Import wiredtiger: ca6eee06ffdacc8e191987e64b3791740dad21e1 from branch mongodb-3.4

ref: 74430da40c..ca6eee06ff
for: 3.4.0

WT-2962       Provide a way to configure builtin extensions
WT-2984       Search of metadata for recently created collection gets WT_NOTFOUND
WT-3000       Missing log records in recovery when crashing after a log file switch
WT-3002       Allow applications to exempt threads from eviction.
WT-3004       lint: declare functions that don't return a value as void
WT-3011       __wt_curjoin_open() saves the wrong URI in the cursor.
WT-3012       Test format hanging on LSM configurations
WT-3015       Test format stuck with 2mb cache
WT-3016       Tests needed for systems without ftruncate
WT-3017       Hazard pointer race with page replace causes error
WT-3018       lint
WT-3020       LSM primary changes impact parallel-pop-lsm load time
WT-3022       LSM operations get stuck in __wt_clsm_await_switch waiting for switch on tree to complete
WT-3023       Test format hang on zSeries
WT-3024       wtperf medium-lsm-compact test can hang

1 files changed, 57 insertions, 23 deletions

diff --git a/src/third_party/wiredtiger/src/log/log.c b/src/third_party/wiredtiger/src/log/log.c
index 00e4ea5f441..8b6c8b2c490 100644
--- a/src/third_party/wiredtiger/src/log/log.c
+++ b/src/third_party/wiredtiger/src/log/log.c
@@ -21,6 +21,60 @@ static int __log_write_internal(
 #define	WT_LOG_OPEN_VERIFY	0x02
 
 /*
+ * __log_wait_for_earlier_slot --
+ *	Wait for write_lsn to catch up to this slot.
+ */
+static void
+__log_wait_for_earlier_slot(WT_SESSION_IMPL *session, WT_LOGSLOT *slot)
+{
+	WT_CONNECTION_IMPL *conn;
+	WT_LOG *log;
+	int yield_count;
+
+	conn = S2C(session);
+	log = conn->log;
+	yield_count = 0;
+
+	while (__wt_log_cmp(&log->write_lsn, &slot->slot_release_lsn) != 0) {
+		/*
+		 * If we're on a locked path and the write LSN is not advancing,
+		 * unlock in case an earlier thread is trying to switch its
+		 * slot and complete its operation.
+		 */
+		if (F_ISSET(session, WT_SESSION_LOCKED_SLOT))
+			__wt_spin_unlock(session, &log->log_slot_lock);
+		__wt_cond_auto_signal(session, conn->log_wrlsn_cond);
+		if (++yield_count < WT_THOUSAND)
+			__wt_yield();
+		else
+			__wt_cond_wait(session, log->log_write_cond, 200);
+		if (F_ISSET(session, WT_SESSION_LOCKED_SLOT))
+			__wt_spin_lock(session, &log->log_slot_lock);
+	}
+}
+
+/*
+ * __log_fs_write --
+ *	Wrapper when writing to a log file.  If we're writing to a new log
+ *	file for the first time wait for writes to the previous log file.
+ */
+static int
+__log_fs_write(WT_SESSION_IMPL *session,
+    WT_LOGSLOT *slot, wt_off_t offset, size_t len, const void *buf)
+{
+	/*
+	 * If we're writing into a new log file, we have to wait for all
+	 * writes to the previous log file to complete otherwise there could
+	 * be a hole at the end of the previous log file that we cannot detect.
+	 */
+	if (slot->slot_release_lsn.l.file < slot->slot_start_lsn.l.file) {
+		__log_wait_for_earlier_slot(session, slot);
+		WT_RET(__wt_log_force_sync(session, &slot->slot_release_lsn));
+	}
+	return (__wt_write(session, slot->slot_fh, offset, len, buf));
+}
+
+/*
  * __wt_log_ckpt --
  *	Record the given LSN as the checkpoint LSN and signal the archive
  *	thread as needed.
@@ -371,8 +425,6 @@ __wt_log_extract_lognum(
 {
 	const char *p;
 
-	WT_UNUSED(session);
-
 	if (id == NULL || name == NULL)
 		return (WT_ERROR);
 	if ((p = strrchr(name, '.')) == NULL ||
@@ -576,7 +628,7 @@ __log_fill(WT_SESSION_IMPL *session,
 		/*
 		 * If this is a force or unbuffered write, write it now.
 		 */
-		WT_ERR(__wt_write(session, myslot->slot->slot_fh,
+		WT_ERR(__log_fs_write(session, myslot->slot,
 		    myslot->offset + myslot->slot->slot_start_offset,
 		    record->size, record->mem));
 
@@ -1352,13 +1404,11 @@ __wt_log_release(WT_SESSION_IMPL *session, WT_LOGSLOT *slot, bool *freep)
 	WT_LSN sync_lsn;
 	int64_t release_buffered, release_bytes;
 	uint64_t fsync_duration_usecs;
-	int yield_count;
 	bool locked;
 
 	conn = S2C(session);
 	log = conn->log;
 	locked = false;
-	yield_count = 0;
 	if (freep != NULL)
 		*freep = 1;
 	release_buffered = WT_LOG_SLOT_RELEASED_BUFFERED(slot->slot_state);
@@ -1379,8 +1429,7 @@ __wt_log_release(WT_SESSION_IMPL *session, WT_LOGSLOT *slot, bool *freep)
 
 	/* Write the buffered records */
 	if (release_buffered != 0)
-		WT_ERR(__wt_write(session,
-		    slot->slot_fh, slot->slot_start_offset,
+		WT_ERR(__log_fs_write(session, slot, slot->slot_start_offset,
 		    (size_t)release_buffered, slot->slot_buf.mem));
 
 	/*
@@ -1411,22 +1460,7 @@ __wt_log_release(WT_SESSION_IMPL *session, WT_LOGSLOT *slot, bool *freep)
 	 * be holes in the log file.
 	 */
 	WT_STAT_CONN_INCR(session, log_release_write_lsn);
-	while (__wt_log_cmp(&log->write_lsn, &slot->slot_release_lsn) != 0) {
-		/*
-		 * If we're on a locked path and the write LSN is not advancing,
-		 * unlock in case an earlier thread is trying to switch its
-		 * slot and complete its operation.
-		 */
-		if (F_ISSET(session, WT_SESSION_LOCKED_SLOT))
-			__wt_spin_unlock(session, &log->log_slot_lock);
-		__wt_cond_auto_signal(session, conn->log_wrlsn_cond);
-		if (++yield_count < WT_THOUSAND)
-			__wt_yield();
-		else
-			__wt_cond_wait(session, log->log_write_cond, 200);
-		if (F_ISSET(session, WT_SESSION_LOCKED_SLOT))
-			__wt_spin_lock(session, &log->log_slot_lock);
-	}
+	__log_wait_for_earlier_slot(session, slot);
 
 	log->write_start_lsn = slot->slot_start_lsn;
 	log->write_lsn = slot->slot_end_lsn;