summaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
authorEric Dumazet <edumazet@google.com>2015-11-18 06:30:50 -0800
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>2017-06-14 13:16:26 +0200
commit52d8b8ad2b4ba478b55e0dfff56a13ab436a6b65 (patch)
tree84d1e0dbb0c2f9d8c9e53b3a028a362f9a3bc441
parent3c0fcb52674afb2a88097510203678fe3d441c71 (diff)
downloadlinux-stable-52d8b8ad2b4ba478b55e0dfff56a13ab436a6b65.tar.gz
linux-stable-52d8b8ad2b4ba478b55e0dfff56a13ab436a6b65.tar.bz2
linux-stable-52d8b8ad2b4ba478b55e0dfff56a13ab436a6b65.zip
net: better skb->sender_cpu and skb->napi_id cohabitation
commit 52bd2d62ce6758d811edcbd2256eb9ea7f6a56cb upstream. skb->sender_cpu and skb->napi_id share a common storage, and we had various bugs about this. We had to call skb_sender_cpu_clear() in some places to not leave a prior skb->napi_id and fool netdev_pick_tx() As suggested by Alexei, we could split the space so that these errors can not happen. 0 value being reserved as the common (not initialized) value, let's reserve [1 .. NR_CPUS] range for valid sender_cpu, and [NR_CPUS+1 .. ~0U] for valid napi_id. This will allow proper busy polling support over tunnels. Signed-off-by: Eric Dumazet <edumazet@google.com> Suggested-by: Alexei Starovoitov <ast@kernel.org> Acked-by: Alexei Starovoitov <ast@kernel.org> Signed-off-by: David S. Miller <davem@davemloft.net> Cc: Paul Menzel <pmenzel@molgen.mpg.de> Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
-rw-r--r--include/linux/skbuff.h3
-rw-r--r--net/core/dev.c33
2 files changed, 16 insertions, 20 deletions
diff --git a/include/linux/skbuff.h b/include/linux/skbuff.h
index d443d9ab0236..3f61c647fc5c 100644
--- a/include/linux/skbuff.h
+++ b/include/linux/skbuff.h
@@ -1084,9 +1084,6 @@ static inline void skb_copy_hash(struct sk_buff *to, const struct sk_buff *from)
static inline void skb_sender_cpu_clear(struct sk_buff *skb)
{
-#ifdef CONFIG_XPS
- skb->sender_cpu = 0;
-#endif
}
#ifdef NET_SKBUFF_DATA_USES_OFFSET
diff --git a/net/core/dev.c b/net/core/dev.c
index 48399d8ce614..87b8754f34ac 100644
--- a/net/core/dev.c
+++ b/net/core/dev.c
@@ -182,7 +182,7 @@ EXPORT_SYMBOL(dev_base_lock);
/* protects napi_hash addition/deletion and napi_gen_id */
static DEFINE_SPINLOCK(napi_hash_lock);
-static unsigned int napi_gen_id;
+static unsigned int napi_gen_id = NR_CPUS;
static DEFINE_HASHTABLE(napi_hash, 8);
static seqcount_t devnet_rename_seq;
@@ -3049,7 +3049,9 @@ struct netdev_queue *netdev_pick_tx(struct net_device *dev,
int queue_index = 0;
#ifdef CONFIG_XPS
- if (skb->sender_cpu == 0)
+ u32 sender_cpu = skb->sender_cpu - 1;
+
+ if (sender_cpu >= (u32)NR_CPUS)
skb->sender_cpu = raw_smp_processor_id() + 1;
#endif
@@ -4726,25 +4728,22 @@ EXPORT_SYMBOL_GPL(napi_by_id);
void napi_hash_add(struct napi_struct *napi)
{
- if (!test_and_set_bit(NAPI_STATE_HASHED, &napi->state)) {
+ if (test_and_set_bit(NAPI_STATE_HASHED, &napi->state))
+ return;
- spin_lock(&napi_hash_lock);
+ spin_lock(&napi_hash_lock);
- /* 0 is not a valid id, we also skip an id that is taken
- * we expect both events to be extremely rare
- */
- napi->napi_id = 0;
- while (!napi->napi_id) {
- napi->napi_id = ++napi_gen_id;
- if (napi_by_id(napi->napi_id))
- napi->napi_id = 0;
- }
+ /* 0..NR_CPUS+1 range is reserved for sender_cpu use */
+ do {
+ if (unlikely(++napi_gen_id < NR_CPUS + 1))
+ napi_gen_id = NR_CPUS + 1;
+ } while (napi_by_id(napi_gen_id));
+ napi->napi_id = napi_gen_id;
- hlist_add_head_rcu(&napi->napi_hash_node,
- &napi_hash[napi->napi_id % HASH_SIZE(napi_hash)]);
+ hlist_add_head_rcu(&napi->napi_hash_node,
+ &napi_hash[napi->napi_id % HASH_SIZE(napi_hash)]);
- spin_unlock(&napi_hash_lock);
- }
+ spin_unlock(&napi_hash_lock);
}
EXPORT_SYMBOL_GPL(napi_hash_add);