net, bpf: Add a warning if NAPI cb missed xdp_do_flush().

A few drivers were missing a xdp_do_flush() invocation after
XDP_REDIRECT.

Add three helper functions each for one of the per-CPU lists. Return
true if the per-CPU list is non-empty and flush the list.

Add xdp_do_check_flushed() which invokes each helper functions and
creates a warning if one of the functions had a non-empty list.

Hide everything behind CONFIG_DEBUG_NET.

Suggested-by: Jesper Dangaard Brouer <[email protected]>
Signed-off-by: Sebastian Andrzej Siewior <[email protected]>
Signed-off-by: Daniel Borkmann <[email protected]>
Reviewed-by: Toke Høiland-Jørgensen <[email protected]>
Acked-by: Jakub Kicinski <[email protected]>
Acked-by: John Fastabend <[email protected]>
Link: https://lore.kernel.org/bpf/[email protected]

Author: Sebastian Andrzej Siewior

include/linux/bpf.h

@@ -2478,6 +2478,9 @@ void bpf_dynptr_init(struct bpf_dynptr_kern *ptr, void *data,
 		     enum bpf_dynptr_type type, u32 offset, u32 size);
 void bpf_dynptr_set_null(struct bpf_dynptr_kern *ptr);
 void bpf_dynptr_set_rdonly(struct bpf_dynptr_kern *ptr);
+
+bool dev_check_flush(void);
+bool cpu_map_check_flush(void);
 #else /* !CONFIG_BPF_SYSCALL */
 static inline struct bpf_prog *bpf_prog_get(u32 ufd)
 {

include/net/xdp_sock.h

@@ -109,4 +109,13 @@ static inline void __xsk_map_flush(void)
 
 #endif /* CONFIG_XDP_SOCKETS */
 
+#if defined(CONFIG_XDP_SOCKETS) && defined(CONFIG_DEBUG_NET)
+bool xsk_map_check_flush(void);
+#else
+static inline bool xsk_map_check_flush(void)
+{
+	return false;
+}
+#endif
+
 #endif /* _LINUX_XDP_SOCK_H */

kernel/bpf/cpumap.c

@@ -764,6 +764,16 @@ void __cpu_map_flush(void)
 	}
 }
 
+#ifdef CONFIG_DEBUG_NET
+bool cpu_map_check_flush(void)
+{
+	if (list_empty(this_cpu_ptr(&cpu_map_flush_list)))
+		return false;
+	__cpu_map_flush();
+	return true;
+}
+#endif
+
 static int __init cpu_map_init(void)
 {
 	int cpu;

kernel/bpf/devmap.c

@@ -418,6 +418,16 @@ void __dev_flush(void)
 	}
 }
 
+#ifdef CONFIG_DEBUG_NET
+bool dev_check_flush(void)
+{
+	if (list_empty(this_cpu_ptr(&dev_flush_list)))
+		return false;
+	__dev_flush();
+	return true;
+}
+#endif
+
 /* Elements are kept alive by RCU; either by rcu_read_lock() (from syscall) or
  * by local_bh_disable() (from XDP calls inside NAPI). The
  * rcu_read_lock_bh_held() below makes lockdep accept both.

net/core/dev.c

@@ -6535,6 +6535,8 @@ static int __napi_poll(struct napi_struct *n, bool *repoll)
 	if (test_bit(NAPI_STATE_SCHED, &n->state)) {
 		work = n->poll(n, weight);
 		trace_napi_poll(n, work, weight);
+
+		xdp_do_check_flushed(n);
 	}
 
 	if (unlikely(work > weight))

net/core/dev.h

@@ -136,4 +136,10 @@ static inline void netif_set_gro_ipv4_max_size(struct net_device *dev,
 }
 
 int rps_cpumask_housekeeping(struct cpumask *mask);
+
+#if defined(CONFIG_DEBUG_NET) && defined(CONFIG_BPF_SYSCALL)
+void xdp_do_check_flushed(struct napi_struct *napi);
+#else
+static inline void xdp_do_check_flushed(struct napi_struct *napi) { }
+#endif
 #endif

net/core/filter.c

@@ -83,6 +83,8 @@
 #include <net/netfilter/nf_conntrack_bpf.h>
 #include <linux/un.h>
 
+#include "dev.h"
+
 static const struct bpf_func_proto *
 bpf_sk_base_func_proto(enum bpf_func_id func_id);
 
@@ -4208,6 +4210,20 @@ void xdp_do_flush(void)
 }
 EXPORT_SYMBOL_GPL(xdp_do_flush);
 
+#if defined(CONFIG_DEBUG_NET) && defined(CONFIG_BPF_SYSCALL)
+void xdp_do_check_flushed(struct napi_struct *napi)
+{
+	bool ret;
+
+	ret = dev_check_flush();
+	ret |= cpu_map_check_flush();
+	ret |= xsk_map_check_flush();
+
+	WARN_ONCE(ret, "Missing xdp_do_flush() invocation after NAPI by %ps\n",
+		  napi->poll);
+}
+#endif
+
 void bpf_clear_redirect_map(struct bpf_map *map)
 {
 	struct bpf_redirect_info *ri;

net/xdp/xsk.c

@@ -391,6 +391,16 @@ void __xsk_map_flush(void)
 	}
 }
 
+#ifdef CONFIG_DEBUG_NET
+bool xsk_map_check_flush(void)
+{
+	if (list_empty(this_cpu_ptr(&xskmap_flush_list)))
+		return false;
+	__xsk_map_flush();
+	return true;
+}
+#endif
+
 void xsk_tx_completed(struct xsk_buff_pool *pool, u32 nb_entries)
 {
 	xskq_prod_submit_n(pool->cq, nb_entries);