summaryrefslogtreecommitdiff
path: root/net/xdp
diff options
context:
space:
mode:
authorDavid S. Miller <davem@davemloft.net>2020-07-14 04:04:05 +0300
committerDavid S. Miller <davem@davemloft.net>2020-07-14 04:04:05 +0300
commit07dd1b7e68e4b83a1004b14dffd7e142c0bc79bd (patch)
tree06a5898494308d55fcecbb3dff647d93c01afe47 /net/xdp
parentbfe4c40333065639cf0f364b8ea9e93270781699 (diff)
parent8afb259a9840fa953efb9a7835356a083ac8ec74 (diff)
downloadlinux-07dd1b7e68e4b83a1004b14dffd7e142c0bc79bd.tar.xz
Merge git://git.kernel.org/pub/scm/linux/kernel/git/bpf/bpf-next
Alexei Starovoitov says: ==================== pull-request: bpf-next 2020-07-13 The following pull-request contains BPF updates for your *net-next* tree. We've added 36 non-merge commits during the last 7 day(s) which contain a total of 62 files changed, 2242 insertions(+), 468 deletions(-). The main changes are: 1) Avoid trace_printk warning banner by switching bpf_trace_printk to use its own tracing event, from Alan. 2) Better libbpf support on older kernels, from Andrii. 3) Additional AF_XDP stats, from Ciara. 4) build time resolution of BTF IDs, from Jiri. 5) BPF_CGROUP_INET_SOCK_RELEASE hook, from Stanislav. ==================== Signed-off-by: David S. Miller <davem@davemloft.net>
Diffstat (limited to 'net/xdp')
-rw-r--r--net/xdp/xsk.c36
-rw-r--r--net/xdp/xsk_buff_pool.c1
-rw-r--r--net/xdp/xsk_diag.c17
-rw-r--r--net/xdp/xsk_queue.h6
4 files changed, 55 insertions, 5 deletions
diff --git a/net/xdp/xsk.c b/net/xdp/xsk.c
index 3700266229f6..26e3bba8c204 100644
--- a/net/xdp/xsk.c
+++ b/net/xdp/xsk.c
@@ -123,7 +123,7 @@ static int __xsk_rcv_zc(struct xdp_sock *xs, struct xdp_buff *xdp, u32 len)
addr = xp_get_handle(xskb);
err = xskq_prod_reserve_desc(xs->rx, addr, len);
if (err) {
- xs->rx_dropped++;
+ xs->rx_queue_full++;
return err;
}
@@ -274,8 +274,10 @@ bool xsk_umem_consume_tx(struct xdp_umem *umem, struct xdp_desc *desc)
rcu_read_lock();
list_for_each_entry_rcu(xs, &umem->xsk_tx_list, list) {
- if (!xskq_cons_peek_desc(xs->tx, desc, umem))
+ if (!xskq_cons_peek_desc(xs->tx, desc, umem)) {
+ xs->tx->queue_empty_descs++;
continue;
+ }
/* This is the backpressure mechanism for the Tx path.
* Reserve space in the completion queue and only proceed
@@ -387,6 +389,8 @@ static int xsk_generic_xmit(struct sock *sk)
sent_frame = true;
}
+ xs->tx->queue_empty_descs++;
+
out:
if (sent_frame)
sk->sk_write_space(sk);
@@ -812,6 +816,12 @@ static void xsk_enter_umem_offsets(struct xdp_ring_offset_v1 *ring)
ring->desc = offsetof(struct xdp_umem_ring, desc);
}
+struct xdp_statistics_v1 {
+ __u64 rx_dropped;
+ __u64 rx_invalid_descs;
+ __u64 tx_invalid_descs;
+};
+
static int xsk_getsockopt(struct socket *sock, int level, int optname,
char __user *optval, int __user *optlen)
{
@@ -831,19 +841,35 @@ static int xsk_getsockopt(struct socket *sock, int level, int optname,
case XDP_STATISTICS:
{
struct xdp_statistics stats;
+ bool extra_stats = true;
+ size_t stats_size;
- if (len < sizeof(stats))
+ if (len < sizeof(struct xdp_statistics_v1)) {
return -EINVAL;
+ } else if (len < sizeof(stats)) {
+ extra_stats = false;
+ stats_size = sizeof(struct xdp_statistics_v1);
+ } else {
+ stats_size = sizeof(stats);
+ }
mutex_lock(&xs->mutex);
stats.rx_dropped = xs->rx_dropped;
+ if (extra_stats) {
+ stats.rx_ring_full = xs->rx_queue_full;
+ stats.rx_fill_ring_empty_descs =
+ xs->umem ? xskq_nb_queue_empty_descs(xs->umem->fq) : 0;
+ stats.tx_ring_empty_descs = xskq_nb_queue_empty_descs(xs->tx);
+ } else {
+ stats.rx_dropped += xs->rx_queue_full;
+ }
stats.rx_invalid_descs = xskq_nb_invalid_descs(xs->rx);
stats.tx_invalid_descs = xskq_nb_invalid_descs(xs->tx);
mutex_unlock(&xs->mutex);
- if (copy_to_user(optval, &stats, sizeof(stats)))
+ if (copy_to_user(optval, &stats, stats_size))
return -EFAULT;
- if (put_user(sizeof(stats), optlen))
+ if (put_user(stats_size, optlen))
return -EFAULT;
return 0;
diff --git a/net/xdp/xsk_buff_pool.c b/net/xdp/xsk_buff_pool.c
index 08b80669f649..a2044c245215 100644
--- a/net/xdp/xsk_buff_pool.c
+++ b/net/xdp/xsk_buff_pool.c
@@ -189,6 +189,7 @@ static struct xdp_buff_xsk *__xp_alloc(struct xsk_buff_pool *pool)
for (;;) {
if (!xskq_cons_peek_addr_unchecked(pool->fq, &addr)) {
+ pool->fq->queue_empty_descs++;
xp_release(xskb);
return NULL;
}
diff --git a/net/xdp/xsk_diag.c b/net/xdp/xsk_diag.c
index 0163b26aaf63..21e9c2d123ee 100644
--- a/net/xdp/xsk_diag.c
+++ b/net/xdp/xsk_diag.c
@@ -76,6 +76,19 @@ static int xsk_diag_put_umem(const struct xdp_sock *xs, struct sk_buff *nlskb)
return err;
}
+static int xsk_diag_put_stats(const struct xdp_sock *xs, struct sk_buff *nlskb)
+{
+ struct xdp_diag_stats du = {};
+
+ du.n_rx_dropped = xs->rx_dropped;
+ du.n_rx_invalid = xskq_nb_invalid_descs(xs->rx);
+ du.n_rx_full = xs->rx_queue_full;
+ du.n_fill_ring_empty = xs->umem ? xskq_nb_queue_empty_descs(xs->umem->fq) : 0;
+ du.n_tx_invalid = xskq_nb_invalid_descs(xs->tx);
+ du.n_tx_ring_empty = xskq_nb_queue_empty_descs(xs->tx);
+ return nla_put(nlskb, XDP_DIAG_STATS, sizeof(du), &du);
+}
+
static int xsk_diag_fill(struct sock *sk, struct sk_buff *nlskb,
struct xdp_diag_req *req,
struct user_namespace *user_ns,
@@ -118,6 +131,10 @@ static int xsk_diag_fill(struct sock *sk, struct sk_buff *nlskb,
sock_diag_put_meminfo(sk, nlskb, XDP_DIAG_MEMINFO))
goto out_nlmsg_trim;
+ if ((req->xdiag_show & XDP_SHOW_STATS) &&
+ xsk_diag_put_stats(xs, nlskb))
+ goto out_nlmsg_trim;
+
mutex_unlock(&xs->mutex);
nlmsg_end(nlskb, nlh);
return 0;
diff --git a/net/xdp/xsk_queue.h b/net/xdp/xsk_queue.h
index 5b5d24d2dd37..bf42cfd74b89 100644
--- a/net/xdp/xsk_queue.h
+++ b/net/xdp/xsk_queue.h
@@ -38,6 +38,7 @@ struct xsk_queue {
u32 cached_cons;
struct xdp_ring *ring;
u64 invalid_descs;
+ u64 queue_empty_descs;
};
/* The structure of the shared state of the rings are the same as the
@@ -354,6 +355,11 @@ static inline u64 xskq_nb_invalid_descs(struct xsk_queue *q)
return q ? q->invalid_descs : 0;
}
+static inline u64 xskq_nb_queue_empty_descs(struct xsk_queue *q)
+{
+ return q ? q->queue_empty_descs : 0;
+}
+
struct xsk_queue *xskq_create(u32 nentries, bool umem_queue);
void xskq_destroy(struct xsk_queue *q_ops);