net_sched: sch_fq: rework fq_gc() to avoid stack canary

Using kmem_cache_free_bulk() in fq_gc() was not optimal.

1) It needs an array.
2) It is only saving cpu cycles for large batches.

The automatic array forces a stack canary, which is expensive.

In practice fq_gc was finding zero, one or two flows at most
per round.

Remove the array, use kmem_cache_free().

This makes fq_enqueue() smaller and faster.

$ scripts/bloat-o-meter -t vmlinux.old vmlinux.new
add/remove: 0/0 grow/shrink: 0/1 up/down: 0/-79 (-79)
Function                                     old     new   delta
fq_enqueue                                  1629    1550     -79
Total: Before=24886583, After=24886504, chg -0.00%

Signed-off-by: Eric Dumazet <edumazet@google.com>
Link: https://patch.msgid.link/20260204190034.76277-1-edumazet@google.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

diff --git a/net/sched/sch_fq.c b/net/sched/sch_fq.c
index d0200ec8ada62e86f10d823556bedcaefb470e6c..80235e85f8440ee83032f171cf28df6f161473db 100644 (file)
--- a/net/sched/sch_fq.c
+++ b/net/sched/sch_fq.c
@@ -245,8 +245,6 @@ static void fq_flow_set_throttled(struct fq_sched_data *q, struct fq_flow *f)
 static struct kmem_cache *fq_flow_cachep __read_mostly;
 
 
-/* limit number of collected flows per round */
-#define FQ_GC_MAX 8
 #define FQ_GC_AGE (3*HZ)
 
 static bool fq_gc_candidate(const struct fq_flow *f)
@@ -259,10 +257,9 @@ static void fq_gc(struct fq_sched_data *q,
                  struct rb_root *root,
                  struct sock *sk)
 {
+       struct fq_flow *f, *tofree = NULL;
        struct rb_node **p, *parent;
-       void *tofree[FQ_GC_MAX];
-       struct fq_flow *f;
-       int i, fcnt = 0;
+       int fcnt;
 
        p = &root->rb_node;
        parent = NULL;
@@ -274,9 +271,8 @@ static void fq_gc(struct fq_sched_data *q,
                        break;
 
                if (fq_gc_candidate(f)) {
-                       tofree[fcnt++] = f;
-                       if (fcnt == FQ_GC_MAX)
-                               break;
+                       f->next = tofree;
+                       tofree = f;
                }
 
                if (f->sk > sk)
@@ -285,18 +281,20 @@ static void fq_gc(struct fq_sched_data *q,
                        p = &parent->rb_left;
        }
 
-       if (!fcnt)
+       if (!tofree)
                return;
 
-       for (i = fcnt; i > 0; ) {
-               f = tofree[--i];
+       fcnt = 0;
+       while (tofree) {
+               f = tofree;
+               tofree = f->next;
                rb_erase(&f->fq_node, root);
+               kmem_cache_free(fq_flow_cachep, f);
+               fcnt++;
        }
        q->flows -= fcnt;
        q->inactive_flows -= fcnt;
        q->stat_gc_flows += fcnt;
-
-       kmem_cache_free_bulk(fq_flow_cachep, fcnt, tofree);
 }
 
 /* Fast path can be used if :