Tarantool development patches archive
 help / color / mirror / Atom feed
* [Tarantool-patches] [PATCH] txn: convert flags to explicit bitfield
@ 2021-02-15 14:01 Cyrill Gorcunov via Tarantool-patches
  2021-02-17 21:15 ` Vladislav Shpilevoy via Tarantool-patches
  0 siblings, 1 reply; 5+ messages in thread
From: Cyrill Gorcunov via Tarantool-patches @ 2021-02-15 14:01 UTC (permalink / raw)
  To: tml; +Cc: Vladislav Shpilevoy

Instead of shifting flags inside txn_x_flag() helpers
lets define precompiled value. Moreover lets drop the
helpers completely: better operate with bitfield directly
which allows to test/set/clear a set of bits at once
instead of one by one.

Closes #5128

Suggested-by: Vladislav Shpilevoy <v.shpilevoy@tarantool.org>
Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>
---
issue https://github.com/tarantool/tarantool/issues/5128
branch gorcunov/gh-5128-txn-flags-4

Guys, take a loook please, once time permit. Helpers are
vanished, code style should be fit I hope.

 src/box/applier.cc             |  4 +--
 src/box/memtx_engine.c         |  2 +-
 src/box/txn.c                  | 54 ++++++++++++++++------------------
 src/box/txn.h                  | 36 ++++++-----------------
 src/box/txn_limbo.c            | 35 ++++++++++------------
 test/unit/snap_quorum_delay.cc |  3 +-
 6 files changed, 55 insertions(+), 79 deletions(-)

diff --git a/src/box/applier.cc b/src/box/applier.cc
index 553db76fc..c7c85d329 100644
--- a/src/box/applier.cc
+++ b/src/box/applier.cc
@@ -237,7 +237,7 @@ apply_snapshot_row(struct xrow_header *row)
 	 * Do not wait for confirmation when fetching a snapshot.
 	 * Master only sends confirmed rows during join.
 	 */
-	txn_set_flag(txn, TXN_FORCE_ASYNC);
+	txn->flags |= TXN_FORCE_ASYNC;
 	if (txn_begin_stmt(txn, space) != 0)
 		goto rollback;
 	/* no access checks here - applier always works with admin privs */
@@ -308,7 +308,7 @@ apply_final_join_row(struct xrow_header *row)
 	 * Do not wait for confirmation while processing final
 	 * join rows. See apply_snapshot_row().
 	 */
-	txn_set_flag(txn, TXN_FORCE_ASYNC);
+	txn->flags |= TXN_FORCE_ASYNC;
 	if (apply_row(row) != 0) {
 		txn_rollback(txn);
 		fiber_gc();
diff --git a/src/box/memtx_engine.c b/src/box/memtx_engine.c
index 1174234d2..8d3cf5bbd 100644
--- a/src/box/memtx_engine.c
+++ b/src/box/memtx_engine.c
@@ -265,7 +265,7 @@ memtx_engine_recover_snapshot_row(struct memtx_engine *memtx,
 	 * Snapshot rows are confirmed by definition. They don't need to go to
 	 * the synchronous transactions limbo.
 	 */
-	txn_set_flag(txn, TXN_FORCE_ASYNC);
+	txn->flags |= TXN_FORCE_ASYNC;
 	rc = txn_commit(txn);
 	/*
 	 * Don't let gc pool grow too much. Yet to
diff --git a/src/box/txn.c b/src/box/txn.c
index a5edbfc60..1958b079c 100644
--- a/src/box/txn.c
+++ b/src/box/txn.c
@@ -282,7 +282,7 @@ txn_begin(void)
 	 * It's a responsibility of an engine to disable yields
 	 * if they are not supported.
 	 */
-	txn_set_flag(txn, TXN_CAN_YIELD);
+	txn->flags |= TXN_CAN_YIELD;
 	return txn;
 }
 
@@ -516,7 +516,7 @@ txn_free_or_wakeup(struct txn *txn)
 	if (txn->fiber == NULL)
 		txn_free(txn);
 	else {
-		txn_set_flag(txn, TXN_IS_DONE);
+		txn->flags |= TXN_IS_DONE;
 		if (txn->fiber != fiber())
 			/* Wake a waiting fiber up. */
 			fiber_wakeup(txn->fiber);
@@ -526,7 +526,7 @@ txn_free_or_wakeup(struct txn *txn)
 void
 txn_complete_fail(struct txn *txn)
 {
-	assert(!txn_has_flag(txn, TXN_IS_DONE));
+	assert((txn->flags & TXN_IS_DONE) == 0);
 	assert(txn->signature < 0);
 	txn->status = TXN_ABORTED;
 	struct txn_stmt *stmt;
@@ -535,7 +535,7 @@ txn_complete_fail(struct txn *txn)
 		txn_rollback_one_stmt(txn, stmt);
 	if (txn->engine != NULL)
 		engine_rollback(txn->engine, txn);
-	if (txn_has_flag(txn, TXN_HAS_TRIGGERS))
+	if (txn->flags & TXN_HAS_TRIGGERS)
 		txn_run_rollback_triggers(txn, &txn->on_rollback);
 	txn_free_or_wakeup(txn);
 }
@@ -543,13 +543,12 @@ txn_complete_fail(struct txn *txn)
 void
 txn_complete_success(struct txn *txn)
 {
-	assert(!txn_has_flag(txn, TXN_IS_DONE));
-	assert(!txn_has_flag(txn, TXN_WAIT_SYNC));
+	assert((txn->flags & (TXN_IS_DONE | TXN_WAIT_SYNC)) == 0);
 	assert(txn->signature >= 0);
 	txn->status = TXN_COMMITTED;
 	if (txn->engine != NULL)
 		engine_commit(txn->engine, txn);
-	if (txn_has_flag(txn, TXN_HAS_TRIGGERS))
+	if (txn->flags & TXN_HAS_TRIGGERS)
 		txn_run_commit_triggers(txn, &txn->on_commit);
 	txn_free_or_wakeup(txn);
 }
@@ -587,9 +586,9 @@ txn_on_journal_write(struct journal_entry *entry)
 				     "%.3f sec", n_rows,
 				     txn->signature - n_rows + 1, delta);
 	}
-	if (txn_has_flag(txn, TXN_HAS_TRIGGERS))
+	if (txn->flags & TXN_HAS_TRIGGERS)
 		txn_run_wal_write_triggers(txn);
-	if (!txn_has_flag(txn, TXN_WAIT_SYNC))
+	if ((txn->flags & TXN_WAIT_SYNC) == 0)
 		txn_complete_success(txn);
 	else if (txn->fiber != NULL && txn->fiber != fiber())
 		fiber_wakeup(txn->fiber);
@@ -640,10 +639,9 @@ txn_journal_entry_new(struct txn *txn)
 	 * space can't be synchronous. So if there is at least one
 	 * synchronous space, the transaction is not local.
 	 */
-	if (!txn_has_flag(txn, TXN_FORCE_ASYNC)) {
+	if ((txn->flags & TXN_FORCE_ASYNC) == 0) {
 		if (is_sync) {
-			txn_set_flag(txn, TXN_WAIT_SYNC);
-			txn_set_flag(txn, TXN_WAIT_ACK);
+			txn->flags |= TXN_WAIT_SYNC | TXN_WAIT_ACK;
 		} else if (!txn_limbo_is_empty(&txn_limbo)) {
 			/*
 			 * There some sync entries on the
@@ -652,7 +650,7 @@ txn_journal_entry_new(struct txn *txn)
 			 * doesn't touch sync space (each sync txn
 			 * should be considered as a barrier).
 			 */
-			txn_set_flag(txn, TXN_WAIT_SYNC);
+			txn->flags |= TXN_WAIT_SYNC;
 		}
 	}
 
@@ -693,8 +691,8 @@ txn_prepare(struct txn *txn)
 {
 	txn->psn = ++txn_last_psn;
 
-	if (txn_has_flag(txn, TXN_IS_ABORTED_BY_YIELD)) {
-		assert(!txn_has_flag(txn, TXN_CAN_YIELD));
+	if (txn->flags & TXN_IS_ABORTED_BY_YIELD) {
+		assert((txn->flags & TXN_CAN_YIELD) == 0);
 		diag_set(ClientError, ER_TRANSACTION_YIELD);
 		diag_log();
 		return -1;
@@ -748,7 +746,7 @@ txn_prepare(struct txn *txn)
 	assert(rlist_empty(&txn->conflicted_by_list));
 
 	trigger_clear(&txn->fiber_on_stop);
-	if (!txn_has_flag(txn, TXN_CAN_YIELD))
+	if ((txn->flags & TXN_CAN_YIELD) == 0)
 		trigger_clear(&txn->fiber_on_yield);
 
 	txn->start_tm = ev_monotonic_now(loop());
@@ -815,7 +813,7 @@ txn_commit_async(struct txn *txn)
 	if (req == NULL)
 		goto rollback;
 
-	bool is_sync = txn_has_flag(txn, TXN_WAIT_SYNC);
+	bool is_sync = txn->flags & TXN_WAIT_SYNC;
 	struct txn_limbo_entry *limbo_entry;
 	if (is_sync) {
 		/*
@@ -838,7 +836,7 @@ txn_commit_async(struct txn *txn)
 		if (limbo_entry == NULL)
 			goto rollback;
 
-		if (txn_has_flag(txn, TXN_WAIT_ACK)) {
+		if (txn->flags & TXN_WAIT_ACK) {
 			int64_t lsn = req->rows[txn->n_applier_rows - 1]->lsn;
 			/*
 			 * Can't tell whether it is local or not -
@@ -895,7 +893,7 @@ txn_commit(struct txn *txn)
 	if (req == NULL)
 		goto rollback;
 
-	bool is_sync = txn_has_flag(txn, TXN_WAIT_SYNC);
+	bool is_sync = txn->flags & TXN_WAIT_SYNC;
 	if (is_sync) {
 		/*
 		 * Remote rows, if any, come before local rows, so
@@ -922,7 +920,7 @@ txn_commit(struct txn *txn)
 		goto rollback;
 	}
 	if (is_sync) {
-		if (txn_has_flag(txn, TXN_WAIT_ACK)) {
+		if (txn->flags & TXN_WAIT_ACK) {
 			int64_t lsn = req->rows[req->n_rows - 1]->lsn;
 			/*
 			 * Use local LSN assignment. Because
@@ -937,7 +935,7 @@ txn_commit(struct txn *txn)
 		if (txn_limbo_wait_complete(&txn_limbo, limbo_entry) < 0)
 			goto rollback;
 	}
-	assert(txn_has_flag(txn, TXN_IS_DONE));
+	assert(txn->flags & TXN_IS_DONE);
 	assert(txn->signature >= 0);
 
 	/* Synchronous transactions are freed by the calling fiber. */
@@ -946,7 +944,7 @@ txn_commit(struct txn *txn)
 
 rollback:
 	assert(txn->fiber != NULL);
-	if (!txn_has_flag(txn, TXN_IS_DONE)) {
+	if ((txn->flags & TXN_IS_DONE) == 0) {
 		fiber_set_txn(fiber(), txn);
 		txn_rollback(txn);
 	} else {
@@ -971,7 +969,7 @@ txn_rollback(struct txn *txn)
 	assert(txn == in_txn());
 	txn->status = TXN_ABORTED;
 	trigger_clear(&txn->fiber_on_stop);
-	if (!txn_has_flag(txn, TXN_CAN_YIELD))
+	if ((txn->flags & TXN_CAN_YIELD) == 0)
 		trigger_clear(&txn->fiber_on_yield);
 	txn->signature = TXN_SIGNATURE_ROLLBACK;
 	txn_complete_fail(txn);
@@ -992,12 +990,12 @@ bool
 txn_can_yield(struct txn *txn, bool set)
 {
 	assert(txn == in_txn());
-	bool could = txn_has_flag(txn, TXN_CAN_YIELD);
+	bool could = txn->flags & TXN_CAN_YIELD;
 	if (set && !could) {
-		txn_set_flag(txn, TXN_CAN_YIELD);
+		txn->flags |= TXN_CAN_YIELD;
 		trigger_clear(&txn->fiber_on_yield);
 	} else if (!set && could) {
-		txn_clear_flag(txn, TXN_CAN_YIELD);
+		txn->flags &= ~TXN_CAN_YIELD;
 		trigger_create(&txn->fiber_on_yield, txn_on_yield, NULL, NULL);
 		trigger_add(&fiber()->on_yield, &txn->fiber_on_yield);
 	}
@@ -1227,8 +1225,8 @@ txn_on_yield(struct trigger *trigger, void *event)
 	(void) event;
 	struct txn *txn = in_txn();
 	assert(txn != NULL);
-	assert(!txn_has_flag(txn, TXN_CAN_YIELD));
+	assert((txn->flags & TXN_CAN_YIELD) == 0);
 	txn_rollback_to_svp(txn, NULL);
-	txn_set_flag(txn, TXN_IS_ABORTED_BY_YIELD);
+	txn->flags |= TXN_IS_ABORTED_BY_YIELD;
 	return 0;
 }
diff --git a/src/box/txn.h b/src/box/txn.h
index fca9bc1d0..fecb98e93 100644
--- a/src/box/txn.h
+++ b/src/box/txn.h
@@ -56,25 +56,25 @@ struct Vdbe;
 
 enum txn_flag {
 	/** Transaction has been processed. */
-	TXN_IS_DONE,
+	TXN_IS_DONE = 0x1,
 	/**
 	 * Transaction has been aborted by fiber yield so
 	 * should be rolled back at commit.
 	 */
-	TXN_IS_ABORTED_BY_YIELD,
+	TXN_IS_ABORTED_BY_YIELD = 0x2,
 	/**
 	 * fiber_yield() is allowed inside the transaction.
 	 * See txn_can_yield() for more details.
 	 */
-	TXN_CAN_YIELD,
+	TXN_CAN_YIELD = 0x4,
 	/** on_commit and/or on_rollback list is not empty. */
-	TXN_HAS_TRIGGERS,
+	TXN_HAS_TRIGGERS = 0x8,
 	/**
 	 * Synchronous transaction touched sync spaces, or an
 	 * asynchronous transaction blocked by a sync one until it
 	 * is confirmed.
 	 */
-	TXN_WAIT_SYNC,
+	TXN_WAIT_SYNC = 0x10,
 	/**
 	 * Synchronous transaction 'waiting for ACKs' state before
 	 * commit. In this state it waits until it is replicated
@@ -82,14 +82,14 @@ enum txn_flag {
 	 * commit and returns success to a user.
 	 * TXN_WAIT_SYNC is always set, if TXN_WAIT_ACK is set.
 	 */
-	TXN_WAIT_ACK,
+	TXN_WAIT_ACK = 0x20,
 	/**
 	 * A transaction may be forced to be asynchronous, not
 	 * wait for any ACKs, and not depend on prepending sync
 	 * transactions. This happens in a few special cases. For
 	 * example, when applier receives snapshot from master.
 	 */
-	TXN_FORCE_ASYNC,
+	TXN_FORCE_ASYNC = 0x40,
 };
 
 enum {
@@ -393,24 +393,6 @@ struct txn {
 	struct rlist read_set;
 };
 
-static inline bool
-txn_has_flag(struct txn *txn, enum txn_flag flag)
-{
-	return (txn->flags & (1 << flag)) != 0;
-}
-
-static inline void
-txn_set_flag(struct txn *txn, enum txn_flag flag)
-{
-	txn->flags |= 1 << flag;
-}
-
-static inline void
-txn_clear_flag(struct txn *txn, enum txn_flag flag)
-{
-	txn->flags &= ~(1 << flag);
-}
-
 /* Pointer to the current transaction (if any) */
 static inline struct txn *
 in_txn(void)
@@ -487,11 +469,11 @@ txn_commit_async(struct txn *txn);
 static inline void
 txn_init_triggers(struct txn *txn)
 {
-	if (!txn_has_flag(txn, TXN_HAS_TRIGGERS)) {
+	if ((txn->flags & TXN_HAS_TRIGGERS) == 0) {
 		rlist_create(&txn->on_commit);
 		rlist_create(&txn->on_rollback);
 		rlist_create(&txn->on_wal_write);
-		txn_set_flag(txn, TXN_HAS_TRIGGERS);
+		txn->flags |= TXN_HAS_TRIGGERS;
 	}
 }
 
diff --git a/src/box/txn_limbo.c b/src/box/txn_limbo.c
index 9c4c3cdf1..0ac59a3a2 100644
--- a/src/box/txn_limbo.c
+++ b/src/box/txn_limbo.c
@@ -60,7 +60,7 @@ txn_limbo_last_synchro_entry(struct txn_limbo *limbo)
 {
 	struct txn_limbo_entry *entry;
 	rlist_foreach_entry_reverse(entry, &limbo->queue, in_queue) {
-		if (txn_has_flag(entry->txn, TXN_WAIT_ACK))
+		if (entry->txn->flags & TXN_WAIT_ACK)
 			return entry;
 	}
 	return NULL;
@@ -69,7 +69,7 @@ txn_limbo_last_synchro_entry(struct txn_limbo *limbo)
 struct txn_limbo_entry *
 txn_limbo_append(struct txn_limbo *limbo, uint32_t id, struct txn *txn)
 {
-	assert(txn_has_flag(txn, TXN_WAIT_SYNC));
+	assert(txn->flags & TXN_WAIT_SYNC);
 	assert(limbo == &txn_limbo);
 	/*
 	 * Transactions should be added to the limbo before WAL write. Limbo
@@ -168,7 +168,7 @@ txn_limbo_assign_remote_lsn(struct txn_limbo *limbo,
 	assert(limbo->owner_id != instance_id);
 	assert(entry->lsn == -1);
 	assert(lsn > 0);
-	assert(txn_has_flag(entry->txn, TXN_WAIT_ACK));
+	assert(entry->txn->flags & TXN_WAIT_ACK);
 	(void) limbo;
 	entry->lsn = lsn;
 }
@@ -181,7 +181,7 @@ txn_limbo_assign_local_lsn(struct txn_limbo *limbo,
 	assert(limbo->owner_id == instance_id);
 	assert(entry->lsn == -1);
 	assert(lsn > 0);
-	assert(txn_has_flag(entry->txn, TXN_WAIT_ACK));
+	assert(entry->txn->flags & TXN_WAIT_ACK);
 
 	entry->lsn = lsn;
 	/*
@@ -215,14 +215,14 @@ txn_limbo_write_rollback(struct txn_limbo *limbo, int64_t lsn);
 int
 txn_limbo_wait_complete(struct txn_limbo *limbo, struct txn_limbo_entry *entry)
 {
-	assert(entry->lsn > 0 || !txn_has_flag(entry->txn, TXN_WAIT_ACK));
+	assert(entry->lsn > 0 || (entry->txn->flags & TXN_WAIT_ACK) == 0);
 	bool cancellable = fiber_set_cancellable(false);
 
 	if (txn_limbo_entry_is_complete(entry))
 		goto complete;
 
-	assert(!txn_has_flag(entry->txn, TXN_IS_DONE));
-	assert(txn_has_flag(entry->txn, TXN_WAIT_SYNC));
+	assert((entry->txn->flags & TXN_IS_DONE) == 0);
+	assert(entry->txn->flags & TXN_WAIT_SYNC);
 	double start_time = fiber_clock();
 	while (true) {
 		double deadline = start_time + replication_synchro_timeout;
@@ -266,8 +266,7 @@ txn_limbo_wait_complete(struct txn_limbo *limbo, struct txn_limbo_entry *entry)
 					 in_queue, tmp) {
 		e->txn->signature = TXN_SIGNATURE_QUORUM_TIMEOUT;
 		txn_limbo_abort(limbo, e);
-		txn_clear_flag(e->txn, TXN_WAIT_SYNC);
-		txn_clear_flag(e->txn, TXN_WAIT_ACK);
+		e->txn->flags &= ~(TXN_WAIT_SYNC | TXN_WAIT_ACK);
 		txn_complete_fail(e->txn);
 		if (e == entry)
 			break;
@@ -289,7 +288,7 @@ txn_limbo_wait_complete(struct txn_limbo *limbo, struct txn_limbo_entry *entry)
 	 * installed the commit/rollback flag.
 	 */
 	assert(rlist_empty(&entry->in_queue));
-	assert(txn_has_flag(entry->txn, TXN_IS_DONE));
+	assert(entry->txn->flags & TXN_IS_DONE);
 	fiber_set_cancellable(cancellable);
 	/*
 	 * The first tx to be rolled back already performed all
@@ -385,7 +384,7 @@ txn_limbo_read_confirm(struct txn_limbo *limbo, int64_t lsn)
 		 * it is last, it does not depend on a not finished sync
 		 * transaction anymore and can be confirmed right away.
 		 */
-		if (txn_has_flag(e->txn, TXN_WAIT_ACK)) {
+		if (e->txn->flags & TXN_WAIT_ACK) {
 			/* Sync transaction not covered by the confirmation. */
 			if (e->lsn > lsn)
 				break;
@@ -399,8 +398,7 @@ txn_limbo_read_confirm(struct txn_limbo *limbo, int64_t lsn)
 		}
 		e->is_commit = true;
 		txn_limbo_remove(limbo, e);
-		txn_clear_flag(e->txn, TXN_WAIT_SYNC);
-		txn_clear_flag(e->txn, TXN_WAIT_ACK);
+		e->txn->flags &= ~(TXN_WAIT_SYNC | TXN_WAIT_ACK);
 		/*
 		 * If already written to WAL by now, finish tx processing.
 		 * Otherwise just clear the sync flags. Tx procesing will finish
@@ -446,7 +444,7 @@ txn_limbo_read_rollback(struct txn_limbo *limbo, int64_t lsn)
 	struct txn_limbo_entry *e, *tmp;
 	struct txn_limbo_entry *last_rollback = NULL;
 	rlist_foreach_entry_reverse(e, &limbo->queue, in_queue) {
-		if (!txn_has_flag(e->txn, TXN_WAIT_ACK))
+		if ((e->txn->flags & TXN_WAIT_ACK) == 0)
 			continue;
 		if (e->lsn < lsn)
 			break;
@@ -456,8 +454,7 @@ txn_limbo_read_rollback(struct txn_limbo *limbo, int64_t lsn)
 		return;
 	rlist_foreach_entry_safe_reverse(e, &limbo->queue, in_queue, tmp) {
 		txn_limbo_abort(limbo, e);
-		txn_clear_flag(e->txn, TXN_WAIT_SYNC);
-		txn_clear_flag(e->txn, TXN_WAIT_ACK);
+		e->txn->flags &= ~(TXN_WAIT_SYNC | TXN_WAIT_ACK);
 		if (e->txn->signature >= 0) {
 			/* Rollback the transaction. */
 			e->txn->signature = TXN_SIGNATURE_SYNC_ROLLBACK;
@@ -530,7 +527,7 @@ txn_limbo_ack(struct txn_limbo *limbo, uint32_t replica_id, int64_t lsn)
 		 * transactions are automatically committed right
 		 * after all the previous sync transactions are.
 		 */
-		if (!txn_has_flag(e->txn, TXN_WAIT_ACK)) {
+		if ((e->txn->flags & TXN_WAIT_ACK) == 0) {
 			assert(e->lsn == -1);
 			if (confirm_lsn == -1)
 				continue;
@@ -664,7 +661,7 @@ txn_limbo_force_empty(struct txn_limbo *limbo, int64_t confirm_lsn)
 	struct txn_limbo_entry *e, *last_quorum = NULL;
 	struct txn_limbo_entry *rollback = NULL;
 	rlist_foreach_entry(e, &limbo->queue, in_queue) {
-		if (txn_has_flag(e->txn, TXN_WAIT_ACK)) {
+		if (e->txn->flags & TXN_WAIT_ACK) {
 			if (e->lsn <= confirm_lsn) {
 				last_quorum = e;
 			} else {
@@ -693,7 +690,7 @@ txn_limbo_on_parameters_change(struct txn_limbo *limbo)
 	int64_t confirm_lsn = -1;
 	rlist_foreach_entry(e, &limbo->queue, in_queue) {
 		assert(e->ack_count <= VCLOCK_MAX);
-		if (!txn_has_flag(e->txn, TXN_WAIT_ACK)) {
+		if ((e->txn->flags & TXN_WAIT_ACK) == 0) {
 			assert(e->lsn == -1);
 			if (confirm_lsn == -1)
 				continue;
diff --git a/test/unit/snap_quorum_delay.cc b/test/unit/snap_quorum_delay.cc
index b9d4cc6c4..ec78f837d 100644
--- a/test/unit/snap_quorum_delay.cc
+++ b/test/unit/snap_quorum_delay.cc
@@ -98,8 +98,7 @@ txn_process_func(va_list ap)
 	struct txn *txn = txn_begin();
 	txn->fiber = fiber();
 	/* Simulate a sync transaction. */
-	txn_set_flag(txn, TXN_WAIT_SYNC);
-	txn_set_flag(txn, TXN_WAIT_ACK);
+	txn->flags |= TXN_WAIT_SYNC | TXN_WAIT_ACK;
 	/*
 	 * The true way to push the transaction to limbo is to call
 	 * txn_commit() for sync transaction. But, if txn_commit()
-- 
2.29.2


^ permalink raw reply	[flat|nested] 5+ messages in thread

* Re: [Tarantool-patches] [PATCH] txn: convert flags to explicit bitfield
  2021-02-15 14:01 [Tarantool-patches] [PATCH] txn: convert flags to explicit bitfield Cyrill Gorcunov via Tarantool-patches
@ 2021-02-17 21:15 ` Vladislav Shpilevoy via Tarantool-patches
  2021-02-17 21:45   ` Cyrill Gorcunov via Tarantool-patches
  0 siblings, 1 reply; 5+ messages in thread
From: Vladislav Shpilevoy via Tarantool-patches @ 2021-02-17 21:15 UTC (permalink / raw)
  To: Cyrill Gorcunov, tml

Hi! Thanks for the patch!

> @@ -535,7 +535,7 @@ txn_complete_fail(struct txn *txn)
>  		txn_rollback_one_stmt(txn, stmt);
>  	if (txn->engine != NULL)
>  		engine_rollback(txn->engine, txn);
> -	if (txn_has_flag(txn, TXN_HAS_TRIGGERS))
> +	if (txn->flags & TXN_HAS_TRIGGERS)

1. Did you do a self-review before sending this? Can you
tell yourself what is wrong here?


2. What was wrong with the idea of having the helpers
but operating on bitfields? And having has_flags() instead
of has_flag() to check presense of all specified flags +
clear_flags() to remove all specified flags.

I don't understand. I think we discussed it already 2 or 3
times and all seemed to agree?

^ permalink raw reply	[flat|nested] 5+ messages in thread

* Re: [Tarantool-patches] [PATCH] txn: convert flags to explicit bitfield
  2021-02-17 21:15 ` Vladislav Shpilevoy via Tarantool-patches
@ 2021-02-17 21:45   ` Cyrill Gorcunov via Tarantool-patches
  2021-02-18  8:32     ` Cyrill Gorcunov via Tarantool-patches
  0 siblings, 1 reply; 5+ messages in thread
From: Cyrill Gorcunov via Tarantool-patches @ 2021-02-17 21:45 UTC (permalink / raw)
  To: Vladislav Shpilevoy; +Cc: tml

On Wed, Feb 17, 2021 at 10:15:30PM +0100, Vladislav Shpilevoy wrote:
> Hi! Thanks for the patch!
> 
> > @@ -535,7 +535,7 @@ txn_complete_fail(struct txn *txn)
> >  		txn_rollback_one_stmt(txn, stmt);
> >  	if (txn->engine != NULL)
> >  		engine_rollback(txn->engine, txn);
> > -	if (txn_has_flag(txn, TXN_HAS_TRIGGERS))
> > +	if (txn->flags & TXN_HAS_TRIGGERS)
> 
> 1. Did you do a self-review before sending this? Can you
> tell yourself what is wrong here?

Yes I did, and I don't get what's wrong with this line? Previously
we have `if (txn_has_flag(txn, TXN_HAS_TRIGGERS))` where txn_has_flag
get expanded to

	return (txn->flags & TXN_HAS_TRIGGERS) != 0;

so where is an error?

> 2. What was wrong with the idea of having the helpers
> but operating on bitfields? And having has_flags() instead
> of has_flag() to check presense of all specified flags +
> clear_flags() to remove all specified flags.
> 
> I don't understand. I think we discussed it already 2 or 3
> times and all seemed to agree?

The initial idea was to operate with flags directly, you pointed
that lines of code as if (!(txn->flags & mask)) violates our
code style. Then we thought of renaming the helpers to txn_x_flags
to operate with multiple flags, which lead us to a vague moment --
it is unclear how exactly txn_has_flag should operate: exact
matching or any matching?

Finlly I simply moved back to direct access to the fields
with explicit `!= 0` or `== 0` form.

^ permalink raw reply	[flat|nested] 5+ messages in thread

* Re: [Tarantool-patches] [PATCH] txn: convert flags to explicit bitfield
  2021-02-17 21:45   ` Cyrill Gorcunov via Tarantool-patches
@ 2021-02-18  8:32     ` Cyrill Gorcunov via Tarantool-patches
  2021-02-23 21:37       ` Vladislav Shpilevoy via Tarantool-patches
  0 siblings, 1 reply; 5+ messages in thread
From: Cyrill Gorcunov via Tarantool-patches @ 2021-02-18  8:32 UTC (permalink / raw)
  To: Vladislav Shpilevoy; +Cc: tml

On Thu, Feb 18, 2021 at 12:45:56AM +0300, Cyrill Gorcunov wrote:
> On Wed, Feb 17, 2021 at 10:15:30PM +0100, Vladislav Shpilevoy wrote:
> > Hi! Thanks for the patch!
> > 
> > > @@ -535,7 +535,7 @@ txn_complete_fail(struct txn *txn)
> > >  		txn_rollback_one_stmt(txn, stmt);
> > >  	if (txn->engine != NULL)
> > >  		engine_rollback(txn->engine, txn);
> > > -	if (txn_has_flag(txn, TXN_HAS_TRIGGERS))
> > > +	if (txn->flags & TXN_HAS_TRIGGERS)
> > 
> > 1. Did you do a self-review before sending this? Can you
> > tell yourself what is wrong here?
> 
> Yes I did, and I don't get what's wrong with this line? Previously
> we have `if (txn_has_flag(txn, TXN_HAS_TRIGGERS))` where txn_has_flag
> get expanded to
> 
> 	return (txn->flags & TXN_HAS_TRIGGERS) != 0;
> 
> so where is an error?
> 
> > 2. What was wrong with the idea of having the helpers
> > but operating on bitfields? And having has_flags() instead
> > of has_flag() to check presense of all specified flags +
> > clear_flags() to remove all specified flags.
> > 
> > I don't understand. I think we discussed it already 2 or 3
> > times and all seemed to agree?
> 
> The initial idea was to operate with flags directly, you pointed
> that lines of code as if (!(txn->flags & mask)) violates our
> code style. Then we thought of renaming the helpers to txn_x_flags
> to operate with multiple flags, which lead us to a vague moment --
> it is unclear how exactly txn_has_flag should operate: exact
> matching or any matching?
> 
> Finlly I simply moved back to direct access to the fields
> with explicit `!= 0` or `== 0` form.

Anyway, here is an another approach, sitting in
gorcunov/gh-5128-txn-flags-5 branch
---
From 0e4484213d53f58e34db16e3cb96c63805bdd5e7 Mon Sep 17 00:00:00 2001
From: Cyrill Gorcunov <gorcunov@gmail.com>
Date: Thu, 4 Feb 2021 14:19:12 +0300
Subject: [PATCH] txn: convert flags to explicit bitfield

Instead of shifting flags inside txn_x_flag() helpers lets
define precompiled value. This allows us to operate with
several flags at onces and shrink code a bit.

Closes #5128

Signed-off-by: Cyrill Gorcunov <gorcunov@gmail.com>
---
 src/box/applier.cc             |  4 ++--
 src/box/memtx_engine.c         |  2 +-
 src/box/txn.c                  | 15 +++++++--------
 src/box/txn.h                  | 27 ++++++++++++++-------------
 src/box/txn_limbo.c            |  9 +++------
 test/unit/snap_quorum_delay.cc |  3 +--
 6 files changed, 28 insertions(+), 32 deletions(-)

diff --git a/src/box/applier.cc b/src/box/applier.cc
index 553db76fc..8d7ce5d99 100644
--- a/src/box/applier.cc
+++ b/src/box/applier.cc
@@ -237,7 +237,7 @@ apply_snapshot_row(struct xrow_header *row)
 	 * Do not wait for confirmation when fetching a snapshot.
 	 * Master only sends confirmed rows during join.
 	 */
-	txn_set_flag(txn, TXN_FORCE_ASYNC);
+	txn_set_flags(txn, TXN_FORCE_ASYNC);
 	if (txn_begin_stmt(txn, space) != 0)
 		goto rollback;
 	/* no access checks here - applier always works with admin privs */
@@ -308,7 +308,7 @@ apply_final_join_row(struct xrow_header *row)
 	 * Do not wait for confirmation while processing final
 	 * join rows. See apply_snapshot_row().
 	 */
-	txn_set_flag(txn, TXN_FORCE_ASYNC);
+	txn_set_flags(txn, TXN_FORCE_ASYNC);
 	if (apply_row(row) != 0) {
 		txn_rollback(txn);
 		fiber_gc();
diff --git a/src/box/memtx_engine.c b/src/box/memtx_engine.c
index 1174234d2..a4cd671f5 100644
--- a/src/box/memtx_engine.c
+++ b/src/box/memtx_engine.c
@@ -265,7 +265,7 @@ memtx_engine_recover_snapshot_row(struct memtx_engine *memtx,
 	 * Snapshot rows are confirmed by definition. They don't need to go to
 	 * the synchronous transactions limbo.
 	 */
-	txn_set_flag(txn, TXN_FORCE_ASYNC);
+	txn_set_flags(txn, TXN_FORCE_ASYNC);
 	rc = txn_commit(txn);
 	/*
 	 * Don't let gc pool grow too much. Yet to
diff --git a/src/box/txn.c b/src/box/txn.c
index a5edbfc60..71c89ce5f 100644
--- a/src/box/txn.c
+++ b/src/box/txn.c
@@ -282,7 +282,7 @@ txn_begin(void)
 	 * It's a responsibility of an engine to disable yields
 	 * if they are not supported.
 	 */
-	txn_set_flag(txn, TXN_CAN_YIELD);
+	txn_set_flags(txn, TXN_CAN_YIELD);
 	return txn;
 }
 
@@ -516,7 +516,7 @@ txn_free_or_wakeup(struct txn *txn)
 	if (txn->fiber == NULL)
 		txn_free(txn);
 	else {
-		txn_set_flag(txn, TXN_IS_DONE);
+		txn_set_flags(txn, TXN_IS_DONE);
 		if (txn->fiber != fiber())
 			/* Wake a waiting fiber up. */
 			fiber_wakeup(txn->fiber);
@@ -642,8 +642,7 @@ txn_journal_entry_new(struct txn *txn)
 	 */
 	if (!txn_has_flag(txn, TXN_FORCE_ASYNC)) {
 		if (is_sync) {
-			txn_set_flag(txn, TXN_WAIT_SYNC);
-			txn_set_flag(txn, TXN_WAIT_ACK);
+			txn_set_flags(txn, TXN_WAIT_SYNC | TXN_WAIT_ACK);
 		} else if (!txn_limbo_is_empty(&txn_limbo)) {
 			/*
 			 * There some sync entries on the
@@ -652,7 +651,7 @@ txn_journal_entry_new(struct txn *txn)
 			 * doesn't touch sync space (each sync txn
 			 * should be considered as a barrier).
 			 */
-			txn_set_flag(txn, TXN_WAIT_SYNC);
+			txn_set_flags(txn, TXN_WAIT_SYNC);
 		}
 	}
 
@@ -994,10 +993,10 @@ txn_can_yield(struct txn *txn, bool set)
 	assert(txn == in_txn());
 	bool could = txn_has_flag(txn, TXN_CAN_YIELD);
 	if (set && !could) {
-		txn_set_flag(txn, TXN_CAN_YIELD);
+		txn_set_flags(txn, TXN_CAN_YIELD);
 		trigger_clear(&txn->fiber_on_yield);
 	} else if (!set && could) {
-		txn_clear_flag(txn, TXN_CAN_YIELD);
+		txn_clear_flags(txn, TXN_CAN_YIELD);
 		trigger_create(&txn->fiber_on_yield, txn_on_yield, NULL, NULL);
 		trigger_add(&fiber()->on_yield, &txn->fiber_on_yield);
 	}
@@ -1229,6 +1228,6 @@ txn_on_yield(struct trigger *trigger, void *event)
 	assert(txn != NULL);
 	assert(!txn_has_flag(txn, TXN_CAN_YIELD));
 	txn_rollback_to_svp(txn, NULL);
-	txn_set_flag(txn, TXN_IS_ABORTED_BY_YIELD);
+	txn_set_flags(txn, TXN_IS_ABORTED_BY_YIELD);
 	return 0;
 }
diff --git a/src/box/txn.h b/src/box/txn.h
index fca9bc1d0..29fe6d5ce 100644
--- a/src/box/txn.h
+++ b/src/box/txn.h
@@ -56,25 +56,25 @@ struct Vdbe;
 
 enum txn_flag {
 	/** Transaction has been processed. */
-	TXN_IS_DONE,
+	TXN_IS_DONE = 0x1,
 	/**
 	 * Transaction has been aborted by fiber yield so
 	 * should be rolled back at commit.
 	 */
-	TXN_IS_ABORTED_BY_YIELD,
+	TXN_IS_ABORTED_BY_YIELD = 0x2,
 	/**
 	 * fiber_yield() is allowed inside the transaction.
 	 * See txn_can_yield() for more details.
 	 */
-	TXN_CAN_YIELD,
+	TXN_CAN_YIELD = 0x4,
 	/** on_commit and/or on_rollback list is not empty. */
-	TXN_HAS_TRIGGERS,
+	TXN_HAS_TRIGGERS = 0x8,
 	/**
 	 * Synchronous transaction touched sync spaces, or an
 	 * asynchronous transaction blocked by a sync one until it
 	 * is confirmed.
 	 */
-	TXN_WAIT_SYNC,
+	TXN_WAIT_SYNC = 0x10,
 	/**
 	 * Synchronous transaction 'waiting for ACKs' state before
 	 * commit. In this state it waits until it is replicated
@@ -82,14 +82,14 @@ enum txn_flag {
 	 * commit and returns success to a user.
 	 * TXN_WAIT_SYNC is always set, if TXN_WAIT_ACK is set.
 	 */
-	TXN_WAIT_ACK,
+	TXN_WAIT_ACK = 0x20,
 	/**
 	 * A transaction may be forced to be asynchronous, not
 	 * wait for any ACKs, and not depend on prepending sync
 	 * transactions. This happens in a few special cases. For
 	 * example, when applier receives snapshot from master.
 	 */
-	TXN_FORCE_ASYNC,
+	TXN_FORCE_ASYNC = 0x40,
 };
 
 enum {
@@ -396,19 +396,20 @@ struct txn {
 static inline bool
 txn_has_flag(struct txn *txn, enum txn_flag flag)
 {
-	return (txn->flags & (1 << flag)) != 0;
+	assert((flag & (flag - 1)) == 0);
+	return (txn->flags & flag) != 0;
 }
 
 static inline void
-txn_set_flag(struct txn *txn, enum txn_flag flag)
+txn_set_flags(struct txn *txn, unsigned int flags)
 {
-	txn->flags |= 1 << flag;
+	txn->flags |= flags;
 }
 
 static inline void
-txn_clear_flag(struct txn *txn, enum txn_flag flag)
+txn_clear_flags(struct txn *txn, unsigned int flags)
 {
-	txn->flags &= ~(1 << flag);
+	txn->flags &= ~flags;
 }
 
 /* Pointer to the current transaction (if any) */
@@ -491,7 +492,7 @@ txn_init_triggers(struct txn *txn)
 		rlist_create(&txn->on_commit);
 		rlist_create(&txn->on_rollback);
 		rlist_create(&txn->on_wal_write);
-		txn_set_flag(txn, TXN_HAS_TRIGGERS);
+		txn_set_flags(txn, TXN_HAS_TRIGGERS);
 	}
 }
 
diff --git a/src/box/txn_limbo.c b/src/box/txn_limbo.c
index 9c4c3cdf1..cf0ad9350 100644
--- a/src/box/txn_limbo.c
+++ b/src/box/txn_limbo.c
@@ -266,8 +266,7 @@ txn_limbo_wait_complete(struct txn_limbo *limbo, struct txn_limbo_entry *entry)
 					 in_queue, tmp) {
 		e->txn->signature = TXN_SIGNATURE_QUORUM_TIMEOUT;
 		txn_limbo_abort(limbo, e);
-		txn_clear_flag(e->txn, TXN_WAIT_SYNC);
-		txn_clear_flag(e->txn, TXN_WAIT_ACK);
+		txn_clear_flags(e->txn, TXN_WAIT_SYNC | TXN_WAIT_ACK);
 		txn_complete_fail(e->txn);
 		if (e == entry)
 			break;
@@ -399,8 +398,7 @@ txn_limbo_read_confirm(struct txn_limbo *limbo, int64_t lsn)
 		}
 		e->is_commit = true;
 		txn_limbo_remove(limbo, e);
-		txn_clear_flag(e->txn, TXN_WAIT_SYNC);
-		txn_clear_flag(e->txn, TXN_WAIT_ACK);
+		txn_clear_flags(e->txn, TXN_WAIT_SYNC | TXN_WAIT_ACK);
 		/*
 		 * If already written to WAL by now, finish tx processing.
 		 * Otherwise just clear the sync flags. Tx procesing will finish
@@ -456,8 +454,7 @@ txn_limbo_read_rollback(struct txn_limbo *limbo, int64_t lsn)
 		return;
 	rlist_foreach_entry_safe_reverse(e, &limbo->queue, in_queue, tmp) {
 		txn_limbo_abort(limbo, e);
-		txn_clear_flag(e->txn, TXN_WAIT_SYNC);
-		txn_clear_flag(e->txn, TXN_WAIT_ACK);
+		txn_clear_flags(e->txn, TXN_WAIT_SYNC | TXN_WAIT_ACK);
 		if (e->txn->signature >= 0) {
 			/* Rollback the transaction. */
 			e->txn->signature = TXN_SIGNATURE_SYNC_ROLLBACK;
diff --git a/test/unit/snap_quorum_delay.cc b/test/unit/snap_quorum_delay.cc
index b9d4cc6c4..803bbbea8 100644
--- a/test/unit/snap_quorum_delay.cc
+++ b/test/unit/snap_quorum_delay.cc
@@ -98,8 +98,7 @@ txn_process_func(va_list ap)
 	struct txn *txn = txn_begin();
 	txn->fiber = fiber();
 	/* Simulate a sync transaction. */
-	txn_set_flag(txn, TXN_WAIT_SYNC);
-	txn_set_flag(txn, TXN_WAIT_ACK);
+	txn_set_flags(txn, TXN_WAIT_SYNC | TXN_WAIT_ACK);
 	/*
 	 * The true way to push the transaction to limbo is to call
 	 * txn_commit() for sync transaction. But, if txn_commit()
-- 
2.29.2


^ permalink raw reply	[flat|nested] 5+ messages in thread

* Re: [Tarantool-patches] [PATCH] txn: convert flags to explicit bitfield
  2021-02-18  8:32     ` Cyrill Gorcunov via Tarantool-patches
@ 2021-02-23 21:37       ` Vladislav Shpilevoy via Tarantool-patches
  0 siblings, 0 replies; 5+ messages in thread
From: Vladislav Shpilevoy via Tarantool-patches @ 2021-02-23 21:37 UTC (permalink / raw)
  To: Cyrill Gorcunov; +Cc: tml

Hi! Thanks for the patch!

LGTM.

> Anyway, here is an another approach, sitting in
> gorcunov/gh-5128-txn-flags-5 branch

^ permalink raw reply	[flat|nested] 5+ messages in thread

end of thread, other threads:[~2021-02-23 21:37 UTC | newest]

Thread overview: 5+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2021-02-15 14:01 [Tarantool-patches] [PATCH] txn: convert flags to explicit bitfield Cyrill Gorcunov via Tarantool-patches
2021-02-17 21:15 ` Vladislav Shpilevoy via Tarantool-patches
2021-02-17 21:45   ` Cyrill Gorcunov via Tarantool-patches
2021-02-18  8:32     ` Cyrill Gorcunov via Tarantool-patches
2021-02-23 21:37       ` Vladislav Shpilevoy via Tarantool-patches

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox