txg_wait_synced_flags: add TXG_WAIT_SUSPEND flag to not wait if pool suspended

This allows a caller to request a wait for txg sync, with an appropriate
error return if the pool is suspended or becomes suspended during the
wait.

To support this, txg_wait_kick() is added to signal the sync condvar,
which wakes up the waiters, causing them to loop and reconsider their
wait conditions again. zio_suspend() now calls this to trigger the break
if the pool suspends while waiting.

Sponsored-by: Klara, Inc.
Sponsored-by: Wasabi Technology, Inc.
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Reviewed-by: Alexander Motin <mav@FreeBSD.org>
Reviewed-by: Paul Dagnelie <paul.dagnelie@klarasystems.com>
Signed-off-by: Rob Norris <rob.norris@klarasystems.com>
Closes #17355

include/sys/txg.h

@@ -25,6 +25,7 @@
  */
 /*
  * Copyright (c) 2012, 2017 by Delphix. All rights reserved.
+ * Copyright (c) 2025, Klara, Inc.
  */
 
 #ifndef _SYS_TXG_H
@@ -78,6 +79,9 @@ typedef enum {
 
 	/* If a signal arrives while waiting, abort and return EINTR */
 	TXG_WAIT_SIGNAL = (1 << 0),
+
+	/* If the pool suspends while waiting, abort and return ESHUTDOWN. */
+	TXG_WAIT_SUSPEND = (1 << 1),
 } txg_wait_flag_t;
 
 struct dsl_pool;
@@ -111,6 +115,11 @@ extern int txg_wait_synced_flags(struct dsl_pool *dp, uint64_t txg,
  */
 extern void txg_wait_synced(struct dsl_pool *dp, uint64_t txg);
 
+/*
+ * Wake all threads waiting in txg_wait_synced_flags() so they can reevaluate.
+ */
+extern void txg_wait_kick(struct dsl_pool *dp);
+
 /*
  * Wait until the given transaction group, or one after it, is
  * the open transaction group.  Try to make this happen as soon

module/zfs/txg.c

@@ -23,6 +23,7 @@
  * Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
  * Portions Copyright 2011 Martin Matuska
  * Copyright (c) 2012, 2019 by Delphix. All rights reserved.
+ * Copyright (c) 2025, Klara, Inc.
  */
 
 #include <sys/zfs_context.h>
@@ -705,7 +706,7 @@ txg_wait_synced_flags(dsl_pool_t *dp, uint64_t txg, txg_wait_flag_t flags)
 	int error = 0;
 	tx_state_t *tx = &dp->dp_tx;
 
-	ASSERT0(flags & ~TXG_WAIT_SIGNAL);
+	ASSERT0(flags & ~(TXG_WAIT_SIGNAL | TXG_WAIT_SUSPEND));
 	ASSERT(!dsl_pool_config_held(dp));
 
 	mutex_enter(&tx->tx_sync_lock);
@@ -723,6 +724,15 @@ txg_wait_synced_flags(dsl_pool_t *dp, uint64_t txg, txg_wait_flag_t flags)
 	 * else interesting happens, we'll set an error and break out.
 	 */
 	while (tx->tx_synced_txg < txg) {
+		if ((flags & TXG_WAIT_SUSPEND) && spa_suspended(dp->dp_spa)) {
+			/*
+			 * Pool suspended and the caller does not want to
+			 * block; inform them immediately.
+			 */
+			error = SET_ERROR(ESHUTDOWN);
+			break;
+		}
+
 		dprintf("broadcasting sync more "
 		    "tx_synced=%llu waiting=%llu dp=%px\n",
 		    (u_longlong_t)tx->tx_synced_txg,
@@ -756,6 +766,15 @@ txg_wait_synced(dsl_pool_t *dp, uint64_t txg)
 	VERIFY0(txg_wait_synced_flags(dp, txg, TXG_WAIT_NONE));
 }
 
+void
+txg_wait_kick(dsl_pool_t *dp)
+{
+	tx_state_t *tx = &dp->dp_tx;
+	mutex_enter(&tx->tx_sync_lock);
+	cv_broadcast(&tx->tx_sync_done_cv);
+	mutex_exit(&tx->tx_sync_lock);
+}
+
 /*
  * Wait for the specified open transaction group.  Set should_quiesce
  * when the current open txg should be quiesced immediately.

module/zfs/zio.c

@@ -2714,6 +2714,8 @@ zio_suspend(spa_t *spa, zio_t *zio, zio_suspend_reason_t reason)
 	}
 
 	mutex_exit(&spa->spa_suspend_lock);
+
+	txg_wait_kick(spa->spa_dsl_pool);
 }
 
 int