arc_evict, arc_evict_ghost: reduce stack usage using kmem_zalloc

With debugging enabled and depending on your kernel config, the size of
arc_buf_hdr_t can blow out the stack of arc_evict() and arc_evict_ghost()
to greater than 1024 bytes. Let's avoid this.

Signed-off-by: Chris Dunlop <[email protected]>
Signed-off-by: Brian Behlendorf <[email protected]>
Closes #3377

1 files changed, 20 insertions, 12 deletions

diff --git a/module/zfs/arc.c b/module/zfs/arc.c
index 421c81e1c..6975ada62 100644
--- a/module/zfs/arc.c
+++ b/module/zfs/arc.c
@@ -1825,13 +1825,15 @@ arc_evict(arc_state_t *state, uint64_t spa, int64_t bytes, boolean_t recycle,
 	kmutex_t *hash_lock;
 	boolean_t have_lock;
 	void *stolen = NULL;
-	arc_buf_hdr_t marker = {{{ 0 }}};
+	arc_buf_hdr_t *marker;
 	int count = 0;
 
 	ASSERT(state == arc_mru || state == arc_mfu);
 
 	evicted_state = (state == arc_mru) ? arc_mru_ghost : arc_mfu_ghost;
 
+	marker = kmem_zalloc(sizeof (arc_buf_hdr_t), KM_SLEEP);
+
 top:
 	mutex_enter(&state->arcs_mtx);
 	mutex_enter(&evicted_state->arcs_mtx);
@@ -1866,14 +1868,14 @@ top:
 		 * the hot code path, so don't sleep.
 		 */
 		if (!recycle && count++ > arc_evict_iterations) {
-			list_insert_after(list, ab, &marker);
+			list_insert_after(list, ab, marker);
 			mutex_exit(&evicted_state->arcs_mtx);
 			mutex_exit(&state->arcs_mtx);
 			kpreempt(KPREEMPT_SYNC);
 			mutex_enter(&state->arcs_mtx);
 			mutex_enter(&evicted_state->arcs_mtx);
-			ab_prev = list_prev(list, &marker);
-			list_remove(list, &marker);
+			ab_prev = list_prev(list, marker);
+			list_remove(list, marker);
 			count = 0;
 			continue;
 		}
@@ -1957,6 +1959,8 @@ top:
 		goto top;
 	}
 
+	kmem_free(marker, sizeof (arc_buf_hdr_t));
+
 	if (bytes_evicted < bytes)
 		dprintf("only evicted %lld bytes from %x\n",
 		    (longlong_t)bytes_evicted, state->arcs_state);
@@ -1986,7 +1990,7 @@ arc_evict_ghost(arc_state_t *state, uint64_t spa, int64_t bytes,
     arc_buf_contents_t type)
 {
 	arc_buf_hdr_t *ab, *ab_prev;
-	arc_buf_hdr_t marker;
+	arc_buf_hdr_t *marker;
 	list_t *list = &state->arcs_list[type];
 	kmutex_t *hash_lock;
 	uint64_t bytes_deleted = 0;
@@ -1994,7 +1998,9 @@ arc_evict_ghost(arc_state_t *state, uint64_t spa, int64_t bytes,
 	int count = 0;
 
 	ASSERT(GHOST_STATE(state));
-	bzero(&marker, sizeof (marker));
+
+	marker = kmem_zalloc(sizeof (arc_buf_hdr_t), KM_SLEEP);
+
 top:
 	mutex_enter(&state->arcs_mtx);
 	for (ab = list_tail(list); ab; ab = ab_prev) {
@@ -2020,12 +2026,12 @@ top:
 		 * before reacquiring the lock.
 		 */
 		if (count++ > arc_evict_iterations) {
-			list_insert_after(list, ab, &marker);
+			list_insert_after(list, ab, marker);
 			mutex_exit(&state->arcs_mtx);
 			kpreempt(KPREEMPT_SYNC);
 			mutex_enter(&state->arcs_mtx);
-			ab_prev = list_prev(list, &marker);
-			list_remove(list, &marker);
+			ab_prev = list_prev(list, marker);
+			list_remove(list, marker);
 			count = 0;
 			continue;
 		}
@@ -2057,13 +2063,13 @@ top:
 			 * hash lock to become available. Once its
 			 * available, restart from where we left off.
 			 */
-			list_insert_after(list, ab, &marker);
+			list_insert_after(list, ab, marker);
 			mutex_exit(&state->arcs_mtx);
 			mutex_enter(hash_lock);
 			mutex_exit(hash_lock);
 			mutex_enter(&state->arcs_mtx);
-			ab_prev = list_prev(list, &marker);
-			list_remove(list, &marker);
+			ab_prev = list_prev(list, marker);
+			list_remove(list, marker);
 		} else {
 			bufs_skipped += 1;
 		}
@@ -2076,6 +2082,8 @@ top:
 		goto top;
 	}
 
+	kmem_free(marker, sizeof (arc_buf_hdr_t));
+
 	if (bufs_skipped) {
 		ARCSTAT_INCR(arcstat_mutex_miss, bufs_skipped);
 		ASSERT(bytes >= 0);