bpf: Add a hint to allocated objects.

To address OOM issue when one cpu is allocating and another cpu is freeing add
a target bpf_mem_cache hint to allocated objects and when local cpu free_llist
overflows free to that bpf_mem_cache. The hint addresses the OOM while
maintaining the same performance for common case when alloc/free are done on the
same cpu.

Note that do_call_rcu_ttrace() now has to check 'draining' flag in one more case,
since do_call_rcu_ttrace() is called not only for current cpu.

Signed-off-by: Alexei Starovoitov <ast@kernel.org>
Signed-off-by: Daniel Borkmann <daniel@iogearbox.net>
Acked-by: Hou Tao <houtao1@huawei.com>
Link: https://lore.kernel.org/bpf/20230706033447.54696-9-alexei.starovoitov@gmail.com

Author: Alexei Starovoitov

kernel/bpf/memalloc.c

@@ -99,6 +99,7 @@ struct bpf_mem_cache {
 	int low_watermark, high_watermark, batch;
 	int percpu_size;
 	bool draining;
+	struct bpf_mem_cache *tgt;
 
 	/* list of objects to be freed after RCU tasks trace GP */
 	struct llist_head free_by_rcu_ttrace;
@@ -199,18 +200,11 @@ static void alloc_bulk(struct bpf_mem_cache *c, int cnt, int node)
 
 	for (i = 0; i < cnt; i++) {
 		/*
-		 * free_by_rcu_ttrace is only manipulated by irq work refill_work().
-		 * IRQ works on the same CPU are called sequentially, so it is
-		 * safe to use __llist_del_first() here. If alloc_bulk() is
-		 * invoked by the initial prefill, there will be no running
-		 * refill_work(), so __llist_del_first() is fine as well.
-		 *
-		 * In most cases, objects on free_by_rcu_ttrace are from the same CPU.
-		 * If some objects come from other CPUs, it doesn't incur any
-		 * harm because NUMA_NO_NODE means the preference for current
-		 * numa node and it is not a guarantee.
+		 * For every 'c' llist_del_first(&c->free_by_rcu_ttrace); is
+		 * done only by one CPU == current CPU. Other CPUs might
+		 * llist_add() and llist_del_all() in parallel.
 		 */
-		obj = __llist_del_first(&c->free_by_rcu_ttrace);
+		obj = llist_del_first(&c->free_by_rcu_ttrace);
 		if (!obj)
 			break;
 		add_obj_to_free_list(c, obj);
@@ -284,18 +278,23 @@ static void enque_to_free(struct bpf_mem_cache *c, void *obj)
 	/* bpf_mem_cache is a per-cpu object. Freeing happens in irq_work.
 	 * Nothing races to add to free_by_rcu_ttrace list.
 	 */
-	__llist_add(llnode, &c->free_by_rcu_ttrace);
+	llist_add(llnode, &c->free_by_rcu_ttrace);
 }
 
 static void do_call_rcu_ttrace(struct bpf_mem_cache *c)
 {
 	struct llist_node *llnode, *t;
 
-	if (atomic_xchg(&c->call_rcu_ttrace_in_progress, 1))
+	if (atomic_xchg(&c->call_rcu_ttrace_in_progress, 1)) {
+		if (unlikely(READ_ONCE(c->draining))) {
+			llnode = llist_del_all(&c->free_by_rcu_ttrace);
+			free_all(llnode, !!c->percpu_size);
+		}
 		return;
+	}
 
 	WARN_ON_ONCE(!llist_empty(&c->waiting_for_gp_ttrace));
-	llist_for_each_safe(llnode, t, __llist_del_all(&c->free_by_rcu_ttrace))
+	llist_for_each_safe(llnode, t, llist_del_all(&c->free_by_rcu_ttrace))
 		/* There is no concurrent __llist_add(waiting_for_gp_ttrace) access.
 		 * It doesn't race with llist_del_all either.
 		 * But there could be two concurrent llist_del_all(waiting_for_gp_ttrace):
@@ -318,10 +317,13 @@ static void do_call_rcu_ttrace(struct bpf_mem_cache *c)
 
 static void free_bulk(struct bpf_mem_cache *c)
 {
+	struct bpf_mem_cache *tgt = c->tgt;
 	struct llist_node *llnode, *t;
 	unsigned long flags;
 	int cnt;
 
+	WARN_ON_ONCE(tgt->unit_size != c->unit_size);
+
 	do {
 		inc_active(c, &flags);
 		llnode = __llist_del_first(&c->free_llist);
@@ -331,13 +333,13 @@ static void free_bulk(struct bpf_mem_cache *c)
 			cnt = 0;
 		dec_active(c, flags);
 		if (llnode)
-			enque_to_free(c, llnode);
+			enque_to_free(tgt, llnode);
 	} while (cnt > (c->high_watermark + c->low_watermark) / 2);
 
 	/* and drain free_llist_extra */
 	llist_for_each_safe(llnode, t, llist_del_all(&c->free_llist_extra))
-		enque_to_free(c, llnode);
-	do_call_rcu_ttrace(c);
+		enque_to_free(tgt, llnode);
+	do_call_rcu_ttrace(tgt);
 }
 
 static void bpf_mem_refill(struct irq_work *work)
@@ -436,6 +438,7 @@ int bpf_mem_alloc_init(struct bpf_mem_alloc *ma, int size, bool percpu)
 			c->unit_size = unit_size;
 			c->objcg = objcg;
 			c->percpu_size = percpu_size;
+			c->tgt = c;
 			prefill_mem_cache(c, cpu);
 		}
 		ma->cache = pc;
@@ -458,6 +461,7 @@ int bpf_mem_alloc_init(struct bpf_mem_alloc *ma, int size, bool percpu)
 			c = &cc->cache[i];
 			c->unit_size = sizes[i];
 			c->objcg = objcg;
+			c->tgt = c;
 			prefill_mem_cache(c, cpu);
 		}
 	}
@@ -476,7 +480,7 @@ static void drain_mem_cache(struct bpf_mem_cache *c)
 	 * Except for waiting_for_gp_ttrace list, there are no concurrent operations
 	 * on these lists, so it is safe to use __llist_del_all().
 	 */
-	free_all(__llist_del_all(&c->free_by_rcu_ttrace), percpu);
+	free_all(llist_del_all(&c->free_by_rcu_ttrace), percpu);
 	free_all(llist_del_all(&c->waiting_for_gp_ttrace), percpu);
 	free_all(__llist_del_all(&c->free_llist), percpu);
 	free_all(__llist_del_all(&c->free_llist_extra), percpu);
@@ -601,8 +605,10 @@ static void notrace *unit_alloc(struct bpf_mem_cache *c)
 	local_irq_save(flags);
 	if (local_inc_return(&c->active) == 1) {
 		llnode = __llist_del_first(&c->free_llist);
-		if (llnode)
+		if (llnode) {
 			cnt = --c->free_cnt;
+			*(struct bpf_mem_cache **)llnode = c;
+		}
 	}
 	local_dec(&c->active);
 	local_irq_restore(flags);
@@ -626,6 +632,12 @@ static void notrace unit_free(struct bpf_mem_cache *c, void *ptr)
 
 	BUILD_BUG_ON(LLIST_NODE_SZ > 8);
 
+	/*
+	 * Remember bpf_mem_cache that allocated this object.
+	 * The hint is not accurate.
+	 */
+	c->tgt = *(struct bpf_mem_cache **)llnode;
+
 	local_irq_save(flags);
 	if (local_inc_return(&c->active) == 1) {
 		__llist_add(llnode, &c->free_llist);