From ebd2b0cde5a4c02e2999fc3dc4a59fdd8a040fb6 Mon Sep 17 00:00:00 2001
From: Philipp Reisner <philipp.reisner@linbit.com>
Date: Wed, 25 May 2011 11:03:04 +0200
Subject: drbd: Lower log priority for an event that is definitely not an error

Signed-off-by: Philipp Reisner <philipp.reisner@linbit.com>
Signed-off-by: Lars Ellenberg <lars.ellenberg@linbit.com>
---
 drivers/block/drbd/drbd_receiver.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/block/drbd/drbd_receiver.c b/drivers/block/drbd/drbd_receiver.c
index 43beaca53179..e7ed0aa93a16 100644
--- a/drivers/block/drbd/drbd_receiver.c
+++ b/drivers/block/drbd/drbd_receiver.c
@@ -957,7 +957,7 @@ static void drbd_flush(struct drbd_conf *mdev)
 		rv = blkdev_issue_flush(mdev->ldev->backing_bdev, GFP_KERNEL,
 					NULL);
 		if (rv) {
-			dev_err(DEV, "local disk flush failed with status %d\n", rv);
+			dev_info(DEV, "local disk flush failed with status %d\n", rv);
 			/* would rather check on EOPNOTSUPP, but that is not reliable.
 			 * don't try again for ANY return value != 0
 			 * if (rv == -EOPNOTSUPP) */
-- 
cgit v1.2.3


From 1381e9a4969d8d24ce7b4bab048f3cb3fc9c1283 Mon Sep 17 00:00:00 2001
From: Lars Ellenberg <lars.ellenberg@linbit.com>
Date: Fri, 3 Jun 2011 21:13:17 +0200
Subject: drbd: cosmetic: fix accidental division instead of modulo when pretty
 printing

For large resync rates, seq_printf_with_thousands_grouping()
accidentally only produced Y,000,00Y, instead of the real numbers.

Signed-off-by: Philipp Reisner <philipp.reisner@linbit.com>
Signed-off-by: Lars Ellenberg <lars.ellenberg@linbit.com>
---
 drivers/block/drbd/drbd_proc.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/block/drbd/drbd_proc.c b/drivers/block/drbd/drbd_proc.c
index 2959cdfb77f5..869bada2ed06 100644
--- a/drivers/block/drbd/drbd_proc.c
+++ b/drivers/block/drbd/drbd_proc.c
@@ -52,7 +52,7 @@ void seq_printf_with_thousands_grouping(struct seq_file *seq, long v)
 	if (unlikely(v >= 1000000)) {
 		/* cool: > GiByte/s */
 		seq_printf(seq, "%ld,", v / 1000000);
-		v /= 1000000;
+		v %= 1000000;
 		seq_printf(seq, "%03ld,%03ld", v/1000, v % 1000);
 	} else if (likely(v >= 1000))
 		seq_printf(seq, "%ld,%03ld", v/1000, v % 1000);
-- 
cgit v1.2.3


From 7948bcdc38b9af9ef3e72199cdea1d775a9537fc Mon Sep 17 00:00:00 2001
From: Lars Ellenberg <lars.ellenberg@linbit.com>
Date: Mon, 6 Jun 2011 15:36:04 +0200
Subject: drbd: spelling fix: too small

It is not "to small", but "too small".

Signed-off-by: Philipp Reisner <philipp.reisner@linbit.com>
Signed-off-by: Lars Ellenberg <lars.ellenberg@linbit.com>
---
 drivers/block/drbd/drbd_nl.c | 8 ++++----
 include/linux/drbd.h         | 4 ++--
 2 files changed, 6 insertions(+), 6 deletions(-)

diff --git a/drivers/block/drbd/drbd_nl.c b/drivers/block/drbd/drbd_nl.c
index abfaacaaf346..f7b5f7e86a57 100644
--- a/drivers/block/drbd/drbd_nl.c
+++ b/drivers/block/drbd/drbd_nl.c
@@ -1032,7 +1032,7 @@ static int drbd_nl_disk_conf(struct drbd_conf *mdev, struct drbd_nl_cfg_req *nlp
 		dev_err(DEV, "max capacity %llu smaller than disk size %llu\n",
 			(unsigned long long) drbd_get_max_capacity(nbc),
 			(unsigned long long) nbc->dc.disk_size);
-		retcode = ERR_DISK_TO_SMALL;
+		retcode = ERR_DISK_TOO_SMALL;
 		goto fail;
 	}
 
@@ -1046,7 +1046,7 @@ static int drbd_nl_disk_conf(struct drbd_conf *mdev, struct drbd_nl_cfg_req *nlp
 	}
 
 	if (drbd_get_capacity(nbc->md_bdev) < min_md_device_sectors) {
-		retcode = ERR_MD_DISK_TO_SMALL;
+		retcode = ERR_MD_DISK_TOO_SMALL;
 		dev_warn(DEV, "refusing attach: md-device too small, "
 		     "at least %llu sectors needed for this meta-disk type\n",
 		     (unsigned long long) min_md_device_sectors);
@@ -1057,7 +1057,7 @@ static int drbd_nl_disk_conf(struct drbd_conf *mdev, struct drbd_nl_cfg_req *nlp
 	 * (we may currently be R_PRIMARY with no local disk...) */
 	if (drbd_get_max_capacity(nbc) <
 	    drbd_get_capacity(mdev->this_bdev)) {
-		retcode = ERR_DISK_TO_SMALL;
+		retcode = ERR_DISK_TOO_SMALL;
 		goto fail;
 	}
 
@@ -1138,7 +1138,7 @@ static int drbd_nl_disk_conf(struct drbd_conf *mdev, struct drbd_nl_cfg_req *nlp
 	if (drbd_md_test_flag(nbc, MDF_CONSISTENT) &&
 	    drbd_new_dev_size(mdev, nbc, 0) < nbc->md.la_size_sect) {
 		dev_warn(DEV, "refusing to truncate a consistent device\n");
-		retcode = ERR_DISK_TO_SMALL;
+		retcode = ERR_DISK_TOO_SMALL;
 		goto force_diskless_dec;
 	}
 
diff --git a/include/linux/drbd.h b/include/linux/drbd.h
index 9e5f5607eba3..cb8728b28432 100644
--- a/include/linux/drbd.h
+++ b/include/linux/drbd.h
@@ -112,8 +112,8 @@ enum drbd_ret_code {
 	ERR_OPEN_MD_DISK	= 105,
 	ERR_DISK_NOT_BDEV	= 107,
 	ERR_MD_NOT_BDEV		= 108,
-	ERR_DISK_TO_SMALL	= 111,
-	ERR_MD_DISK_TO_SMALL	= 112,
+	ERR_DISK_TOO_SMALL	= 111,
+	ERR_MD_DISK_TOO_SMALL	= 112,
 	ERR_BDCLAIM_DISK	= 114,
 	ERR_BDCLAIM_MD_DISK	= 115,
 	ERR_MD_IDX_INVALID	= 116,
-- 
cgit v1.2.3


From 5f138ce01ae6430db2e2cebd0a945dff75581d62 Mon Sep 17 00:00:00 2001
From: David Howells <dhowells@redhat.com>
Date: Wed, 15 Jun 2011 00:59:04 +0100
Subject: DRBD: Fix comparison always false warning due to long/long long
 compare

Fix warnings of the following nature in the drbd header:

In file included from drivers/block/drbd/drbd_bitmap.c:32:
drivers/block/drbd/drbd_int.h: In function 'drbd_get_syncer_progress':
drivers/block/drbd/drbd_int.h:2234: warning: comparison is always false due to limited range of data

where mdev->rs_total (an unsigned long) is being compared to 1ULL << 32, which
is always false on a 32-bit machine.

Signed-off-by: David Howells <dhowells@redhat.com>
Signed-off-by: Lars Ellenberg <lars.ellenberg@linbit.com>
Signed-off-by: Philipp Reisner <philipp.reisner@linbit.com>
---
 drivers/block/drbd/drbd_int.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/block/drbd/drbd_int.h b/drivers/block/drbd/drbd_int.h
index 8d680562ba73..2be057b8f568 100644
--- a/drivers/block/drbd/drbd_int.h
+++ b/drivers/block/drbd/drbd_int.h
@@ -2230,7 +2230,7 @@ static inline void drbd_get_syncer_progress(struct drbd_conf *mdev,
 		 * 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 int shift = mdev->rs_total > UINT_MAX ? 16 : 10;
 		unsigned long left = *bits_left >> shift;
 		unsigned long total = 1UL + (mdev->rs_total >> shift);
 		unsigned long tmp = 1000UL - left * 1000UL/total;
-- 
cgit v1.2.3


From 071cf1c9a7043660be08358e4b508f62120f224b Mon Sep 17 00:00:00 2001
From: Lars Ellenberg <lars.ellenberg@linbit.com>
Date: Mon, 20 Jun 2011 22:21:19 +0200
Subject: drbd: allow ping-timeout of up to 30 seconds

Allow up to 300 centi-seconds to be configured for the "ping timeout".
There may be setups where heavy congestion, huge buffers, and asymmetric
bandwidth limitations may need a "huge" ping-timeout as work-around
for "spurious connection loss" problems.

Signed-off-by: Philipp Reisner <philipp.reisner@linbit.com>
Signed-off-by: Lars Ellenberg <lars.ellenberg@linbit.com>
---
 include/linux/drbd_limits.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/include/linux/drbd_limits.h b/include/linux/drbd_limits.h
index 447c36752385..928c84dfaf42 100644
--- a/include/linux/drbd_limits.h
+++ b/include/linux/drbd_limits.h
@@ -60,7 +60,7 @@
 
  /* timeout for the ping packets.*/
 #define DRBD_PING_TIMEO_MIN  1
-#define DRBD_PING_TIMEO_MAX  100
+#define DRBD_PING_TIMEO_MAX  300
 #define DRBD_PING_TIMEO_DEF  5
 
   /* max number of write requests between write barriers */
-- 
cgit v1.2.3


From 07667347c89e5d971243a411ca4ed83ad14f9f9e Mon Sep 17 00:00:00 2001
From: Lars Ellenberg <lars.ellenberg@linbit.com>
Date: Tue, 21 Jun 2011 01:13:37 +0200
Subject: drbd: downgraded error printk to info

Signed-off-by: Philipp Reisner <philipp.reisner@linbit.com>
Signed-off-by: Lars Ellenberg <lars.ellenberg@linbit.com>
---
 drivers/block/drbd/drbd_main.c | 6 ++----
 1 file changed, 2 insertions(+), 4 deletions(-)

diff --git a/drivers/block/drbd/drbd_main.c b/drivers/block/drbd/drbd_main.c
index 211fc44f84be..906aca9b6659 100644
--- a/drivers/block/drbd/drbd_main.c
+++ b/drivers/block/drbd/drbd_main.c
@@ -1530,9 +1530,7 @@ static void after_state_ch(struct drbd_conf *mdev, union drbd_state os,
 				drbd_disk_str(mdev->state.disk));
 
 		if (drbd_send_state(mdev))
-			dev_warn(DEV, "Notified peer that I am detaching my disk\n");
-		else
-			dev_err(DEV, "Sending state for detaching disk failed\n");
+			dev_info(DEV, "Notified peer that I am detaching my disk\n");
 
 		drbd_rs_cancel_all(mdev);
 
@@ -1562,7 +1560,7 @@ static void after_state_ch(struct drbd_conf *mdev, union drbd_state os,
                 atomic_set(&mdev->rs_pending_cnt, 0);
 
 		if (drbd_send_state(mdev))
-			dev_warn(DEV, "Notified peer that I'm now diskless.\n");
+			dev_info(DEV, "Notified peer that I'm now diskless.\n");
 		/* corresponding get_ldev in __drbd_set_state
 		 * this may finally trigger drbd_ldev_destroy. */
 		put_ldev(mdev);
-- 
cgit v1.2.3


From 77e8fdfc188f0490754f703f80d2853c69052c12 Mon Sep 17 00:00:00 2001
From: Philipp Reisner <philipp.reisner@linbit.com>
Date: Wed, 29 Jun 2011 10:49:13 +0200
Subject: drbd: Only print sanitize state's warnings, if the state change
 happens

The reason for this change is that, with when doing
'drbdadm invalidate' on a disconnected resource caused
an "implicitly set pdsk from UpToDate to DUnknown" message,
which was missleading.

Signed-off-by: Philipp Reisner <philipp.reisner@linbit.com>
Signed-off-by: Lars Ellenberg <lars.ellenberg@linbit.com>
---
 drivers/block/drbd/drbd_main.c | 55 ++++++++++++++++++++++++++++++------------
 1 file changed, 40 insertions(+), 15 deletions(-)

diff --git a/drivers/block/drbd/drbd_main.c b/drivers/block/drbd/drbd_main.c
index 906aca9b6659..8e83a402ed5c 100644
--- a/drivers/block/drbd/drbd_main.c
+++ b/drivers/block/drbd/drbd_main.c
@@ -509,8 +509,16 @@ static enum drbd_state_rv is_valid_state(struct drbd_conf *, union drbd_state);
 static enum drbd_state_rv is_valid_state_transition(struct drbd_conf *,
 						    union drbd_state,
 						    union drbd_state);
+enum sanitize_state_warnings {
+	NO_WARNING,
+	ABORTED_ONLINE_VERIFY,
+	ABORTED_RESYNC,
+	CONNECTION_LOST_NEGOTIATING,
+	IMPLICITLY_UPGRADED_DISK,
+	IMPLICITLY_UPGRADED_PDSK,
+};
 static union drbd_state sanitize_state(struct drbd_conf *mdev, union drbd_state os,
-				       union drbd_state ns, const char **warn_sync_abort);
+				       union drbd_state ns, enum sanitize_state_warnings *warn);
 int drbd_send_state_req(struct drbd_conf *,
 			union drbd_state, union drbd_state);
 
@@ -803,6 +811,21 @@ is_valid_state_transition(struct drbd_conf *mdev, union drbd_state ns,
 	return rv;
 }
 
+static void print_sanitize_warnings(struct drbd_conf *mdev, enum sanitize_state_warnings warn)
+{
+	static const char *msg_table[] = {
+		[NO_WARNING] = "",
+		[ABORTED_ONLINE_VERIFY] = "Online-verify aborted.",
+		[ABORTED_RESYNC] = "Resync aborted.",
+		[CONNECTION_LOST_NEGOTIATING] = "Connection lost while negotiating, no data!",
+		[IMPLICITLY_UPGRADED_DISK] = "Implicitly upgraded disk",
+		[IMPLICITLY_UPGRADED_PDSK] = "Implicitly upgraded pdsk",
+	};
+
+	if (warn != NO_WARNING)
+		dev_warn(DEV, "%s\n", msg_table[warn]);
+}
+
 /**
  * sanitize_state() - Resolves implicitly necessary additional changes to a state transition
  * @mdev:	DRBD device.
@@ -814,11 +837,14 @@ is_valid_state_transition(struct drbd_conf *mdev, union drbd_state ns,
  * to D_UNKNOWN. This rule and many more along those lines are in this function.
  */
 static union drbd_state sanitize_state(struct drbd_conf *mdev, union drbd_state os,
-				       union drbd_state ns, const char **warn_sync_abort)
+				       union drbd_state ns, enum sanitize_state_warnings *warn)
 {
 	enum drbd_fencing_p fp;
 	enum drbd_disk_state disk_min, disk_max, pdsk_min, pdsk_max;
 
+	if (warn)
+		*warn = NO_WARNING;
+
 	fp = FP_DONT_CARE;
 	if (get_ldev(mdev)) {
 		fp = mdev->ldev->dc.fencing;
@@ -863,10 +889,9 @@ static union drbd_state sanitize_state(struct drbd_conf *mdev, union drbd_state
 	/* Abort resync if a disk fails/detaches */
 	if (os.conn > C_CONNECTED && ns.conn > C_CONNECTED &&
 	    (ns.disk <= D_FAILED || ns.pdsk <= D_FAILED)) {
-		if (warn_sync_abort)
-			*warn_sync_abort =
-				os.conn == C_VERIFY_S || os.conn == C_VERIFY_T ?
-				"Online-verify" : "Resync";
+		if (warn)
+			*warn =	os.conn == C_VERIFY_S || os.conn == C_VERIFY_T ?
+				ABORTED_ONLINE_VERIFY : ABORTED_RESYNC;
 		ns.conn = C_CONNECTED;
 	}
 
@@ -877,7 +902,8 @@ static union drbd_state sanitize_state(struct drbd_conf *mdev, union drbd_state
 			ns.disk = mdev->new_state_tmp.disk;
 			ns.pdsk = mdev->new_state_tmp.pdsk;
 		} else {
-			dev_alert(DEV, "Connection lost while negotiating, no data!\n");
+			if (warn)
+				*warn = CONNECTION_LOST_NEGOTIATING;
 			ns.disk = D_DISKLESS;
 			ns.pdsk = D_UNKNOWN;
 		}
@@ -959,16 +985,16 @@ static union drbd_state sanitize_state(struct drbd_conf *mdev, union drbd_state
 		ns.disk = disk_max;
 
 	if (ns.disk < disk_min) {
-		dev_warn(DEV, "Implicitly set disk from %s to %s\n",
-			 drbd_disk_str(ns.disk), drbd_disk_str(disk_min));
+		if (warn)
+			*warn = IMPLICITLY_UPGRADED_DISK;
 		ns.disk = disk_min;
 	}
 	if (ns.pdsk > pdsk_max)
 		ns.pdsk = pdsk_max;
 
 	if (ns.pdsk < pdsk_min) {
-		dev_warn(DEV, "Implicitly set pdsk from %s to %s\n",
-			 drbd_disk_str(ns.pdsk), drbd_disk_str(pdsk_min));
+		if (warn)
+			*warn = IMPLICITLY_UPGRADED_PDSK;
 		ns.pdsk = pdsk_min;
 	}
 
@@ -1045,12 +1071,12 @@ __drbd_set_state(struct drbd_conf *mdev, union drbd_state ns,
 {
 	union drbd_state os;
 	enum drbd_state_rv rv = SS_SUCCESS;
-	const char *warn_sync_abort = NULL;
+	enum sanitize_state_warnings ssw;
 	struct after_state_chg_work *ascw;
 
 	os = mdev->state;
 
-	ns = sanitize_state(mdev, os, ns, &warn_sync_abort);
+	ns = sanitize_state(mdev, os, ns, &ssw);
 
 	if (ns.i == os.i)
 		return SS_NOTHING_TO_DO;
@@ -1076,8 +1102,7 @@ __drbd_set_state(struct drbd_conf *mdev, union drbd_state ns,
 		return rv;
 	}
 
-	if (warn_sync_abort)
-		dev_warn(DEV, "%s aborted.\n", warn_sync_abort);
+	print_sanitize_warnings(mdev, ssw);
 
 	{
 	char *pbp, pb[300];
-- 
cgit v1.2.3


From 6809384c7152c34e74e29a4033826a300eb94f11 Mon Sep 17 00:00:00 2001
From: Philipp Reisner <philipp.reisner@linbit.com>
Date: Thu, 30 Jun 2011 15:43:06 +0200
Subject: drbd: Improve compatibility with drbd's older than 8.3.7

Regression introduced with 8.3.11 commit:
drbd: Take a more conservative approach when deciding max_bio_size

Never ever tell an older drbd, that we support more than 32KiB
in a single data request (packet).
Never believe an older drbd, that is supports more than 32KiB
in a single data request (packet)

Signed-off-by: Philipp Reisner <philipp.reisner@linbit.com>
Signed-off-by: Lars Ellenberg <lars.ellenberg@linbit.com>
---
 drivers/block/drbd/drbd_main.c | 4 ++++
 drivers/block/drbd/drbd_nl.c   | 7 ++++---
 2 files changed, 8 insertions(+), 3 deletions(-)

diff --git a/drivers/block/drbd/drbd_main.c b/drivers/block/drbd/drbd_main.c
index 8e83a402ed5c..a15135b9b60a 100644
--- a/drivers/block/drbd/drbd_main.c
+++ b/drivers/block/drbd/drbd_main.c
@@ -2112,6 +2112,10 @@ int drbd_send_sizes(struct drbd_conf *mdev, int trigger_reply, enum dds_flags fl
 		max_bio_size = DRBD_MAX_BIO_SIZE; /* ... multiple BIOs per peer_request */
 	}
 
+	/* Never allow old drbd (up to 8.3.7) to see more than 32KiB */
+	if (mdev->agreed_pro_version <= 94)
+		max_bio_size = min_t(int, max_bio_size, DRBD_MAX_SIZE_H80_PACKET);
+
 	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));
diff --git a/drivers/block/drbd/drbd_nl.c b/drivers/block/drbd/drbd_nl.c
index f7b5f7e86a57..cabad39f908c 100644
--- a/drivers/block/drbd/drbd_nl.c
+++ b/drivers/block/drbd/drbd_nl.c
@@ -845,9 +845,10 @@ void drbd_reconsider_max_bio_size(struct drbd_conf *mdev)
 	   Because new from 8.3.8 onwards the peer can use multiple
 	   BIOs for a single peer_request */
 	if (mdev->state.conn >= C_CONNECTED) {
-		if (mdev->agreed_pro_version < 94)
-			peer = mdev->peer_max_bio_size;
-		else if (mdev->agreed_pro_version == 94)
+		if (mdev->agreed_pro_version < 94) {
+			peer = min_t(int, mdev->peer_max_bio_size, DRBD_MAX_SIZE_H80_PACKET);
+			/* Correct old drbd (up to 8.3.7) if it believes it can do more than 32KiB */
+		} else if (mdev->agreed_pro_version == 94)
 			peer = DRBD_MAX_SIZE_H80_PACKET;
 		else /* drbd 8.3.8 onwards */
 			peer = DRBD_MAX_BIO_SIZE;
-- 
cgit v1.2.3


From 6d7e32f56806ad58006720ed98a433b2047444da Mon Sep 17 00:00:00 2001
From: Philipp Reisner <philipp.reisner@linbit.com>
Date: Tue, 15 Mar 2011 10:25:18 +0100
Subject: drbd: Keep a reference to barrier acked requests

Signed-off-by: Philipp Reisner <philipp.reisner@linbit.com>
Signed-off-by: Lars Ellenberg <lars.ellenberg@linbit.com>
---
 drivers/block/drbd/drbd_int.h  |  1 +
 drivers/block/drbd/drbd_main.c | 23 +++++++++++++++++++++--
 drivers/block/drbd/drbd_req.h  |  1 +
 3 files changed, 23 insertions(+), 2 deletions(-)

diff --git a/drivers/block/drbd/drbd_int.h b/drivers/block/drbd/drbd_int.h
index 2be057b8f568..5bb5114dd23c 100644
--- a/drivers/block/drbd/drbd_int.h
+++ b/drivers/block/drbd/drbd_int.h
@@ -1022,6 +1022,7 @@ struct drbd_conf {
 	struct drbd_tl_epoch *newest_tle;
 	struct drbd_tl_epoch *oldest_tle;
 	struct list_head out_of_sequence_requests;
+	struct list_head barrier_acked_requests;
 	struct hlist_head *tl_hash;
 	unsigned int tl_hash_s;
 
diff --git a/drivers/block/drbd/drbd_main.c b/drivers/block/drbd/drbd_main.c
index a15135b9b60a..8e489a6d022e 100644
--- a/drivers/block/drbd/drbd_main.c
+++ b/drivers/block/drbd/drbd_main.c
@@ -208,6 +208,7 @@ static int tl_init(struct drbd_conf *mdev)
 	mdev->oldest_tle = b;
 	mdev->newest_tle = b;
 	INIT_LIST_HEAD(&mdev->out_of_sequence_requests);
+	INIT_LIST_HEAD(&mdev->barrier_acked_requests);
 
 	mdev->tl_hash = NULL;
 	mdev->tl_hash_s = 0;
@@ -311,7 +312,7 @@ void tl_release(struct drbd_conf *mdev, unsigned int barrier_nr,
 	   These have been list_move'd to the out_of_sequence_requests list in
 	   _req_mod(, barrier_acked) above.
 	   */
-	list_del_init(&b->requests);
+	list_splice_init(&b->requests, &mdev->barrier_acked_requests);
 
 	nob = b->next;
 	if (test_and_clear_bit(CREATE_BARRIER, &mdev->flags)) {
@@ -343,7 +344,7 @@ bail:
  * @what:       The action/event to perform with all request objects
  *
  * @what might be one of connection_lost_while_pending, resend, fail_frozen_disk_io,
- * restart_frozen_disk_io.
+ * restart_frozen_disk_io, abort_disk_io.
  */
 static void _tl_restart(struct drbd_conf *mdev, enum drbd_req_event what)
 {
@@ -411,6 +412,24 @@ static void _tl_restart(struct drbd_conf *mdev, enum drbd_req_event what)
 		b = tmp;
 		list_splice(&carry_reads, &b->requests);
 	}
+
+	/* Actions operating on the disk state, also want to work on
+	   requests that got barrier acked. */
+	switch (what) {
+	case abort_disk_io:
+	case fail_frozen_disk_io:
+	case restart_frozen_disk_io:
+		list_for_each_safe(le, tle, &mdev->barrier_acked_requests) {
+			req = list_entry(le, struct drbd_request, tl_requests);
+			_req_mod(req, what);
+		}
+
+	case connection_lost_while_pending:
+	case resend:
+		break;
+	default:
+		dev_err(DEV, "what = %d in _tl_restart()\n", what);
+	}
 }
 
 
diff --git a/drivers/block/drbd/drbd_req.h b/drivers/block/drbd/drbd_req.h
index 68a234a5fdc5..74c5b9f14d61 100644
--- a/drivers/block/drbd/drbd_req.h
+++ b/drivers/block/drbd/drbd_req.h
@@ -105,6 +105,7 @@ enum drbd_req_event {
 	read_completed_with_error,
 	read_ahead_completed_with_error,
 	write_completed_with_error,
+	abort_disk_io,
 	completed_ok,
 	resend,
 	fail_frozen_disk_io,
-- 
cgit v1.2.3


From 2b4dd36fbae7203a0d503a6cede1f4ce17aa72ac Mon Sep 17 00:00:00 2001
From: Philipp Reisner <philipp.reisner@linbit.com>
Date: Mon, 14 Mar 2011 13:01:50 +0100
Subject: drbd: Immediately allow completion of IOs, that wait for IO
 completions on a failed disk

Signed-off-by: Philipp Reisner <philipp.reisner@linbit.com>
Signed-off-by: Lars Ellenberg <lars.ellenberg@linbit.com>
---
 drivers/block/drbd/drbd_main.c | 10 ++++++++++
 drivers/block/drbd/drbd_req.c  | 20 ++++++++++++++++----
 drivers/block/drbd/drbd_req.h  | 18 +++++++++++-------
 3 files changed, 37 insertions(+), 11 deletions(-)

diff --git a/drivers/block/drbd/drbd_main.c b/drivers/block/drbd/drbd_main.c
index 8e489a6d022e..16969c2b96cd 100644
--- a/drivers/block/drbd/drbd_main.c
+++ b/drivers/block/drbd/drbd_main.c
@@ -368,6 +368,12 @@ static void _tl_restart(struct drbd_conf *mdev, enum drbd_req_event what)
 		}
 		tmp = b->next;
 
+		if (what == abort_disk_io) {
+			/* Only walk the TL, leave barrier objects in place */
+			b = tmp;
+			continue;
+		}
+
 		if (n_writes) {
 			if (what == resend) {
 				b->n_writes = n_writes;
@@ -1565,6 +1571,10 @@ static void after_state_ch(struct drbd_conf *mdev, union drbd_state os,
 		eh = mdev->ldev->dc.on_io_error;
 		was_io_error = test_and_clear_bit(WAS_IO_ERROR, &mdev->flags);
 
+		/* Immediately allow completion of all application IO, that waits
+		   for completion from the local disk. */
+		tl_restart(mdev, abort_disk_io);
+
 		/* current state still has to be D_FAILED,
 		 * there is only one way out: to D_DISKLESS,
 		 * and that may only happen after our put_ldev below. */
diff --git a/drivers/block/drbd/drbd_req.c b/drivers/block/drbd/drbd_req.c
index 4a0f314086e5..1a8aac4b0c2f 100644
--- a/drivers/block/drbd/drbd_req.c
+++ b/drivers/block/drbd/drbd_req.c
@@ -214,8 +214,7 @@ void _req_may_be_done(struct drbd_request *req, struct bio_and_error *m)
 {
 	const unsigned long s = req->rq_state;
 	struct drbd_conf *mdev = req->mdev;
-	/* only WRITES may end up here without a master bio (on barrier ack) */
-	int rw = req->master_bio ? bio_data_dir(req->master_bio) : WRITE;
+	int rw = req->rq_state & RQ_WRITE ? WRITE : READ;
 
 	/* we must not complete the master bio, while it is
 	 *	still being processed by _drbd_send_zc_bio (drbd_send_dblock)
@@ -230,7 +229,7 @@ void _req_may_be_done(struct drbd_request *req, struct bio_and_error *m)
 		return;
 	if (s & RQ_NET_PENDING)
 		return;
-	if (s & RQ_LOCAL_PENDING)
+	if (s & RQ_LOCAL_PENDING && !(s & RQ_LOCAL_ABORTED))
 		return;
 
 	if (req->master_bio) {
@@ -277,6 +276,9 @@ void _req_may_be_done(struct drbd_request *req, struct bio_and_error *m)
 		req->master_bio = NULL;
 	}
 
+	if (s & RQ_LOCAL_PENDING)
+		return;
+
 	if ((s & RQ_NET_MASK) == 0 || (s & RQ_NET_DONE)) {
 		/* this is disconnected (local only) operation,
 		 * or protocol C P_WRITE_ACK,
@@ -429,7 +431,7 @@ int __req_mod(struct drbd_request *req, enum drbd_req_event what,
 		break;
 
 	case completed_ok:
-		if (bio_data_dir(req->master_bio) == WRITE)
+		if (req->rq_state & RQ_WRITE)
 			mdev->writ_cnt += req->size>>9;
 		else
 			mdev->read_cnt += req->size>>9;
@@ -441,6 +443,14 @@ int __req_mod(struct drbd_request *req, enum drbd_req_event what,
 		put_ldev(mdev);
 		break;
 
+	case abort_disk_io:
+		req->rq_state |= RQ_LOCAL_ABORTED;
+		if (req->rq_state & RQ_WRITE)
+			_req_may_be_done_not_susp(req, m);
+		else
+			goto goto_queue_for_net_read;
+		break;
+
 	case write_completed_with_error:
 		req->rq_state |= RQ_LOCAL_COMPLETED;
 		req->rq_state &= ~RQ_LOCAL_PENDING;
@@ -469,6 +479,8 @@ int __req_mod(struct drbd_request *req, enum drbd_req_event what,
 		__drbd_chk_io_error(mdev, false);
 		put_ldev(mdev);
 
+	goto_queue_for_net_read:
+
 		/* no point in retrying if there is no good remote data,
 		 * or we have no connection. */
 		if (mdev->state.pdsk != D_UP_TO_DATE) {
diff --git a/drivers/block/drbd/drbd_req.h b/drivers/block/drbd/drbd_req.h
index 74c5b9f14d61..3d2111919486 100644
--- a/drivers/block/drbd/drbd_req.h
+++ b/drivers/block/drbd/drbd_req.h
@@ -119,18 +119,21 @@ enum drbd_req_event {
  * same time, so we should hold the request lock anyways.
  */
 enum drbd_req_state_bits {
-	/* 210
-	 * 000: no local possible
-	 * 001: to be submitted
+	/* 3210
+	 * 0000: no local possible
+	 * 0001: to be submitted
 	 *    UNUSED, we could map: 011: submitted, completion still pending
-	 * 110: completed ok
-	 * 010: completed with error
+	 * 0110: completed ok
+	 * 0010: completed with error
+	 * 1001: Aborted (before completion)
+	 * 1x10: Aborted and completed -> free
 	 */
 	__RQ_LOCAL_PENDING,
 	__RQ_LOCAL_COMPLETED,
 	__RQ_LOCAL_OK,
+	__RQ_LOCAL_ABORTED,
 
-	/* 76543
+	/* 87654
 	 * 00000: no network possible
 	 * 00001: to be send
 	 * 00011: to be send, on worker queue
@@ -200,8 +203,9 @@ enum drbd_req_state_bits {
 #define RQ_LOCAL_PENDING   (1UL << __RQ_LOCAL_PENDING)
 #define RQ_LOCAL_COMPLETED (1UL << __RQ_LOCAL_COMPLETED)
 #define RQ_LOCAL_OK        (1UL << __RQ_LOCAL_OK)
+#define RQ_LOCAL_ABORTED   (1UL << __RQ_LOCAL_ABORTED)
 
-#define RQ_LOCAL_MASK      ((RQ_LOCAL_OK << 1)-1) /* 0x07 */
+#define RQ_LOCAL_MASK      ((RQ_LOCAL_ABORTED << 1)-1)
 
 #define RQ_NET_PENDING     (1UL << __RQ_NET_PENDING)
 #define RQ_NET_QUEUED      (1UL << __RQ_NET_QUEUED)
-- 
cgit v1.2.3


From cc94c65015022e7329e80e057e20848581d3f2a5 Mon Sep 17 00:00:00 2001
From: Philipp Reisner <philipp.reisner@linbit.com>
Date: Sun, 26 Jun 2011 11:20:27 +0200
Subject: drbd: moved md_io into mdev

Signed-off-by: Philipp Reisner <philipp.reisner@linbit.com>
Signed-off-by: Lars Ellenberg <lars.ellenberg@linbit.com>
---
 drivers/block/drbd/drbd_actlog.c | 12 +++++-------
 drivers/block/drbd/drbd_int.h    |  4 ++--
 drivers/block/drbd/drbd_worker.c |  3 +++
 3 files changed, 10 insertions(+), 9 deletions(-)

diff --git a/drivers/block/drbd/drbd_actlog.c b/drivers/block/drbd/drbd_actlog.c
index cf0e63dd97da..c64ca6b956cf 100644
--- a/drivers/block/drbd/drbd_actlog.c
+++ b/drivers/block/drbd/drbd_actlog.c
@@ -71,12 +71,10 @@ static int _drbd_md_sync_page_io(struct drbd_conf *mdev,
 				 int rw, int size)
 {
 	struct bio *bio;
-	struct drbd_md_io md_io;
 	int ok;
 
-	md_io.mdev = mdev;
-	init_completion(&md_io.event);
-	md_io.error = 0;
+	init_completion(&mdev->md_io.event);
+	mdev->md_io.error = 0;
 
 	if ((rw & WRITE) && !test_bit(MD_NO_FUA, &mdev->flags))
 		rw |= REQ_FUA | REQ_FLUSH;
@@ -88,7 +86,7 @@ static int _drbd_md_sync_page_io(struct drbd_conf *mdev,
 	ok = (bio_add_page(bio, page, size, 0) == size);
 	if (!ok)
 		goto out;
-	bio->bi_private = &md_io;
+	bio->bi_private = &mdev->md_io;
 	bio->bi_end_io = drbd_md_io_complete;
 	bio->bi_rw = rw;
 
@@ -96,8 +94,8 @@ static int _drbd_md_sync_page_io(struct drbd_conf *mdev,
 		bio_endio(bio, -EIO);
 	else
 		submit_bio(rw, bio);
-	wait_for_completion(&md_io.event);
-	ok = bio_flagged(bio, BIO_UPTODATE) && md_io.error == 0;
+	wait_for_completion(&mdev->md_io.event);
+	ok = bio_flagged(bio, BIO_UPTODATE) && mdev->md_io.error == 0;
 
  out:
 	bio_put(bio);
diff --git a/drivers/block/drbd/drbd_int.h b/drivers/block/drbd/drbd_int.h
index 5bb5114dd23c..91e69ffd5566 100644
--- a/drivers/block/drbd/drbd_int.h
+++ b/drivers/block/drbd/drbd_int.h
@@ -938,7 +938,6 @@ struct drbd_backing_dev {
 };
 
 struct drbd_md_io {
-	struct drbd_conf *mdev;
 	struct completion event;
 	int error;
 };
@@ -1095,7 +1094,8 @@ struct drbd_conf {
 	wait_queue_head_t ee_wait;
 	struct page *md_io_page;	/* one page buffer for md_io */
 	struct page *md_io_tmpp;	/* for logical_block_size != 512 */
-	struct mutex md_io_mutex;	/* protects the md_io_buffer */
+	struct drbd_md_io md_io;
+	struct mutex md_io_mutex;	/* protects the md_io, md_io_page and md_io_tmpp */
 	spinlock_t al_lock;
 	wait_queue_head_t al_wait;
 	struct lru_cache *act_log;	/* activity log */
diff --git a/drivers/block/drbd/drbd_worker.c b/drivers/block/drbd/drbd_worker.c
index 4d3e6f6213ba..eedfe311c403 100644
--- a/drivers/block/drbd/drbd_worker.c
+++ b/drivers/block/drbd/drbd_worker.c
@@ -70,8 +70,11 @@ rwlock_t global_state_lock;
 void drbd_md_io_complete(struct bio *bio, int error)
 {
 	struct drbd_md_io *md_io;
+	struct drbd_conf *mdev;
 
 	md_io = (struct drbd_md_io *)bio->bi_private;
+	mdev = container_of(md_io, struct drbd_conf, md_io);
+
 	md_io->error = error;
 
 	complete(&md_io->event);
-- 
cgit v1.2.3


From e17117310b73ce6d2340ad46a539d3896a2d6de8 Mon Sep 17 00:00:00 2001
From: Philipp Reisner <philipp.reisner@linbit.com>
Date: Mon, 27 Jun 2011 11:51:46 +0200
Subject: drbd: Replaced md_io_mutex by an atomic: md_io_in_use

The new function drbd_md_get_buffer() aborts waiting for the buffer
in case the disk failes in the meantime.

Signed-off-by: Philipp Reisner <philipp.reisner@linbit.com>
Signed-off-by: Lars Ellenberg <lars.ellenberg@linbit.com>
---
 drivers/block/drbd/drbd_actlog.c | 44 +++++++++++++++++++++++++++++++---------
 drivers/block/drbd/drbd_int.h    |  6 ++++--
 drivers/block/drbd/drbd_main.c   | 19 ++++++++++-------
 drivers/block/drbd/drbd_worker.c |  1 +
 4 files changed, 51 insertions(+), 19 deletions(-)

diff --git a/drivers/block/drbd/drbd_actlog.c b/drivers/block/drbd/drbd_actlog.c
index c64ca6b956cf..4271352dd72b 100644
--- a/drivers/block/drbd/drbd_actlog.c
+++ b/drivers/block/drbd/drbd_actlog.c
@@ -65,6 +65,23 @@ struct drbd_atodb_wait {
 
 int w_al_write_transaction(struct drbd_conf *, struct drbd_work *, int);
 
+void *drbd_md_get_buffer(struct drbd_conf *mdev)
+{
+	int r;
+
+	wait_event(mdev->misc_wait,
+		   (r = atomic_cmpxchg(&mdev->md_io_in_use, 0, 1)) == 0 ||
+		   mdev->state.disk <= D_FAILED);
+
+	return r ? NULL : page_address(mdev->md_io_page);
+}
+
+void drbd_md_put_buffer(struct drbd_conf *mdev)
+{
+	if (atomic_dec_and_test(&mdev->md_io_in_use))
+		wake_up(&mdev->misc_wait);
+}
+
 static int _drbd_md_sync_page_io(struct drbd_conf *mdev,
 				 struct drbd_backing_dev *bdev,
 				 struct page *page, sector_t sector,
@@ -90,6 +107,7 @@ static int _drbd_md_sync_page_io(struct drbd_conf *mdev,
 	bio->bi_end_io = drbd_md_io_complete;
 	bio->bi_rw = rw;
 
+	atomic_inc(&mdev->md_io_in_use); /* drbd_md_put_buffer() is in the completion handler */
 	if (drbd_insert_fault(mdev, (rw & WRITE) ? DRBD_FAULT_MD_WR : DRBD_FAULT_MD_RD))
 		bio_endio(bio, -EIO);
 	else
@@ -109,7 +127,7 @@ int drbd_md_sync_page_io(struct drbd_conf *mdev, struct drbd_backing_dev *bdev,
 	int offset = 0;
 	struct page *iop = mdev->md_io_page;
 
-	D_ASSERT(mutex_is_locked(&mdev->md_io_mutex));
+	D_ASSERT(atomic_read(&mdev->md_io_in_use) == 1);
 
 	BUG_ON(!bdev->md_bdev);
 
@@ -326,8 +344,13 @@ w_al_write_transaction(struct drbd_conf *mdev, struct drbd_work *w, int unused)
 		return 1;
 	}
 
-	mutex_lock(&mdev->md_io_mutex); /* protects md_io_buffer, al_tr_cycle, ... */
-	buffer = (struct al_transaction *)page_address(mdev->md_io_page);
+	buffer = drbd_md_get_buffer(mdev); /* protects md_io_buffer, al_tr_cycle, ... */
+	if (!buffer) {
+		dev_err(DEV, "disk failed while waiting for md_io buffer\n");
+		complete(&((struct update_al_work *)w)->event);
+		put_ldev(mdev);
+		return 1;
+	}
 
 	buffer->magic = __constant_cpu_to_be32(DRBD_MAGIC);
 	buffer->tr_number = cpu_to_be32(mdev->al_tr_number);
@@ -372,7 +395,7 @@ w_al_write_transaction(struct drbd_conf *mdev, struct drbd_work *w, int unused)
 	D_ASSERT(mdev->al_tr_pos < MD_AL_MAX_SIZE);
 	mdev->al_tr_number++;
 
-	mutex_unlock(&mdev->md_io_mutex);
+	drbd_md_put_buffer(mdev);
 
 	complete(&((struct update_al_work *)w)->event);
 	put_ldev(mdev);
@@ -441,8 +464,9 @@ int drbd_al_read_log(struct drbd_conf *mdev, struct drbd_backing_dev *bdev)
 	/* lock out all other meta data io for now,
 	 * and make sure the page is mapped.
 	 */
-	mutex_lock(&mdev->md_io_mutex);
-	buffer = page_address(mdev->md_io_page);
+	buffer = drbd_md_get_buffer(mdev);
+	if (!buffer)
+		return 0;
 
 	/* Find the valid transaction in the log */
 	for (i = 0; i <= mx; i++) {
@@ -450,7 +474,7 @@ int drbd_al_read_log(struct drbd_conf *mdev, struct drbd_backing_dev *bdev)
 		if (rv == 0)
 			continue;
 		if (rv == -1) {
-			mutex_unlock(&mdev->md_io_mutex);
+			drbd_md_put_buffer(mdev);
 			return 0;
 		}
 		cnr = be32_to_cpu(buffer->tr_number);
@@ -476,7 +500,7 @@ int drbd_al_read_log(struct drbd_conf *mdev, struct drbd_backing_dev *bdev)
 
 	if (!found_valid) {
 		dev_warn(DEV, "No usable activity log found.\n");
-		mutex_unlock(&mdev->md_io_mutex);
+		drbd_md_put_buffer(mdev);
 		return 1;
 	}
 
@@ -491,7 +515,7 @@ int drbd_al_read_log(struct drbd_conf *mdev, struct drbd_backing_dev *bdev)
 		rv = drbd_al_read_tr(mdev, bdev, buffer, i);
 		ERR_IF(rv == 0) goto cancel;
 		if (rv == -1) {
-			mutex_unlock(&mdev->md_io_mutex);
+			drbd_md_put_buffer(mdev);
 			return 0;
 		}
 
@@ -532,7 +556,7 @@ cancel:
 		mdev->al_tr_pos = 0;
 
 	/* ok, we are done with it */
-	mutex_unlock(&mdev->md_io_mutex);
+	drbd_md_put_buffer(mdev);
 
 	dev_info(DEV, "Found %d transactions (%d active extents) in activity log.\n",
 	     transactions, active_extents);
diff --git a/drivers/block/drbd/drbd_int.h b/drivers/block/drbd/drbd_int.h
index 91e69ffd5566..55cae74911e8 100644
--- a/drivers/block/drbd/drbd_int.h
+++ b/drivers/block/drbd/drbd_int.h
@@ -1095,7 +1095,7 @@ struct drbd_conf {
 	struct page *md_io_page;	/* one page buffer for md_io */
 	struct page *md_io_tmpp;	/* for logical_block_size != 512 */
 	struct drbd_md_io md_io;
-	struct mutex md_io_mutex;	/* protects the md_io, md_io_page and md_io_tmpp */
+	atomic_t md_io_in_use;		/* protects the md_io, md_io_page and md_io_tmpp */
 	spinlock_t al_lock;
 	wait_queue_head_t al_wait;
 	struct lru_cache *act_log;	/* activity log */
@@ -1537,8 +1537,10 @@ extern void resume_next_sg(struct drbd_conf *mdev);
 extern void suspend_other_sg(struct drbd_conf *mdev);
 extern int drbd_resync_finished(struct drbd_conf *mdev);
 /* maybe rather drbd_main.c ? */
+extern void *drbd_md_get_buffer(struct drbd_conf *mdev);
+extern void drbd_md_put_buffer(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);
+				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);
 
diff --git a/drivers/block/drbd/drbd_main.c b/drivers/block/drbd/drbd_main.c
index 16969c2b96cd..1b59ab3ab9c7 100644
--- a/drivers/block/drbd/drbd_main.c
+++ b/drivers/block/drbd/drbd_main.c
@@ -3043,8 +3043,8 @@ void drbd_init_set_defaults(struct drbd_conf *mdev)
 	atomic_set(&mdev->rs_sect_in, 0);
 	atomic_set(&mdev->rs_sect_ev, 0);
 	atomic_set(&mdev->ap_in_flight, 0);
+	atomic_set(&mdev->md_io_in_use, 0);
 
-	mutex_init(&mdev->md_io_mutex);
 	mutex_init(&mdev->data.mutex);
 	mutex_init(&mdev->meta.mutex);
 	sema_init(&mdev->data.work.s, 0);
@@ -3722,8 +3722,10 @@ void drbd_md_sync(struct drbd_conf *mdev)
 	if (!get_ldev_if_state(mdev, D_FAILED))
 		return;
 
-	mutex_lock(&mdev->md_io_mutex);
-	buffer = (struct meta_data_on_disk *)page_address(mdev->md_io_page);
+	buffer = drbd_md_get_buffer(mdev);
+	if (!buffer)
+		goto out;
+
 	memset(buffer, 0, 512);
 
 	buffer->la_size = cpu_to_be64(drbd_get_capacity(mdev->this_bdev));
@@ -3754,7 +3756,8 @@ void drbd_md_sync(struct drbd_conf *mdev)
 	 * since we updated it on metadata. */
 	mdev->ldev->md.la_size_sect = drbd_get_capacity(mdev->this_bdev);
 
-	mutex_unlock(&mdev->md_io_mutex);
+	drbd_md_put_buffer(mdev);
+out:
 	put_ldev(mdev);
 }
 
@@ -3774,8 +3777,9 @@ int drbd_md_read(struct drbd_conf *mdev, struct drbd_backing_dev *bdev)
 	if (!get_ldev_if_state(mdev, D_ATTACHING))
 		return ERR_IO_MD_DISK;
 
-	mutex_lock(&mdev->md_io_mutex);
-	buffer = (struct meta_data_on_disk *)page_address(mdev->md_io_page);
+	buffer = drbd_md_get_buffer(mdev);
+	if (!buffer)
+		goto out;
 
 	if (!drbd_md_sync_page_io(mdev, bdev, bdev->md.md_offset, READ)) {
 		/* NOTE: can't do normal error processing here as this is
@@ -3836,7 +3840,8 @@ int drbd_md_read(struct drbd_conf *mdev, struct drbd_backing_dev *bdev)
 		mdev->sync_conf.al_extents = 127;
 
  err:
-	mutex_unlock(&mdev->md_io_mutex);
+	drbd_md_put_buffer(mdev);
+ out:
 	put_ldev(mdev);
 
 	return rv;
diff --git a/drivers/block/drbd/drbd_worker.c b/drivers/block/drbd/drbd_worker.c
index eedfe311c403..94a128b49f4f 100644
--- a/drivers/block/drbd/drbd_worker.c
+++ b/drivers/block/drbd/drbd_worker.c
@@ -78,6 +78,7 @@ void drbd_md_io_complete(struct bio *bio, int error)
 	md_io->error = error;
 
 	complete(&md_io->event);
+	drbd_md_put_buffer(mdev);
 }
 
 /* reads on behalf of the partner,
-- 
cgit v1.2.3


From 0c464425158482647226fb30708c68fffc061585 Mon Sep 17 00:00:00 2001
From: Philipp Reisner <philipp.reisner@linbit.com>
Date: Sun, 26 Jun 2011 22:26:31 +0200
Subject: drbd: Implemented wait_until_done_or_disk_failure()

Signed-off-by: Philipp Reisner <philipp.reisner@linbit.com>
Signed-off-by: Lars Ellenberg <lars.ellenberg@linbit.com>
---
 drivers/block/drbd/drbd_actlog.c | 17 ++++++++++++++---
 drivers/block/drbd/drbd_int.h    |  3 ++-
 drivers/block/drbd/drbd_worker.c |  3 ++-
 3 files changed, 18 insertions(+), 5 deletions(-)

diff --git a/drivers/block/drbd/drbd_actlog.c b/drivers/block/drbd/drbd_actlog.c
index 4271352dd72b..aee599fad960 100644
--- a/drivers/block/drbd/drbd_actlog.c
+++ b/drivers/block/drbd/drbd_actlog.c
@@ -82,6 +82,17 @@ void drbd_md_put_buffer(struct drbd_conf *mdev)
 		wake_up(&mdev->misc_wait);
 }
 
+static bool md_io_allowed(struct drbd_conf *mdev)
+{
+	enum drbd_disk_state ds = mdev->state.disk;
+	return ds >= D_NEGOTIATING || ds == D_ATTACHING;
+}
+
+void wait_until_done_or_disk_failure(struct drbd_conf *mdev, unsigned int *done)
+{
+	wait_event(mdev->misc_wait, *done || !md_io_allowed(mdev));
+}
+
 static int _drbd_md_sync_page_io(struct drbd_conf *mdev,
 				 struct drbd_backing_dev *bdev,
 				 struct page *page, sector_t sector,
@@ -90,8 +101,8 @@ static int _drbd_md_sync_page_io(struct drbd_conf *mdev,
 	struct bio *bio;
 	int ok;
 
-	init_completion(&mdev->md_io.event);
-	mdev->md_io.error = 0;
+	mdev->md_io.done = 0;
+	mdev->md_io.error = -ENODEV;
 
 	if ((rw & WRITE) && !test_bit(MD_NO_FUA, &mdev->flags))
 		rw |= REQ_FUA | REQ_FLUSH;
@@ -112,7 +123,7 @@ static int _drbd_md_sync_page_io(struct drbd_conf *mdev,
 		bio_endio(bio, -EIO);
 	else
 		submit_bio(rw, bio);
-	wait_for_completion(&mdev->md_io.event);
+	wait_until_done_or_disk_failure(mdev, &mdev->md_io.done);
 	ok = bio_flagged(bio, BIO_UPTODATE) && mdev->md_io.error == 0;
 
  out:
diff --git a/drivers/block/drbd/drbd_int.h b/drivers/block/drbd/drbd_int.h
index 55cae74911e8..10ea9e388246 100644
--- a/drivers/block/drbd/drbd_int.h
+++ b/drivers/block/drbd/drbd_int.h
@@ -938,7 +938,7 @@ struct drbd_backing_dev {
 };
 
 struct drbd_md_io {
-	struct completion event;
+	unsigned int done;
 	int error;
 };
 
@@ -1541,6 +1541,7 @@ extern void *drbd_md_get_buffer(struct drbd_conf *mdev);
 extern void drbd_md_put_buffer(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 wait_until_done_or_disk_failure(struct drbd_conf *mdev, unsigned int *done);
 extern void drbd_ov_oos_found(struct drbd_conf*, sector_t, int);
 extern void drbd_rs_controller_reset(struct drbd_conf *mdev);
 
diff --git a/drivers/block/drbd/drbd_worker.c b/drivers/block/drbd/drbd_worker.c
index 94a128b49f4f..ee8303680dd8 100644
--- a/drivers/block/drbd/drbd_worker.c
+++ b/drivers/block/drbd/drbd_worker.c
@@ -77,7 +77,8 @@ void drbd_md_io_complete(struct bio *bio, int error)
 
 	md_io->error = error;
 
-	complete(&md_io->event);
+	md_io->done = 1;
+	wake_up(&mdev->misc_wait);
 	drbd_md_put_buffer(mdev);
 }
 
-- 
cgit v1.2.3


From 4a2fe568b5428abc56d7d172e3571e33d8ab7265 Mon Sep 17 00:00:00 2001
From: Philipp Reisner <philipp.reisner@linbit.com>
Date: Mon, 4 Jul 2011 11:14:31 +0200
Subject: drbd: Keep a reference to the bio until the completion handler
 finished

Signed-off-by: Philipp Reisner <philipp.reisner@linbit.com>
Signed-off-by: Lars Ellenberg <lars.ellenberg@linbit.com>
---
 drivers/block/drbd/drbd_actlog.c | 1 +
 drivers/block/drbd/drbd_worker.c | 1 +
 2 files changed, 2 insertions(+)

diff --git a/drivers/block/drbd/drbd_actlog.c b/drivers/block/drbd/drbd_actlog.c
index aee599fad960..528342c82ba1 100644
--- a/drivers/block/drbd/drbd_actlog.c
+++ b/drivers/block/drbd/drbd_actlog.c
@@ -118,6 +118,7 @@ static int _drbd_md_sync_page_io(struct drbd_conf *mdev,
 	bio->bi_end_io = drbd_md_io_complete;
 	bio->bi_rw = rw;
 
+	bio_get(bio); /* one bio_put() is in the completion handler */
 	atomic_inc(&mdev->md_io_in_use); /* drbd_md_put_buffer() is in the completion handler */
 	if (drbd_insert_fault(mdev, (rw & WRITE) ? DRBD_FAULT_MD_WR : DRBD_FAULT_MD_RD))
 		bio_endio(bio, -EIO);
diff --git a/drivers/block/drbd/drbd_worker.c b/drivers/block/drbd/drbd_worker.c
index ee8303680dd8..6dcd9f6e78c8 100644
--- a/drivers/block/drbd/drbd_worker.c
+++ b/drivers/block/drbd/drbd_worker.c
@@ -79,6 +79,7 @@ void drbd_md_io_complete(struct bio *bio, int error)
 
 	md_io->done = 1;
 	wake_up(&mdev->misc_wait);
+	bio_put(bio);
 	drbd_md_put_buffer(mdev);
 }
 
-- 
cgit v1.2.3


From b2057629ea96c33e4ae38102ecd0f27ed9a3c3ef Mon Sep 17 00:00:00 2001
From: Philipp Reisner <philipp.reisner@linbit.com>
Date: Mon, 27 Jun 2011 12:23:33 +0200
Subject: drbd: Hold a reference to ldev while doing meta-data IO

Signed-off-by: Philipp Reisner <philipp.reisner@linbit.com>
Signed-off-by: Lars Ellenberg <lars.ellenberg@linbit.com>
---
 drivers/block/drbd/drbd_actlog.c | 6 ++++++
 drivers/block/drbd/drbd_worker.c | 1 +
 2 files changed, 7 insertions(+)

diff --git a/drivers/block/drbd/drbd_actlog.c b/drivers/block/drbd/drbd_actlog.c
index 528342c82ba1..3d7c2153daca 100644
--- a/drivers/block/drbd/drbd_actlog.c
+++ b/drivers/block/drbd/drbd_actlog.c
@@ -118,6 +118,12 @@ static int _drbd_md_sync_page_io(struct drbd_conf *mdev,
 	bio->bi_end_io = drbd_md_io_complete;
 	bio->bi_rw = rw;
 
+	if (!get_ldev_if_state(mdev, D_ATTACHING)) {  /* Corresponding put_ldev in drbd_md_io_complete() */
+		dev_err(DEV, "ASSERT FAILED: get_ldev_if_state() == 1 in _drbd_md_sync_page_io()\n");
+		ok = 0;
+		goto out;
+	}
+
 	bio_get(bio); /* one bio_put() is in the completion handler */
 	atomic_inc(&mdev->md_io_in_use); /* drbd_md_put_buffer() is in the completion handler */
 	if (drbd_insert_fault(mdev, (rw & WRITE) ? DRBD_FAULT_MD_WR : DRBD_FAULT_MD_RD))
diff --git a/drivers/block/drbd/drbd_worker.c b/drivers/block/drbd/drbd_worker.c
index 6dcd9f6e78c8..933091ffefca 100644
--- a/drivers/block/drbd/drbd_worker.c
+++ b/drivers/block/drbd/drbd_worker.c
@@ -81,6 +81,7 @@ void drbd_md_io_complete(struct bio *bio, int error)
 	wake_up(&mdev->misc_wait);
 	bio_put(bio);
 	drbd_md_put_buffer(mdev);
+	put_ldev(mdev);
 }
 
 /* reads on behalf of the partner,
-- 
cgit v1.2.3


From d1f3779bbeae70b9552c9ac70d6ec8c4d3545615 Mon Sep 17 00:00:00 2001
From: Philipp Reisner <philipp.reisner@linbit.com>
Date: Mon, 27 Jun 2011 14:56:52 +0200
Subject: drbd: Added a kref to bm_aio_ctx

Signed-off-by: Philipp Reisner <philipp.reisner@linbit.com>
Signed-off-by: Lars Ellenberg <lars.ellenberg@linbit.com>
---
 drivers/block/drbd/drbd_bitmap.c | 84 ++++++++++++++++++++++++++++------------
 1 file changed, 59 insertions(+), 25 deletions(-)

diff --git a/drivers/block/drbd/drbd_bitmap.c b/drivers/block/drbd/drbd_bitmap.c
index 3030201c69d8..a2c337b38d6e 100644
--- a/drivers/block/drbd/drbd_bitmap.c
+++ b/drivers/block/drbd/drbd_bitmap.c
@@ -890,8 +890,16 @@ struct bm_aio_ctx {
 	unsigned flags;
 #define BM_AIO_COPY_PAGES	1
 	int error;
+	struct kref kref;
 };
 
+static void bm_aio_ctx_destroy(struct kref *kref)
+{
+	struct bm_aio_ctx *ctx = container_of(kref, struct bm_aio_ctx, kref);
+
+	kfree(ctx);
+}
+
 /* bv_page may be a copy, or may be the original */
 static void bm_async_io_complete(struct bio *bio, int error)
 {
@@ -936,8 +944,10 @@ static void bm_async_io_complete(struct bio *bio, int error)
 
 	bio_put(bio);
 
-	if (atomic_dec_and_test(&ctx->in_flight))
+	if (atomic_dec_and_test(&ctx->in_flight)) {
 		complete(&ctx->done);
+		kref_put(&ctx->kref, &bm_aio_ctx_destroy);
+	}
 }
 
 static void bm_page_io_async(struct bm_aio_ctx *ctx, int page_nr, int rw) __must_hold(local)
@@ -1001,12 +1011,7 @@ static void bm_page_io_async(struct bm_aio_ctx *ctx, int page_nr, int rw) __must
  */
 static int bm_rw(struct drbd_conf *mdev, int rw, unsigned lazy_writeout_upper_idx) __must_hold(local)
 {
-	struct bm_aio_ctx ctx = {
-		.mdev = mdev,
-		.in_flight = ATOMIC_INIT(1),
-		.done = COMPLETION_INITIALIZER_ONSTACK(ctx.done),
-		.flags = lazy_writeout_upper_idx ? BM_AIO_COPY_PAGES : 0,
-	};
+	struct bm_aio_ctx *ctx;
 	struct drbd_bitmap *b = mdev->bitmap;
 	int num_pages, i, count = 0;
 	unsigned long now;
@@ -1021,7 +1026,21 @@ static int bm_rw(struct drbd_conf *mdev, int rw, unsigned lazy_writeout_upper_id
 	 * For lazy writeout, we don't care for ongoing changes to the bitmap,
 	 * as we submit copies of pages anyways.
 	 */
-	if (!ctx.flags)
+
+	ctx = kmalloc(sizeof(struct bm_aio_ctx), GFP_KERNEL);
+	if (!ctx)
+		return -ENOMEM;
+
+	*ctx = (struct bm_aio_ctx) {
+		.mdev = mdev,
+		.in_flight = ATOMIC_INIT(1),
+		.done = COMPLETION_INITIALIZER(ctx->done),
+		.flags = lazy_writeout_upper_idx ? BM_AIO_COPY_PAGES : 0,
+		.error = 0,
+		.kref = { ATOMIC_INIT(2) },
+	};
+
+	if (!ctx->flags)
 		WARN_ON(!(BM_LOCKED_MASK & b->bm_flags));
 
 	num_pages = b->bm_number_of_pages;
@@ -1046,27 +1065,28 @@ static int bm_rw(struct drbd_conf *mdev, int rw, unsigned lazy_writeout_upper_id
 				continue;
 			}
 		}
-		atomic_inc(&ctx.in_flight);
-		bm_page_io_async(&ctx, i, rw);
+		atomic_inc(&ctx->in_flight);
+		bm_page_io_async(ctx, i, rw);
 		++count;
 		cond_resched();
 	}
 
 	/*
-	 * We initialize ctx.in_flight to one to make sure bm_async_io_complete
+	 * We initialize ctx->in_flight to one to make sure bm_async_io_complete
 	 * will not complete() early, and decrement / test it here.  If there
 	 * are still some bios in flight, we need to wait for them here.
 	 */
-	if (!atomic_dec_and_test(&ctx.in_flight))
-		wait_for_completion(&ctx.done);
+	if (!atomic_dec_and_test(&ctx->in_flight))
+		wait_for_completion(&ctx->done);
+
 	dev_info(DEV, "bitmap %s of %u pages took %lu jiffies\n",
 			rw == WRITE ? "WRITE" : "READ",
 			count, jiffies - now);
 
-	if (ctx.error) {
+	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; /* ctx.error ? */
+		err = -EIO; /* ctx->error ? */
 	}
 
 	now = jiffies;
@@ -1082,6 +1102,8 @@ static int bm_rw(struct drbd_conf *mdev, int rw, unsigned lazy_writeout_upper_id
 	dev_info(DEV, "%s (%lu bits) marked out-of-sync by on disk bit-map.\n",
 	     ppsize(ppb, now << (BM_BLOCK_SHIFT-10)), now);
 
+	kref_put(&ctx->kref, &bm_aio_ctx_destroy);
+
 	return err;
 }
 
@@ -1130,28 +1152,40 @@ int drbd_bm_write_lazy(struct drbd_conf *mdev, unsigned upper_idx) __must_hold(l
  */
 int drbd_bm_write_page(struct drbd_conf *mdev, unsigned int idx) __must_hold(local)
 {
-	struct bm_aio_ctx ctx = {
-		.mdev = mdev,
-		.in_flight = ATOMIC_INIT(1),
-		.done = COMPLETION_INITIALIZER_ONSTACK(ctx.done),
-		.flags = BM_AIO_COPY_PAGES,
-	};
+	struct bm_aio_ctx *ctx;
+	int err;
 
 	if (bm_test_page_unchanged(mdev->bitmap->bm_pages[idx])) {
 		dynamic_dev_dbg(DEV, "skipped bm page write for idx %u\n", idx);
 		return 0;
 	}
 
-	bm_page_io_async(&ctx, idx, WRITE_SYNC);
-	wait_for_completion(&ctx.done);
+	ctx = kmalloc(sizeof(struct bm_aio_ctx), GFP_KERNEL);
+	if (!ctx)
+		return -ENOMEM;
+
+	*ctx = (struct bm_aio_ctx) {
+		.mdev = mdev,
+		.in_flight = ATOMIC_INIT(1),
+		.done = COMPLETION_INITIALIZER(ctx->done),
+		.flags = BM_AIO_COPY_PAGES,
+		.error = 0,
+		.kref = { ATOMIC_INIT(2) },
+	};
+
+	bm_page_io_async(ctx, idx, WRITE_SYNC);
+	wait_for_completion(&ctx->done);
 
-	if (ctx.error)
+	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++;
-	return ctx.error;
+	err = ctx->error;
+	kref_put(&ctx->kref, &bm_aio_ctx_destroy);
+
+	return err;
 }
 
 /* NOTE
-- 
cgit v1.2.3


From 9e58c4dad70bfba86f016bcc98fa19f468d0b777 Mon Sep 17 00:00:00 2001
From: Philipp Reisner <philipp.reisner@linbit.com>
Date: Mon, 27 Jun 2011 15:29:16 +0200
Subject: drbd: Bitmap IO functions can now return prematurely if the disk
 breaks

Signed-off-by: Philipp Reisner <philipp.reisner@linbit.com>
Signed-off-by: Lars Ellenberg <lars.ellenberg@linbit.com>
---
 drivers/block/drbd/drbd_bitmap.c | 40 ++++++++++++++++++++++++++++++----------
 1 file changed, 30 insertions(+), 10 deletions(-)

diff --git a/drivers/block/drbd/drbd_bitmap.c b/drivers/block/drbd/drbd_bitmap.c
index a2c337b38d6e..e5e756dbc434 100644
--- a/drivers/block/drbd/drbd_bitmap.c
+++ b/drivers/block/drbd/drbd_bitmap.c
@@ -886,7 +886,7 @@ void drbd_bm_clear_all(struct drbd_conf *mdev)
 struct bm_aio_ctx {
 	struct drbd_conf *mdev;
 	atomic_t in_flight;
-	struct completion done;
+	unsigned int done;
 	unsigned flags;
 #define BM_AIO_COPY_PAGES	1
 	int error;
@@ -897,6 +897,7 @@ static void bm_aio_ctx_destroy(struct kref *kref)
 {
 	struct bm_aio_ctx *ctx = container_of(kref, struct bm_aio_ctx, kref);
 
+	put_ldev(ctx->mdev);
 	kfree(ctx);
 }
 
@@ -945,7 +946,8 @@ static void bm_async_io_complete(struct bio *bio, int error)
 	bio_put(bio);
 
 	if (atomic_dec_and_test(&ctx->in_flight)) {
-		complete(&ctx->done);
+		ctx->done = 1;
+		wake_up(&mdev->misc_wait);
 		kref_put(&ctx->kref, &bm_aio_ctx_destroy);
 	}
 }
@@ -1034,12 +1036,18 @@ static int bm_rw(struct drbd_conf *mdev, int rw, unsigned lazy_writeout_upper_id
 	*ctx = (struct bm_aio_ctx) {
 		.mdev = mdev,
 		.in_flight = ATOMIC_INIT(1),
-		.done = COMPLETION_INITIALIZER(ctx->done),
+		.done = 0,
 		.flags = lazy_writeout_upper_idx ? BM_AIO_COPY_PAGES : 0,
 		.error = 0,
 		.kref = { ATOMIC_INIT(2) },
 	};
 
+	if (!get_ldev_if_state(mdev, D_ATTACHING)) {  /* put is in bm_aio_ctx_destroy() */
+		dev_err(DEV, "ASSERT FAILED: get_ldev_if_state() == 1 in bm_rw()\n");
+		kfree(ctx);
+		return -ENODEV;
+	}
+
 	if (!ctx->flags)
 		WARN_ON(!(BM_LOCKED_MASK & b->bm_flags));
 
@@ -1073,11 +1081,16 @@ static int bm_rw(struct drbd_conf *mdev, int rw, unsigned lazy_writeout_upper_id
 
 	/*
 	 * We initialize ctx->in_flight to one to make sure bm_async_io_complete
-	 * will not complete() early, and decrement / test it here.  If there
+	 * will not set ctx->done early, and decrement / test it here.  If there
 	 * are still some bios in flight, we need to wait for them here.
+	 * If all IO is done already (or nothing had been submitted), there is
+	 * no need to wait.  Still, we need to put the kref associated with the
+	 * "in_flight reached zero, all done" event.
 	 */
 	if (!atomic_dec_and_test(&ctx->in_flight))
-		wait_for_completion(&ctx->done);
+		wait_until_done_or_disk_failure(mdev, &ctx->done);
+	else
+		kref_put(&ctx->kref, &bm_aio_ctx_destroy);
 
 	dev_info(DEV, "bitmap %s of %u pages took %lu jiffies\n",
 			rw == WRITE ? "WRITE" : "READ",
@@ -1089,6 +1102,9 @@ static int bm_rw(struct drbd_conf *mdev, int rw, unsigned lazy_writeout_upper_id
 		err = -EIO; /* ctx->error ? */
 	}
 
+	if (atomic_read(&ctx->in_flight))
+		err = -EIO; /* Disk failed during IO... */
+
 	now = jiffies;
 	if (rw == WRITE) {
 		drbd_md_flush(mdev);
@@ -1103,7 +1119,6 @@ static int bm_rw(struct drbd_conf *mdev, int rw, unsigned lazy_writeout_upper_id
 	     ppsize(ppb, now << (BM_BLOCK_SHIFT-10)), now);
 
 	kref_put(&ctx->kref, &bm_aio_ctx_destroy);
-
 	return err;
 }
 
@@ -1167,14 +1182,20 @@ int drbd_bm_write_page(struct drbd_conf *mdev, unsigned int idx) __must_hold(loc
 	*ctx = (struct bm_aio_ctx) {
 		.mdev = mdev,
 		.in_flight = ATOMIC_INIT(1),
-		.done = COMPLETION_INITIALIZER(ctx->done),
+		.done = 0,
 		.flags = BM_AIO_COPY_PAGES,
 		.error = 0,
 		.kref = { ATOMIC_INIT(2) },
 	};
 
+	if (!get_ldev_if_state(mdev, D_ATTACHING)) {  /* put is in bm_aio_ctx_destroy() */
+		dev_err(DEV, "ASSERT FAILED: get_ldev_if_state() == 1 in drbd_bm_write_page()\n");
+		kfree(ctx);
+		return -ENODEV;
+	}
+
 	bm_page_io_async(ctx, idx, WRITE_SYNC);
-	wait_for_completion(&ctx->done);
+	wait_until_done_or_disk_failure(mdev, &ctx->done);
 
 	if (ctx->error)
 		drbd_chk_io_error(mdev, 1, true);
@@ -1182,9 +1203,8 @@ int drbd_bm_write_page(struct drbd_conf *mdev, unsigned int idx) __must_hold(loc
 		 * gone in a moment as well. */
 
 	mdev->bm_writ_cnt++;
-	err = ctx->error;
+	err = atomic_read(&ctx->in_flight) ? -EIO : ctx->error;
 	kref_put(&ctx->kref, &bm_aio_ctx_destroy);
-
 	return err;
 }
 
-- 
cgit v1.2.3


From 5ca1de0384dafe843de10fed843de26de740bca1 Mon Sep 17 00:00:00 2001
From: Philipp Reisner <philipp.reisner@linbit.com>
Date: Tue, 28 Jun 2011 17:01:19 +0200
Subject: drbd: Allow new IOs while the local disk in in FAILED state

The last bunch of commits prepared the 'detach from tar pit' feature.
With that we can be for long time in disk state FAILED. We need
to accept new IO requests during that time.

Signed-off-by: Philipp Reisner <philipp.reisner@linbit.com>
Signed-off-by: Lars Ellenberg <lars.ellenberg@linbit.com>
---
 drivers/block/drbd/drbd_int.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/block/drbd/drbd_int.h b/drivers/block/drbd/drbd_int.h
index 10ea9e388246..c7976a77dfba 100644
--- a/drivers/block/drbd/drbd_int.h
+++ b/drivers/block/drbd/drbd_int.h
@@ -2310,12 +2310,12 @@ static inline int drbd_state_is_stable(struct drbd_conf *mdev)
 	case D_OUTDATED:
 	case D_CONSISTENT:
 	case D_UP_TO_DATE:
+	case D_FAILED:
 		/* disk state is stable as well. */
 		break;
 
 	/* no new io accepted during tansitional states */
 	case D_ATTACHING:
-	case D_FAILED:
 	case D_NEGOTIATING:
 	case D_UNKNOWN:
 	case D_MASK:
-- 
cgit v1.2.3


From 02ee8f95fadf7c94b3d28df436a095152f6392b2 Mon Sep 17 00:00:00 2001
From: Philipp Reisner <philipp.reisner@linbit.com>
Date: Mon, 14 Mar 2011 11:54:47 +0100
Subject: drbd: Force flag for the detach operation

Signed-off-by: Philipp Reisner <philipp.reisner@linbit.com>
Signed-off-by: Lars Ellenberg <lars.ellenberg@linbit.com>
---
 drivers/block/drbd/drbd_main.c |  2 +-
 drivers/block/drbd/drbd_nl.c   | 15 +++++++++++++++
 include/linux/drbd_nl.h        |  4 +++-
 3 files changed, 19 insertions(+), 2 deletions(-)

diff --git a/drivers/block/drbd/drbd_main.c b/drivers/block/drbd/drbd_main.c
index 1b59ab3ab9c7..bc8a8a7556da 100644
--- a/drivers/block/drbd/drbd_main.c
+++ b/drivers/block/drbd/drbd_main.c
@@ -495,7 +495,7 @@ static int cl_wide_st_chg(struct drbd_conf *mdev,
 		 ((os.role != R_PRIMARY && ns.role == R_PRIMARY) ||
 		  (os.conn != C_STARTING_SYNC_T && ns.conn == C_STARTING_SYNC_T) ||
 		  (os.conn != C_STARTING_SYNC_S && ns.conn == C_STARTING_SYNC_S) ||
-		  (os.disk != D_DISKLESS && ns.disk == D_DISKLESS))) ||
+		  (os.disk != D_FAILED && ns.disk == D_FAILED))) ||
 		(os.conn >= C_CONNECTED && ns.conn == C_DISCONNECTING) ||
 		(os.conn == C_CONNECTED && ns.conn == C_VERIFY_S);
 }
diff --git a/drivers/block/drbd/drbd_nl.c b/drivers/block/drbd/drbd_nl.c
index cabad39f908c..6d116a2b2321 100644
--- a/drivers/block/drbd/drbd_nl.c
+++ b/drivers/block/drbd/drbd_nl.c
@@ -1337,17 +1337,32 @@ static int drbd_nl_detach(struct drbd_conf *mdev, struct drbd_nl_cfg_req *nlp,
 {
 	enum drbd_ret_code retcode;
 	int ret;
+	struct detach dt = {};
+
+	if (!detach_from_tags(mdev, nlp->tag_list, &dt)) {
+		reply->ret_code = ERR_MANDATORY_TAG;
+		goto out;
+	}
+
+	if (dt.detach_force) {
+		drbd_force_state(mdev, NS(disk, D_FAILED));
+		reply->ret_code = SS_SUCCESS;
+		goto out;
+	}
+
 	drbd_suspend_io(mdev); /* so no-one is stuck in drbd_al_begin_io */
 	retcode = drbd_request_state(mdev, NS(disk, D_FAILED));
 	/* D_FAILED will transition to DISKLESS. */
 	ret = wait_event_interruptible(mdev->misc_wait,
 			mdev->state.disk != D_FAILED);
 	drbd_resume_io(mdev);
+
 	if ((int)retcode == (int)SS_IS_DISKLESS)
 		retcode = SS_NOTHING_TO_DO;
 	if (ret)
 		retcode = ERR_INTR;
 	reply->ret_code = retcode;
+out:
 	return 0;
 }
 
diff --git a/include/linux/drbd_nl.h b/include/linux/drbd_nl.h
index ab6159e4fcf0..7203c9ead233 100644
--- a/include/linux/drbd_nl.h
+++ b/include/linux/drbd_nl.h
@@ -33,7 +33,9 @@ NL_PACKET(disk_conf, 3,
 	NL_BIT(		58,	T_MAY_IGNORE,	no_disk_drain)
 )
 
-NL_PACKET(detach, 4, )
+NL_PACKET(detach, 4,
+	NL_BIT(		88,	T_MANDATORY,	detach_force)
+)
 
 NL_PACKET(net_conf, 5,
 	NL_STRING(	8,	T_MANDATORY,	my_addr,	128)
-- 
cgit v1.2.3


From dfa8bedbfe881caf6676704ab1aae18dfe8e430a Mon Sep 17 00:00:00 2001
From: Philipp Reisner <philipp.reisner@linbit.com>
Date: Wed, 29 Jun 2011 14:06:08 +0200
Subject: drbd: Implemented the disk-timeout option

When the disk-timeout is active, and it expires for a single request,
we consider the local disk as D_FAILED. Note: With this change,
I made both timeout based state transitions HARD state transitions.

Signed-off-by: Philipp Reisner <philipp.reisner@linbit.com>
Signed-off-by: Lars Ellenberg <lars.ellenberg@linbit.com>
---
 drivers/block/drbd/drbd_main.c     |  5 +++++
 drivers/block/drbd/drbd_receiver.c |  2 --
 drivers/block/drbd/drbd_req.c      | 32 ++++++++++++++++++++------------
 include/linux/drbd_limits.h        |  5 +++++
 include/linux/drbd_nl.h            |  1 +
 5 files changed, 31 insertions(+), 14 deletions(-)

diff --git a/drivers/block/drbd/drbd_main.c b/drivers/block/drbd/drbd_main.c
index bc8a8a7556da..4bd636524dd1 100644
--- a/drivers/block/drbd/drbd_main.c
+++ b/drivers/block/drbd/drbd_main.c
@@ -1404,6 +1404,9 @@ static void after_state_ch(struct drbd_conf *mdev, union drbd_state os,
 	/* Here we have the actions that are performed after a
 	   state change. This function might sleep */
 
+	if (os.disk <= D_NEGOTIATING && ns.disk > D_NEGOTIATING)
+		mod_timer(&mdev->request_timer, jiffies + HZ);
+
 	nsm.i = -1;
 	if (ns.susp_nod) {
 		if (os.conn < C_CONNECTED && ns.conn >= C_CONNECTED)
@@ -3318,6 +3321,8 @@ static void drbd_delete_device(unsigned int minor)
 	if (!mdev)
 		return;
 
+	del_timer_sync(&mdev->request_timer);
+
 	/* paranoia asserts */
 	if (mdev->open_cnt != 0)
 		dev_err(DEV, "open_cnt = %d in %s:%u", mdev->open_cnt,
diff --git a/drivers/block/drbd/drbd_receiver.c b/drivers/block/drbd/drbd_receiver.c
index e7ed0aa93a16..a85bbe1bbc2b 100644
--- a/drivers/block/drbd/drbd_receiver.c
+++ b/drivers/block/drbd/drbd_receiver.c
@@ -3803,8 +3803,6 @@ static void drbd_disconnect(struct drbd_conf *mdev)
 	atomic_set(&mdev->rs_pending_cnt, 0);
 	wake_up(&mdev->misc_wait);
 
-	del_timer(&mdev->request_timer);
-
 	/* make sure syncer is stopped and w_resume_next_sg queued */
 	del_timer_sync(&mdev->resync_timer);
 	resync_timer_fn((unsigned long)mdev);
diff --git a/drivers/block/drbd/drbd_req.c b/drivers/block/drbd/drbd_req.c
index 1a8aac4b0c2f..ef145e33a647 100644
--- a/drivers/block/drbd/drbd_req.c
+++ b/drivers/block/drbd/drbd_req.c
@@ -1208,13 +1208,19 @@ void request_timer_fn(unsigned long data)
 	struct drbd_conf *mdev = (struct drbd_conf *) data;
 	struct drbd_request *req; /* oldest request */
 	struct list_head *le;
-	unsigned long et = 0; /* effective timeout = ko_count * timeout */
+	unsigned long ent = 0, dt = 0, et, nt; /* effective timeout = ko_count * timeout */
 
 	if (get_net_conf(mdev)) {
-		et = mdev->net_conf->timeout*HZ/10 * mdev->net_conf->ko_count;
+		ent = mdev->net_conf->timeout*HZ/10 * mdev->net_conf->ko_count;
 		put_net_conf(mdev);
 	}
-	if (!et || mdev->state.conn < C_WF_REPORT_PARAMS)
+	if (get_ldev(mdev)) {
+		dt = mdev->ldev->dc.disk_timeout * HZ / 10;
+		put_ldev(mdev);
+	}
+	et = min_not_zero(dt, ent);
+
+	if (!et || (mdev->state.conn < C_WF_REPORT_PARAMS && mdev->state.disk <= D_FAILED))
 		return; /* Recurring timer stopped */
 
 	spin_lock_irq(&mdev->req_lock);
@@ -1227,17 +1233,19 @@ void request_timer_fn(unsigned long data)
 
 	le = le->prev;
 	req = list_entry(le, struct drbd_request, tl_requests);
-	if (time_is_before_eq_jiffies(req->start_time + et)) {
-		if (req->rq_state & RQ_NET_PENDING) {
+	if (ent && req->rq_state & RQ_NET_PENDING) {
+		if (time_is_before_eq_jiffies(req->start_time + ent)) {
 			dev_warn(DEV, "Remote failed to finish a request within ko-count * timeout\n");
-			_drbd_set_state(_NS(mdev, conn, C_TIMEOUT), CS_VERBOSE, NULL);
-		} else {
-			dev_warn(DEV, "Local backing block device frozen?\n");
-			mod_timer(&mdev->request_timer, jiffies + et);
+			_drbd_set_state(_NS(mdev, conn, C_TIMEOUT), CS_VERBOSE | CS_HARD, NULL);
 		}
-	} else {
-		mod_timer(&mdev->request_timer, req->start_time + et);
 	}
-
+	if (dt && req->rq_state & RQ_LOCAL_PENDING) {
+		if (time_is_before_eq_jiffies(req->start_time + dt)) {
+			dev_warn(DEV, "Local backing device failed to meet the disk-timeout\n");
+			__drbd_chk_io_error(mdev, 1);
+		}
+	}
+	nt = (time_is_before_eq_jiffies(req->start_time + et) ? jiffies : req->start_time) + et;
 	spin_unlock_irq(&mdev->req_lock);
+	mod_timer(&mdev->request_timer, nt);
 }
diff --git a/include/linux/drbd_limits.h b/include/linux/drbd_limits.h
index 928c84dfaf42..fb670bf603f7 100644
--- a/include/linux/drbd_limits.h
+++ b/include/linux/drbd_limits.h
@@ -48,6 +48,11 @@
 #define DRBD_TIMEOUT_MAX 600
 #define DRBD_TIMEOUT_DEF 60       /* 6 seconds */
 
+ /* If backing disk takes longer than disk_timeout, mark the disk as failed */
+#define DRBD_DISK_TIMEOUT_MIN 0    /* 0 = disabled */
+#define DRBD_DISK_TIMEOUT_MAX 6000 /* 10 Minutes */
+#define DRBD_DISK_TIMEOUT_DEF 0    /* disabled */
+
   /* active connection retries when C_WF_CONNECTION */
 #define DRBD_CONNECT_INT_MIN 1
 #define DRBD_CONNECT_INT_MAX 120
diff --git a/include/linux/drbd_nl.h b/include/linux/drbd_nl.h
index 7203c9ead233..a8706f08ab36 100644
--- a/include/linux/drbd_nl.h
+++ b/include/linux/drbd_nl.h
@@ -31,6 +31,7 @@ NL_PACKET(disk_conf, 3,
 	NL_INTEGER(	56,	T_MAY_IGNORE,	max_bio_bvecs)
 	NL_BIT(		57,	T_MAY_IGNORE,	no_disk_barrier)
 	NL_BIT(		58,	T_MAY_IGNORE,	no_disk_drain)
+	NL_INTEGER(	89,	T_MAY_IGNORE,	disk_timeout)
 )
 
 NL_PACKET(detach, 4,
-- 
cgit v1.2.3


From 22f46ce2ef94151f806f84ab0f9ee43a72dfb1f1 Mon Sep 17 00:00:00 2001
From: Lars Ellenberg <lars.ellenberg@linbit.com>
Date: Mon, 11 Jul 2011 17:32:26 +0200
Subject: drbd: change some GFP_KERNEL to GFP_NOIO

Bitmap IO may happend in the context of an application write,
in the generic block IO path.  We need to use GFP_NOIO.

Signed-off-by: Philipp Reisner <philipp.reisner@linbit.com>
Signed-off-by: Lars Ellenberg <lars.ellenberg@linbit.com>
---
 drivers/block/drbd/drbd_bitmap.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/drivers/block/drbd/drbd_bitmap.c b/drivers/block/drbd/drbd_bitmap.c
index e5e756dbc434..9611db43cc7a 100644
--- a/drivers/block/drbd/drbd_bitmap.c
+++ b/drivers/block/drbd/drbd_bitmap.c
@@ -955,7 +955,7 @@ static void bm_async_io_complete(struct bio *bio, int error)
 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 bio *bio = bio_alloc(GFP_NOIO, 1);
 	struct drbd_conf *mdev = ctx->mdev;
 	struct drbd_bitmap *b = mdev->bitmap;
 	struct page *page;
@@ -1029,7 +1029,7 @@ static int bm_rw(struct drbd_conf *mdev, int rw, unsigned lazy_writeout_upper_id
 	 * as we submit copies of pages anyways.
 	 */
 
-	ctx = kmalloc(sizeof(struct bm_aio_ctx), GFP_KERNEL);
+	ctx = kmalloc(sizeof(struct bm_aio_ctx), GFP_NOIO);
 	if (!ctx)
 		return -ENOMEM;
 
@@ -1175,7 +1175,7 @@ int drbd_bm_write_page(struct drbd_conf *mdev, unsigned int idx) __must_hold(loc
 		return 0;
 	}
 
-	ctx = kmalloc(sizeof(struct bm_aio_ctx), GFP_KERNEL);
+	ctx = kmalloc(sizeof(struct bm_aio_ctx), GFP_NOIO);
 	if (!ctx)
 		return -ENOMEM;
 
-- 
cgit v1.2.3


From bca482e90bf87569ffb87cba2c3a777ac23e5c86 Mon Sep 17 00:00:00 2001
From: Philipp Reisner <philipp.reisner@linbit.com>
Date: Fri, 15 Jul 2011 12:14:27 +0200
Subject: drbd: Fixed current UUID generation

Now, the new edition of the clause only fires if a diskless
peer gets promoted.

This is a fixup for "drbd: Delayed creation of current-UUID".

Signed-off-by: Philipp Reisner <philipp.reisner@linbit.com>
Signed-off-by: Lars Ellenberg <lars.ellenberg@linbit.com>
---
 drivers/block/drbd/drbd_main.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/drivers/block/drbd/drbd_main.c b/drivers/block/drbd/drbd_main.c
index 4bd636524dd1..b62e3559b50c 100644
--- a/drivers/block/drbd/drbd_main.c
+++ b/drivers/block/drbd/drbd_main.c
@@ -1494,11 +1494,11 @@ static void after_state_ch(struct drbd_conf *mdev, union drbd_state os,
 	}
 
 	if (ns.pdsk < D_INCONSISTENT && get_ldev(mdev)) {
-		if (ns.peer == R_PRIMARY && mdev->ldev->md.uuid[UI_BITMAP] == 0) {
+		if (os.peer == R_SECONDARY && ns.peer == R_PRIMARY &&
+		    mdev->ldev->md.uuid[UI_BITMAP] == 0 && ns.disk >= D_UP_TO_DATE) {
 			drbd_uuid_new_current(mdev);
 			drbd_send_uuids(mdev);
 		}
-
 		/* D_DISKLESS Peer becomes secondary */
 		if (os.peer == R_PRIMARY && ns.peer == R_SECONDARY)
 			/* We may still be Primary ourselves.
-- 
cgit v1.2.3


From 79f16f5dbc95da372c25afddac80f4adef3cfce1 Mon Sep 17 00:00:00 2001
From: Philipp Reisner <phil