Merge https://git.kernel.org/pub/scm/linux/kernel/git/bpf/bpf

Daniel Borkmann says:

====================
pull-request: bpf 2022-10-03

We've added 10 non-merge commits during the last 23 day(s) which contain
a total of 14 files changed, 130 insertions(+), 69 deletions(-).

The main changes are:

1) Fix dynptr helper API to gate behind CAP_BPF given it was not intended
   for unprivileged BPF programs, from Kumar Kartikeya Dwivedi.

2) Fix need_wakeup flag inheritance from umem buffer pool for shared xsk
   sockets, from Jalal Mostafa.

3) Fix truncated last_member_type_id in btf_struct_resolve() which had a
   wrong storage type, from Lorenz Bauer.

4) Fix xsk back-pressure mechanism on tx when amount of produced
   descriptors to CQ is lower than what was grabbed from xsk tx ring,
   from Maciej Fijalkowski.

5) Fix wrong cgroup attach flags being displayed to effective progs,
   from Pu Lehui.

* https://git.kernel.org/pub/scm/linux/kernel/git/bpf/bpf:
  xsk: Inherit need_wakeup flag for shared sockets
  bpf: Gate dynptr API behind CAP_BPF
  selftests/bpf: Adapt cgroup effective query uapi change
  bpftool: Fix wrong cgroup attach flags being assigned to effective progs
  bpf, cgroup: Reject prog_attach_flags array when effective query
  bpf: Ensure correct locking around vulnerable function find_vpid()
  bpf: btf: fix truncated last_member_type_id in btf_struct_resolve
  selftests/xsk: Add missing close() on netns fd
  xsk: Fix backpressure mechanism on Tx
  MAINTAINERS: Add include/linux/tnum.h to BPF CORE
====================

Link: https://lore.kernel.org/r/20221003201957.13149-1-daniel@iogearbox.net
Signed-off-by: Jakub Kicinski <kuba@kernel.org>

3 files changed, 26 insertions, 27 deletions

diff --git a/net/xdp/xsk.c b/net/xdp/xsk.c
index 5b4ce6ba1bc7..9f0561b67c12 100644
--- a/net/xdp/xsk.c
+++ b/net/xdp/xsk.c
@@ -355,16 +355,15 @@ static u32 xsk_tx_peek_release_fallback(struct xsk_buff_pool *pool, u32 max_entr
 	return nb_pkts;
 }
 
-u32 xsk_tx_peek_release_desc_batch(struct xsk_buff_pool *pool, u32 max_entries)
+u32 xsk_tx_peek_release_desc_batch(struct xsk_buff_pool *pool, u32 nb_pkts)
 {
 	struct xdp_sock *xs;
-	u32 nb_pkts;
 
 	rcu_read_lock();
 	if (!list_is_singular(&pool->xsk_tx_list)) {
 		/* Fallback to the non-batched version */
 		rcu_read_unlock();
-		return xsk_tx_peek_release_fallback(pool, max_entries);
+		return xsk_tx_peek_release_fallback(pool, nb_pkts);
 	}
 
 	xs = list_first_or_null_rcu(&pool->xsk_tx_list, struct xdp_sock, tx_list);
@@ -373,12 +372,7 @@ u32 xsk_tx_peek_release_desc_batch(struct xsk_buff_pool *pool, u32 max_entries)
 		goto out;
 	}
 
-	max_entries = xskq_cons_nb_entries(xs->tx, max_entries);
-	nb_pkts = xskq_cons_read_desc_batch(xs->tx, pool, max_entries);
-	if (!nb_pkts) {
-		xs->tx->queue_empty_descs++;
-		goto out;
-	}
+	nb_pkts = xskq_cons_nb_entries(xs->tx, nb_pkts);
 
 	/* This is the backpressure mechanism for the Tx path. Try to
 	 * reserve space in the completion queue for all packets, but
@@ -386,12 +380,18 @@ u32 xsk_tx_peek_release_desc_batch(struct xsk_buff_pool *pool, u32 max_entries)
 	 * packets. This avoids having to implement any buffering in
 	 * the Tx path.
 	 */
-	nb_pkts = xskq_prod_reserve_addr_batch(pool->cq, pool->tx_descs, nb_pkts);
+	nb_pkts = xskq_prod_nb_free(pool->cq, nb_pkts);
 	if (!nb_pkts)
 		goto out;
 
-	xskq_cons_release_n(xs->tx, max_entries);
+	nb_pkts = xskq_cons_read_desc_batch(xs->tx, pool, nb_pkts);
+	if (!nb_pkts) {
+		xs->tx->queue_empty_descs++;
+		goto out;
+	}
+
 	__xskq_cons_release(xs->tx);
+	xskq_prod_write_addr_batch(pool->cq, pool->tx_descs, nb_pkts);
 	xs->sk.sk_write_space(&xs->sk);
 
 out:
@@ -954,8 +954,8 @@ static int xsk_bind(struct socket *sock, struct sockaddr *addr, int addr_len)
 				goto out_unlock;
 			}
 
-			err = xp_assign_dev_shared(xs->pool, umem_xs->umem,
-						   dev, qid);
+			err = xp_assign_dev_shared(xs->pool, umem_xs, dev,
+						   qid);
 			if (err) {
 				xp_destroy(xs->pool);
 				xs->pool = NULL;
diff --git a/net/xdp/xsk_buff_pool.c b/net/xdp/xsk_buff_pool.c
index a71a8c6edf55..ed6c71826d31 100644
--- a/net/xdp/xsk_buff_pool.c
+++ b/net/xdp/xsk_buff_pool.c
@@ -212,17 +212,18 @@ err_unreg_pool:
 	return err;
 }
 
-int xp_assign_dev_shared(struct xsk_buff_pool *pool, struct xdp_umem *umem,
+int xp_assign_dev_shared(struct xsk_buff_pool *pool, struct xdp_sock *umem_xs,
 			 struct net_device *dev, u16 queue_id)
 {
 	u16 flags;
+	struct xdp_umem *umem = umem_xs->umem;
 
 	/* One fill and completion ring required for each queue id. */
 	if (!pool->fq || !pool->cq)
 		return -EINVAL;
 
 	flags = umem->zc ? XDP_ZEROCOPY : XDP_COPY;
-	if (pool->uses_need_wakeup)
+	if (umem_xs->pool->uses_need_wakeup)
 		flags |= XDP_USE_NEED_WAKEUP;
 
 	return xp_assign_dev(pool, dev, queue_id, flags);
diff --git a/net/xdp/xsk_queue.h b/net/xdp/xsk_queue.h
index fb20bf7207cf..c6fb6b763658 100644
--- a/net/xdp/xsk_queue.h
+++ b/net/xdp/xsk_queue.h
@@ -205,6 +205,11 @@ static inline bool xskq_cons_read_desc(struct xsk_queue *q,
 	return false;
 }
 
+static inline void xskq_cons_release_n(struct xsk_queue *q, u32 cnt)
+{
+	q->cached_cons += cnt;
+}
+
 static inline u32 xskq_cons_read_desc_batch(struct xsk_queue *q, struct xsk_buff_pool *pool,
 					    u32 max)
 {
@@ -226,6 +231,8 @@ static inline u32 xskq_cons_read_desc_batch(struct xsk_queue *q, struct xsk_buff
 		cached_cons++;
 	}
 
+	/* Release valid plus any invalid entries */
+	xskq_cons_release_n(q, cached_cons - q->cached_cons);
 	return nb_entries;
 }
 
@@ -291,11 +298,6 @@ static inline void xskq_cons_release(struct xsk_queue *q)
 	q->cached_cons++;
 }
 
-static inline void xskq_cons_release_n(struct xsk_queue *q, u32 cnt)
-{
-	q->cached_cons += cnt;
-}
-
 static inline u32 xskq_cons_present_entries(struct xsk_queue *q)
 {
 	/* No barriers needed since data is not accessed */
@@ -350,21 +352,17 @@ static inline int xskq_prod_reserve_addr(struct xsk_queue *q, u64 addr)
 	return 0;
 }
 
-static inline u32 xskq_prod_reserve_addr_batch(struct xsk_queue *q, struct xdp_desc *descs,
-					       u32 max)
+static inline void xskq_prod_write_addr_batch(struct xsk_queue *q, struct xdp_desc *descs,
+					      u32 nb_entries)
 {
 	struct xdp_umem_ring *ring = (struct xdp_umem_ring *)q->ring;
-	u32 nb_entries, i, cached_prod;
-
-	nb_entries = xskq_prod_nb_free(q, max);
+	u32 i, cached_prod;
 
 	/* A, matches D */
 	cached_prod = q->cached_prod;
 	for (i = 0; i < nb_entries; i++)
 		ring->desc[cached_prod++ & q->ring_mask] = descs[i].addr;
 	q->cached_prod = cached_prod;
-
-	return nb_entries;
 }
 
 static inline int xskq_prod_reserve_desc(struct xsk_queue *q,