L2ARC: Implement persistent markers with consistent tail scanning

This commit introduces per-sublist persistent markers that eliminate
redundant tail scanning between L2ARC iterations, providing significant
CPU efficiency improvements. Markers are pre-allocated during device
initialization and properly cleaned up during device removal.

The implementation uses conditional behavior based on device capacity:
small devices (capacity < arc_c) retain original HEAD/TAIL scanning
based on ARC warmup state, while large devices (capacity >= arc_c)
use the persistent marker approach for optimal CPU efficiency.

Reviewed-by: Alexander Motin <alexander.motin@TrueNAS.com>
Reviewed-by: Brian Behlendorf <behlendorf1@llnl.gov>
Signed-off-by: Ameer Hamza <ahamza@ixsystems.com>
Closes #18093

include/sys/arc_impl.h

@@ -41,6 +41,16 @@
 extern "C" {
 #endif
 
+/*
+ * L2ARC state and statistics for persistent marker management.
+ */
+typedef struct l2arc_info {
+	arc_buf_hdr_t	***l2arc_markers;	/* persistent markers */
+	uint64_t	l2arc_total_writes;	/* total writes for reset */
+	uint64_t	l2arc_total_capacity;	/* total L2ARC capacity */
+	uint64_t	l2arc_smallest_capacity; /* smallest device capacity */
+} l2arc_info_t;
+
 /*
  * Note that buffers can be in one of 6 states:
  *	ARC_anon	- anonymous (discussed below)

include/sys/spa_impl.h

@@ -52,6 +52,7 @@
 #include <sys/dsl_crypt.h>
 #include <sys/zfeature.h>
 #include <sys/zthr.h>
+#include <sys/arc_impl.h>
 #include <sys/dsl_deadlist.h>
 #include <zfeature_common.h>
 
@@ -283,6 +284,7 @@ struct spa {
 	spa_aux_vdev_t	spa_spares;		/* hot spares */
 	spa_aux_vdev_t	spa_l2cache;		/* L2ARC cache devices */
 	boolean_t	spa_aux_sync_uber;	/* need to sync aux uber */
+	l2arc_info_t	spa_l2arc_info;		/* L2ARC state and stats */
 	nvlist_t	*spa_label_features;	/* Features for reading MOS */
 	uint64_t	spa_config_object;	/* MOS object for pool config */
 	uint64_t	spa_config_generation;	/* config generation number */