbd_conf *mdev, sector_t sector, int size, const char *file, const unsigned int line); #define drbd_set_in_sync(mdev, sector, size) \ __drbd_set_in_sync(mdev, sector, size, __FILE__, __LINE__) -extern void __drbd_set_out_of_sync(struct drbd_conf *mdev, sector_t sector, +extern int __drbd_set_out_of_sync(struct drbd_conf *mdev, sector_t sector, int size, const char *file, const unsigned int line); #define drbd_set_out_of_sync(mdev, sector, size) \ __drbd_set_out_of_sync(mdev, sector, size, __FILE__, __LINE__) diff --git a/drivers/block/drbd/drbd_main.c b/drivers/block/drbd/drbd_main.c index 46f27d6c0b2..0dc93f43a47 100644 --- a/drivers/block/drbd/drbd_main.c +++ b/drivers/block/drbd/drbd_main.c @@ -2634,6 +2634,16 @@ int drbd_send_block(struct drbd_conf *mdev, enum drbd_packets cmd, return ok; } +int drbd_send_oos(struct drbd_conf *mdev, struct drbd_request *req) +{ + struct p_block_desc p; + + p.sector = cpu_to_be64(req->sector); + p.blksize = cpu_to_be32(req->size); + + return drbd_send_cmd(mdev, USE_DATA_SOCKET, P_OUT_OF_SYNC, &p.head, sizeof(p)); +} + /* drbd_send distinguishes two cases: diff --git a/drivers/block/drbd/drbd_receiver.c b/drivers/block/drbd/drbd_receiver.c index b19e8b2c4ce..04a08e7541c 100644 --- a/drivers/block/drbd/drbd_receiver.c +++ b/drivers/block/drbd/drbd_receiver.c @@ -3562,6 +3562,15 @@ static int receive_UnplugRemote(struct drbd_conf *mdev, enum drbd_packets cmd, u 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; + + 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); struct data_cmd { @@ -3592,6 +3601,7 @@ 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 }, diff --git a/drivers/block/drbd/drbd_req.c b/drivers/block/drbd/drbd_req.c index 60288fb3c4d..a8d1ff2bda2 100644 --- a/drivers/block/drbd/drbd_req.c +++ b/drivers/block/drbd/drbd_req.c @@ -142,7 +142,7 @@ static void _about_to_complete_local_write(struct drbd_conf *mdev, /* before we can signal completion to the upper layers, * we may need to close the current epoch */ - if (mdev->state.conn >= C_CONNECTED && + if (mdev->state.conn >= C_CONNECTED && mdev->state.conn < C_AHEAD && req->epoch == mdev->newest_tle->br_number) queue_barrier(mdev); @@ -545,6 +545,14 @@ int __req_mod(struct drbd_request *req, enum drbd_req_event what, break; + case queue_for_send_oos: + req->rq_state |= RQ_NET_QUEUED; + req->w.cb = w_send_oos; + drbd_queue_work(&mdev->data.work, &req->w); + break; + + case oos_handed_to_network: + /* actually the same */ case send_canceled: /* treat it the same */ case send_failed: @@ -756,7 +764,7 @@ static int drbd_make_request_common(struct drbd_conf *mdev, struct bio *bio) const sector_t sector = bio->bi_sector; struct drbd_tl_epoch *b = NULL; struct drbd_request *req; - int local, remote; + int local, remote, send_oos = 0; int err = -EIO; int ret = 0; @@ -820,8 +828,11 @@ static int drbd_make_request_common(struct drbd_conf *mdev, struct bio *bio) } remote = remote && (mdev->state.pdsk == D_UP_TO_DATE || - (mdev->state.pdsk == D_INCONSISTENT && - mdev->state.conn >= C_CONNECTED)); + (mdev->state.pdsk >= D_INCONSISTENT && + mdev->state.conn >= C_CONNECTED && + mdev->state.conn < C_AHEAD)); + send_oos = (rw == WRITE && mdev->state.conn == C_AHEAD && + mdev->state.pdsk >= D_INCONSISTENT); if (!(local || remote) && !is_susp(mdev->state)) { if (__ratelimit(&drbd_ratelimit_state)) @@ -835,7 +846,7 @@ static int drbd_make_request_common(struct drbd_conf *mdev, struct bio *bio) * but there is a race between testing the bit and pointer outside the * spinlock, and grabbing the spinlock. * if we lost that race, we retry. */ - if (rw == WRITE && remote && + if (rw == WRITE && (remote || send_oos) && mdev->unused_spare_tle == NULL && test_bit(CREATE_BARRIER, &mdev->flags)) { allocate_barrier: @@ -860,11 +871,15 @@ allocate_barrier: goto fail_free_complete; } - if (remote) { + if (remote || send_oos) { remote = (mdev->state.pdsk == D_UP_TO_DATE || - (mdev->state.pdsk == D_INCONSISTENT && - mdev->state.conn >= C_CONNECTED)); - if (!remote) + (mdev->state.pdsk >= D_INCONSISTENT && + mdev->state.conn >= C_CONNECTED && + mdev->state.conn < C_AHEAD)); + send_oos = (rw == WRITE && mdev->state.conn == C_AHEAD && + mdev->state.pdsk >= D_INCONSISTENT); + + if (!(remote || send_oos)) dev_warn(DEV, "lost connection while grabbing the req_lock!\n"); if (!(local || remote)) { dev_err(DEV, "IO ERROR: neither local nor remote disk\n"); @@ -877,7 +892,7 @@ allocate_barrier: mdev->unused_spare_tle = b; b = NULL; } - if (rw == WRITE && remote && + if (rw == WRITE && (remote || send_oos) && mdev->unused_spare_tle == NULL && test_bit(CREATE_BARRIER, &mdev->flags)) { /* someone closed the current epoch @@ -900,7 +915,7 @@ allocate_barrier: * barrier packet. To get the write ordering right, we only have to * make sure that, if this is a write request and it triggered a * barrier packet, this request is queued within the same spinlock. */ - if (remote && mdev->unused_spare_tle && + if ((remote || send_oos) && mdev->unused_spare_tle && test_and_clear_bit(CREATE_BARRIER, &mdev->flags)) { _tl_add_barrier(mdev, mdev->unused_spare_tle); mdev->unused_spare_tle = NULL; @@ -948,8 +963,11 @@ allocate_barrier: ? queue_for_net_write : queue_for_net_read); } + if (send_oos && drbd_set_out_of_sync(mdev, sector, size)) + _req_mod(req, queue_for_send_oos); - if (remote && mdev->net_conf->on_congestion != OC_BLOCK) { + if (remote && + mdev->net_conf->on_congestion != OC_BLOCK && mdev->agreed_pro_version >= 96) { int congested = 0; if (mdev->net_conf->cong_fill && @@ -964,6 +982,8 @@ allocate_barrier: } if (congested) { + queue_barrier(mdev); + if (mdev->net_conf->on_congestion == OC_PULL_AHEAD) _drbd_set_state(_NS(mdev, conn, C_AHEAD), 0, NULL); else /*mdev->net_conf->on_congestion == OC_DISCONNECT */ diff --git a/drivers/block/drbd/drbd_req.h b/drivers/block/drbd/drbd_req.h index 69d350fe7c1..40d3dcd8fc8 100644 --- a/drivers/block/drbd/drbd_req.h +++ b/drivers/block/drbd/drbd_req.h @@ -82,14 +82,16 @@ enum drbd_req_event { to_be_submitted, /* XXX yes, now I am inconsistent... - * these two are not "events" but "actions" + * these are not "events" but "actions" * oh, well... */ queue_for_net_write, queue_for_net_read, + queue_for_send_oos, send_canceled, send_failed, handed_over_to_network, + oos_handed_to_network, connection_lost_while_pending, read_retry_remote_canceled, recv_acked_by_peer, diff --git a/drivers/block/drbd/drbd_worker.c b/drivers/block/drbd/drbd_worker.c index 782d87237cb..67499077c48 100644 --- a/drivers/block/drbd/drbd_worker.c +++ b/drivers/block/drbd/drbd_worker.c @@ -1237,6 +1237,22 @@ int w_send_write_hint(struct drbd_conf *mdev, struct drbd_work *w, int cancel) return drbd_send_short_cmd(mdev, P_UNPLUG_REMOTE); } +int w_send_oos(struct drbd_conf *mdev, struct drbd_work *w, int cancel) +{ + struct drbd_request *req = container_of(w, struct drbd_request, w); + int ok; + + if (unlikely(cancel)) { + req_mod(req, send_canceled); + return 1; + } + + ok = drbd_send_oos(mdev, req); + req_mod(req, oos_handed_to_network); + + return ok; +} + /** * w_send_dblock() - Worker callback to send a P_DATA packet in order to mirror a write request * @mdev: DRBD device. diff --git a/include/linux/drbd.h b/include/linux/drbd.h index 23f31be6f00..41da654cc0b 100644 --- a/include/linux/drbd.h +++ b/include/linux/drbd.h @@ -56,7 +56,7 @@ extern const char *drbd_buildtag(void); #define REL_VERSION "8.3.9" #define API_VERSION 88 #define PRO_VERSION_MIN 86 -#define PRO_VERSION_MAX 95 +#define PRO_VERSION_MAX 96 enum drbd_io_error_p { -- cgit v1.2.3-70-g09d2 From c4752ef1284519c3baa1c3b19df34a80b4905245 Mon Sep 17 00:00:00 2001 From: Philipp Reisner Date: Wed, 27 Oct 2010 17:32:36 +0200 Subject: drbd: When proxy's buffer drained off go into regular resync mode Signed-off-by: Philipp Reisner Signed-off-by: Lars Ellenberg --- drivers/block/drbd/drbd_int.h | 4 +++- drivers/block/drbd/drbd_main.c | 1 + drivers/block/drbd/drbd_receiver.c | 9 +++++++++ drivers/block/drbd/drbd_worker.c | 10 +++++++++- 4 files changed, 22 insertions(+), 2 deletions(-) (limited to 'drivers/block/drbd/drbd_worker.c') diff --git a/drivers/block/drbd/drbd_int.h b/drivers/block/drbd/drbd_int.h index 47133123682..da02cce374c 100644 --- a/drivers/block/drbd/drbd_int.h +++ b/drivers/block/drbd/drbd_int.h @@ -973,7 +973,8 @@ struct drbd_conf { struct drbd_work resync_work, unplug_work, go_diskless, - md_sync_work; + md_sync_work, + start_resync_work; struct timer_list resync_timer; struct timer_list md_sync_timer; #ifdef DRBD_DEBUG_MD_SYNC @@ -1546,6 +1547,7 @@ 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 int w_send_oos(struct drbd_conf *, struct drbd_work *, int); +extern int w_start_resync(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 0dc93f43a47..f49505cf8d0 100644 --- a/drivers/block/drbd/drbd_main.c +++ b/drivers/block/drbd/drbd_main.c @@ -2846,6 +2846,7 @@ void drbd_init_set_defaults(struct drbd_conf *mdev) INIT_LIST_HEAD(&mdev->unplug_work.list); INIT_LIST_HEAD(&mdev->go_diskless.list); INIT_LIST_HEAD(&mdev->md_sync_work.list); + INIT_LIST_HEAD(&mdev->start_resync_work.list); INIT_LIST_HEAD(&mdev->bm_io_work.w.list); mdev->resync_work.cb = w_resync_inactive; diff --git a/drivers/block/drbd/drbd_receiver.c b/drivers/block/drbd/drbd_receiver.c index 04a08e7541c..ee9238e5932 100644 --- a/drivers/block/drbd/drbd_receiver.c +++ b/drivers/block/drbd/drbd_receiver.c @@ -3279,6 +3279,7 @@ 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); @@ -4338,6 +4339,14 @@ static int got_BarrierAck(struct drbd_conf *mdev, struct p_header80 *h) tl_release(mdev, p->barrier, be32_to_cpu(p->set_size)); + if (mdev->state.conn == C_AHEAD && + atomic_read(&mdev->ap_in_flight) == 0 && + list_empty(&mdev->start_resync_work.list)) { + struct drbd_work *w = &mdev->start_resync_work; + w->cb = w_start_resync; + drbd_queue_work_front(&mdev->data.work, w); + } + return TRUE; } diff --git a/drivers/block/drbd/drbd_worker.c b/drivers/block/drbd/drbd_worker.c index 67499077c48..c9e7cb7c788 100644 --- a/drivers/block/drbd/drbd_worker.c +++ b/drivers/block/drbd/drbd_worker.c @@ -742,6 +742,14 @@ static int w_make_ov_request(struct drbd_conf *mdev, struct drbd_work *w, int ca return 1; } + +int w_start_resync(struct drbd_conf *mdev, struct drbd_work *w, int cancel) +{ + drbd_start_resync(mdev, C_SYNC_SOURCE); + + return 1; +} + int w_ov_finished(struct drbd_conf *mdev, struct drbd_work *w, int cancel) { kfree(w); @@ -1472,7 +1480,7 @@ void drbd_start_resync(struct drbd_conf *mdev, enum drbd_conns side) union drbd_state ns; int r; - if (mdev->state.conn >= C_SYNC_SOURCE) { + if (mdev->state.conn >= C_SYNC_SOURCE && mdev->state.conn < C_AHEAD) { dev_err(DEV, "Resync already running!\n"); return; } -- cgit v1.2.3-70-g09d2 From 59817f4fab6a165ba83ce399464ba38432db8233 Mon Sep 17 00:00:00 2001 From: Philipp Reisner Date: Fri, 29 Oct 2010 12:44:20 +0200 Subject: drbd: Do not cleanup resync LRU for the Ahead/Behind SyncSource/SyncTarget transitions This one should be replaced with moving this cleanup to the 'right' position. Signed-off-by: Philipp Reisner Signed-off-by: Lars Ellenberg --- drivers/block/drbd/drbd_worker.c | 9 +++++++-- 1 file changed, 7 insertions(+), 2 deletions(-) (limited to 'drivers/block/drbd/drbd_worker.c') diff --git a/drivers/block/drbd/drbd_worker.c b/drivers/block/drbd/drbd_worker.c index c9e7cb7c788..982d68432a0 100644 --- a/drivers/block/drbd/drbd_worker.c +++ b/drivers/block/drbd/drbd_worker.c @@ -1485,8 +1485,13 @@ void drbd_start_resync(struct drbd_conf *mdev, enum drbd_conns side) return; } - /* In case a previous resync run was aborted by an IO error/detach on the peer. */ - drbd_rs_cancel_all(mdev); + if (mdev->state.conn < C_AHEAD) { + /* In case a previous resync run was aborted by an IO error/detach on the peer. */ + drbd_rs_cancel_all(mdev); + /* This should be done when we abort the resync. We definitely do not + want to have this for connections going back and forth between + Ahead/Behind and SyncSource/SyncTarget */ + } if (side == C_SYNC_TARGET) { /* Since application IO was locked out during C_WF_BITMAP_T and -- cgit v1.2.3-70-g09d2 From e3555d8545976703938d1b59e2db509426dbe02c Mon Sep 17 00:00:00 2001 From: Philipp Reisner Date: Sun, 7 Nov 2010 15:56:29 +0100 Subject: drbd: Implemented priority inheritance for resync requests We only issue resync requests if there is no significant application IO going on. = Application IO has higher priority than resnyc IO. If application IO can not be started because the resync process locked an resync_lru entry, start the IO operations necessary to release the lock ASAP. Signed-off-by: Philipp Reisner Signed-off-by: Lars Ellenberg --- drivers/block/drbd/drbd_actlog.c | 4 ++-- drivers/block/drbd/drbd_int.h | 3 ++- drivers/block/drbd/drbd_receiver.c | 22 ++++++++++++++++++---- drivers/block/drbd/drbd_worker.c | 18 +++++------------- 4 files changed, 27 insertions(+), 20 deletions(-) (limited to 'drivers/block/drbd/drbd_worker.c') diff --git a/drivers/block/drbd/drbd_actlog.c b/drivers/block/drbd/drbd_actlog.c index 33f6cc537d0..28f85d95078 100644 --- a/drivers/block/drbd/drbd_actlog.c +++ b/drivers/block/drbd/drbd_actlog.c @@ -182,6 +182,7 @@ static struct lc_element *_al_get(struct drbd_conf *mdev, unsigned int enr) if (unlikely(tmp != NULL)) { struct bm_extent *bm_ext = lc_entry(tmp, struct bm_extent, lce); if (test_bit(BME_NO_WRITES, &bm_ext->flags)) { + set_bit(BME_PRIORITY, &bm_ext->flags); spin_unlock_irq(&mdev->al_lock); return NULL; } @@ -1297,8 +1298,7 @@ void drbd_rs_complete_io(struct drbd_conf *mdev, sector_t sector) } if (lc_put(mdev->resync, &bm_ext->lce) == 0) { - clear_bit(BME_LOCKED, &bm_ext->flags); - clear_bit(BME_NO_WRITES, &bm_ext->flags); + bm_ext->flags = 0; /* clear BME_LOCKED, BME_NO_WRITES and BME_PRIORITY */ mdev->resync_locked--; wake_up(&mdev->al_wait); } diff --git a/drivers/block/drbd/drbd_int.h b/drivers/block/drbd/drbd_int.h index da02cce374c..366873d661b 100644 --- a/drivers/block/drbd/drbd_int.h +++ b/drivers/block/drbd/drbd_int.h @@ -1326,6 +1326,7 @@ struct bm_extent { #define BME_NO_WRITES 0 /* bm_extent.flags: no more requests on this one! */ #define BME_LOCKED 1 /* bm_extent.flags: syncer active on this one. */ +#define BME_PRIORITY 2 /* finish resync IO on this extent ASAP! App IO waiting! */ /* drbd_bitmap.c */ /* @@ -1552,7 +1553,7 @@ extern int w_start_resync(struct drbd_conf *, struct drbd_work *, int); extern void resync_timer_fn(unsigned long data); /* drbd_receiver.c */ -extern int drbd_rs_should_slow_down(struct drbd_conf *mdev); +extern int drbd_rs_should_slow_down(struct drbd_conf *mdev, sector_t sector); extern int drbd_submit_ee(struct drbd_conf *mdev, struct drbd_epoch_entry *e, const unsigned rw, const int fault_type); extern int drbd_release_ee(struct drbd_conf *mdev, struct list_head *list); diff --git a/drivers/block/drbd/drbd_receiver.c b/drivers/block/drbd/drbd_receiver.c index ee9238e5932..0630a2e122d 100644 --- a/drivers/block/drbd/drbd_receiver.c +++ b/drivers/block/drbd/drbd_receiver.c @@ -1862,10 +1862,11 @@ 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; @@ -1873,9 +1874,22 @@ 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; @@ -2060,9 +2074,9 @@ 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: diff --git a/drivers/block/drbd/drbd_worker.c b/drivers/block/drbd/drbd_worker.c index 982d68432a0..4008130f2b2 100644 --- a/drivers/block/drbd/drbd_worker.c +++ b/drivers/block/drbd/drbd_worker.c @@ -355,7 +355,7 @@ static int read_for_csum(struct drbd_conf *mdev, sector_t sector, int size) if (!get_ldev(mdev)) return -EIO; - if (drbd_rs_should_slow_down(mdev)) + if (drbd_rs_should_slow_down(mdev, sector)) goto defer; /* GFP_TRY, because if there is no memory available right now, this may @@ -503,16 +503,6 @@ int drbd_rs_number_requests(struct drbd_conf *mdev) number = SLEEP_TIME * mdev->c_sync_rate / ((BM_BLOCK_SIZE / 1024) * HZ); } - /* Throttle resync on lower level disk activity, which may also be - * caused by application IO on Primary/SyncTarget. - * Keep this after the call to drbd_rs_controller, as that assumes - * to be called as precisely as possible every SLEEP_TIME, - * and would be confused otherwise. */ - if (number && drbd_rs_should_slow_down(mdev)) { - mdev->c_sync_rate = 1; - number = 0; - } - /* ignore the amount of pending requests, the resync controller should * throttle down to incoming reply rate soon enough anyways. */ return number; @@ -594,7 +584,8 @@ next_sector: sector = BM_BIT_TO_SECT(bit); - if (drbd_try_rs_begin_io(mdev, sector)) { + if (drbd_rs_should_slow_down(mdev, sector) || + drbd_try_rs_begin_io(mdev, sector)) { mdev->bm_resync_fo = bit; goto requeue; } @@ -719,7 +710,8 @@ static int w_make_ov_request(struct drbd_conf *mdev, struct drbd_work *w, int ca size = BM_BLOCK_SIZE; - if (drbd_try_rs_begin_io(mdev, sector)) { + if (drbd_rs_should_slow_down(mdev, sector) || + drbd_try_rs_begin_io(mdev, sector)) { mdev->ov_position = sector; goto requeue; } -- cgit v1.2.3-70-g09d2 From 9d77a5fee9d2a1ea4cd9a841d27b107df5913b33 Mon Sep 17 00:00:00 2001 From: Philipp Reisner Date: Sun, 7 Nov 2010 18:02:56 +0100 Subject: drbd: Make some functions static Signed-off-by: Philipp Reisner Signed-off-by: Lars Ellenberg --- drivers/block/drbd/drbd_int.h | 1 - drivers/block/drbd/drbd_worker.c | 10 ++++++---- 2 files changed, 6 insertions(+), 5 deletions(-) (limited to 'drivers/block/drbd/drbd_worker.c') diff --git a/drivers/block/drbd/drbd_int.h b/drivers/block/drbd/drbd_int.h index 366873d661b..77ac6765fd5 100644 --- a/drivers/block/drbd/drbd_int.h +++ b/drivers/block/drbd/drbd_int.h @@ -1540,7 +1540,6 @@ extern int w_ov_finished(struct drbd_conf *, struct drbd_work *, int); extern int w_resync_inactive(struct drbd_conf *, struct drbd_work *, int); extern int w_resume_next_sg(struct drbd_conf *, struct drbd_work *, int); extern int w_send_write_hint(struct drbd_conf *, struct drbd_work *, int); -extern int w_make_resync_request(struct drbd_conf *, struct drbd_work *, int); extern int w_send_dblock(struct drbd_conf *, struct drbd_work *, int); 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); diff --git a/drivers/block/drbd/drbd_worker.c b/drivers/block/drbd/drbd_worker.c index 4008130f2b2..ab5cff502f5 100644 --- a/drivers/block/drbd/drbd_worker.c +++ b/drivers/block/drbd/drbd_worker.c @@ -39,6 +39,8 @@ #include "drbd_req.h" static int w_make_ov_request(struct drbd_conf *mdev, struct drbd_work *w, int cancel); +static int w_make_resync_request(struct drbd_conf *mdev, + struct drbd_work *w, int cancel); @@ -438,7 +440,7 @@ static void fifo_add_val(struct fifo_buffer *fb, int value) fb->values[i] += value; } -int drbd_rs_controller(struct drbd_conf *mdev) +static int drbd_rs_controller(struct drbd_conf *mdev) { unsigned int sect_in; /* Number of sectors that came in since the last turn */ unsigned int want; /* The number of sectors we want in the proxy */ @@ -492,7 +494,7 @@ int drbd_rs_controller(struct drbd_conf *mdev) return req_sect; } -int drbd_rs_number_requests(struct drbd_conf *mdev) +static int drbd_rs_number_requests(struct drbd_conf *mdev) { int number; if (mdev->rs_plan_s.size) { /* mdev->sync_conf.c_plan_ahead */ @@ -508,8 +510,8 @@ int drbd_rs_number_requests(struct drbd_conf *mdev) return number; } -int w_make_resync_request(struct drbd_conf *mdev, - struct drbd_work *w, int cancel) +static int w_make_resync_request(struct drbd_conf *mdev, + struct drbd_work *w, int cancel) { unsigned long bit; sector_t sector; -- cgit v1.2.3-70-g09d2 From 09b9e7979378fe070784de20e50bb1d42aa643ab Mon Sep 17 00:00:00 2001 From: Philipp Reisner Date: Fri, 3 Dec 2010 16:04:24 +0100 Subject: drbd: Implemented the before-resync-source handler Signed-off-by: Philipp Reisner Signed-off-by: Lars Ellenberg --- drivers/block/drbd/drbd_worker.c | 14 ++++++++++++++ 1 file changed, 14 insertions(+) (limited to 'drivers/block/drbd/drbd_worker.c') diff --git a/drivers/block/drbd/drbd_worker.c b/drivers/block/drbd/drbd_worker.c index ab5cff502f5..e201f6f82c0 100644 --- a/drivers/block/drbd/drbd_worker.c +++ b/drivers/block/drbd/drbd_worker.c @@ -1499,6 +1499,20 @@ void drbd_start_resync(struct drbd_conf *mdev, enum drbd_conns side) drbd_force_state(mdev, NS(conn, C_DISCONNECTING)); return; } + } else /* C_SYNC_SOURCE */ { + r = drbd_khelper(mdev, "before-resync-source"); + r = (r >> 8) & 0xff; + if (r > 0) { + if (r == 3) { + dev_info(DEV, "before-resync-source handler returned %d, " + "ignoring. Old userland tools?", r); + } else { + dev_info(DEV, "before-resync-source handler returned %d, " + "dropping connection.\n", r); + drbd_force_state(mdev, NS(conn, C_DISCONNECTING)); + return; + } + } } drbd_state_lock(mdev); -- cgit v1.2.3-70-g09d2 From 81e84650c200de0695372461964dd960365696db Mon Sep 17 00:00:00 2001 From: Andreas Gruenbacher Date: Thu, 9 Dec 2010 15:03:57 +0100 Subject: drbd: Use the standard bool, true, and false keywords Signed-off-by: Philipp Reisner Signed-off-by: Lars Ellenberg --- drivers/block/drbd/drbd_actlog.c | 8 +-- drivers/block/drbd/drbd_bitmap.c | 4 +- drivers/block/drbd/drbd_int.h | 13 +--- drivers/block/drbd/drbd_main.c | 30 ++++----- drivers/block/drbd/drbd_nl.c | 4 +- drivers/block/drbd/drbd_receiver.c | 128 ++++++++++++++++++------------------- drivers/block/drbd/drbd_req.c | 4 +- drivers/block/drbd/drbd_worker.c | 4 +- 8 files changed, 94 insertions(+), 101 deletions(-) (limited to 'drivers/block/drbd/drbd_worker.c') diff --git a/drivers/block/drbd/drbd_actlog.c b/drivers/block/drbd/drbd_actlog.c index 2e8a95ce79b..e3f0f4d31d7 100644 --- a/drivers/block/drbd/drbd_actlog.c +++ b/drivers/block/drbd/drbd_actlog.c @@ -338,7 +338,7 @@ w_al_write_transaction(struct drbd_conf *mdev, struct drbd_work *w, int unused) + mdev->ldev->md.al_offset + mdev->al_tr_pos; if (!drbd_md_sync_page_io(mdev, mdev->ldev, sector, WRITE)) - drbd_chk_io_error(mdev, 1, TRUE); + drbd_chk_io_error(mdev, 1, true); if (++mdev->al_tr_pos > div_ceil(mdev->act_log->nr_elements, AL_EXTENTS_PT)) @@ -528,7 +528,7 @@ static void atodb_endio(struct bio *bio, int error) if (!error && !uptodate) error = -EIO; - drbd_chk_io_error(mdev, error, TRUE); + drbd_chk_io_error(mdev, error, true); if (error && wc->error == 0) wc->error = error; @@ -991,7 +991,7 @@ void __drbd_set_in_sync(struct drbd_conf *mdev, sector_t sector, int size, if (count && get_ldev(mdev)) { drbd_advance_rs_marks(mdev, drbd_bm_total_weight(mdev)); spin_lock_irqsave(&mdev->al_lock, flags); - drbd_try_clear_on_disk_bm(mdev, sector, count, TRUE); + drbd_try_clear_on_disk_bm(mdev, sector, count, true); spin_unlock_irqrestore(&mdev->al_lock, flags); /* just wake_up unconditional now, various lc_chaged(), @@ -1441,7 +1441,7 @@ void drbd_rs_failed_io(struct drbd_conf *mdev, sector_t sector, int size) mdev->rs_failed += count; if (get_ldev(mdev)) { - drbd_try_clear_on_disk_bm(mdev, sector, count, FALSE); + drbd_try_clear_on_disk_bm(mdev, sector, count, false); put_ldev(mdev); } diff --git a/drivers/block/drbd/drbd_bitmap.c b/drivers/block/drbd/drbd_bitmap.c index 5dafbabe961..9390e952678 100644 --- a/drivers/block/drbd/drbd_bitmap.c +++ b/drivers/block/drbd/drbd_bitmap.c @@ -844,7 +844,7 @@ static int bm_rw(struct drbd_conf *mdev, int rw) __must_hold(local) if (test_bit(BM_MD_IO_ERROR, &b->bm_flags)) { dev_alert(DEV, "we had at least one MD IO ERROR during bitmap IO\n"); - drbd_chk_io_error(mdev, 1, TRUE); + drbd_chk_io_error(mdev, 1, true); err = -EIO; } @@ -916,7 +916,7 @@ int drbd_bm_write_sect(struct drbd_conf *mdev, unsigned long enr) __must_hold(lo dev_err(DEV, "IO ERROR writing bitmap sector %lu " "(meta-disk sector %llus)\n", enr, (unsigned long long)on_disk_sector); - drbd_chk_io_error(mdev, 1, TRUE); + drbd_chk_io_error(mdev, 1, true); for (i = 0; i < AL_EXT_PER_BM_SECT; i++) drbd_bm_ALe_set_all(mdev, enr*AL_EXT_PER_BM_SECT+i); } diff --git a/drivers/block/drbd/drbd_int.h b/drivers/block/drbd/drbd_int.h index 749607a494e..0855934e19d 100644 --- a/drivers/block/drbd/drbd_int.h +++ b/drivers/block/drbd/drbd_int.h @@ -72,13 +72,6 @@ extern int fault_devs; extern char usermode_helper[]; -#ifndef TRUE -#define TRUE 1 -#endif -#ifndef FALSE -#define FALSE 0 -#endif - /* I don't remember why XCPU ... * This is used to wake the asender, * and to interrupt sending the sending task @@ -2002,17 +1995,17 @@ static inline int drbd_send_ping_ack(struct drbd_conf *mdev) static inline void drbd_thread_stop(struct drbd_thread *thi) { - _drbd_thread_stop(thi, FALSE, TRUE); + _drbd_thread_stop(thi, false, true); } static inline void drbd_thread_stop_nowait(struct drbd_thread *thi) { - _drbd_thread_stop(thi, FALSE, FALSE); + _drbd_thread_stop(thi, false, false); } static inline void drbd_thread_restart_nowait(struct drbd_thread *thi) { - _drbd_thread_stop(thi, TRUE, FALSE); + _drbd_thread_stop(thi, true, false); } /* counts how many answer packets packets we expect from our peer, diff --git a/drivers/block/drbd/drbd_main.c b/drivers/block/drbd/drbd_main.c index a101dceb2d4..7eb447d20cc 100644 --- a/drivers/block/drbd/drbd_main.c +++ b/drivers/block/drbd/drbd_main.c @@ -456,7 +456,7 @@ void tl_restart(struct drbd_conf *mdev, enum drbd_req_event what) } /** - * cl_wide_st_chg() - TRUE if the state change is a cluster wide one + * cl_wide_st_chg() - true if the state change is a cluster wide one * @mdev: DRBD device. * @os: old (current) state. * @ns: new (wanted) state. @@ -1623,7 +1623,7 @@ int drbd_thread_start(struct drbd_thread *thi) if (!try_module_get(THIS_MODULE)) { dev_err(DEV, "Failed to get module reference in drbd_thread_start\n"); spin_unlock_irqrestore(&thi->t_lock, flags); - return FALSE; + return false; } init_completion(&thi->stop); @@ -1640,7 +1640,7 @@ int drbd_thread_start(struct drbd_thread *thi) dev_err(DEV, "Couldn't start thread\n"); module_put(THIS_MODULE); - return FALSE; + return false; } spin_lock_irqsave(&thi->t_lock, flags); thi->task = nt; @@ -1660,7 +1660,7 @@ int drbd_thread_start(struct drbd_thread *thi) break; } - return TRUE; + return true; } @@ -1758,8 +1758,8 @@ int _drbd_send_cmd(struct drbd_conf *mdev, struct socket *sock, { int sent, ok; - ERR_IF(!h) return FALSE; - ERR_IF(!size) return FALSE; + ERR_IF(!h) return false; + ERR_IF(!size) return false; h->magic = BE_DRBD_MAGIC; h->command = cpu_to_be16(cmd); @@ -2196,14 +2196,14 @@ int _drbd_send_bitmap(struct drbd_conf *mdev) struct p_header80 *p; int ret; - ERR_IF(!mdev->bitmap) return FALSE; + ERR_IF(!mdev->bitmap) return false; /* maybe we should use some per thread scratch page, * and allocate that during initial device creation? */ p = (struct p_header80 *) __get_free_page(GFP_NOIO); if (!p) { dev_err(DEV, "failed to allocate one page buffer in %s\n", __func__); - return FALSE; + return false; } if (get_ldev(mdev)) { @@ -2256,7 +2256,7 @@ int drbd_send_b_ack(struct drbd_conf *mdev, u32 barrier_nr, u32 set_size) p.set_size = cpu_to_be32(set_size); if (mdev->state.conn < C_CONNECTED) - return FALSE; + return false; ok = drbd_send_cmd(mdev, USE_META_SOCKET, P_BARRIER_ACK, (struct p_header80 *)&p, sizeof(p)); return ok; @@ -2284,7 +2284,7 @@ static int _drbd_send_ack(struct drbd_conf *mdev, enum drbd_packets cmd, p.seq_num = cpu_to_be32(atomic_add_return(1, &mdev->packet_seq)); if (!mdev->meta.socket || mdev->state.conn < C_CONNECTED) - return FALSE; + return false; ok = drbd_send_cmd(mdev, USE_META_SOCKET, cmd, (struct p_header80 *)&p, sizeof(p)); return ok; @@ -2390,8 +2390,8 @@ int drbd_send_ov_request(struct drbd_conf *mdev, sector_t sector, int size) } /* called on sndtimeo - * returns FALSE if we should retry, - * TRUE if we think connection is dead + * returns false if we should retry, + * true if we think connection is dead */ static int we_should_drop_the_connection(struct drbd_conf *mdev, struct socket *sock) { @@ -2404,7 +2404,7 @@ static int we_should_drop_the_connection(struct drbd_conf *mdev, struct socket * || mdev->state.conn < C_CONNECTED; if (drop_it) - return TRUE; + return true; drop_it = !--mdev->ko_count; if (!drop_it) { @@ -3283,7 +3283,7 @@ struct drbd_conf *drbd_new_device(unsigned int minor) goto out_no_disk; mdev->vdisk = disk; - set_disk_ro(disk, TRUE); + set_disk_ro(disk, true); disk->queue = q; disk->major = DRBD_MAJOR; @@ -3560,7 +3560,7 @@ void drbd_md_sync(struct drbd_conf *mdev) if (!drbd_md_sync_page_io(mdev, mdev->ldev, sector, WRITE)) { /* this was a try anyways ... */ dev_err(DEV, "meta data update failed!\n"); - drbd_chk_io_error(mdev, 1, TRUE); + drbd_chk_io_error(mdev, 1, true); } /* Update mdev->ldev->md.la_size_sect, diff --git a/drivers/block/drbd/drbd_nl.c b/drivers/block/drbd/drbd_nl.c index 7a2faf6d48e..9e94c348c42 100644 --- a/drivers/block/drbd/drbd_nl.c +++ b/drivers/block/drbd/drbd_nl.c @@ -385,7 +385,7 @@ drbd_set_role(struct drbd_conf *mdev, enum drbd_role new_role, int force) wait_event(mdev->misc_wait, atomic_read(&mdev->ap_pending_cnt) == 0); if (new_role == R_SECONDARY) { - set_disk_ro(mdev->vdisk, TRUE); + set_disk_ro(mdev->vdisk, true); if (get_ldev(mdev)) { mdev->ldev->md.uuid[UI_CURRENT] &= ~(u64)1; put_ldev(mdev); @@ -395,7 +395,7 @@ drbd_set_role(struct drbd_conf *mdev, enum drbd_role new_role, int force) mdev->net_conf->want_lose = 0; put_net_conf(mdev); } - set_disk_ro(mdev->vdisk, FALSE); + set_disk_ro(mdev->vdisk, false); if (get_ldev(mdev)) { if (((mdev->state.conn < C_CONNECTED || mdev->state.pdsk <= D_FAILED) diff --git a/drivers/block/drbd/drbd_receiver.c b/drivers/block/drbd/drbd_receiver.c index 00dcb1172ca..732aacb46a3 100644 --- a/drivers/block/drbd/drbd_receiver.c +++ b/drivers/block/drbd/drbd_receiver.c @@ -725,16 +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; } } @@ -933,7 +933,7 @@ 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)) { @@ -947,11 +947,11 @@ 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) @@ -1160,7 +1160,7 @@ 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 */ @@ -1188,10 +1188,10 @@ 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; @@ -1209,7 +1209,7 @@ 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) @@ -1303,7 +1303,7 @@ 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); @@ -1426,7 +1426,7 @@ 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. @@ -1438,7 +1438,7 @@ 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) @@ -1455,7 +1455,7 @@ 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 @@ -1655,7 +1655,7 @@ 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; @@ -1774,7 +1774,7 @@ 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)) { @@ -1830,7 +1830,7 @@ 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. @@ -1848,7 +1848,7 @@ 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, @@ -1934,12 +1934,12 @@ static int receive_DataRequest(struct drbd_conf *mdev, enum drbd_packets cmd, un 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)) { @@ -1976,7 +1976,7 @@ 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) { @@ -2089,7 +2089,7 @@ 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. @@ -2102,7 +2102,7 @@ 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) @@ -2690,7 +2690,7 @@ 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)) { @@ -2701,11 +2701,11 @@ 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 @@ -2737,7 +2737,7 @@ 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; @@ -2755,7 +2755,7 @@ 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) { @@ -2775,7 +2775,7 @@ 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); @@ -2785,11 +2785,11 @@ 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 */ @@ -2883,7 +2883,7 @@ 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) @@ -2920,7 +2920,7 @@ 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. @@ -2957,7 +2957,7 @@ 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); } @@ -2967,7 +2967,7 @@ static int receive_sizes(struct drbd_conf *mdev, enum drbd_packets cmd, unsigned 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. */ @@ -3014,7 +3014,7 @@ 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) @@ -3038,7 +3038,7 @@ 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)) { @@ -3073,7 +3073,7 @@ static int receive_uuids(struct drbd_conf *mdev, enum drbd_packets cmd, unsigned if (mdev->state.conn >= C_CONNECTED && mdev->state.disk < D_INCONSISTENT) drbd_set_ed_uuid(mdev, p_uuid[UI_CURRENT]); - return TRUE; + return true; } /** @@ -3118,7 +3118,7 @@ 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); @@ -3129,7 +3129,7 @@ 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) @@ -3174,7 +3174,7 @@ 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; } } @@ -3227,10 +3227,10 @@ 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; } } } @@ -3255,7 +3255,7 @@ 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; @@ -3263,7 +3263,7 @@ 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) { @@ -3281,7 +3281,7 @@ 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) @@ -3308,7 +3308,7 @@ static int receive_sync_uuid(struct drbd_conf *mdev, enum drbd_packets cmd, unsi } else dev_err(DEV, "Ignoring SyncUUID packet!\n"); - return TRUE; + return true; } enum receive_bitmap_ret { OK, DONE, FAILED }; @@ -3462,7 +3462,7 @@ 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 ok = false; struct p_header80 *h = &mdev->data.rbuf.header.h80; /* drbd_bm_lock(mdev, "receive bitmap"); By intention no bm_lock */ @@ -3535,7 +3535,7 @@ 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); by intention no lock */ if (ok && mdev->state.conn == C_WF_BITMAP_S) @@ -3569,7 +3569,7 @@ static int receive_UnplugRemote(struct drbd_conf *mdev, enum drbd_packets cmd, u * 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) @@ -3578,7 +3578,7 @@ static int receive_out_of_sync(struct drbd_conf *mdev, enum drbd_packets cmd, un drbd_set_out_of_sync(mdev, be64_to_cpu(p->sector), be32_to_cpu(p->blksize)); - return TRUE; + return true; } typedef int (*drbd_cmd_handler_f)(struct drbd_conf *, enum drbd_packets cmd, unsigned int to_receive); @@ -4147,7 +4147,7 @@ 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) @@ -4163,7 +4163,7 @@ 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) @@ -4186,7 +4186,7 @@ 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, @@ -4230,14 +4230,14 @@ static int validate_req_change_req_state(struct drbd_conf *mdev, if (unlikely(!req)) { spin_unlock_irq(&mdev->req_lock); dev_err(DEV, "%s: got a corrupt block_id/sector pair\n", func); - return FALSE; + 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) @@ -4252,7 +4252,7 @@ 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: @@ -4273,7 +4273,7 @@ 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, @@ -4294,7 +4294,7 @@ static int got_NegAck(struct drbd_conf *mdev, struct p_header80 *h) 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); @@ -4332,7 +4332,7 @@ static int got_NegRSDReply(struct drbd_conf *mdev, struct p_header80 *h) put_ldev(mdev); } - return TRUE; + return true; } static int got_BarrierAck(struct drbd_conf *mdev, struct p_header80 *h) @@ -4349,7 +4349,7 @@ static int got_BarrierAck(struct drbd_conf *mdev, struct p_header80 *h) drbd_queue_work_front(&mdev->data.work, w); } - return TRUE; + return true; } static int got_OVResult(struct drbd_conf *mdev, struct p_header80 *h) @@ -4370,7 +4370,7 @@ 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); @@ -4393,12 +4393,12 @@ 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 { diff --git a/drivers/block/drbd/drbd_req.c b/drivers/block/drbd/drbd_req.c index 53e7cc50645..528909090df 100644 --- a/drivers/block/drbd/drbd_req.c +++ b/drivers/block/drbd/drbd_req.c @@ -445,7 +445,7 @@ int __req_mod(struct drbd_request *req, enum drbd_req_event what, req->rq_state |= RQ_LOCAL_COMPLETED; req->rq_state &= ~RQ_LOCAL_PENDING; - __drbd_chk_io_error(mdev, FALSE); + __drbd_chk_io_error(mdev, false); _req_may_be_done_not_susp(req, m); put_ldev(mdev); break; @@ -466,7 +466,7 @@ int __req_mod(struct drbd_request *req, enum drbd_req_event what, D_ASSERT(!(req->rq_state & RQ_NET_MASK)); - __drbd_chk_io_error(mdev, FALSE); + __drbd_chk_io_error(mdev, false); put_ldev(mdev); /* no point in retrying if there is no good remote data, diff --git a/drivers/block/drbd/drbd_worker.c b/drivers/block/drbd/drbd_worker.c index e201f6f82c0..9fe3e890da0 100644 --- a/drivers/block/drbd/drbd_worker.c +++ b/drivers/block/drbd/drbd_worker.c @@ -96,7 +96,7 @@ void drbd_endio_read_sec_final(struct drbd_epoch_entry *e) __releases(local) if (list_empty(&mdev->read_ee)) wake_up(&mdev->ee_wait); if (test_bit(__EE_WAS_ERROR, &e->flags)) - __drbd_chk_io_error(mdev, FALSE); + __drbd_chk_io_error(mdev, false); spin_unlock_irqrestore(&mdev->req_lock, flags); drbd_queue_work(&mdev->data.work, &e->w); @@ -139,7 +139,7 @@ static void drbd_endio_write_sec_final(struct drbd_epoch_entry *e) __releases(lo : list_empty(&mdev->active_ee); if (test_bit(__EE_WAS_ERROR, &e->flags)) - __drbd_chk_io_error(mdev, FALSE); + __drbd_chk_io_error(mdev, false); spin_unlock_irqrestore(&mdev->req_lock, flags); if (is_syncer_req) -- cgit v1.2.3-70-g09d2 From 19f843aa08e2d8f87a09b4c2edc43b00638423a8 Mon Sep 17 00:00:00 2001 From: Lars Ellenberg Date: Wed, 15 Dec 2010 08:59:11 +0100 Subject: drbd: bitmap keep track of changes vs on-disk bitmap When we set or clear bits in a bitmap page, also set a flag in the page->private pointer. This allows us to skip writes of unchanged pages. Signed-off-by: Philipp Reisner Signed-off-by: Lars Ellenberg --- drivers/block/drbd/drbd_actlog.c | 132 +++---------- drivers/block/drbd/drbd_bitmap.c | 409 +++++++++++++++++++++++++++++---------- drivers/block/drbd/drbd_int.h | 7 +- drivers/block/drbd/drbd_main.c | 27 ++- drivers/block/drbd/drbd_nl.c | 11 +- drivers/block/drbd/drbd_worker.c | 7 +- 6 files changed, 373 insertions(+), 220 deletions(-) (limited to 'drivers/block/drbd/drbd_worker.c') diff --git a/drivers/block/drbd/drbd_actlog.c b/drivers/block/drbd/drbd_actlog.c index e3f0f4d31d7..090fc2ce0df 100644 --- a/drivers/block/drbd/drbd_actlog.c +++ b/drivers/block/drbd/drbd_actlog.c @@ -262,6 +262,33 @@ void drbd_al_complete_io(struct drbd_conf *mdev, sector_t sector) spin_unlock_irqrestore(&mdev->al_lock, flags); } +#if (PAGE_SHIFT + 3) < (AL_EXTENT_SHIFT - BM_BLOCK_SHIFT) +/* Currently BM_BLOCK_SHIFT, BM_EXT_SHIFT and AL_EXTENT_SHIFT + * are still coupled, or assume too much about their relation. + * Code below will not work if this is violated. + * Will be cleaned up with some followup patch. + */ +# error FIXME +#endif + +static unsigned int al_extent_to_bm_page(unsigned int al_enr) +{ + return al_enr >> + /* bit to page */ + ((PAGE_SHIFT + 3) - + /* al extent number to bit */ + (AL_EXTENT_SHIFT - BM_BLOCK_SHIFT)); +} + +static unsigned int rs_extent_to_bm_page(unsigned int rs_enr) +{ + return rs_enr >> + /* bit to page */ + ((PAGE_SHIFT + 3) - + /* al extent number to bit */ + (BM_EXT_SHIFT - BM_BLOCK_SHIFT)); +} + int w_al_write_transaction(struct drbd_conf *mdev, struct drbd_work *w, int unused) { @@ -289,7 +316,7 @@ w_al_write_transaction(struct drbd_conf *mdev, struct drbd_work *w, int unused) * For now, we must not write the transaction, * if we cannot write out the bitmap of the evicted extent. */ if (mdev->state.conn < C_CONNECTED && evicted != LC_FREE) - drbd_bm_write_sect(mdev, evicted/AL_EXT_PER_BM_SECT); + drbd_bm_write_page(mdev, al_extent_to_bm_page(evicted)); /* The bitmap write may have failed, causing a state change. */ if (mdev->state.disk < D_INCONSISTENT) { @@ -635,105 +662,6 @@ out_bio_put: return err; } -/** - * drbd_al_to_on_disk_bm() - * Writes bitmap parts covered by active AL extents - * @mdev: DRBD device. - * - * Called when we detach (unconfigure) local storage, - * or when we go from R_PRIMARY to R_SECONDARY role. - */ -void drbd_al_to_on_disk_bm(struct drbd_conf *mdev) -{ - int i, nr_elements; - unsigned int enr; - struct bio **bios; - struct drbd_atodb_wait wc; - - ERR_IF (!get_ldev_if_state(mdev, D_ATTACHING)) - return; /* sorry, I don't have any act_log etc... */ - - wait_event(mdev->al_wait, lc_try_lock(mdev->act_log)); - - nr_elements = mdev->act_log->nr_elements; - - /* GFP_KERNEL, we are not in anyone's write-out path */ - bios = kzalloc(sizeof(struct bio *) * nr_elements, GFP_KERNEL); - if (!bios) - goto submit_one_by_one; - - atomic_set(&wc.count, 0); - init_completion(&wc.io_done); - wc.mdev = mdev; - wc.error = 0; - - for (i = 0; i < nr_elements; i++) { - enr = lc_element_by_index(mdev->act_log, i)->lc_number; - if (enr == LC_FREE) - continue; - /* next statement also does atomic_inc wc.count and local_cnt */ - if (atodb_prepare_unless_covered(mdev, bios, - enr/AL_EXT_PER_BM_SECT, - &wc)) - goto free_bios_submit_one_by_one; - } - - /* unnecessary optimization? */ - lc_unlock(mdev->act_log); - wake_up(&mdev->al_wait); - - /* all prepared, submit them */ - for (i = 0; i < nr_elements; i++) { - if (bios[i] == NULL) - break; - if (drbd_insert_fault(mdev, DRBD_FAULT_MD_WR)) { - bios[i]->bi_rw = WRITE; - bio_endio(bios[i], -EIO); - } else { - submit_bio(WRITE, bios[i]); - } - } - - /* always (try to) flush bitmap to stable storage */ - drbd_md_flush(mdev); - - /* In case we did not submit a single IO do not wait for - * them to complete. ( Because we would wait forever here. ) - * - * In case we had IOs and they are already complete, there - * is not point in waiting anyways. - * Therefore this if () ... */ - if (atomic_read(&wc.count)) - wait_for_completion(&wc.io_done); - - put_ldev(mdev); - - kfree(bios); - return; - - free_bios_submit_one_by_one: - /* free everything by calling the endio callback directly. */ - for (i = 0; i < nr_elements && bios[i]; i++) - bio_endio(bios[i], 0); - - kfree(bios); - - submit_one_by_one: - dev_warn(DEV, "Using the slow drbd_al_to_on_disk_bm()\n"); - - for (i = 0; i < mdev->act_log->nr_elements; i++) { - enr = lc_element_by_index(mdev->act_log, i)->lc_number; - if (enr == LC_FREE) - continue; - /* Really slow: if we have al-extents 16..19 active, - * sector 4 will be written four times! Synchronous! */ - drbd_bm_write_sect(mdev, enr/AL_EXT_PER_BM_SECT); - } - - lc_unlock(mdev->act_log); - wake_up(&mdev->al_wait); - put_ldev(mdev); -} - /** * drbd_al_apply_to_bm() - Sets the bitmap to diry(1) where covered ba active AL extents * @mdev: DRBD device. @@ -813,7 +741,7 @@ static int w_update_odbm(struct drbd_conf *mdev, struct drbd_work *w, int unused return 1; } - drbd_bm_write_sect(mdev, udw->enr); + drbd_bm_write_page(mdev, rs_extent_to_bm_page(udw->enr)); put_ldev(mdev); kfree(udw); @@ -893,7 +821,6 @@ static void drbd_try_clear_on_disk_bm(struct drbd_conf *mdev, sector_t sector, dev_warn(DEV, "Kicking resync_lru element enr=%u " "out with rs_failed=%d\n", ext->lce.lc_number, ext->rs_failed); - set_bit(WRITE_BM_AFTER_RESYNC, &mdev->flags); } ext->rs_left = rs_left; ext->rs_failed = success ? 0 : count; @@ -912,7 +839,6 @@ static void drbd_try_clear_on_disk_bm(struct drbd_conf *mdev, sector_t sector, drbd_queue_work_front(&mdev->data.work, &udw->w); } else { dev_warn(DEV, "Could not kmalloc an udw\n"); - set_bit(WRITE_BM_AFTER_RESYNC, &mdev->flags); } } } else { diff --git a/drivers/block/drbd/drbd_bitmap.c b/drivers/block/drbd/drbd_bitmap.c index 8d959ed6c2c..72cd41a96ef 100644 --- a/drivers/block/drbd/drbd_bitmap.c +++ b/drivers/block/drbd/drbd_bitmap.c @@ -70,8 +70,7 @@ struct drbd_bitmap { sector_t bm_dev_capacity; struct mutex bm_change; /* serializes resize operations */ - atomic_t bm_async_io; - wait_queue_head_t bm_io_wait; + wait_queue_head_t bm_io_wait; /* used to serialize IO of single pages */ unsigned long bm_flags; @@ -82,7 +81,7 @@ struct drbd_bitmap { /* definition of bits in bm_flags */ #define BM_LOCKED 0 -#define BM_MD_IO_ERROR 1 +// #define BM_MD_IO_ERROR 1 unused now. #define BM_P_VMALLOCED 2 static int __bm_change_bits_to(struct drbd_conf *mdev, const unsigned long s, @@ -155,26 +154,117 @@ void drbd_bm_unlock(struct drbd_conf *mdev) mutex_unlock(&b->bm_change); } -static unsigned int bm_bit_to_page_idx(struct drbd_bitmap *b, u64 bitnr) +/* we store some "meta" info about our pages in page->private */ +/* at a granularity of 4k storage per bitmap bit: + * one peta byte storage: 1<<50 byte, 1<<38 * 4k storage blocks + * 1<<38 bits, + * 1<<23 4k bitmap pages. + * Use 24 bits as page index, covers 2 peta byte storage + * at a granularity of 4k per bit. + * Used to report the failed page idx on io error from the endio handlers. + */ +#define BM_PAGE_IDX_MASK ((1UL<<24)-1) +/* this page is currently read in, or written back */ +#define BM_PAGE_IO_LOCK 31 +/* if there has been an IO error for this page */ +#define BM_PAGE_IO_ERROR 30 +/* this is to be able to intelligently skip disk IO, + * set if bits have been set since last IO. */ +#define BM_PAGE_NEED_WRITEOUT 29 +/* to mark for lazy writeout once syncer cleared all clearable bits, + * we if bits have been cleared since last IO. */ +#define BM_PAGE_LAZY_WRITEOUT 28 + +/* store_page_idx uses non-atomic assingment. It is only used directly after + * allocating the page. All other bm_set_page_* and bm_clear_page_* need to + * use atomic bit manipulation, as set_out_of_sync (and therefore bitmap + * changes) may happen from various contexts, and wait_on_bit/wake_up_bit + * requires it all to be atomic as well. */ +static void bm_store_page_idx(struct page *page, unsigned long idx) { - /* page_nr = (bitnr/8) >> PAGE_SHIFT; */ - unsigned int page_nr = bitnr >> (PAGE_SHIFT + 3); - BUG_ON(page_nr >= b->bm_number_of_pages); - return page_nr; + BUG_ON(0 != (idx & ~BM_PAGE_IDX_MASK)); + page_private(page) |= idx; } -/* word offset to long pointer */ -static unsigned long *__bm_map_paddr(struct drbd_bitmap *b, unsigned long offset, const enum km_type km) +static unsigned long bm_page_to_idx(struct page *page) { - struct page *page; - unsigned long page_nr; + return page_private(page) & BM_PAGE_IDX_MASK; +} + +/* As is very unlikely that the same page is under IO from more than one + * context, we can get away with a bit per page and one wait queue per bitmap. + */ +static void bm_page_lock_io(struct drbd_conf *mdev, int page_nr) +{ + struct drbd_bitmap *b = mdev->bitmap; + void *addr = &page_private(b->bm_pages[page_nr]); + wait_event(b->bm_io_wait, !test_and_set_bit(BM_PAGE_IO_LOCK, addr)); +} + +static void bm_page_unlock_io(struct drbd_conf *mdev, int page_nr) +{ + struct drbd_bitmap *b = mdev->bitmap; + void *addr = &page_private(b->bm_pages[page_nr]); + clear_bit(BM_PAGE_IO_LOCK, addr); + smp_mb__after_clear_bit(); + wake_up(&mdev->bitmap->bm_io_wait); +} + +/* set _before_ submit_io, so it may be reset due to being changed + * while this page is in flight... will get submitted later again */ +static void bm_set_page_unchanged(struct page *page) +{ + /* use cmpxchg? */ + clear_bit(BM_PAGE_NEED_WRITEOUT, &page_private(page)); + clear_bit(BM_PAGE_LAZY_WRITEOUT, &page_private(page)); +} + +static void bm_set_page_need_writeout(struct page *page) +{ + set_bit(BM_PAGE_NEED_WRITEOUT, &page_private(page)); +} + +static int bm_test_page_unchanged(struct page *page) +{ + volatile const unsigned long *addr = &page_private(page); + return (*addr & ((1UL<> PAGE_SHIFT; */ - page_nr = offset >> (PAGE_SHIFT - LN2_BPL + 3); + unsigned int page_nr = long_nr >> (PAGE_SHIFT - LN2_BPL + 3); BUG_ON(page_nr >= b->bm_number_of_pages); - page = b->bm_pages[page_nr]; + return page_nr; +} - return (unsigned long *) kmap_atomic(page, km); +static unsigned int bm_bit_to_page_idx(struct drbd_bitmap *b, u64 bitnr) +{ + /* page_nr = (bitnr/8) >> PAGE_SHIFT; */ + unsigned int page_nr = bitnr >> (PAGE_SHIFT + 3); + BUG_ON(page_nr >= b->bm_number_of_pages); + return page_nr; } static unsigned long *__bm_map_pidx(struct drbd_bitmap *b, unsigned int idx, const enum km_type km) @@ -188,11 +278,6 @@ static unsigned long *bm_map_pidx(struct drbd_bitmap *b, unsigned int idx) return __bm_map_pidx(b, idx, KM_IRQ1); } -static unsigned long * bm_map_paddr(struct drbd_bitmap *b, unsigned long offset) -{ - return __bm_map_paddr(b, offset, KM_IRQ1); -} - static void __bm_unmap(unsigned long *p_addr, const enum km_type km) { kunmap_atomic(p_addr, km); @@ -222,6 +307,7 @@ static void bm_unmap(unsigned long *p_addr) * to be able to report device specific. */ + static void bm_free_pages(struct page **pages, unsigned long number) { unsigned long i; @@ -289,6 +375,9 @@ static struct page **bm_realloc_pages(struct drbd_bitmap *b, unsigned long want) bm_vk_free(new_pages, vmalloced); return NULL; } + /* we want to know which page it is + * from the endio handlers */ + bm_store_page_idx(page, i); new_pages[i] = page; } } else { @@ -443,7 +532,7 @@ static unsigned long bm_count_bits(struct drbd_bitmap *b) while (offset < words) { i = do_now = min_t(size_t, words-offset, LWPP); - p_addr = __bm_map_paddr(b, offset, KM_USER0); + p_addr = __bm_map_pidx(b, bm_word_to_page_idx(b, offset), KM_USER0); bm = p_addr + MLPP(offset); while (i--) { bits += hweight_long(*bm++); @@ -472,6 +561,7 @@ static unsigned long bm_count_bits(struct drbd_bitmap *b) static void bm_memset(struct drbd_bitmap *b, size_t offset, int c, size_t len) { unsigned long *p_addr, *bm; + unsigned int idx; size_t do_now, end; #define BM_SECTORS_PER_BIT (BM_BLOCK_SIZE/512) @@ -485,7 +575,8 @@ static void bm_memset(struct drbd_bitmap *b, size_t offset, int c, size_t len) while (offset < end) { do_now = min_t(size_t, ALIGN(offset + 1, LWPP), end) - offset; - p_addr = bm_map_paddr(b, offset); + idx = bm_word_to_page_idx(b, offset); + p_addr = bm_map_pidx(b, idx); bm = p_addr + MLPP(offset); if (bm+do_now > p_addr + LWPP) { printk(KERN_ALERT "drbd: BUG BUG BUG! p_addr:%p bm:%p do_now:%d\n", @@ -494,6 +585,7 @@ static void bm_memset(struct drbd_bitmap *b, size_t offset, int c, size_t len) } memset(bm, c, do_now * sizeof(long)); bm_unmap(p_addr); + bm_set_page_need_writeout(b->bm_pages[idx]); offset += do_now; } } @@ -604,7 +696,7 @@ int drbd_bm_resize(struct drbd_conf *mdev, sector_t capacity, int set_new_bits) bm_free_pages(opages + want, have - want); } - p_addr = bm_map_paddr(b, words); + p_addr = bm_map_pidx(b, bm_word_to_page_idx(b, words)); bm = p_addr + MLPP(words); *bm = DRBD_MAGIC; bm_unmap(p_addr); @@ -616,7 +708,7 @@ int drbd_bm_resize(struct drbd_conf *mdev, sector_t capacity, int set_new_bits) bm_vk_free(opages, opages_vmalloced); if (!growing) b->bm_set = bm_count_bits(b); - dev_info(DEV, "resync bitmap: bits=%lu words=%lu\n", bits, words); + dev_info(DEV, "resync bitmap: bits=%lu words=%lu pages=%lu\n", bits, words, want); out: drbd_bm_unlock(mdev); @@ -686,6 +778,7 @@ void drbd_bm_merge_lel(struct drbd_conf *mdev, size_t offset, size_t number, struct drbd_bitmap *b = mdev->bitmap; unsigned long *p_addr, *bm; unsigned long word, bits; + unsigned int idx; size_t end, do_now; end = offset + number; @@ -700,7 +793,8 @@ void drbd_bm_merge_lel(struct drbd_conf *mdev, size_t offset, size_t number, spin_lock_irq(&b->bm_lock); while (offset < end) { do_now = min_t(size_t, ALIGN(offset+1, LWPP), end) - offset; - p_addr = bm_map_paddr(b, offset); + idx = bm_word_to_page_idx(b, offset); + p_addr = bm_map_pidx(b, idx); bm = p_addr + MLPP(offset); offset += do_now; while (do_now--) { @@ -710,6 +804,7 @@ void drbd_bm_merge_lel(struct drbd_conf *mdev, size_t offset, size_t number, b->bm_set += hweight_long(word) - bits; } bm_unmap(p_addr); + bm_set_page_need_writeout(b->bm_pages[idx]); } /* with 32bit <-> 64bit cross-platform connect * this is only correct for current usage, @@ -748,7 +843,7 @@ void drbd_bm_get_lel(struct drbd_conf *mdev, size_t offset, size_t number, else { while (offset < end) { do_now = min_t(size_t, ALIGN(offset+1, LWPP), end) - offset; - p_addr = bm_map_paddr(b, offset); + p_addr = bm_map_pidx(b, bm_word_to_page_idx(b, offset)); bm = p_addr + MLPP(offset); offset += do_now; while (do_now--) @@ -786,9 +881,22 @@ void drbd_bm_clear_all(struct drbd_conf *mdev) spin_unlock_irq(&b->bm_lock); } +struct bm_aio_ctx { + struct drbd_conf *mdev; + atomic_t in_flight; + wait_queue_head_t io_wait; + unsigned flags; +#define BM_AIO_COPY_PAGES 1 + int error; +}; + +/* bv_page may be a copy, or may be the original */ static void bm_async_io_complete(struct bio *bio, int error) { - struct drbd_bitmap *b = bio->bi_private; + struct bm_aio_ctx *ctx = bio->bi_private; + struct drbd_conf *mdev = ctx->mdev; + struct drbd_bitmap *b = mdev->bitmap; + unsigned int idx = bm_page_to_idx(bio->bi_io_vec[0].bv_page); int uptodate = bio_flagged(bio, BIO_UPTODATE); @@ -799,35 +907,79 @@ static void bm_async_io_complete(struct bio *bio, int error) if (!error && !uptodate) error = -EIO; + if (!bm_test_page_unchanged(b->bm_pages[idx])) + dev_info(DEV, "bitmap page idx %u changed during IO!\n", idx); + if (error) { - /* doh. what now? - * for now, set all bits, and flag MD_IO_ERROR */ - __set_bit(BM_MD_IO_ERROR, &b->bm_flags); + /* ctx error will hold the completed-last non-zero error code, + * in case error codes differ. */ + ctx->error = error; + bm_set_page_io_err(b->bm_pages[idx]); + /* Not identical to on disk version of it. + * Is BM_PAGE_IO_ERROR enough? */ + if (__ratelimit(&drbd_ratelimit_state)) + dev_err(DEV, "IO ERROR %d on bitmap page idx %u\n", + error, idx); + } else { + bm_clear_page_io_err(b->bm_pages[idx]); + dynamic_dev_dbg(DEV, "bitmap page idx %u completed\n", idx); } - if (atomic_dec_and_test(&b->bm_async_io)) - wake_up(&b->bm_io_wait); + + bm_page_unlock_io(mdev, idx); + + /* FIXME give back to page pool */ + if (ctx->flags & BM_AIO_COPY_PAGES) + put_page(bio->bi_io_vec[0].bv_page); bio_put(bio); + + if (atomic_dec_and_test(&ctx->in_flight)) + wake_up(&ctx->io_wait); } -static void bm_page_io_async(struct drbd_conf *mdev, struct drbd_bitmap *b, int page_nr, int rw) __must_hold(local) +static void bm_page_io_async(struct bm_aio_ctx *ctx, int page_nr, int rw) __must_hold(local) { /* we are process context. we always get a bio */ struct bio *bio = bio_alloc(GFP_KERNEL, 1); + struct drbd_conf *mdev = ctx->mdev; + struct drbd_bitmap *b = mdev->bitmap; + struct page *page; unsigned int len; + sector_t on_disk_sector = mdev->ldev->md.md_offset + mdev->ldev->md.bm_offset; on_disk_sector += ((sector_t)page_nr) << (PAGE_SHIFT-9); /* this might happen with very small - * flexible external meta data device */ + * flexible external meta data device, + * or with PAGE_SIZE > 4k */ len = min_t(unsigned int, PAGE_SIZE, (drbd_md_last_sector(mdev->ldev) - on_disk_sector + 1)<<9); + /* serialize IO on this page */ + bm_page_lock_io(mdev, page_nr); + /* before memcpy and submit, + * so it can be redirtied any time */ + bm_set_page_unchanged(b->bm_pages[page_nr]); + + if (ctx->flags & BM_AIO_COPY_PAGES) { + /* FIXME alloc_page is good enough for now, but actually needs + * to use pre-allocated page pool */ + void *src, *dest; + page = alloc_page(__GFP_HIGHMEM|__GFP_WAIT); + dest = kmap_atomic(page, KM_USER0); + src = kmap_atomic(b->bm_pages[page_nr], KM_USER1); + memcpy(dest, src, PAGE_SIZE); + kunmap_atomic(src, KM_USER1); + kunmap_atomic(dest, KM_USER0); + bm_store_page_idx(page, page_nr); + } else + page = b->bm_pages[page_nr]; + bio->bi_bdev = mdev->ldev->md_bdev; bio->bi_sector = on_disk_sector; - bio_add_page(bio, b->bm_pages[page_nr], len, 0); - bio->bi_private = b; + bio_add_page(bio, page, len, 0); + bio->bi_private = ctx; bio->bi_end_io = bm_async_io_complete; if (drbd_insert_fault(mdev, (rw & WRITE) ? DRBD_FAULT_MD_WR : DRBD_FAULT_MD_RD)) { @@ -841,36 +993,72 @@ static void bm_page_io_async(struct drbd_conf *mdev, struct drbd_bitmap *b, int /* * bm_rw: read/write the whole bitmap from/to its on disk location. */ -static int bm_rw(struct drbd_conf *mdev, int rw) __must_hold(local) +static int bm_rw(struct drbd_conf *mdev, int rw, unsigned lazy_writeout_upper_idx) __must_hold(local) { + struct bm_aio_ctx ctx = + { .flags = lazy_writeout_upper_idx ? BM_AIO_COPY_PAGES : 0 }; struct drbd_bitmap *b = mdev->bitmap; - /* sector_t sector; */ - int bm_words, num_pages, i; + int last_page, i, count = 0; unsigned long now; char ppb[10]; int err = 0; - WARN_ON(!bm_is_locked(b)); - - /* no spinlock here, the drbd_bm_lock should be enough! */ + /* + * We are protected against bitmap disappearing/resizing by holding an + * ldev reference (caller must have called get_ldev()). + * For read/write, we are protected against changes to the bitmap by + * the bitmap lock (see drbd_bitmap_io). + * For lazy writeout, we don't care for ongoing changes to the bitmap, + * as we submit copies of pages anyways. + */ + if (!ctx.flags) + WARN_ON(!bm_is_locked(b)); - bm_words = drbd_bm_words(mdev); - num_pages = (bm_words*sizeof(long) + PAGE_SIZE-1) >> PAGE_SHIFT; + /* because of the "extra long to catch oob access" we allocate in + * drbd_bm_resize, bm_number_of_pages -1 is not necessarily the page + * containing the last _relevant_ bitmap word */ + last_page = bm_word_to_page_idx(b, b->bm_words - 1); now = jiffies; - atomic_set(&b->bm_async_io, num_pages); - __clear_bit(BM_MD_IO_ERROR, &b->bm_flags); + ctx.mdev = mdev; + atomic_set(&ctx.in_flight, 1); /* one extra ref */ + init_waitqueue_head(&ctx.io_wait); + ctx.error = 0; /* let the layers below us try to merge these bios... */ - for (i = 0; i < num_pages; i++) - bm_page_io_async(mdev, b, i, rw); + for (i = 0; i <= last_page; i++) { + /* ignore completely unchanged pages */ + if (lazy_writeout_upper_idx && i == lazy_writeout_upper_idx) + break; + if (rw & WRITE) { + if (bm_test_page_unchanged(b->bm_pages[i])) { + dynamic_dev_dbg(DEV, "skipped bm write for idx %u\n", i); + continue; + } + /* during lazy writeout, + * ignore those pages not marked for lazy writeout. */ + if (lazy_writeout_upper_idx && + !bm_test_page_lazy_writeout(b->bm_pages[i])) { + dynamic_dev_dbg(DEV, "skipped bm lazy write for idx %u\n", i); + continue; + } + } + atomic_inc(&ctx.in_flight); + bm_page_io_async(&ctx, i, rw); + ++count; + cond_resched(); + } - wait_event(b->bm_io_wait, atomic_read(&b->bm_async_io) == 0); + atomic_dec(&ctx.in_flight); /* drop the extra ref */ + wait_event(ctx.io_wait, atomic_read(&ctx.in_flight) == 0); + dev_info(DEV, "bitmap %s of %u pages took %lu jiffies\n", + rw == WRITE ? "WRITE" : "READ", + count, jiffies - now); - if (test_bit(BM_MD_IO_ERROR, &b->bm_flags)) { + if (ctx.error) { dev_alert(DEV, "we had at least one MD IO ERROR during bitmap IO\n"); drbd_chk_io_error(mdev, 1, true); - err = -EIO; + err = -EIO; /* ctx.error ? */ } now = jiffies; @@ -895,55 +1083,63 @@ static int bm_rw(struct drbd_conf *mdev, int rw) __must_hold(local) */ int drbd_bm_read(struct drbd_conf *mdev) __must_hold(local) { - return bm_rw(mdev, READ); + return bm_rw(mdev, READ, 0); } /** * drbd_bm_write() - Write the whole bitmap to its on disk location. * @mdev: DRBD device. + * + * Will only write pages that have changed since last IO. */ int drbd_bm_write(struct drbd_conf *mdev) __must_hold(local) { - return bm_rw(mdev, WRITE); + return bm_rw(mdev, WRITE, 0); } /** - * drbd_bm_write_sect: Writes a 512 (MD_SECTOR_SIZE) byte piece of the bitmap + * drbd_bm_lazy_write_out() - Write bitmap pages 0 to @upper_idx-1, if they have changed. * @mdev: DRBD device. - * @enr: Extent number in the resync lru (happens to be sector offset) + * @upper_idx: 0: write all changed pages; +ve: page index to stop scanning for changed pages + */ +int drbd_bm_write_lazy(struct drbd_conf *mdev, unsigned upper_idx) __must_hold(local) +{ + return bm_rw(mdev, WRITE, upper_idx); +} + + +/** + * drbd_bm_write_page: Writes a PAGE_SIZE aligned piece of bitmap + * @mdev: DRBD device. + * @idx: bitmap page index * - * The BM_EXT_SIZE is on purpose exactly the amount of the bitmap covered - * by a single sector write. Therefore enr == sector offset from the - * start of the bitmap. + * We don't want to special case on logical_block_size of the underlaying + * device, so we submit PAGE_SIZE aligned pieces containing the requested enr. + * Note that on "most" systems, PAGE_SIZE is 4k. */ -int drbd_bm_write_sect(struct drbd_conf *mdev, unsigned long enr) __must_hold(local) +int drbd_bm_write_page(struct drbd_conf *mdev, unsigned int idx) __must_hold(local) { - sector_t on_disk_sector = enr + mdev->ldev->md.md_offset - + mdev->ldev->md.bm_offset; - int bm_words, num_words, offset; - int err = 0; + struct bm_aio_ctx ctx = { .flags = BM_AIO_COPY_PAGES, }; - mutex_lock(&mdev->md_io_mutex); - bm_words = drbd_bm_words(mdev); - offset = S2W(enr); /* word offset into bitmap */ - num_words = min(S2W(1), bm_words - offset); - if (num_words < S2W(1)) - memset(page_address(mdev->md_io_page), 0, MD_SECTOR_SIZE); - drbd_bm_get_lel(mdev, offset, num_words, - page_address(mdev->md_io_page)); - if (!drbd_md_sync_page_io(mdev, mdev->ldev, on_disk_sector, WRITE)) { - int i; - err = -EIO; - dev_err(DEV, "IO ERROR writing bitmap sector %lu " - "(meta-disk sector %llus)\n", - enr, (unsigned long long)on_disk_sector); - drbd_chk_io_error(mdev, 1, true); - for (i = 0; i < AL_EXT_PER_BM_SECT; i++) - drbd_bm_ALe_set_all(mdev, enr*AL_EXT_PER_BM_SECT+i); + if (bm_test_page_unchanged(mdev->bitmap->bm_pages[idx])) { + dev_info(DEV, "skipped bm page write for idx %u\n", idx); + return 0; } + + ctx.mdev = mdev; + atomic_set(&ctx.in_flight, 1); + init_waitqueue_head(&ctx.io_wait); + + bm_page_io_async(&ctx, idx, WRITE_SYNC); + wait_event(ctx.io_wait, atomic_read(&ctx.in_flight) == 0); + + if (ctx.error) + drbd_chk_io_error(mdev, 1, true); + /* that should force detach, so the in memory bitmap will be + * gone in a moment as well. */ + mdev->bm_writ_cnt++; - mutex_unlock(&mdev->md_io_mutex); - return err; + return ctx.error; } /* NOTE @@ -965,10 +1161,9 @@ static unsigned long __bm_find_next(struct drbd_conf *mdev, unsigned long bm_fo, dev_err(DEV, "bm_fo=%lu bm_bits=%lu\n", bm_fo, b->bm_bits); } else { while (bm_fo < b->bm_bits) { - unsigned long offset; - bit_offset = bm_fo & ~BPP_MASK; /* bit offset of the page */ - offset = bit_offset >> LN2_BPL; /* word offset of the page */ - p_addr = __bm_map_paddr(b, offset, km); + /* bit offset of the first bit in the page */ + bit_offset = bm_fo & ~BPP_MASK; + p_addr = __bm_map_pidx(b, bm_bit_to_page_idx(b, bm_fo), km); if (find_zero_bit) i = generic_find_next_zero_le_bit(p_addr, PAGE_SIZE*8, bm_fo & BPP_MASK); @@ -1048,8 +1243,9 @@ static int __bm_change_bits_to(struct drbd_conf *mdev, const unsigned long s, struct drbd_bitmap *b = mdev->bitmap; unsigned long *p_addr = NULL; unsigned long bitnr; - unsigned long last_page_nr = -1UL; + unsigned int last_page_nr = -1U; int c = 0; + int changed_total = 0; if (e >= b->bm_bits) { dev_err(DEV, "ASSERT FAILED: bit_s=%lu bit_e=%lu bm_bits=%lu\n", @@ -1057,12 +1253,17 @@ static int __bm_change_bits_to(struct drbd_conf *mdev, const unsigned long s, e = b->bm_bits ? b->bm_bits -1 : 0; } for (bitnr = s; bitnr <= e; bitnr++) { - unsigned long offset = bitnr>>LN2_BPL; - unsigned long page_nr = offset >> (PAGE_SHIFT - LN2_BPL + 3); + unsigned int page_nr = bm_bit_to_page_idx(b, bitnr); if (page_nr != last_page_nr) { if (p_addr) __bm_unmap(p_addr, km); - p_addr = __bm_map_paddr(b, offset, km); + if (c < 0) + bm_set_page_lazy_writeout(b->bm_pages[last_page_nr]); + else if (c > 0) + bm_set_page_need_writeout(b->bm_pages[last_page_nr]); + changed_total += c; + c = 0; + p_addr = __bm_map_pidx(b, page_nr, km); last_page_nr = page_nr; } if (val) @@ -1072,8 +1273,13 @@ static int __bm_change_bits_to(struct drbd_conf *mdev, const unsigned long s, } if (p_addr) __bm_unmap(p_addr, km); - b->bm_set += c; - return c; + if (c < 0) + bm_set_page_lazy_writeout(b->bm_pages[last_page_nr]); + else if (c > 0) + bm_set_page_need_writeout(b->bm_pages[last_page_nr]); + changed_total += c; + b->bm_set += changed_total; + return changed_total; } /* returns number of bits actually changed. @@ -1211,8 +1417,7 @@ int drbd_bm_test_bit(struct drbd_conf *mdev, const unsigned long bitnr) if (bm_is_locked(b)) bm_print_lock_info(mdev); if (bitnr < b->bm_bits) { - unsigned long offset = bitnr>>LN2_BPL; - p_addr = bm_map_paddr(b, offset); + p_addr = bm_map_pidx(b, bm_bit_to_page_idx(b, bitnr)); i = generic_test_le_bit(bitnr & BPP_MASK, p_addr) ? 1 : 0; bm_unmap(p_addr); } else if (bitnr == b->bm_bits) { @@ -1231,10 +1436,10 @@ int drbd_bm_count_bits(struct drbd_conf *mdev, const unsigned long s, const unsi { unsigned long flags; struct drbd_bitmap *b = mdev->bitmap; - unsigned long *p_addr = NULL, page_nr = -1; + unsigned long *p_addr = NULL; unsigned long bitnr; + unsigned int page_nr = -1U; int c = 0; - size_t w; /* If this is called without a bitmap, that is a bug. But just to be * robust in case we screwed up elsewhere, in that case pretend there @@ -1247,12 +1452,12 @@ int drbd_bm_count_bits(struct drbd_conf *mdev, const unsigned long s, const unsi if (bm_is_locked(b)) bm_print_lock_info(mdev); for (bitnr = s; bitnr <= e; bitnr++) { - w = bitnr >> LN2_BPL; - if (page_nr != w >> (PAGE_SHIFT - LN2_BPL + 3)) { - page_nr = w >> (PAGE_SHIFT - LN2_BPL + 3); + unsigned int idx = bm_bit_to_page_idx(b, bitnr); + if (page_nr != idx) { + page_nr = idx; if (p_addr) bm_unmap(p_addr); - p_addr = bm_map_paddr(b, w); + p_addr = bm_map_pidx(b, idx); } ERR_IF (bitnr >= b->bm_bits) { dev_err(DEV, "bitnr=%lu bm_bits=%lu\n", bitnr, b->bm_bits); @@ -1300,7 +1505,7 @@ int drbd_bm_e_weight(struct drbd_conf *mdev, unsigned long enr) count = 0; if (s < b->bm_words) { int n = e-s; - p_addr = bm_map_paddr(b, s); + p_addr = bm_map_pidx(b, bm_word_to_page_idx(b, s)); bm = p_addr + MLPP(s); while (n--) count += hweight_long(*bm++); @@ -1335,7 +1540,7 @@ unsigned long drbd_bm_ALe_set_all(struct drbd_conf *mdev, unsigned long al_enr) count = 0; if (s < b->bm_words) { i = do_now = e-s; - p_addr = bm_map_paddr(b, s); + p_addr = bm_map_pidx(b, bm_word_to_page_idx(b, s)); bm = p_addr + MLPP(s); while (i--) { count += hweight_long(*bm); diff --git a/drivers/block/drbd/drbd_int.h b/drivers/block/drbd/drbd_int.h index f6da48bb8c7..74cc50a2182 100644 --- a/drivers/block/drbd/drbd_int.h +++ b/drivers/block/drbd/drbd_int.h @@ -833,7 +833,7 @@ enum { CRASHED_PRIMARY, /* This node was a crashed primary. * Gets cleared when the state.conn * goes into C_CONNECTED state. */ - WRITE_BM_AFTER_RESYNC, /* A kmalloc() during resync failed */ + NO_BARRIER_SUPP, /* underlying block device doesn't implement barriers */ CONSIDER_RESYNC, MD_NO_FUA, /* Users wants us to not use FUA/FLUSH on meta data dev */ @@ -1428,7 +1428,7 @@ extern void _drbd_bm_set_bits(struct drbd_conf *mdev, const unsigned long s, const unsigned long e); extern int drbd_bm_test_bit(struct drbd_conf *mdev, unsigned long bitnr); extern int drbd_bm_e_weight(struct drbd_conf *mdev, unsigned long enr); -extern int drbd_bm_write_sect(struct drbd_conf *mdev, unsigned long enr) __must_hold(local); +extern int drbd_bm_write_page(struct drbd_conf *mdev, unsigned int idx) __must_hold(local); extern int drbd_bm_read(struct drbd_conf *mdev) __must_hold(local); extern int drbd_bm_write(struct drbd_conf *mdev) __must_hold(local); extern unsigned long drbd_bm_ALe_set_all(struct drbd_conf *mdev, @@ -1446,7 +1446,7 @@ extern int drbd_bm_rs_done(struct drbd_conf *mdev); /* for receive_bitmap */ extern void drbd_bm_merge_lel(struct drbd_conf *mdev, size_t offset, size_t number, unsigned long *buffer); -/* for _drbd_send_bitmap and drbd_bm_write_sect */ +/* for _drbd_send_bitmap */ extern void drbd_bm_get_lel(struct drbd_conf *mdev, size_t offset, size_t number, unsigned long *buffer); @@ -1641,7 +1641,6 @@ extern int __drbd_set_out_of_sync(struct drbd_conf *mdev, sector_t sector, #define drbd_set_out_of_sync(mdev, sector, size) \ __drbd_set_out_of_sync(mdev, sector, size, __FILE__, __LINE__) extern void drbd_al_apply_to_bm(struct drbd_conf *mdev); -extern void drbd_al_to_on_disk_bm(struct drbd_conf *mdev); extern void drbd_al_shrink(struct drbd_conf *mdev); diff --git a/drivers/block/drbd/drbd_main.c b/drivers/block/drbd/drbd_main.c index 67fffad213e..57ed7181742 100644 --- a/drivers/block/drbd/drbd_main.c +++ b/drivers/block/drbd/drbd_main.c @@ -1289,6 +1289,26 @@ static void abw_start_sync(struct drbd_conf *mdev, int rv) } } +int drbd_bitmap_io_from_worker(struct drbd_conf *mdev, int (*io_fn)(struct drbd_conf *), char *why) +{ + int rv; + + D_ASSERT(current == mdev->worker.task); + + /* open coded non-blocking drbd_suspend_io(mdev); */ + set_bit(SUSPEND_IO, &mdev->flags); + if (!is_susp(mdev->state)) + D_ASSERT(atomic_read(&mdev->ap_bio_cnt) == 0); + + drbd_bm_lock(mdev, why); + rv = io_fn(mdev); + drbd_bm_unlock(mdev); + + drbd_resume_io(mdev); + + return rv; +} + /** * after_state_ch() - Perform after state change actions that may sleep * @mdev: DRBD device. @@ -1404,7 +1424,12 @@ static void after_state_ch(struct drbd_conf *mdev, union drbd_state os, /* D_DISKLESS Peer becomes secondary */ if (os.peer == R_PRIMARY && ns.peer == R_SECONDARY) - drbd_al_to_on_disk_bm(mdev); + drbd_bitmap_io_from_worker(mdev, &drbd_bm_write, "demote diskless peer"); + put_ldev(mdev); + } + + if (os.role == R_PRIMARY && ns.role == R_SECONDARY && get_ldev(mdev)) { + drbd_bitmap_io_from_worker(mdev, &drbd_bm_write, "demote"); put_ldev(mdev); } diff --git a/drivers/block/drbd/drbd_nl.c b/drivers/block/drbd/drbd_nl.c index 2f072498214..77dc022eaf6 100644 --- a/drivers/block/drbd/drbd_nl.c +++ b/drivers/block/drbd/drbd_nl.c @@ -407,10 +407,8 @@ drbd_set_role(struct drbd_conf *mdev, enum drbd_role new_role, int force) } } - if ((new_role == R_SECONDARY) && get_ldev(mdev)) { - drbd_al_to_on_disk_bm(mdev); - put_ldev(mdev); - } + /* writeout of activity log covered areas of the bitmap + * to stable storage done in after state change already */ if (mdev->state.conn >= C_WF_REPORT_PARAMS) { /* if this was forced, we should consider sync */ @@ -1174,7 +1172,10 @@ static int drbd_nl_disk_conf(struct drbd_conf *mdev, struct drbd_nl_cfg_req *nlp if (cp_discovered) { drbd_al_apply_to_bm(mdev); - drbd_al_to_on_disk_bm(mdev); + if (drbd_bitmap_io(mdev, &drbd_bm_write, "crashed primary apply AL")) { + retcode = ERR_IO_MD_DISK; + goto force_diskless_dec; + } } if (_drbd_bm_total_weight(mdev) == drbd_bm_bits(mdev)) diff --git a/drivers/block/drbd/drbd_worker.c b/drivers/block/drbd/drbd_worker.c index 9fe3e890da0..d17f2ed777c 100644 --- a/drivers/block/drbd/drbd_worker.c +++ b/drivers/block/drbd/drbd_worker.c @@ -907,10 +907,8 @@ out: drbd_md_sync(mdev); - if (test_and_clear_bit(WRITE_BM_AFTER_RESYNC, &mdev->flags)) { - dev_info(DEV, "Writing the whole bitmap\n"); - drbd_queue_bitmap_io(mdev, &drbd_bm_write, NULL, "write from resync_finished"); - } + dev_info(DEV, "Writing changed bitmap pages\n"); + drbd_queue_bitmap_io(mdev, &drbd_bm_write, NULL, "write from resync_finished"); if (khelper_cmd) drbd_khelper(mdev, khelper_cmd); @@ -1127,7 +1125,6 @@ void drbd_ov_oos_found(struct drbd_conf *mdev, sector_t sector, int size) mdev->ov_last_oos_size = size>>9; } drbd_set_out_of_sync(mdev, sector, size); - set_bit(WRITE_BM_AFTER_RESYNC, &mdev->flags); } int w_e_end_ov_reply(struct drbd_conf *mdev, struct drbd_work *w, int cancel) -- cgit v1.2.3-70-g09d2 From 4b0715f09655e76ca24c35a9e25e7c464c2f7346 Mon Sep 17 00:00:00 2001 From: Lars Ellenberg Date: Tue, 14 Dec 2010 15:13:04 +0100 Subject: drbd: allow petabyte storage on 64bit arch Signed-off-by: Philipp Reisner Signed-off-by: Lars Ellenberg --- drivers/block/drbd/drbd_bitmap.c | 170 ++++++++++++++++++++++++--------------- drivers/block/drbd/drbd_int.h | 41 ++++++---- drivers/block/drbd/drbd_nl.c | 8 +- drivers/block/drbd/drbd_proc.c | 6 +- drivers/block/drbd/drbd_worker.c | 2 +- 5 files changed, 142 insertions(+), 85 deletions(-) (limited to 'drivers/block/drbd/drbd_worker.c') diff --git a/drivers/block/drbd/drbd_bitmap.c b/drivers/block/drbd/drbd_bitmap.c index 72cd41a96ef..0e31e573af7 100644 --- a/drivers/block/drbd/drbd_bitmap.c +++ b/drivers/block/drbd/drbd_bitmap.c @@ -37,10 +37,46 @@ * convention: * function name drbd_bm_... => used elsewhere, "public". * function name bm_... => internal to implementation, "private". + */ + - * Note that since find_first_bit returns int, at the current granularity of - * the bitmap (4KB per byte), this implementation "only" supports up to - * 1<<(32+12) == 16 TB... +/* + * LIMITATIONS: + * We want to support >= peta byte of backend storage, while for now still using + * a granularity of one bit per 4KiB of storage. + * 1 << 50 bytes backend storage (1 PiB) + * 1 << (50 - 12) bits needed + * 38 --> we need u64 to index and count bits + * 1 << (38 - 3) bitmap bytes needed + * 35 --> we still need u64 to index and count bytes + * (that's 32 GiB of bitmap for 1 PiB storage) + * 1 << (35 - 2) 32bit longs needed + * 33 --> we'd even need u64 to index and count 32bit long words. + * 1 << (35 - 3) 64bit longs needed + * 32 --> we could get away with a 32bit unsigned int to index and count + * 64bit long words, but I rather stay with unsigned long for now. + * We probably should neither count nor point to bytes or long words + * directly, but either by bitnumber, or by page index and offset. + * 1 << (35 - 12) + * 22 --> we need that much 4KiB pages of bitmap. + * 1 << (22 + 3) --> on a 64bit arch, + * we need 32 MiB to store the array of page pointers. + * + * Because I'm lazy, and because the resulting patch was too large, too ugly + * and still incomplete, on 32bit we still "only" support 16 TiB (minus some), + * (1 << 32) bits * 4k storage. + * + + * bitmap storage and IO: + * Bitmap is stored little endian on disk, and is kept little endian in + * core memory. Currently we still hold the full bitmap in core as long + * as we are "attached" to a local disk, which at 32 GiB for 1PiB storage + * seems excessive. + * + * We plan to reduce the amount of in-core bitmap pages by pageing them in + * and out against their on-disk location as necessary, but need to make + * sure we don't cause too much meta data IO, and must not deadlock in + * tight memory situations. This needs some more work. */ /* @@ -56,13 +92,9 @@ struct drbd_bitmap { struct page **bm_pages; spinlock_t bm_lock; - /* WARNING unsigned long bm_*: - * 32bit number of bit offset is just enough for 512 MB bitmap. - * it will blow up if we make the bitmap bigger... - * not that it makes much sense to have a bitmap that large, - * rather change the granularity to 16k or 64k or something. - * (that implies other problems, however...) - */ + + /* see LIMITATIONS: above */ + unsigned long bm_set; /* nr of set bits; THINK maybe atomic_t? */ unsigned long bm_bits; size_t bm_words; @@ -517,43 +549,39 @@ static void bm_set_surplus(struct drbd_bitmap *b) bm_unmap(p_addr); } +/* you better not modify the bitmap while this is running, + * or its results will be stale */ static unsigned long bm_count_bits(struct drbd_bitmap *b) { - unsigned long *p_addr, *bm, offset = 0; + unsigned long *p_addr; unsigned long bits = 0; - unsigned long i, do_now; - unsigned long words; - - /* due to 64bit alignment, the last long on a 32bit arch - * may be not used at all. The last used long will likely - * be only partially used, always. Don't count those bits, - * but mask them out. */ - words = (b->bm_bits + BITS_PER_LONG - 1) >> LN2_BPL; - - while (offset < words) { - i = do_now = min_t(size_t, words-offset, LWPP); - p_addr = __bm_map_pidx(b, bm_word_to_page_idx(b, offset), KM_USER0); - bm = p_addr + MLPP(offset); - while (i--) { - bits += hweight_long(*bm++); - } - offset += do_now; - if (offset == words) { - /* last word may only be partially used, - * see also bm_clear_surplus. */ - i = (1UL << (b->bm_bits & (BITS_PER_LONG-1))) -1; - if (i) { - bits -= hweight_long(p_addr[do_now-1] & ~i); - p_addr[do_now-1] &= i; - } - /* 32bit arch, may have an unused padding long */ - if (words != b->bm_words) - p_addr[do_now] = 0; - } + unsigned long mask = (1UL << (b->bm_bits & BITS_PER_LONG_MASK)) -1; + int idx, last_page, i, last_word; + + /* because of the "extra long to catch oob access" we allocate in + * drbd_bm_resize, bm_number_of_pages -1 is not necessarily the page + * containing the last _relevant_ bitmap word */ + last_page = bm_bit_to_page_idx(b, b->bm_bits-1); + + /* all but last page */ + for (idx = 0; idx < last_page; idx++) { + p_addr = __bm_map_pidx(b, idx, KM_USER0); + for (i = 0; i < LWPP; i++) + bits += hweight_long(p_addr[i]); __bm_unmap(p_addr, KM_USER0); cond_resched(); } - + /* last (or only) page */ + last_word = ((b->bm_bits - 1) & BITS_PER_PAGE_MASK) >> LN2_BPL; + p_addr = __bm_map_pidx(b, idx, KM_USER0); + for (i = 0; i < last_word; i++) + bits += hweight_long(p_addr[i]); + p_addr[last_word] &= cpu_to_lel(mask); + bits += hweight_long(p_addr[last_word]); + /* 32bit arch, may have an unused padding long */ + if (BITS_PER_LONG == 32 && (last_word & 1) == 0) + p_addr[last_word+1] = 0; + __bm_unmap(p_addr, KM_USER0); return bits; } @@ -564,8 +592,6 @@ static void bm_memset(struct drbd_bitmap *b, size_t offset, int c, size_t len) unsigned int idx; size_t do_now, end; -#define BM_SECTORS_PER_BIT (BM_BLOCK_SIZE/512) - end = offset + len; if (end > b->bm_words) { @@ -645,8 +671,14 @@ int drbd_bm_resize(struct drbd_conf *mdev, sector_t capacity, int set_new_bits) words = ALIGN(bits, 64) >> LN2_BPL; if (get_ldev(mdev)) { - D_ASSERT((u64)bits <= (((u64)mdev->ldev->md.md_size_sect-MD_BM_OFFSET) << 12)); + u64 bits_on_disk = ((u64)mdev->ldev->md.md_size_sect-MD_BM_OFFSET) << 12; put_ldev(mdev); + if (bits > bits_on_disk) { + dev_info(DEV, "bits = %lu\n", bits); + dev_info(DEV, "bits_on_disk = %llu\n", bits_on_disk); + err = -ENOSPC; + goto out; + } } /* one extra long to catch off by one errors */ @@ -1113,9 +1145,12 @@ int drbd_bm_write_lazy(struct drbd_conf *mdev, unsigned upper_idx) __must_hold(l * @mdev: DRBD device. * @idx: bitmap page index * - * We don't want to special case on logical_block_size of the underlaying - * device, so we submit PAGE_SIZE aligned pieces containing the requested enr. + * We don't want to special case on logical_block_size of the backend device, + * so we submit PAGE_SIZE aligned pieces. * Note that on "most" systems, PAGE_SIZE is 4k. + * + * In case this becomes an issue on systems with larger PAGE_SIZE, + * we may want to change this again to write 4k aligned 4k pieces. */ int drbd_bm_write_page(struct drbd_conf *mdev, unsigned int idx) __must_hold(local) { @@ -1144,52 +1179,57 @@ int drbd_bm_write_page(struct drbd_conf *mdev, unsigned int idx) __must_hold(loc /* NOTE * find_first_bit returns int, we return unsigned long. - * should not make much difference anyways, but ... + * For this to work on 32bit arch with bitnumbers > (1<<32), + * we'd need to return u64, and get a whole lot of other places + * fixed where we still use unsigned long. * * this returns a bit number, NOT a sector! */ -#define BPP_MASK ((1UL << (PAGE_SHIFT+3)) - 1) static unsigned long __bm_find_next(struct drbd_conf *mdev, unsigned long bm_fo, const int find_zero_bit, const enum km_type km) { struct drbd_bitmap *b = mdev->bitmap; - unsigned long i = -1UL; unsigned long *p_addr; - unsigned long bit_offset; /* bit offset of the mapped page. */ + unsigned long bit_offset; + unsigned i; + if (bm_fo > b->bm_bits) { dev_err(DEV, "bm_fo=%lu bm_bits=%lu\n", bm_fo, b->bm_bits); + bm_fo = DRBD_END_OF_BITMAP; } else { while (bm_fo < b->bm_bits) { /* bit offset of the first bit in the page */ - bit_offset = bm_fo & ~BPP_MASK; + bit_offset = bm_fo & ~BITS_PER_PAGE_MASK; p_addr = __bm_map_pidx(b, bm_bit_to_page_idx(b, bm_fo), km); if (find_zero_bit) - i = generic_find_next_zero_le_bit(p_addr, PAGE_SIZE*8, bm_fo & BPP_MASK); + i = generic_find_next_zero_le_bit(p_addr, + PAGE_SIZE*8, bm_fo & BITS_PER_PAGE_MASK); else - i = generic_find_next_le_bit(p_addr, PAGE_SIZE*8, bm_fo & BPP_MASK); + i = generic_find_next_le_bit(p_addr, + PAGE_SIZE*8, bm_fo & BITS_PER_PAGE_MASK); __bm_unmap(p_addr, km); if (i < PAGE_SIZE*8) { - i = bit_offset + i; - if (i >= b->bm_bits) + bm_fo = bit_offset + i; + if (bm_fo >= b->bm_bits) break; goto found; } bm_fo = bit_offset + PAGE_SIZE*8; } - i = -1UL; + bm_fo = DRBD_END_OF_BITMAP; } found: - return i; + return bm_fo; } static unsigned long bm_find_next(struct drbd_conf *mdev, unsigned long bm_fo, const int find_zero_bit) { struct drbd_bitmap *b = mdev->bitmap; - unsigned long i = -1UL; + unsigned long i = DRBD_END_OF_BITMAP; ERR_IF(!b) return i; ERR_IF(!b->bm_pages) return i; @@ -1267,9 +1307,9 @@ static int __bm_change_bits_to(struct drbd_conf *mdev, const unsigned long s, last_page_nr = page_nr; } if (val) - c += (0 == generic___test_and_set_le_bit(bitnr & BPP_MASK, p_addr)); + c += (0 == generic___test_and_set_le_bit(bitnr & BITS_PER_PAGE_MASK, p_addr)); else - c -= (0 != generic___test_and_clear_le_bit(bitnr & BPP_MASK, p_addr)); + c -= (0 != generic___test_and_clear_le_bit(bitnr & BITS_PER_PAGE_MASK, p_addr)); } if (p_addr) __bm_unmap(p_addr, km); @@ -1418,7 +1458,7 @@ int drbd_bm_test_bit(struct drbd_conf *mdev, const unsigned long bitnr) bm_print_lock_info(mdev); if (bitnr < b->bm_bits) { p_addr = bm_map_pidx(b, bm_bit_to_page_idx(b, bitnr)); - i = generic_test_le_bit(bitnr & BPP_MASK, p_addr) ? 1 : 0; + i = generic_test_le_bit(bitnr & BITS_PER_PAGE_MASK, p_addr) ? 1 : 0; bm_unmap(p_addr); } else if (bitnr == b->bm_bits) { i = -1; @@ -1517,13 +1557,15 @@ int drbd_bm_e_weight(struct drbd_conf *mdev, unsigned long enr) return count; } -/* set all bits covered by the AL-extent al_enr */ +/* Set all bits covered by the AL-extent al_enr. + * Returns number of bits changed. */ unsigned long drbd_bm_ALe_set_all(struct drbd_conf *mdev, unsigned long al_enr) { struct drbd_bitmap *b = mdev->bitmap; unsigned long *p_addr, *bm; unsigned long weight; - int count, s, e, i, do_now; + unsigned long s, e; + int count, i, do_now; ERR_IF(!b) return 0; ERR_IF(!b->bm_pages) return 0; @@ -1552,7 +1594,7 @@ unsigned long drbd_bm_ALe_set_all(struct drbd_conf *mdev, unsigned long al_enr) if (e == b->bm_words) b->bm_set -= bm_clear_surplus(b); } else { - dev_err(DEV, "start offset (%d) too large in drbd_bm_ALe_set_all\n", s); + dev_err(DEV, "start offset (%lu) too large in drbd_bm_ALe_set_all\n", s); } weight = b->bm_set - weight; spin_unlock_irq(&b->bm_lock); diff --git a/drivers/block/drbd/drbd_int.h b/drivers/block/drbd/drbd_int.h index 74cc50a2182..5a2d0ec72b3 100644 --- a/drivers/block/drbd/drbd_int.h +++ b/drivers/block/drbd/drbd_int.h @@ -1003,9 +1003,9 @@ struct drbd_conf { struct hlist_head *tl_hash; unsigned int tl_hash_s; - /* blocks to sync in this run [unit BM_BLOCK_SIZE] */ + /* blocks to resync in this run [unit BM_BLOCK_SIZE] */ unsigned long rs_total; - /* number of sync IOs that failed in this run */ + /* number of resync blocks that failed in this run */ unsigned long rs_failed; /* Syncer's start time [unit jiffies] */ unsigned long rs_start; @@ -1399,7 +1399,9 @@ struct bm_extent { * you should use 64bit OS for that much storage, anyways. */ #define DRBD_MAX_SECTORS_FLEX BM_BIT_TO_SECT(0xffff7fff) #else -#define DRBD_MAX_SECTORS_FLEX BM_BIT_TO_SECT(0x1LU << 32) +/* we allow up to 1 PiB now on 64bit architecture with "flexible" meta data */ +#define DRBD_MAX_SECTORS_FLEX (1UL << 51) +/* corresponds to (1UL << 38) bits right now. */ #endif #endif @@ -1419,11 +1421,15 @@ extern int drbd_bm_resize(struct drbd_conf *mdev, sector_t sectors, int set_new extern void drbd_bm_cleanup(struct drbd_conf *mdev); extern void drbd_bm_set_all(struct drbd_conf *mdev); extern void drbd_bm_clear_all(struct drbd_conf *mdev); +/* set/clear/test only a few bits at a time */ extern int drbd_bm_set_bits( struct drbd_conf *mdev, unsigned long s, unsigned long e); extern int drbd_bm_clear_bits( struct drbd_conf *mdev, unsigned long s, unsigned long e); -/* bm_set_bits variant for use while holding drbd_bm_lock */ +extern int drbd_bm_count_bits( + struct drbd_conf *mdev, const unsigned long s, const unsigned long e); +/* bm_set_bits variant for use while holding drbd_bm_lock, + * may process the whole bitmap in one go */ extern void _drbd_bm_set_bits(struct drbd_conf *mdev, const unsigned long s, const unsigned long e); extern int drbd_bm_test_bit(struct drbd_conf *mdev, unsigned long bitnr); @@ -1436,6 +1442,8 @@ extern unsigned long drbd_bm_ALe_set_all(struct drbd_conf *mdev, extern size_t drbd_bm_words(struct drbd_conf *mdev); extern unsigned long drbd_bm_bits(struct drbd_conf *mdev); extern sector_t drbd_bm_capacity(struct drbd_conf *mdev); + +#define DRBD_END_OF_BITMAP (~(unsigned long)0) extern unsigned long drbd_bm_find_next(struct drbd_conf *mdev, unsigned long bm_fo); /* bm_find_next variants for use while you hold drbd_bm_lock() */ extern unsigned long _drbd_bm_find_next(struct drbd_conf *mdev, unsigned long bm_fo); @@ -1452,8 +1460,6 @@ extern void drbd_bm_get_lel(struct drbd_conf *mdev, size_t offset, extern void drbd_bm_lock(struct drbd_conf *mdev, char *why); extern void drbd_bm_unlock(struct drbd_conf *mdev); - -extern int drbd_bm_count_bits(struct drbd_conf *mdev, const unsigned long s, const unsigned long e); /* drbd_main.c */ extern struct kmem_cache *drbd_request_cache; @@ -2158,10 +2164,8 @@ extern int _get_ldev_if_state(struct drbd_conf *mdev, enum drbd_disk_state mins) static inline void drbd_get_syncer_progress(struct drbd_conf *mdev, unsigned long *bits_left, unsigned int *per_mil_done) { - /* - * this is to break it at compile time when we change that - * (we may feel 4TB maximum storage per drbd is not enough) - */ + /* this is to break it at compile time when we change that, in case we + * want to support more than (1<<32) bits on a 32bit arch. */ typecheck(unsigned long, mdev->rs_total); /* note: both rs_total and rs_left are in bits, i.e. in @@ -2186,10 +2190,19 @@ static inline void drbd_get_syncer_progress(struct drbd_conf *mdev, *bits_left, mdev->rs_total, mdev->rs_failed); *per_mil_done = 0; } else { - /* make sure the calculation happens in long context */ - unsigned long tmp = 1000UL - - (*bits_left >> 10)*1000UL - / ((mdev->rs_total >> 10) + 1UL); + /* Make sure the division happens in long context. + * We allow up to one petabyte storage right now, + * at a granularity of 4k per bit that is 2**38 bits. + * After shift right and multiplication by 1000, + * this should still fit easily into a 32bit long, + * so we don't need a 64bit division on 32bit arch. + * Note: currently we don't support such large bitmaps on 32bit + * arch anyways, but no harm done to be prepared for it here. + */ + unsigned int shift = mdev->rs_total >= (1ULL << 32) ? 16 : 10; + unsigned long left = *bits_left >> shift; + unsigned long total = 1UL + (mdev->rs_total >> shift); + unsigned long tmp = 1000UL - left * 1000UL/total; *per_mil_done = tmp; } } diff --git a/drivers/block/drbd/drbd_nl.c b/drivers/block/drbd/drbd_nl.c index 77dc022eaf6..a46bc0287e2 100644 --- a/drivers/block/drbd/drbd_nl.c +++ b/drivers/block/drbd/drbd_nl.c @@ -527,17 +527,19 @@ static void drbd_md_set_sector_offsets(struct drbd_conf *mdev, } } +/* input size is expected to be in KB */ char *ppsize(char *buf, unsigned long long size) { - /* Needs 9 bytes at max. */ + /* Needs 9 bytes at max including trailing NUL: + * -1ULL ==> "16384 EB" */ static char units[] = { 'K', 'M', 'G', 'T', 'P', 'E' }; int base = 0; - while (size >= 10000) { + while (size >= 10000 && base < sizeof(units)-1) { /* shift + round */ size = (size >> 10) + !!(size & (1<<9)); base++; } - sprintf(buf, "%lu %cB", (long)size, units[base]); + sprintf(buf, "%u %cB", (unsigned)size, units[base]); return buf; } diff --git a/drivers/block/drbd/drbd_proc.c b/drivers/block/drbd/drbd_proc.c index efba62cd2e5..2959cdfb77f 100644 --- a/drivers/block/drbd/drbd_proc.c +++ b/drivers/block/drbd/drbd_proc.c @@ -91,9 +91,9 @@ static void drbd_syncer_progress(struct drbd_conf *mdev, struct seq_file *seq) seq_printf(seq, "sync'ed:"); seq_printf(seq, "%3u.%u%% ", res / 10, res % 10); - /* if more than 1 GB display in MB */ - if (mdev->rs_total > 0x100000L) - seq_printf(seq, "(%lu/%lu)M\n\t", + /* if more than a few GB, display in MB */ + if (mdev->rs_total > (4UL << (30 - BM_BLOCK_SHIFT))) + seq_printf(seq, "(%lu/%lu)M", (unsigned long) Bit2KB(rs_left >> 10), (unsigned long) Bit2KB(mdev->rs_total >> 10)); else diff --git a/drivers/block/drbd/drbd_worker.c b/drivers/block/drbd/drbd_worker.c index d17f2ed777c..be46084c254 100644 --- a/drivers/block/drbd/drbd_worker.c +++ b/drivers/block/drbd/drbd_worker.c @@ -577,7 +577,7 @@ next_sector: size = BM_BLOCK_SIZE; bit = drbd_bm_find_next(mdev, mdev->bm_resync_fo); - if (bit == -1UL) { + if (bit == DRBD_END_OF_BITMAP) { mdev->bm_resync_fo = drbd_bm_bits(mdev); mdev->resync_work.cb = w_resync_inactive; put_ldev(mdev); -- cgit v1.2.3-70-g09d2 From 02851e9f00d78dbc8ded0aacbf9bf3b631d627b3 Mon Sep 17 00:00:00 2001 From: Lars Ellenberg Date: Thu, 16 Dec 2010 14:47:39 +0100 Subject: drbd: move bitmap write from resync_finished to after_state_change We must not call it directly from resync_finished, as we may be in either receiver or worker context there. Signed-off-by: Philipp Reisner Signed-off-by: Lars Ellenberg --- drivers/block/drbd/drbd_main.c | 14 ++++++++++---- drivers/block/drbd/drbd_worker.c | 3 --- 2 files changed, 10 insertions(+), 7 deletions(-) (limited to 'drivers/block/drbd/drbd_worker.c') diff --git a/drivers/block/drbd/drbd_main.c b/drivers/block/drbd/drbd_main.c index 57ed7181742..c6c7e3e6dc2 100644 --- a/drivers/block/drbd/drbd_main.c +++ b/drivers/block/drbd/drbd_main.c @@ -1548,6 +1548,9 @@ static void after_state_ch(struct drbd_conf *mdev, union drbd_state os, if (os.disk < D_UP_TO_DATE && os.conn >= C_SYNC_SOURCE && ns.conn == C_CONNECTED) drbd_send_state(mdev); + if (os.conn > C_CONNECTED && ns.conn == C_CONNECTED) + drbd_queue_bitmap_io(mdev, &drbd_bm_write, NULL, "write from resync_finished"); + /* free tl_hash if we Got thawed and are C_STANDALONE */ if (ns.conn == C_STANDALONE && !is_susp(ns) && mdev->tl_hash) drbd_free_tl_hash(mdev); @@ -3860,13 +3863,16 @@ int drbd_bmio_clear_n_write(struct drbd_conf *mdev) static int w_bitmap_io(struct drbd_conf *mdev, struct drbd_work *w, int unused) { struct bm_io_work *work = container_of(w, struct bm_io_work, w); - int rv; + int rv = -EIO; D_ASSERT(atomic_read(&mdev->ap_bio_cnt) == 0); - drbd_bm_lock(mdev, work->why); - rv = work->io_fn(mdev); - drbd_bm_unlock(mdev); + if (get_ldev(mdev)) { + drbd_bm_lock(mdev, work->why); + rv = work->io_fn(mdev); + drbd_bm_unlock(mdev); + put_ldev(mdev); + } clear_bit(BITMAP_IO, &mdev->flags); smp_mb__after_clear_bit(); diff --git a/drivers/block/drbd/drbd_worker.c b/drivers/block/drbd/drbd_worker.c index be46084c254..2374454cdf1 100644 --- a/drivers/block/drbd/drbd_worker.c +++ b/drivers/block/drbd/drbd_worker.c @@ -907,9 +907,6 @@ out: drbd_md_sync(mdev); - dev_info(DEV, "Writing changed bitmap pages\n"); - drbd_queue_bitmap_io(mdev, &drbd_bm_write, NULL, "write from resync_finished"); - if (khelper_cmd) drbd_khelper(mdev, khelper_cmd); -- cgit v1.2.3-70-g09d2 From 5a22db8968a69bec835d1ed9a96ab3381719e0c0 Mon Sep 17 00:00:00 2001 From: Lars Ellenberg Date: Fri, 17 Dec 2010 21:14:23 +0100 Subject: drbd: serialize sending of resync uuid with pending w_send_oos To improve the latency of IO requests during bitmap exchange, we recently allowed writes while waiting for the bitmap, sending "set out-of-sync" information packets for any newly dirtied bits. We have to make sure that the new resync-uuid does not overtake these "set oos" packets. Once the resync-uuid is received, the sync target starts the resync process, and expects the bitmap to only be cleared, not re-set. If we use this protocol extension, we queue the generation and sending of the resync-uuid on the worker, which naturally serializes with all previously queued packets. Signed-off-by: Philipp Reisner Signed-off-by: Lars Ellenberg --- drivers/block/drbd/drbd_int.h | 2 +- drivers/block/drbd/drbd_main.c | 22 +++++++++++++++++++--- drivers/block/drbd/drbd_worker.c | 27 ++++++++++++++------------- 3 files changed, 34 insertions(+), 17 deletions(-) (limited to 'drivers/block/drbd/drbd_worker.c') diff --git a/drivers/block/drbd/drbd_int.h b/drivers/block/drbd/drbd_int.h index 5a2d0ec72b3..ec06e744be4 100644 --- a/drivers/block/drbd/drbd_int.h +++ b/drivers/block/drbd/drbd_int.h @@ -1204,7 +1204,7 @@ extern int drbd_send(struct drbd_conf *mdev, struct socket *sock, extern int drbd_send_protocol(struct drbd_conf *mdev); extern int drbd_send_uuids(struct drbd_conf *mdev); extern int drbd_send_uuids_skip_initial_sync(struct drbd_conf *mdev); -extern int drbd_send_sync_uuid(struct drbd_conf *mdev, u64 val); +extern int drbd_gen_and_send_sync_uuid(struct drbd_conf *mdev); extern int drbd_send_sizes(struct drbd_conf *mdev, int trigger_reply, enum dds_flags flags); extern int _drbd_send_state(struct drbd_conf *mdev); extern int drbd_send_state(struct drbd_conf *mdev); diff --git a/drivers/block/drbd/drbd_main.c b/drivers/block/drbd/drbd_main.c index 4da6f11cc82..2190064d59b 100644 --- a/drivers/block/drbd/drbd_main.c +++ b/drivers/block/drbd/drbd_main.c @@ -1387,6 +1387,17 @@ static void after_state_ch(struct drbd_conf *mdev, union drbd_state os, spin_unlock_irq(&mdev->req_lock); } + /* Became sync source. With protocol >= 96, we still need to send out + * the sync uuid now. Need to do that before any drbd_send_state, or + * the other side may go "paused sync" before receiving the sync uuids, + * which is unexpected. */ + if ((os.conn != C_SYNC_SOURCE && os.conn != C_PAUSED_SYNC_S) && + (ns.conn == C_SYNC_SOURCE || ns.conn == C_PAUSED_SYNC_S) && + mdev->agreed_pro_version >= 96 && get_ldev(mdev)) { + drbd_gen_and_send_sync_uuid(mdev); + put_ldev(mdev); + } + /* Do not change the order of the if above and the two below... */ if (os.pdsk == D_DISKLESS && ns.pdsk > D_DISKLESS) { /* attach on the peer */ drbd_send_uuids(mdev); @@ -1980,12 +1991,17 @@ int drbd_send_uuids_skip_initial_sync(struct drbd_conf *mdev) return _drbd_send_uuids(mdev, 8); } - -int drbd_send_sync_uuid(struct drbd_conf *mdev, u64 val) +int drbd_gen_and_send_sync_uuid(struct drbd_conf *mdev) { struct p_rs_uuid p; + u64 uuid; + + D_ASSERT(mdev->state.disk == D_UP_TO_DATE); - p.uuid = cpu_to_be64(val); + get_random_bytes(&uuid, sizeof(u64)); + drbd_uuid_set(mdev, UI_BITMAP, uuid); + drbd_md_sync(mdev); + p.uuid = cpu_to_be64(uuid); return drbd_send_cmd(mdev, USE_DATA_SOCKET, P_SYNC_UUID, (struct p_header80 *)&p, sizeof(p)); diff --git a/drivers/block/drbd/drbd_worker.c b/drivers/block/drbd/drbd_worker.c index 2374454cdf1..3df37e65c11 100644 --- a/drivers/block/drbd/drbd_worker.c +++ b/drivers/block/drbd/drbd_worker.c @@ -1516,18 +1516,6 @@ void drbd_start_resync(struct drbd_conf *mdev, enum drbd_conns side) return; } - if (side == C_SYNC_TARGET) { - mdev->bm_resync_fo = 0; - } else /* side == C_SYNC_SOURCE */ { - u64 uuid; - - get_random_bytes(&uuid, sizeof(u64)); - drbd_uuid_set(mdev, UI_BITMAP, uuid); - drbd_send_sync_uuid(mdev, uuid); - - D_ASSERT(mdev->state.disk == D_UP_TO_DATE); - } - write_lock_irq(&global_state_lock); ns = mdev->state; @@ -1565,7 +1553,19 @@ void drbd_start_resync(struct drbd_conf *mdev, enum drbd_conns side) _drbd_pause_after(mdev); } write_unlock_irq(&global_state_lock); - put_ldev(mdev); + + if (side == C_SYNC_TARGET) + mdev->bm_resync_fo = 0; + + /* Since protocol 96, we must serialize drbd_gen_and_send_sync_uuid + * with w_send_oos, or the sync target will get confused as to + * how much bits to resync. We cannot do that always, because for an + * empty resync and protocol < 95, we need to do it here, as we call + * drbd_resync_finished from here in that case. + * We drbd_gen_and_send_sync_uuid here for protocol < 96, + * and from after_state_ch otherwise. */ + if (side == C_SYNC_SOURCE && mdev->agreed_pro_version < 96) + drbd_gen_and_send_sync_uuid(mdev); if (r == SS_SUCCESS) { dev_info(DEV, "Began resync as %s (will sync %lu KB [%lu bits set]).\n", @@ -1601,6 +1601,7 @@ void drbd_start_resync(struct drbd_conf *mdev, enum drbd_conns side) drbd_md_sync(mdev); } + put_ldev(mdev); drbd_state_unlock(mdev); } -- cgit v1.2.3-70-g09d2 From 071942727824bab03b1a3f6b6eeb5b269697b333 Mon Sep 17 00:00:00 2001 From: Lars Ellenberg Date: Mon, 20 Dec 2010 15:38:07 +0100 Subject: drbd: ratelimit io error messages Signed-off-by: Philipp Reisner Signed-off-by: Lars Ellenberg --- drivers/block/drbd/drbd_worker.c | 9 +++++---- 1 file changed, 5 insertions(+), 4 deletions(-) (limited to 'drivers/block/drbd/drbd_worker.c') diff --git a/drivers/block/drbd/drbd_worker.c b/drivers/block/drbd/drbd_worker.c index 3df37e65c11..7bfeb79e710 100644 --- a/drivers/block/drbd/drbd_worker.c +++ b/drivers/block/drbd/drbd_worker.c @@ -165,14 +165,15 @@ void drbd_endio_sec(struct bio *bio, int error) int uptodate = bio_flagged(bio, BIO_UPTODATE); int is_write = bio_data_dir(bio) == WRITE; - if (error) + if (error && __ratelimit(&drbd_ratelimit_state)) dev_warn(DEV, "%s: error=%d s=%llus\n", is_write ? "write" : "read", error, (unsigned long long)e->sector); if (!error && !uptodate) { - dev_warn(DEV, "%s: setting error to -EIO s=%llus\n", - is_write ? "write" : "read", - (unsigned long long)e->sector); + if (__ratelimit(&drbd_ratelimit_state)) + dev_warn(DEV, "%s: setting error to -EIO s=%llus\n", + is_write ? "write" : "read", + (unsigned long long)e->sector); /* strange behavior of some lower level drivers... * fail the request by clearing the uptodate flag, * but do not return any error?! */ -- cgit v1.2.3-70-g09d2 From d612d309e4c8401ad94c531678b59c4a8b7c41ce Mon Sep 17 00:00:00 2001 From: Philipp Reisner Date: Mon, 27 Dec 2010 10:53:28 +0100 Subject: drbd: No longer answer P_RS_DATA_REQUEST packets when in C_AHEAD mode When the sync source node replies to a P_RS_DATA_REQUEST packet when it is already in ahead mode. I.e. those two packets crossed each other on the wire, that may lead to diverging bitmaps. This never happens in a well-tuned-system. In a well-tuned- system the resync controller has reduced the resync speed to zero long before we got into ahead-mode. But we have to be prepared for the not-well-tuned-system of course as well. Because -> diverging bitmaps = non terminating resync. Signed-off-by: Philipp Reisner Signed-off-by: Lars Ellenberg --- drivers/block/drbd/drbd_int.h | 3 ++- drivers/block/drbd/drbd_receiver.c | 12 +++++++++++- drivers/block/drbd/drbd_worker.c | 4 +++- 3 files changed, 16 insertions(+), 3 deletions(-) (limited to 'drivers/block/drbd/drbd_worker.c') diff --git a/drivers/block/drbd/drbd_int.h b/drivers/block/drbd/drbd_int.h index ec06e744be4..a529285b0cd 100644 --- a/drivers/block/drbd/drbd_int.h +++ b/drivers/block/drbd/drbd_int.h @@ -205,8 +205,9 @@ enum drbd_packets { /* P_CKPT_DISABLE_REQ = 0x26, * currently reserved for protocol D */ P_DELAY_PROBE = 0x27, /* is used on BOTH sockets */ P_OUT_OF_SYNC = 0x28, /* Mark as out of sync (Outrunning), data socket */ + P_RS_CANCEL = 0x29, /* meta: Used to cancel RS_DATA_REQUEST packet by SyncSource */ - P_MAX_CMD = 0x28, + P_MAX_CMD = 0x2A, P_MAY_IGNORE = 0x100, /* Flag to test if (cmd > P_MAY_IGNORE) ... */ P_MAX_OPT_CMD = 0x101, diff --git a/drivers/block/drbd/drbd_receiver.c b/drivers/block/drbd/drbd_receiver.c index f0a0f66fbe6..bf865bd8341 100644 --- a/drivers/block/drbd/drbd_receiver.c +++ b/drivers/block/drbd/drbd_receiver.c @@ -4361,7 +4361,16 @@ 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); } @@ -4459,6 +4468,7 @@ 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) diff --git a/drivers/block/drbd/drbd_worker.c b/drivers/block/drbd/drbd_worker.c index 7bfeb79e710..1d7510ebaa4 100644 --- a/drivers/block/drbd/drbd_worker.c +++ b/drivers/block/drbd/drbd_worker.c @@ -988,7 +988,9 @@ int w_e_end_rsdata_req(struct drbd_conf *mdev, struct drbd_work *w, int cancel) put_ldev(mdev); } - if (likely((e->flags & EE_WAS_ERROR) == 0)) { + if (mdev->state.conn == C_AHEAD) { + ok = drbd_send_ack(mdev, P_RS_CANCEL, e); + } else if (likely((e->flags & EE_WAS_ERROR) == 0)) { if (likely(mdev->state.pdsk >= D_INCONSISTENT)) { inc_rs_pending(mdev); ok = drbd_send_block(mdev, P_RS_DATA_REPLY, e); -- cgit v1.2.3-70-g09d2 From 794abb753e29e85949b3719dbc2ab6a98711a47e Mon Sep 17 00:00:00 2001 From: Philipp Reisner Date: Mon, 27 Dec 2010 11:51:23 +0100 Subject: drbd: Cleaned up the resync timer logic Besides removed a few lines of code, this moves the inspection of the state from before the queuing process to after the queuing. I.e. more closely to the actual invocation of the work. Signed-off-by: Philipp Reisner Signed-off-by: Lars Ellenberg --- drivers/block/drbd/drbd_int.h | 2 +- drivers/block/drbd/drbd_main.c | 2 +- drivers/block/drbd/drbd_worker.c | 48 +++++++++------------------------------- 3 files changed, 13 insertions(+), 39 deletions(-) (limited to 'drivers/block/drbd/drbd_worker.c') diff --git a/drivers/block/drbd/drbd_int.h b/drivers/block/drbd/drbd_int.h index a529285b0cd..dc11b7070ba 100644 --- a/drivers/block/drbd/drbd_int.h +++ b/drivers/block/drbd/drbd_int.h @@ -1540,7 +1540,7 @@ extern int w_e_end_csum_rs_req(struct drbd_conf *, struct drbd_work *, int); extern int w_e_end_ov_reply(struct drbd_conf *, struct drbd_work *, int); extern int w_e_end_ov_req(struct drbd_conf *, struct drbd_work *, int); extern int w_ov_finished(struct drbd_conf *, struct drbd_work *, int); -extern int w_resync_inactive(struct drbd_conf *, struct drbd_work *, int); +extern int w_resync_timer(struct drbd_conf *, struct drbd_work *, int); extern int w_resume_next_sg(struct drbd_conf *, struct drbd_work *, int); extern int w_send_write_hint(struct drbd_conf *, struct drbd_work *, int); extern int w_send_dblock(struct drbd_conf *, struct drbd_work *, int); diff --git a/drivers/block/drbd/drbd_main.c b/drivers/block/drbd/drbd_main.c index 90050ab7adf..4074d669930 100644 --- a/drivers/block/drbd/drbd_main.c +++ b/drivers/block/drbd/drbd_main.c @@ -2978,7 +2978,7 @@ void drbd_init_set_defaults(struct drbd_conf *mdev) INIT_LIST_HEAD(&mdev->start_resync_work.list); INIT_LIST_HEAD(&mdev->bm_io_work.w.list); - mdev->resync_work.cb = w_resync_inactive; + mdev->resync_work.cb = w_resync_timer; mdev->unplug_work.cb = w_send_write_hint; mdev->go_diskless.cb = w_go_diskless; mdev->md_sync_work.cb = w_md_sync; diff --git a/drivers/block/drbd/drbd_worker.c b/drivers/block/drbd/drbd_worker.c index 1d7510ebaa4..147f76b2681 100644 --- a/drivers/block/drbd/drbd_worker.c +++ b/drivers/block/drbd/drbd_worker.c @@ -253,13 +253,6 @@ int w_read_retry_remote(struct drbd_conf *mdev, struct drbd_work *w, int cancel) return w_send_read_req(mdev, w, 0); } -int w_resync_inactive(struct drbd_conf *mdev, struct drbd_work *w, int cancel) -{ - ERR_IF(cancel) return 1; - dev_err(DEV, "resync inactive, but callback triggered??\n"); - return 1; /* Simply ignore this! */ -} - void drbd_csum_ee(struct drbd_conf *mdev, struct crypto_hash *tfm, struct drbd_epoch_entry *e, void *digest) { struct hash_desc desc; @@ -389,26 +382,25 @@ defer: return -EAGAIN; } -void resync_timer_fn(unsigned long data) +int w_resync_timer(struct drbd_conf *mdev, struct drbd_work *w, int cancel) { - struct drbd_conf *mdev = (struct drbd_conf *) data; - int queue; - - queue = 1; switch (mdev->state.conn) { case C_VERIFY_S: - mdev->resync_work.cb = w_make_ov_request; + w_make_ov_request(mdev, w, cancel); break; case C_SYNC_TARGET: - mdev->resync_work.cb = w_make_resync_request; + w_make_resync_request(mdev, w, cancel); break; - default: - queue = 0; - mdev->resync_work.cb = w_resync_inactive; } - /* harmless race: list_empty outside data.work.q_lock */ - if (list_empty(&mdev->resync_work.list) && queue) + return 1; +} + +void resync_timer_fn(unsigned long data) +{ + struct drbd_conf *mdev = (struct drbd_conf *) data; + + if (list_empty(&mdev->resync_work.list)) drbd_queue_work(&mdev->data.work, &mdev->resync_work); } @@ -525,15 +517,6 @@ static int w_make_resync_request(struct drbd_conf *mdev, if (unlikely(cancel)) return 1; - if (unlikely(mdev->state.conn < C_CONNECTED)) { - dev_err(DEV, "Confused in w_make_resync_request()! cstate < Connected"); - return 0; - } - - if (mdev->state.conn != C_SYNC_TARGET) - dev_err(DEV, "%s in w_make_resync_request\n", - drbd_conn_str(mdev->state.conn)); - if (mdev->rs_total == 0) { /* empty resync? */ drbd_resync_finished(mdev); @@ -546,7 +529,6 @@ static int w_make_resync_request(struct drbd_conf *mdev, to continue resync with a broken disk makes no sense at all */ dev_err(DEV, "Disk broke down during resync!\n"); - mdev->resync_work.cb = w_resync_inactive; return 1; } @@ -580,7 +562,6 @@ next_sector: if (bit == DRBD_END_OF_BITMAP) { mdev->bm_resync_fo = drbd_bm_bits(mdev); - mdev->resync_work.cb = w_resync_inactive; put_ldev(mdev); return 1; } @@ -676,7 +657,6 @@ next_sector: * resync data block, and the last bit is cleared. * until then resync "work" is "inactive" ... */ - mdev->resync_work.cb = w_resync_inactive; put_ldev(mdev); return 1; } @@ -697,17 +677,11 @@ static int w_make_ov_request(struct drbd_conf *mdev, struct drbd_work *w, int ca if (unlikely(cancel)) return 1; - if (unlikely(mdev->state.conn < C_CONNECTED)) { - dev_err(DEV, "Confused in w_make_ov_request()! cstate < Connected"); - return 0; - } - number = drbd_rs_number_requests(mdev); sector = mdev->ov_position; for (i = 0; i < number; i++) { if (sector >= capacity) { - mdev->resync_work.cb = w_resync_inactive; return 1; } -- cgit v1.2.3-70-g09d2 From 370a43e7982dd497822097e0ae6022947ac2e7d4 Mon Sep 17 00:00:00 2001 From: Philipp Reisner Date: Fri, 14 Jan 2011 16:03:11 +0100 Subject: drbd: Work on the Ahead -> SyncSource transition The test if rs_pending_cnt == 0 was too weak. Using Test for unacked_cnt == 0 instead. Moved that into the worker. Since unacked_cnt gets already increased when an P_RS_DATA_REQ comes in. Also using a timer to make Ahead -> SyncSource -> Ahead cycles slower... Signed-off-by: Philipp Reisner Signed-off-by: Lars Ellenberg --- drivers/block/drbd/drbd_int.h | 3 +++ drivers/block/drbd/drbd_main.c | 4 ++++ drivers/block/drbd/drbd_receiver.c | 8 +++----- drivers/block/drbd/drbd_worker.c | 16 +++++++++++++++- 4 files changed, 25 insertions(+), 6 deletions(-) (limited to 'drivers/block/drbd/drbd_worker.c') diff --git a/drivers/block/drbd/drbd_int.h b/drivers/block/drbd/drbd_int.h index 3c8eecd9666..cfe7fff459e 100644 --- a/drivers/block/drbd/drbd_int.h +++ b/drivers/block/drbd/drbd_int.h @@ -850,6 +850,7 @@ enum { GOT_PING_ACK, /* set when we receive a ping_ack packet, misc wait gets woken */ NEW_CUR_UUID, /* Create new current UUID when thawing IO */ AL_SUSPENDED, /* Activity logging is currently suspended. */ + AHEAD_TO_SYNC_SOURCE, /* Ahead -> SyncSource queued */ }; struct drbd_bitmap; /* opaque for drbd_conf */ @@ -961,6 +962,7 @@ struct drbd_conf { start_resync_work; struct timer_list resync_timer; struct timer_list md_sync_timer; + struct timer_list start_resync_timer; #ifdef DRBD_DEBUG_MD_SYNC struct { unsigned int line; @@ -1544,6 +1546,7 @@ extern int w_send_oos(struct drbd_conf *, struct drbd_work *, int); extern int w_start_resync(struct drbd_conf *, struct drbd_work *, int); extern void resync_timer_fn(unsigned long data); +extern void start_resync_timer_fn(unsigned long data); /* drbd_receiver.c */ extern int drbd_rs_should_slow_down(struct drbd_conf *mdev, sector_t sector); diff --git a/drivers/block/drbd/drbd_main.c b/drivers/block/drbd/drbd_main.c index b3b6d3190f6..8ec7c65988b 100644 --- a/drivers/block/drbd/drbd_main.c +++ b/drivers/block/drbd/drbd_main.c @@ -2984,12 +2984,16 @@ void drbd_init_set_defaults(struct drbd_conf *mdev) mdev->go_diskless.cb = w_go_diskless; mdev->md_sync_work.cb = w_md_sync; mdev->bm_io_work.w.cb = w_bitmap_io; + mdev->start_resync_work.cb = w_start_resync; init_timer(&mdev->resync_timer); init_timer(&mdev->md_sync_timer); + init_timer(&mdev->start_resync_timer); mdev->resync_timer.function = resync_timer_fn; mdev->resync_timer.data = (unsigned long) mdev; mdev->md_sync_timer.function = md_sync_timer_fn; mdev->md_sync_timer.data = (unsigned long) mdev; + mdev->start_resync_timer.function = start_resync_timer_fn; + mdev->start_resync_timer.data = (unsigned long) mdev; init_waitqueue_head(&mdev->misc_wait); init_waitqueue_head(&mdev->state_wait); diff --git a/drivers/block/drbd/drbd_receiver.c b/drivers/block/drbd/drbd_receiver.c index 7991183749e..18cb8b609fe 100644 --- a/drivers/block/drbd/drbd_receiver.c +++ b/drivers/block/drbd/drbd_receiver.c @@ -4392,11 +4392,9 @@ static int got_BarrierAck(struct drbd_conf *mdev, struct p_header80 *h) if (mdev->state.conn == C_AHEAD && atomic_read(&mdev->ap_in_flight) == 0 && - atomic_read(&mdev->rs_pending_cnt) == 0 && - list_empty(&mdev->start_resync_work.list)) { - struct drbd_work *w = &mdev->start_resync_work; - w->cb = w_start_resync; - drbd_queue_work(&mdev->data.work, w); + !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; diff --git a/drivers/block/drbd/drbd_worker.c b/drivers/block/drbd/drbd_worker.c index 147f76b2681..3a95b701b5d 100644 --- a/drivers/block/drbd/drbd_worker.c +++ b/drivers/block/drbd/drbd_worker.c @@ -712,10 +712,24 @@ static int w_make_ov_request(struct drbd_conf *mdev, struct drbd_work *w, int ca } +void start_resync_timer_fn(unsigned long data) +{ + struct drbd_conf *mdev = (struct drbd_conf *) data; + + drbd_queue_work(&mdev->data.work, &mdev->start_resync_work); +} + int w_start_resync(struct drbd_conf *mdev, struct drbd_work *w, int cancel) { - drbd_start_resync(mdev, C_SYNC_SOURCE); + if (atomic_read(&mdev->unacked_cnt) || atomic_read(&mdev->rs_pending_cnt)) { + dev_warn(DEV, "w_start_resync later...\n"); + mdev->start_resync_timer.expires = jiffies + HZ/10; + add_timer(&mdev->start_resync_timer); + return 1; + } + drbd_start_resync(mdev, C_SYNC_SOURCE); + clear_bit(AHEAD_TO_SYNC_SOURCE, &mdev->current_epoch->flags); return 1; } -- cgit v1.2.3-70-g09d2 From 20ee639024e3d33111df0e343050b218c656bf16 Mon Sep 17 00:00:00 2001 From: Philipp Reisner Date: Tue, 18 Jan 2011 15:28:59 +0100 Subject: drbd: cleaned up __set_current_state() followed by schedule_timeout() calls Signed-off-by: Philipp Reisner Signed-off-by: Lars Ellenberg --- drivers/block/drbd/drbd_nl.c | 3 +-- drivers/block/drbd/drbd_receiver.c | 9 +++------ drivers/block/drbd/drbd_worker.c | 3 +-- 3 files changed, 5 insertions(+), 10 deletions(-) (limited to 'drivers/block/drbd/drbd_worker.c') diff --git a/drivers/block/drbd/drbd_nl.c b/drivers/block/drbd/drbd_nl.c index a46bc0287e2..42e16e4edfa 100644 --- a/drivers/block/drbd/drbd_nl.c +++ b/drivers/block/drbd/drbd_nl.c @@ -360,8 +360,7 @@ drbd_set_role(struct drbd_conf *mdev, enum drbd_role new_role, int force) if (rv == SS_TWO_PRIMARIES) { /* Maybe the peer is detected as dead very soon... retry at most once more in this case. */ - __set_current_state(TASK_INTERRUPTIBLE); - schedule_timeout((mdev->net_conf->ping_timeo+1)*HZ/10); + schedule_timeout_interruptible((mdev->net_conf->ping_timeo+1)*HZ/10); if (try < max_tries) try = max_tries - 1; continue; diff --git a/drivers/block/drbd/drbd_receiver.c b/drivers/block/drbd/drbd_receiver.c index a7f5b6d134e..3ccc6c33a33 100644 --- a/drivers/block/drbd/drbd_receiver.c +++ b/drivers/block/drbd/drbd_receiver.c @@ -768,8 +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) { @@ -788,8 +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) @@ -4142,8 +4140,7 @@ 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"); diff --git a/drivers/block/drbd/drbd_worker.c b/drivers/block/drbd/drbd_worker.c index 3a95b701b5d..7b3d4dc4af8 100644 --- a/drivers/block/drbd/drbd_worker.c +++ b/drivers/block/drbd/drbd_worker.c @@ -777,8 +777,7 @@ int drbd_resync_finished(struct drbd_conf *mdev) * queue (or even the read operations for those packets * is not finished by now). Retry in 100ms. */ - __set_current_state(TASK_INTERRUPTIBLE); - schedule_timeout(HZ / 10); + schedule_timeout_interruptible(HZ / 10); w = kmalloc(sizeof(struct drbd_work), GFP_ATOMIC); if (w) { w->cb = w_resync_finished; -- cgit v1.2.3-70-g09d2 From 6c922ed543bee1bc6685ade07be59f3fa49a7288 Mon Sep 17 00:00:00 2001 From: Lars Ellenberg Date: Wed, 12 Jan 2011 11:51:13 +0100 Subject: drbd: only generate and send a new sync uuid after a successful state change Signed-off-by: Philipp Reisner Signed-off-by: Lars Ellenberg --- drivers/block/drbd/drbd_worker.c | 25 ++++++++++++------------- 1 file changed, 12 insertions(+), 13 deletions(-) (limited to 'drivers/block/drbd/drbd_worker.c') diff --git a/drivers/block/drbd/drbd_worker.c b/drivers/block/drbd/drbd_worker.c index 7b3d4dc4af8..ec42e04bb51 100644 --- a/drivers/block/drbd/drbd_worker.c +++ b/drivers/block/drbd/drbd_worker.c @@ -1544,24 +1544,23 @@ void drbd_start_resync(struct drbd_conf *mdev, enum drbd_conns side) } write_unlock_irq(&global_state_lock); - if (side == C_SYNC_TARGET) - mdev->bm_resync_fo = 0; - - /* Since protocol 96, we must serialize drbd_gen_and_send_sync_uuid - * with w_send_oos, or the sync target will get confused as to - * how much bits to resync. We cannot do that always, because for an - * empty resync and protocol < 95, we need to do it here, as we call - * drbd_resync_finished from here in that case. - * We drbd_gen_and_send_sync_uuid here for protocol < 96, - * and from after_state_ch otherwise. */ - if (side == C_SYNC_SOURCE && mdev->agreed_pro_version < 96) - drbd_gen_and_send_sync_uuid(mdev); - if (r == SS_SUCCESS) { dev_info(DEV, "Began resync as %s (will sync %lu KB [%lu bits set]).\n", drbd_conn_str(ns.conn), (unsigned long) mdev->rs_total << (BM_BLOCK_SHIFT-10), (unsigned long) mdev->rs_total); + if (side == C_SYNC_TARGET) + mdev->bm_resync_fo = 0; + + /* Since protocol 96, we must serialize drbd_gen_and_send_sync_uuid + * with w_send_oos, or the sync target will get confused as to + * how much bits to resync. We cannot do that always, because for an + * empty resync and protocol < 95, we need to do it here, as we call + * drbd_resync_finished from here in that case. + * We drbd_gen_and_send_sync_uuid here for protocol < 96, + * and from after_state_ch otherwise. */ + if (side == C_SYNC_SOURCE && mdev->agreed_pro_version < 96) + drbd_gen_and_send_sync_uuid(mdev); if (mdev->agreed_pro_version < 95 && mdev->rs_total == 0) { /* This still has a race (about when exactly the peers -- cgit v1.2.3-70-g09d2 From 62b0da3a244ac33d25a77861ef1cc0080103f2ff Mon Sep 17 00:00:00 2001 From: Lars Ellenberg Date: Thu, 20 Jan 2011 13:25:21 +0100 Subject: drbd: log UUIDs whenever they change All decisions about sync, sync direction, and wether or not to allow a connect or attach are based on our set of UUIDs to tag a data generation. Log changes to the UUIDs whenever they occur, logging "new current UUID P:Q:R:S" is more useful than "Creating new current UUID". Signed-off-by: Philipp Reisner Signed-off-by: Lars Ellenberg --- drivers/block/drbd/drbd_int.h | 6 ++-- drivers/block/drbd/drbd_main.c | 71 ++++++++++++++++++-------------------- drivers/block/drbd/drbd_nl.c | 1 + drivers/block/drbd/drbd_receiver.c | 11 ++++-- drivers/block/drbd/drbd_worker.c | 20 ++++++----- 5 files changed, 58 insertions(+), 51 deletions(-) (limited to 'drivers/block/drbd/drbd_worker.c') diff --git a/drivers/block/drbd/drbd_int.h b/drivers/block/drbd/drbd_int.h index cfe7fff459e..0a9059eb94d 100644 --- a/drivers/block/drbd/drbd_int.h +++ b/drivers/block/drbd/drbd_int.h @@ -1240,11 +1240,11 @@ extern int _drbd_send_bitmap(struct drbd_conf *mdev); extern int drbd_send_sr_reply(struct drbd_conf *mdev, enum drbd_state_rv retcode); extern void drbd_free_bc(struct drbd_backing_dev *ldev); extern void drbd_mdev_cleanup(struct drbd_conf *mdev); +void drbd_print_uuids(struct drbd_conf *mdev, const char *text); /* drbd_meta-data.c (still in drbd_main.c) */ extern void drbd_md_sync(struct drbd_conf *mdev); extern int drbd_md_read(struct drbd_conf *mdev, struct drbd_backing_dev *bdev); -/* maybe define them below as inline? */ extern void drbd_uuid_set(struct drbd_conf *mdev, int idx, u64 val) __must_hold(local); extern void _drbd_uuid_set(struct drbd_conf *mdev, int idx, u64 val) __must_hold(local); extern void drbd_uuid_new_current(struct drbd_conf *mdev) __must_hold(local); @@ -2360,9 +2360,11 @@ static inline void dec_ap_bio(struct drbd_conf *mdev) } } -static inline void drbd_set_ed_uuid(struct drbd_conf *mdev, u64 val) +static inline int drbd_set_ed_uuid(struct drbd_conf *mdev, u64 val) { + int changed = mdev->ed_uuid != val; mdev->ed_uuid = val; + return changed; } static inline int seq_cmp(u32 a, u32 b) diff --git a/drivers/block/drbd/drbd_main.c b/drivers/block/drbd/drbd_main.c index e0be4077d56..b68332a0e73 100644 --- a/drivers/block/drbd/drbd_main.c +++ b/drivers/block/drbd/drbd_main.c @@ -1159,6 +1159,10 @@ __drbd_set_state(struct drbd_conf *mdev, union drbd_state ns, atomic_inc(&mdev->local_cnt); mdev->state = ns; + + if (os.disk == D_ATTACHING && ns.disk >= D_NEGOTIATING) + drbd_print_uuids(mdev, "attached to UUIDs"); + wake_up(&mdev->misc_wait); wake_up(&mdev->state_wait); @@ -2035,6 +2039,24 @@ int drbd_send_uuids_skip_initial_sync(struct drbd_conf *mdev) return _drbd_send_uuids(mdev, 8); } +void drbd_print_uuids(struct drbd_conf *mdev, const char *text) +{ + if (get_ldev_if_state(mdev, D_NEGOTIATING)) { + u64 *uuid = mdev->ldev->md.uuid; + dev_info(DEV, "%s %016llX:%016llX:%016llX:%016llX\n", + text, + (unsigned long long)uuid[UI_CURRENT], + (unsigned long long)uuid[UI_BITMAP], + (unsigned long long)uuid[UI_HISTORY_START], + (unsigned long long)uuid[UI_HISTORY_END]); + put_ldev(mdev); + } else { + dev_info(DEV, "%s effective data uuid: %016llX\n", + text, + (unsigned long long)mdev->ed_uuid); + } +} + int drbd_gen_and_send_sync_uuid(struct drbd_conf *mdev) { struct p_rs_uuid p; @@ -2044,6 +2066,7 @@ int drbd_gen_and_send_sync_uuid(struct drbd_conf *mdev) uuid = mdev->ldev->md.uuid[UI_BITMAP] + UUID_NEW_BM_OFFSET; drbd_uuid_set(mdev, UI_BITMAP, uuid); + drbd_print_uuids(mdev, "updated sync UUID"); drbd_md_sync(mdev); p.uuid = cpu_to_be64(uuid); @@ -3749,28 +3772,6 @@ int drbd_md_read(struct drbd_conf *mdev, struct drbd_backing_dev *bdev) return rv; } -static void debug_drbd_uuid(struct drbd_conf *mdev, enum drbd_uuid_index index) -{ - static char *uuid_str[UI_EXTENDED_SIZE] = { - [UI_CURRENT] = "CURRENT", - [UI_BITMAP] = "BITMAP", - [UI_HISTORY_START] = "HISTORY_START", - [UI_HISTORY_END] = "HISTORY_END", - [UI_SIZE] = "SIZE", - [UI_FLAGS] = "FLAGS", - }; - - if (index >= UI_EXTENDED_SIZE) { - dev_warn(DEV, " uuid_index >= EXTENDED_SIZE\n"); - return; - } - - dynamic_dev_dbg(DEV, " uuid[%s] now %016llX\n", - uuid_str[index], - (unsigned long long)mdev->ldev->md.uuid[index]); -} - - /** * drbd_md_mark_dirty() - Mark meta data super block as dirty * @mdev: DRBD device. @@ -3800,10 +3801,8 @@ static void drbd_uuid_move_history(struct drbd_conf *mdev) __must_hold(local) { int i; - for (i = UI_HISTORY_START; i < UI_HISTORY_END; i++) { + for (i = UI_HISTORY_START; i < UI_HISTORY_END; i++) mdev->ldev->md.uuid[i+1] = mdev->ldev->md.uuid[i]; - debug_drbd_uuid(mdev, i+1); - } } void _drbd_uuid_set(struct drbd_conf *mdev, int idx, u64 val) __must_hold(local) @@ -3818,7 +3817,6 @@ void _drbd_uuid_set(struct drbd_conf *mdev, int idx, u64 val) __must_hold(local) } mdev->ldev->md.uuid[idx] = val; - debug_drbd_uuid(mdev, idx); drbd_md_mark_dirty(mdev); } @@ -3828,7 +3826,6 @@ void drbd_uuid_set(struct drbd_conf *mdev, int idx, u64 val) __must_hold(local) if (mdev->ldev->md.uuid[idx]) { drbd_uuid_move_history(mdev); mdev->ldev->md.uuid[UI_HISTORY_START] = mdev->ldev->md.uuid[idx]; - debug_drbd_uuid(mdev, UI_HISTORY_START); } _drbd_uuid_set(mdev, idx, val); } @@ -3843,14 +3840,16 @@ void drbd_uuid_set(struct drbd_conf *mdev, int idx, u64 val) __must_hold(local) void drbd_uuid_new_current(struct drbd_conf *mdev) __must_hold(local) { u64 val; + unsigned long long bm_uuid = mdev->ldev->md.uuid[UI_BITMAP]; + + if (bm_uuid) + dev_warn(DEV, "bm UUID was already set: %llX\n", bm_uuid); - dev_info(DEV, "Creating new current UUID\n"); - D_ASSERT(mdev->ldev->md.uuid[UI_BITMAP] == 0); mdev->ldev->md.uuid[UI_BITMAP] = mdev->ldev->md.uuid[UI_CURRENT]; - debug_drbd_uuid(mdev, UI_BITMAP); get_random_bytes(&val, sizeof(u64)); _drbd_uuid_set(mdev, UI_CURRENT, val); + drbd_print_uuids(mdev, "new current UUID"); /* get it to stable storage _now_ */ drbd_md_sync(mdev); } @@ -3864,16 +3863,12 @@ void drbd_uuid_set_bm(struct drbd_conf *mdev, u64 val) __must_hold(local) drbd_uuid_move_history(mdev); mdev->ldev->md.uuid[UI_HISTORY_START] = mdev->ldev->md.uuid[UI_BITMAP]; mdev->ldev->md.uuid[UI_BITMAP] = 0; - debug_drbd_uuid(mdev, UI_HISTORY_START); - debug_drbd_uuid(mdev, UI_BITMAP); } else { - if (mdev->ldev->md.uuid[UI_BITMAP]) - dev_warn(DEV, "bm UUID already set"); - - mdev->ldev->md.uuid[UI_BITMAP] = val; - mdev->ldev->md.uuid[UI_BITMAP] &= ~((u64)1); + unsigned long long bm_uuid = mdev->ldev->md.uuid[UI_BITMAP]; + if (bm_uuid) + dev_warn(DEV, "bm UUID was already set: %llX\n", bm_uuid); - debug_drbd_uuid(mdev, UI_BITMAP); + mdev->ldev->md.uuid[UI_BITMAP] = val & ~((u64)1); } drbd_md_mark_dirty(mdev); } diff --git a/drivers/block/drbd/drbd_nl.c b/drivers/block/drbd/drbd_nl.c index ffe3a97fef9..ce6f2fe8085 100644 --- a/drivers/block/drbd/drbd_nl.c +++ b/drivers/block/drbd/drbd_nl.c @@ -2151,6 +2151,7 @@ static int drbd_nl_new_c_uuid(struct drbd_conf *mdev, struct drbd_nl_cfg_req *nl if (skip_initial_sync) { drbd_send_uuids_skip_initial_sync(mdev); _drbd_uuid_set(mdev, UI_BITMAP, 0); + drbd_print_uuids(mdev, "cleared bitmap UUID"); spin_lock_irq(&mdev->req_lock); _drbd_set_state(_NS2(mdev, disk, D_UP_TO_DATE, pdsk, D_UP_TO_DATE), CS_VERBOSE, NULL); diff --git a/drivers/block/drbd/drbd_receiver.c b/drivers/block/drbd/drbd_receiver.c index e9354931eac..e5686a81f42 100644 --- a/drivers/block/drbd/drbd_receiver.c +++ b/drivers/block/drbd/drbd_receiver.c @@ -3024,7 +3024,7 @@ static int receive_uuids(struct drbd_conf *mdev, enum drbd_packets cmd, unsigned { 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); @@ -3059,13 +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 @@ -3074,7 +3075,10 @@ 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; } @@ -3305,6 +3309,7 @@ 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); diff --git a/drivers/block/drbd/drbd_worker.c b/drivers/block/drbd/drbd_worker.c index ec42e04bb51..ff0eb308ee4 100644 --- a/drivers/block/drbd/drbd_worker.c +++ b/drivers/block/drbd/drbd_worker.c @@ -871,14 +871,18 @@ int drbd_resync_finished(struct drbd_conf *mdev) } } - drbd_uuid_set_bm(mdev, 0UL); - - if (mdev->p_uuid) { - /* Now the two UUID sets are equal, update what we - * know of the peer. */ - int i; - for (i = UI_CURRENT ; i <= UI_HISTORY_END ; i++) - mdev->p_uuid[i] = mdev->ldev->md.uuid[i]; + if (!(os.conn == C_VERIFY_S || os.conn == C_VERIFY_T)) { + /* for verify runs, we don't update uuids here, + * so there would be nothing to report. */ + drbd_uuid_set_bm(mdev, 0UL); + drbd_print_uuids(mdev, "updated UUIDs"); + if (mdev->p_uuid) { + /* Now the two UUID sets are equal, update what we + * know of the peer. */ + int i; + for (i = UI_CURRENT ; i <= UI_HISTORY_END ; i++) + mdev->p_uuid[i] = mdev->ldev->md.uuid[i]; + } } } -- cgit v1.2.3-70-g09d2 From 10f6d9926cd17afff9dc03c967706419798b4929 Mon Sep 17 00:00:00 2001 From: Lars Ellenberg Date: Mon, 24 Jan 2011 14:47:09 +0100 Subject: drbd: don't BUG_ON, if bio_add_page of a single page to an empty bio fails Just deal with it more gracefully, if we fail to add even a single page to an empty bio. We used to BUG_ON() there, but it has been observed in some Xen deployment, so we need to handle that case more robustly now. Signed-off-by: Philipp Reisner Signed-off-by: Lars Ellenberg --- drivers/block/drbd/drbd_receiver.c | 45 +++++++++++++++++++++++++------------- drivers/block/drbd/drbd_worker.c | 7 +++--- 2 files changed, 34 insertions(+), 18 deletions(-) (limited to 'drivers/block/drbd/drbd_worker.c') diff --git a/drivers/block/drbd/drbd_receiver.c b/drivers/block/drbd/drbd_receiver.c index a56b107e01e..9e9fc341313 100644 --- a/drivers/block/drbd/drbd_receiver.c +++ b/drivers/block/drbd/drbd_receiver.c @@ -1073,6 +1073,16 @@ void drbd_bump_write_ordering(struct drbd_conf *mdev, enum write_ordering_e wo) * @mdev: DRBD device. * @e: epoch entry * @rw: flag field, see bio->bi_rw + * + * May spread the pages to multiple bios, + * depending on bio_add_page restrictions. + * + * Returns 0 if all bios have been submitted, + * -ENOMEM if we could not allocate enough bios, + * -ENOSPC (any better suggestion?) if we have not been able to bio_add_page a + * single page to an empty bio (which should never happen and likely indicates + * that the lower level IO stack is in some way broken). This has been observed + * on certain Xen deployments. */ /* TODO allocate from our own bio_set. */ int drbd_submit_ee(struct drbd_conf *mdev, struct drbd_epoch_entry *e, @@ -1085,6 +1095,7 @@ int drbd_submit_ee(struct drbd_conf *mdev, struct drbd_epoch_entry *e, unsigned ds = e->size; unsigned n_bios = 0; unsigned nr_pages = (ds + PAGE_SIZE -1) >> PAGE_SHIFT; + int err = -ENOMEM; /* In most cases, we will only need one bio. But in case the lower * level restrictions happen to be different at this offset on this @@ -1110,8 +1121,17 @@ next_bio: page_chain_for_each(page) { unsigned len = min_t(unsigned, ds, PAGE_SIZE); if (!bio_add_page(bio, page, len, 0)) { - /* a single page must always be possible! */ - BUG_ON(bio->bi_vcnt == 0); + /* A single page must always be possible! + * But in case it fails anyways, + * we deal with it, and complain (below). */ + if (bio->bi_vcnt == 0) { + dev_err(DEV, + "bio_add_page failed for len=%u, " + "bi_vcnt=0 (bi_sector=%llu)\n", + len, (unsigned long long)bio->bi_sector); + err = -ENOSPC; + goto fail; + } goto next_bio; } ds -= len; @@ -1137,7 +1157,7 @@ fail: bios = bios->bi_next; bio_put(bio); } - return -ENOMEM; + return err; } static int receive_Barrier(struct drbd_conf *mdev, enum drbd_packets cmd, unsigned int data_size) @@ -1436,9 +1456,8 @@ static int recv_resync_read(struct drbd_conf *mdev, sector_t sector, int data_si if (drbd_submit_ee(mdev, e, WRITE, DRBD_FAULT_RS_WR) == 0) return true; - /* drbd_submit_ee currently fails for one reason only: - * not being able to allocate enough bios. - * Is dropping the connection going to help? */ + /* don't care for the reason here */ + dev_err(DEV, "submit failed, triggering re-connect\n"); spin_lock_irq(&mdev->req_lock); list_del(&e->w.list); spin_unlock_irq(&mdev->req_lock); @@ -1837,9 +1856,8 @@ 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; - /* drbd_submit_ee currently fails for one reason only: - * not being able to allocate enough bios. - * Is dropping the connection going to help? */ + /* don't care for the reason here */ + dev_err(DEV, "submit failed, triggering re-connect\n"); spin_lock_irq(&mdev->req_lock); list_del(&e->w.list); hlist_del_init(&e->colision); @@ -1848,9 +1866,7 @@ static int receive_Data(struct drbd_conf *mdev, enum drbd_packets cmd, unsigned drbd_al_complete_io(mdev, e->sector); out_interrupted: - /* yes, the epoch_size now is imbalanced. - * but we drop the connection anyways, so we don't have a chance to - * receive a barrier... atomic_inc(&mdev->epoch_size); */ + drbd_may_finish_epoch(mdev, e->epoch, EV_PUT + EV_CLEANUP); put_ldev(mdev); drbd_free_ee(mdev, e); return false; @@ -2096,9 +2112,8 @@ submit: if (drbd_submit_ee(mdev, e, READ, fault_type) == 0) return true; - /* drbd_submit_ee currently fails for one reason only: - * not being able to allocate enough bios. - * Is dropping the connection going to help? */ + /* don't care for the reason here */ + dev_err(DEV, "submit failed, triggering re-connect\n"); spin_lock_irq(&mdev->req_lock); list_del(&e->w.list); spin_unlock_irq(&mdev->req_lock); diff --git a/drivers/block/drbd/drbd_worker.c b/drivers/block/drbd/drbd_worker.c index ff0eb308ee4..cfd324b9f95 100644 --- a/drivers/block/drbd/drbd_worker.c +++ b/drivers/block/drbd/drbd_worker.c @@ -369,9 +369,10 @@ static int read_for_csum(struct drbd_conf *mdev, sector_t sector, int size) if (drbd_submit_ee(mdev, e, READ, DRBD_FAULT_RS_RD) == 0) return 0; - /* drbd_submit_ee currently fails for one reason only: - * not being able to allocate enough bios. - * Is dropping the connection going to help? */ + /* If it failed because of ENOMEM, retry should help. If it failed + * because bio_add_page failed (probably broken lower level driver), + * retry may or may not help. + * If it does not, you may need to force disconnect. */ spin_lock_irq(&mdev->req_lock); list_del(&e->w.list); spin_unlock_irq(&mdev->req_lock); -- cgit v1.2.3-70-g09d2 From c5a91619793d444e5103ec5841045bf878718398 Mon Sep 17 00:00:00 2001 From: Andreas Gruenbacher Date: Tue, 25 Jan 2011 17:33:38 +0100 Subject: drbd: Remove unused function atodb_endio() Signed-off-by: Philipp Reisner Signed-off-by: Lars Ellenberg --- drivers/block/drbd/drbd_actlog.c | 27 --------------------------- drivers/block/drbd/drbd_worker.c | 15 ++++++--------- include/linux/drbd.h | 2 +- 3 files changed, 7 insertions(+), 37 deletions(-) (limited to 'drivers/block/drbd/drbd_worker.c') diff --git a/drivers/block/drbd/drbd_actlog.c b/drivers/block/drbd/drbd_actlog.c index a6050791401..2a1642bc451 100644 --- a/drivers/block/drbd/drbd_actlog.c +++ b/drivers/block/drbd/drbd_actlog.c @@ -542,33 +542,6 @@ cancel: return 1; } -static void atodb_endio(struct bio *bio, int error) -{ - struct drbd_atodb_wait *wc = bio->bi_private; - struct drbd_conf *mdev = wc->mdev; - struct page *page; - int uptodate = bio_flagged(bio, BIO_UPTODATE); - - /* strange behavior of some lower level drivers... - * fail the request by clearing the uptodate flag, - * but do not return any error?! */ - if (!error && !uptodate) - error = -EIO; - - drbd_chk_io_error(mdev, error, true); - if (error && wc->error == 0) - wc->error = error; - - if (atomic_dec_and_test(&wc->count)) - complete(&wc->io_done); - - page = bio->bi_io_vec[0].bv_page; - put_page(page); - bio_put(bio); - mdev->bm_writ_cnt++; - put_ldev(mdev); -} - /** * drbd_al_apply_to_bm() - Sets the bitmap to diry(1) where covered ba active AL extents * @mdev: DRBD device. diff --git a/drivers/block/drbd/drbd_worker.c b/drivers/block/drbd/drbd_worker.c index cfd324b9f95..3d70d8d015d 100644 --- a/drivers/block/drbd/drbd_worker.c +++ b/drivers/block/drbd/drbd_worker.c @@ -44,15 +44,12 @@ static int w_make_resync_request(struct drbd_conf *mdev, -/* defined here: - drbd_md_io_complete - drbd_endio_sec - drbd_endio_pri - - * more endio handlers: - atodb_endio in drbd_actlog.c - drbd_bm_async_io_complete in drbd_bitmap.c - +/* endio handlers: + * drbd_md_io_complete (defined here) + * drbd_endio_pri (defined here) + * drbd_endio_sec (defined here) + * bm_async_io_complete (defined in drbd_bitmap.c) + * * For all these callbacks, note the following: * The callbacks will be called in irq context by the IDE drivers, * and in Softirqs/Tasklets/BH context by the SCSI drivers. diff --git a/include/linux/drbd.h b/include/linux/drbd.h index ba5c785d3f7..d18d673ebc7 100644 --- a/include/linux/drbd.h +++ b/include/linux/drbd.h @@ -53,7 +53,7 @@ extern const char *drbd_buildtag(void); -#define REL_VERSION "8.3.9" +#define REL_VERSION "8.3.10" #define API_VERSION 88 #define PRO_VERSION_MIN 86 #define PRO_VERSION_MAX 96 -- cgit v1.2.3-70-g09d2 From 8f21420ebd5ca5a751e2f606b49b0acd2a2af314 Mon Sep 17 00:00:00 2001 From: Philipp Reisner Date: Tue, 1 Mar 2011 15:52:35 +0100 Subject: drbd: Fixed handling of read errors on a 'VerifyT' node Signed-off-by: Philipp Reisner Signed-off-by: Lars Ellenberg --- drivers/block/drbd/drbd_worker.c | 28 +++++++++++++++------------- 1 file changed, 15 insertions(+), 13 deletions(-) (limited to 'drivers/block/drbd/drbd_worker.c') diff --git a/drivers/block/drbd/drbd_worker.c b/drivers/block/drbd/drbd_worker.c index 3d70d8d015d..7db29080f36 100644 --- a/drivers/block/drbd/drbd_worker.c +++ b/drivers/block/drbd/drbd_worker.c @@ -1081,25 +1081,27 @@ int w_e_end_ov_req(struct drbd_conf *mdev, struct drbd_work *w, int cancel) if (unlikely(cancel)) goto out; - if (unlikely((e->flags & EE_WAS_ERROR) != 0)) - goto out; - digest_size = crypto_hash_digestsize(mdev->verify_tfm); - /* FIXME if this allocation fails, online verify will not terminate! */ digest = kmalloc(digest_size, GFP_NOIO); - if (digest) { - drbd_csum_ee(mdev, mdev->verify_tfm, e, digest); - inc_rs_pending(mdev); - ok = drbd_send_drequest_csum(mdev, e->sector, e->size, - digest, digest_size, P_OV_REPLY); - if (!ok) - dec_rs_pending(mdev); - kfree(digest); + if (!digest) { + ok = 0; /* terminate the connection in case the allocation failed */ + goto out; } + if (likely(!(e->flags & EE_WAS_ERROR))) + drbd_csum_ee(mdev, mdev->verify_tfm, e, digest); + else + memset(digest, 0, digest_size); + + inc_rs_pending(mdev); + ok = drbd_send_drequest_csum(mdev, e->sector, e->size, + digest, digest_size, P_OV_REPLY); + if (!ok) + dec_rs_pending(mdev); + kfree(digest); + out: drbd_free_ee(mdev, e); - dec_unacked(mdev); return ok; -- cgit v1.2.3-70-g09d2 From 7961243b7bdd62d72b47eb2c0bee776c51a8a8e2 Mon Sep 17 00:00:00 2001 From: Philipp Reisner Date: Wed, 2 Mar 2011 23:14:44 +0100 Subject: drbd: Fixed handling of read errors on a 'VerifyS' node Signed-off-by: Philipp Reisner Signed-off-by: Lars Ellenberg --- drivers/block/drbd/drbd_worker.c | 4 ---- 1 file changed, 4 deletions(-) (limited to 'drivers/block/drbd/drbd_worker.c') diff --git a/drivers/block/drbd/drbd_worker.c b/drivers/block/drbd/drbd_worker.c index 7db29080f36..f7e6c92f8d0 100644 --- a/drivers/block/drbd/drbd_worker.c +++ b/drivers/block/drbd/drbd_worker.c @@ -1151,10 +1151,6 @@ int w_e_end_ov_reply(struct drbd_conf *mdev, struct drbd_work *w, int cancel) eq = !memcmp(digest, di->digest, digest_size); kfree(digest); } - } else { - ok = drbd_send_ack(mdev, P_NEG_RS_DREPLY, e); - if (__ratelimit(&drbd_ratelimit_state)) - dev_err(DEV, "Sending NegDReply. I guess it gets messy.\n"); } dec_unacked(mdev); -- cgit v1.2.3-70-g09d2