The current semantic for napi_consume_skb() is that caller need
to provide non-zero budget when calling from NAPI context, and
breaking this semantic will cause hard to debug problem, because
_kfree_skb_defer() need to run in atomic context in order to push
the skb to the particular cpu' napi_alloc_cache atomically.
So add a in_softirq() debug checking in napi_consume_skb() to catch
this kind of error.
Suggested-by: Eric Dumazet <eduma...@google.com>
Signed-off-by: Yunsheng Lin <linyunsh...@huawei.com>
---
v1: drop RFC in the title
---
include/linux/netdevice.h | 6 ++++++
net/Kconfig | 7 +++++++
net/core/skbuff.c | 4 ++++
3 files changed, 17 insertions(+)
diff --git a/include/linux/netdevice.h b/include/linux/netdevice.h
index 964b494..8042bf1 100644
--- a/include/linux/netdevice.h
+++ b/include/linux/netdevice.h
@@ -5158,6 +5158,12 @@ do {
\
})
#endif
+#if defined(CONFIG_DEBUG_NET)
+#define DEBUG_NET_WARN(condition, ...) WARN(condition, ##__VA_ARGS__)
+#else
+#define DEBUG_NET_WARN(condition, ...)
+#endif
+
/*
* The list of packet types we will receive (as opposed to discard)
* and the routines to invoke.
diff --git a/net/Kconfig b/net/Kconfig
index d656716..82e69b0 100644
--- a/net/Kconfig
+++ b/net/Kconfig
@@ -459,6 +459,13 @@ config ETHTOOL_NETLINK
netlink. It provides better extensibility and some new features,
e.g. notification messages.
+config DEBUG_NET
+ bool "Net debugging and diagnostics"
+ depends on DEBUG_KERNEL
+ default n
+ help
+ Say Y here to add some extra checks and diagnostics to networking.
+
endif # if NET
# Used by archs to tell that they support BPF JIT compiler plus which flavour.
diff --git a/net/core/skbuff.c b/net/core/skbuff.c
index 1ba8f01..1834007 100644
--- a/net/core/skbuff.c
+++ b/net/core/skbuff.c
@@ -897,6 +897,10 @@ void napi_consume_skb(struct sk_buff *skb, int budget)
return;
}
+ DEBUG_NET_WARN(!in_softirq(),
+ "%s is called with non-zero budget outside softirq
context.\n",
+ __func__);
+
if (!skb_unref(skb))
return;
--
2.8.1
